Introduction to Cell signaling

All living cells perceive signals from the outside environment and adjust their behavior accordingly. If you think back to the earliest living cells, it is easy to imagine the incredible pressure they were under to evolve the ability to sense features of the environment and to change in response to these signals. The ability to sense and move toward nutrients, or to sense and avoid stresses and toxins, would give a unicellular organism a huge competitive advantage. This ability to respond to environmental cues is important for single cells, but it is also absolutely essential for the normal development and functioning of multicellular organisms, which depend on a continuous and extensive exchange of information to coordinate the activities of many individual cells. Furthermore, when this cellular com- munication goes awry, it can result in diseases such as cancer. In this chapter, we will introduce the basic principles of cell signaling and the molecular mechanisms that underlie it. What Is Cell sIgnalIng? Cells are the smallest fundamental units of life. Part of what makes them so distinctly “living” is their remarkable ability to sense stimuli and to respond to them in a dynamic manner. This ability of cells to detect or receive information and process it to make decisions can also be considered from the broader perspective of information processing. Here we can draw analogies to the engineering and design principles of other, more familiar information-processing systems, such as human- made electronic devices. It is this interface between the unique proper- ties of living systems and the more universal properties of any system that processes information that makes the study of cellular signaling mechanisms so compelling. Figure 1.1 Cell signaling systems process information. (a) the role of cell signaling systems is to receive input from the environment and, on the basis of that input, generate an appropriate output response. Information processing performed by the cell is conceptually similar to that of other, familiar information-processing systems, such as the brain or a computer. Cellular information processing must be accomplished by a densely packed and diverse collection of biomolecules. (b, adapted from D. s. goodsell, Trends Biochem. Sci. 16: 203–206, 1991. With permission from elsevier.) All cells have the ability to respond to their environment While biologists and philosophers may not all agree on the precise defini- tion of life, most definitions include a number of common properties, such as autonomy, the ability to generate energy, and the ability to reproduce. One of these common properties is adaptability—the ability to respond to changes in the environment. We all understand that one way to test whether a thing is animate or inanimate, or living or dead, is to poke it and see if it responds. Single-celled organisms display the ability to detect diverse molecular species and stresses in their environment, and are able to change aspects of their gene expression, growth, structure, and metabolism in response, usually to improve their ability to survive under changing conditions. With the emergence of multicellular organisms, individual cells within the organism evolved the highly specialized ability to sense specific sig- nals transmitted from other cells in the organism, allowing for extraor- dinary levels of communication within the organism. The coordinated regulation of growth, death, morphology, and metabolism is absolutely essential for many individual cells to function in concert as an integrated organism. Moreover, cells have the ability to monitor aspects of their own internal state, and to respond in a self-correcting way—the foundation of cellular homeostasis and repair. Thus, cell signaling, which encompasses the study of this wide range of stimulus–response behaviors observed in cells, is central to all of biology. Today, as our knowledge of biological systems increases rapidly, we have begun to view cell signaling from the perspective of a more general question: how do cells process information? How a cell receives diverse signaling inputs, processes and integrates these signals, and converts them into responses is in many ways analogous to how other systems, at widely varying scales, process information. We can think about informa- tion processing and storage in the cell, just as we focus on these concepts when we consider a brain or a computer (Figure 1.1a). At the level of the organism, for example, we can marvel at the ability of an athlete to detect the movement of a ball, calculate its trajectory, and mount an effective response to intercept it that involves the coordinated action of hundreds of different muscles, all within a fraction of a second. Individual cells per- form similarly remarkable feats, such as detecting the presence of just a few specific molecules in the outside environment, and responding by set- ting in motion elaborate cellular programs that culminate in behaviors such as proliferation, directed migration, or cell death. At each of these scales, a common series of challenges must be overcome. Just a few exam- ples include detecting, amplifying, and robustly responding to a faint incoming signal; integrating and responding coherently to diverse and contradictory signals; and adapting to the strength or duration of a sig- nal and shutting down the response when appropriate. These represent (a) (b) universal information-processing tasks that a cell or an organism must be able to perform. In this book, we will focus on information processing by the smallest unit of life, the cell. It is particularly interesting to consider how such tasks are accomplished by the molecules that make up a cell. Unlike the tidy systems of wires, transistors, and other components that make up human-made signaling devices like mobile phones or computers, the cel- lular signal-processing machinery consists of a densely packed mixture of rapidly diffusing proteins, lipids, nucleic acids, and other biomolecules, all surrounded by a water-impermeable membrane (Figure 1.1b). How this genetically encoded set of molecules can perform complex information- processing behaviors is one of the most exciting and fundamental ques- tions in modern biology. Cells must perceive and respond to a wide range of signals To get a better sense of the scope of the challenges faced by cellular sig- naling systems, we will briefly consider the types of inputs to which they must react. Some of these inputs, along with the types of responses that can be evoked, are illustrated in Figure 1.2. The most basic inputs, those common both to free-living unicellular organ- isms and the cells of multicellular organisms, are nutrients and other raw materials useful to the cell, along with various environmental stresses. In the case of nutrients, the cell is likely either to take advantage of a hospi- table environment by staying in place, or to move to where resources are more plentiful. On the other hand, noxious physical or chemical stimuli may cause the cell to migrate away, or otherwise adapt to endure the hard times until conditions improve. A yeast cell that is starved for phosphate, for example, mounts a complex response in which phosphate utilization is minimized, phosphate transport is increased, and enzymes (known as phosphatases) are secreted by the cell to release phosphate from environ- mental sources. In multicellular organisms, a whole new set of signals needs to be detected by each of the many cells that make up the various tissues and organs. For example, cells signal to nearby or adjacent cells to regulate the develop- ment and function of organs and tissues. This permits the cells to work INPUTS OUTPUTS Figure 1.2 Cell signaling involves a diversity of inputs and outputs. Cells must read diverse external and intracellular signals, and use this information to generate many output responses. some common inputs (green arrows) and outputs (orange arrows) are indicated. In many cases, the output of the system will change the response to future inputs (feedback). together as an integrated unit, rather than a collection of individual cells, each marching to its own drummer. These localized signals include solu- ble signals that diffuse only a short distance, signals that are attached to the surface of the cell and thus can be detected only by cells they directly contact, and signals from the extracellular matrix that provides the physi- cal framework to which the cells are attached. Longer-range signals of various types are also secreted and transported throughout the organism, allowing spatially separated cells to act col- lectively. A familiar example is the hormone insulin, which is secreted by the pancreatic islet cells and circulates through the bloodstream to regulate metabolic activity in tissues throughout the body. Other exam- ples include epinephrine secreted by the adrenal gland, which induces an organism-wide and almost immediate “fight or flight” response; and sex hormones secreted by the gonads, which orchestrate the many physical changes in the body during puberty. Finally, cells must continuously monitor their internal state to adjust to changing circumstances and respond to damage. The term homeosta- sis refers to the ability of living systems to adjust their behavior sponta- neously to maintain a stable intracellular environment, despite varying environmental conditions. While this may sound like the antithesis of adaptability, it is really just another manifestation of the ability to detect changes in conditions and to modify cellular activities in concert with those changes. Typically, these homeostatic mechanisms involve feedback, which is the ability of the output from a system to adjust the incoming sig- nal. For example, high levels of a certain cellular metabolite might cause the biosynthetic enzymes that generate the metabolite to shut down, or might close channels that regulate transport of the metabolite or its pre- cursor into the cell. Feedback is an important component of almost all cell signaling systems. Examples of the internal states that cells must monitor include physical or chemical damage (for example, to the genomic DNA) and the progres- sion of cell division (the cell cycle). For example, cells that are preparing to go into mitosis (the process by which the cell, including its genomic DNA, is divided into two daughter cells) must ensure that DNA replication is completed before the physical events of mitosis are initiated. Signaling systems need to solve a number of common problems Cellular signaling systems must be able to overcome a number of chal- lenges in order to reliably process information and mount an appropriate response. Some of these are listed in Table 1.1. While our understanding of the details of how this is accomplished remains far from complete, in this book we will discuss a number of solutions to each of these problems. Here, we briefly provide a sense of the magnitude and complexity of just a few of the challenges that must be overcome in cell signaling. The issue of specificity is of particular importance considering the intra- cellular environment where cellular information processing must occur. The cytosol (and the extracellular environment for most of the cells in multicellular organisms) is a highly concentrated solution of organic mol- ecules, in which the molar concentrations of individual components range over many orders of magnitude. Being able to pick out a specific signal in such a densely packed environment is a formidable challenge. The sheer variety of different molecules that are present is enormous: the protein products of many thousands of genes, along with the numerous variants of each that might be generated by alternative mRNA splicing or by various Molecules must be able to sense the presence of stimuli, both inside and outside of the cell this sensed information must be stored and transmitted, so that it can eventually alter the core physiological processes in the cell, such as gene expression, cell morphology, and metabolism Information must be amplified so that small inputs can yield major changes in out- put, while random fluctuations (noise) are filtered out Information must be integrated, so that outputs can be shaped by multiple inputs Output information must be transmitted and processed to allow feedback control of behavior Information must be transmitted through the impermeable plasma membrane that surrounds the cell Information processing must be coordinated to give responses that are correctly shaped over both time and space signaling responses must be robust to a variety of conditions, such as changing ambient conditions or variations in the concentration of molecular components the cell must achieve specificity in signaling, despite the many behaviors it must simultaneously coordinate and the vast number and variety of molecules diffusing about in the cell evolution must be able to give rise to new response behaviors and optimize/tune existing ones Table 1.1 Fundamental challenges in cellular information processing types of post-translational modification and processing, plus numerous lipids, carbohydrates, nucleic acids, and simpler organic molecules. Yet cellular signaling systems can easily discriminate between two biomol- ecules that differ from each other in the minutest of details—a kink in a polypeptide chain, the addition of a phosphate group—and reliably detect slight differences in their concentration. Within this complex molecular milieu, cellular signaling systems must also be able to respond strongly to the faintest of incoming signals. Some- times just one or a few input molecules, such as of a hormone, are suffi- cient to induce wholesale changes in the cell receiving the signal, tipping the balance between life and death, or between quiescence and a round of cell division. This ability to amplify an incoming signal (signal amplifica- tion) is even more remarkable when one considers that the system must at the same time be resistant to various sorts of background noise, such as the random fluctuations in the conformation, activity, and local concentra- tion of cellular components as they jostle about in the cell. Another challenge faced by cell signaling systems is signal integration. A typical cell is simultaneously subjected to an enormous variety of dif- ferent inputs, and often the response depends not on just one but many of these signals. For example, a cell may make the decision to proliferate only when nutrients are plentiful and specific pro-proliferative signaling molecules are present, when the cell is firmly attached to a particular type of extracellular matrix, when cell volume is sufficiently large to support two daughter cells, and when the genomic DNA is intact. Each of these individual conditions may be necessary but not sufficient on its own to switch on the proliferation program. Thus, cellular information processing is much more complicated than the simple conversion of one kind of input into another kind of output. Finally, cells must modulate their output over time in various ways. For example, it is often useful for a cell to respond to a constant stimulus with an initial burst of output activity, then to turn off the response. This down-regulation is a specialized kind of feedback, in which output from the system suppresses future output. More complex kinds of output mod- ulation allow a system to adjust its sensitivity to respond to widely dif- ferent levels of input (adaptation), or to generate waves or oscillations of output activity. the FunDamental Role oF sIgnalIng In BIologICal PRoCesses Our current understanding of signaling mechanisms is the result of many years of research in seemingly unrelated fields, using an array of different experimental approaches. It is really only recently, with the remarkable advances in our ability to identify, clone, and sequence key genes involved in a process, that we have discovered that the same or closely related types of communication molecules are utilized in a wide range of physi- ological information-processing functions. This synthesis, which has led to the emergence of the field of cell signaling, is one of the major scientific accomplishments of the last few decades. Work in many different fields converged to reveal the underlying mechanisms of signaling The field of cell signaling emerged from a number of disciplines that have historically been considered distinct (Figure 1.3). Indeed, the diversity of areas of inquiry that ultimately led to the field of cell signaling under- lines the central role of signaling across biology. For example, because signaling is so important to normal physiology, the disruption or misregu- lation of signaling mechanisms is the basis for many human diseases, and thus these mechanisms are of interest in the areas of medicine and human health. Similarly, because normal development depends on the precise coordination of cell behaviors such as differentiation and move- ment, research on developmental events necessarily sheds light on the underlying signaling mechanisms. And because the signaling apparatus is comprised of biomolecules such as proteins, which are encoded by the genetic material, signaling mechanisms are amenable to the experimen- tal approaches and analytic tools of biochemistry and genetics. endocrinology developmental biology immunology growth & differentiation pathways hormone response pathways immune activation pathways cancer biology neurobiology Figure 1.3 Many different fields of research contributed to the current understanding of cell signaling. Research in a wide variety of different disciplines revealed a common set of mechanisms and pathways that provide the basis for diverse biological activities. oncogenes neuronal signaling cell signaling Cancer biology played a particularly important role in the emergence of the field of cell signaling. Our understanding of the molecular basis of can- cer was revolutionized by the discovery of oncogenes, genes that when mutated or overexpressed induce cells to respond inappropriately to nor- mal signals and therefore proliferate uncontrollably, potentially leading to the formation of a malignant tumor (cancer). These oncogenes, once cloned and biochemically characterized, were in most cases found to be constitutively active or otherwise misregulated forms of signaling pro- teins. By understanding how signaling mistakes could lead to cancer, we learned a great deal about the normal signaling mechanisms that regu- late cell proliferation and differentiation. The field of endocrinology focuses on how hormones secreted into the blood, such as insulin, coordinate physiological communication between the different organs and glands that comprise an organism. As research- ers delved into the biochemical basis by which target cells respond to hormones, a similar group of signaling molecules to those found in can- cer biology was uncovered. For example, the receptor for insulin, because of its enormous significance to the common human disease diabetes, was one of the first hormone receptors to be cloned. The fact that it was highly related to some of the oncogenic proteins that cause cancer, and had the same biochemical activity (the ability to phosphorylate tyrosine residues on its target substrates), underscored the commonality of sign- aling mechanisms. In the field of developmental biology, powerful genetic methods were used to identify mutations that disrupt the patterns of distinct cell fates that emerge within a multicellular organism. Model organisms such as the fruit fly Drosophila melanogaster and the roundworm Caenorhabditis ele- gans were screened for genetic mutations that affected normal develop- ment. As in cancer research, identifying genes whose mutation disrupted normal physiology was a key tool for understanding the normal signaling mechanisms. Embryologists using other approaches, such as manipulat- ing frog or chick embryos, also zeroed in on regulatory molecules involved in cell fate determination and differentiation. As the genes that regulated development were identified and characterized, many were found to be similar or identical to signaling molecules implicated in oncogenesis. Important contributions to our understanding of signaling were also pro- vided by other specialized fields, such as neurobiology and immunology. The nervous system and the immune system are both clearly involved in information processing at the organismal and physiological levels. Research in these fields demonstrated that the development and proper function of these systems involve, at the molecular level, similar sets of signaling molecules and modules. Thus, despite the fact that one system is embodied by neurons that communicate to process higher-order cogni- tive function, and the other is embodied by immune cells that communi- cate to detect and mobilize a response to foreign invading pathogens, at the molecular level they share many common types of molecular parts and molecular network architectures. Despite the diversity of signaling pathways and mechanisms, fundamental commonalities have emerged In the years around 1990, studies in Drosophila and C. elegans converged with ongoing work in cancer biology to identify a key signaling pathway used in many different cellular contexts (Figure 1.4). In Drosophila, developmental biologists had used genetic screens to identify a number of genes important for the development of a specific retinal cell, the R7 Figure 1.4 A common signaling pathway in three different cell systems. Research on the misregulated signaling that drives cancer in humans (left), on the determination of retinal cell fate in the fruit fly (middle), and on the determination of vulval cell fate in roundworms (right) converged on the same signaling pathway. the names of the individual genes or gene products are given for each organism, with the function of each indicated on the far left. In humans, the epidermal growth factor receptor (egFR), Ras, and Raf have all been shown to function as oncogenes (pink). geF, guanine nucleotide exchange factor; maPK, mitogen-activated protein kinase; maPKK, maPK kinase; maPKKK, maPK kinase kinase. human cancer tumor proliferation fruit fly eye development eye cell fate roundworm vulval development vulval cell fate photoreceptor. This particular cell was a good choice because it was not essential for viability, simplifying the design of genetic screens, and its presence or absence could be detected relatively easily in the compound eye of the fly. These genetic studies began to sketch out the signaling path- way that specified the R7 cell fate. This pathway began with a cell-surface receptor with tyrosine kinase activity (a receptor tyrosine kinase), led to a G protein closely related to Ras, which was already known as an onco- gene in vertebrates, then on to a cascade of three serine/threonine protein kinases (termed the MAP kinase cascade). The vertebrate homolog of one of these kinases, Raf, was also a known oncogene. At the same time, inves- tigators were using similar genetic approaches in C. elegans to identify the genes involved in specifying the cells that would develop into the vulva. Again, this particular system offered advantages for genetic screens, as defects were easily visualized and were not lethal. These experiments identified components of a signaling pathway very similar to that found in Drosophila, involving a receptor tyrosine kinase, a Ras-like G protein, and the MAP kinase cascade. A key question in the field then became how the receptor might activate Ras, as was predicted by genetic experiments. In flies, a protein called Sos was discovered that seemed to function downstream of the receptor and upstream of Ras; furthermore, it had homology to a protein that was thought to be a direct G protein activator in yeast. Meanwhile, in C. ele- gans, genetic screens uncovered another gene, Sem-5, that also seemed to act between the receptor and Ras. The Sem-5 protein was unrelated to any known catalytic domains, but did contain small regions of sequence similarity to a number of other signaling proteins including several ver- tebrate oncogenes. Biochemical studies were then able to flesh out the physical and mecha- nistic details of the interactions that were implied by the genetic relation- ships. Sos was found to activate Ras by causing it to release the nucleotide GDP and bind GTP instead (Sos is an enzyme referred to as a guanine nucleotide exchange factor or GEF). The GTP-bound state of Ras that was generated was found to bind to and activate Raf, the first kinase in the MAP kinase cascade. Sem-5 (and its Drosophila and human homologs, Drk and Grb2, respectively) was found to function as an adaptor, binding to the activated receptor tyrosine kinase and bringing along Sos, which then activated Ras. Thus, a complete step-by-step pathway from external ligand to receptor to Ras and the MAP kinase pathway was elucidated, the end result of which was MAP kinase-mediated phosphorylation of nuclear transcription factors and changes in the genes transcribed in the cell. Remarkably, this same basic signaling framework was used to deter- mine cell fate in fruit flies and roundworms, and to stimulate proliferation of human cells. Mapping of this pathway was a beautiful convergence of developmental genetics and cancer biology. This convergence has led to a new apprecia- tion of signaling, one that emphasizes common elements and strategies that are used in many different systems. Signaling is no longer simply a description of one specific physiological pathway (or set of pathways), nor is it associated with only one particular class of physiological processes. Rather, it can be viewed from the perspective of a more general set of information-processing tasks and solutions. In this book, we emphasize the common features of signaling mechanisms: how the same kinds of strategies and molecular components are used over and over to solve simi- lar biological problems. Signaling must operate at multiple scales in space and time We can roughly divide signaling pathways into the events that occur mostly on the plasma membrane and in the cytosol, and the nuclear events that lead to increased or decreased transcription of specific genes (Figure 1.5). Changes in gene expression are the end result of many sig- naling pathways. This makes sense, as stable, wholesale changes in cell behavior are likely to require the synthesis of new sets of proteins that allow the cell to adapt to its changing environment. Those new proteins can only be generated by changes in the transcription of the messenger cell signaling enables transmission from outside of cell to nucleus fast ON and OFF (seconds to minutes) transient changes (minutes to hours) spatial/directional responses and organization energetically cheap (no protein synthesis) gene expression slow ON and OFF (minutes to hours) stable changes (hours to years) limited spatial responses energetically costly (transcription and translation) Figure 1.5 Cellular regulation integrates signaling and gene expression networks. a typical cell signaling system is illustrated, in which binding of ligands to cell-surface receptors ultimately leads to changes in gene expression in the nucleus. Key properties of the cell signaling and gene expressions systems are compared to highlight important differences. RNAs (mRNAs) that encode them. These transcriptional responses, how- ever, are rather different in character than the information transfer and processing events that cause them. Here we consider some of these dif- ferences in terms of timing, spatial control, and energetic cost to the cell. Let us first consider the consequences of changes in gene expression. The process of transcribing a message into mRNA and translating it into pro- tein is relatively slow, occurring on the time scale of minutes to hours. The effects are slower still, as it takes considerable time for new gene prod- ucts to accumulate and for existing proteins to disappear. Furthermore, changes in transcriptional patterns are often highly stable, as when a cell terminally differentiates into a specialized cell type such as a muscle cell or neuron. In most cases, the effects of transcription are cell-wide, not lim- ited spatially to specific sites in the cell. Finally, transcriptional changes require significant investment of cell resources, both in energy and in raw materials, to transcribe new mRNAs and to translate new proteins. Such investments are not to be undertaken lightly by a cell. By contrast, the events that precipitate transcriptional changes are usually faster, more spatially organized, and less costly in terms of energy expendi- ture. First, unlike transcription, which is limited to the nucleus where the genomic DNA resides, the cell signaling machinery spans the cell from the external surface of the cell membrane into the nucleus. Indeed, one of the major tasks of many signaling systems is to transduce signals from outside the cell to the interior, where they can modulate intracellular proc- esses such as transcription. A second major difference is in the speed of the response. Many cell signaling events are extremely rapid, capable of switching states in fractions of a second, and the changes induced are often transient and reversible. Multiple time scales are often involved. For example, activation of a receptor may occur virtually instantaneously upon binding of its ligand; the activated receptor may then lead to the activa- tion of a second intracellular enzyme, which may occur within a few sec- onds. This second enzyme may remain active for a few minutes, modifying downstream effectors, which may include transcription factors (there can be many intermediate steps, depending on the particular pathway). The modified transcription factors then become active in inducing or repressing transcription of certain genes, ultimately changing protein constituents of the cell at even longer time scales. Another difference between cell signaling mechanisms and gene expres- sion is that signaling can be spatially organized and localized. Again, this organization can occur over different scales, from the distance of a few molecules to the diameter of the cell and beyond. For example, certain receptors can cluster together when activated and dramatically change the protein and lipid composition in the immediate vicinity of the cluster, compared to the rest of the membrane. On a larger scale, one end of a cell may respond to an environmental cue by extending actin-rich protru- sions (lamellipodia) that move the front edge of the cell forward, while at the other end the cytosol retracts to push the cell body ahead, leading to directed motion. Overall, the complex morphologies that we observe in cells are usually organized or directed by signaling proteins and circuits that regulate components of the cytoskeleton. Finally, most signaling transactions do not expend anywhere near the resources needed to make new proteins, usually requiring the equivalent of a molecule or two of ATP to transmit information from one molecule to another. Because signaling reactions are relatively cheap, it is not a seri- ous disadvantage for the cell to continuously monitor the environment. In this book, we concentrate on the relatively fast, spatially organized, and cheap mechanisms that allow a cell to respond nimbly to environmental cues. We will devote much less attention to the ultimate long-term conse- quences of such signaling, such as gene expression. the moleCulaR CuRRenCIes oF InFoRmatIon PRoCessIng How are molecules in the cell used to store and transmit information? At its most fundamental level, the transfer of information requires some sort of change in state in the components of the signaling apparatus. In the cell, which is limited in the molecular raw materials from which the sig- naling apparatus can be built, a relatively small number of state changes are used over and over for signaling. In this section, we will consider these basic “currencies” of signal transduction, and how they are often combined in signaling mechanisms. Information is transferred by changes in the state of proteins We can think of information transfer as a series of switches or nodes that can change state in response to input signals; when a node changes state, a signaling output is generated (Figure 1.6a). When multiple nodes are linked together, information can be processed in a sophisticated fashion. Indeed, this kind of simple architecture provides the basis for electronic information-processing devices like a computer, which may contain mil- lions of transistor-based switches linked together. The idea that some kind of change is essential for information process- ing cannot be emphasized too strongly. We intuitively understand this concept from our own experience. For example, we notice and react to Figure 1.6 changes in sound—a loud noise breaking the silence, or a change in pitch and rhythm—whereas a sound of a constant volume and frequency rap- idly fades into the background. Similarly, a constant radio signal has lit- tle information content, but if the amplitude of the radio waves changes over time, or their frequency, then an enormous amount of information can be conveyed (these correspond to amplitude modulation, or AM, and frequency modulation, or FM, on the radio dial). What is the molecular basis for cellular information-processing machines? This is where cellular systems diverge significantly from the example of electronic circuits. In electronic circuits, there is one universal currency for information transmission, which is the electrons flowing through the circuit. There are relatively few devices that process information, and their inputs and outputs all involve this single currency. In cell signal- ing, there is a wider variety of signaling currencies, and therefore a wider variety of molecular devices or systems that are able to read one currency and convert it to other output currencies. Changes in state link signal inputs to outputs. (a) a generic cell signaling system is illustrated on the left, and a single node of that system on the right. the node can be thought of as a switch that can exist in two states, oFF and on. signal inputs (green arrow) can switch the state, leading to a change in output (orange arrow). In cases where the change of state is induced by a post-translational modification (pink “X”), enzymes that add or remove the modification are inputs that can be thought of as “writers” and “erasers.” (c) often, changes in state are affected by multiple different inputs, some of which promote switching (arrows) while others inhibit it (t-shaped arrow). (b) (c) “WRITER” “ERASER” MULTIPLE INPUTS OUTPUT In the cell, proteins are the workhorses of signal processing. Proteins are enormously versatile in the types of physical structures that they can form and in the chemical reactions that they can perform. One class of proteins, the enzymes, act as catalysts that can enhance the rates of spe- cific, useful biochemical reactions enormously, providing the basis for energy metabolism, replication, motility, and other behaviors associated with life. It is certainly then no surprise that proteins are essential for cell signaling mechanisms. Other biomolecules such as lipids, nucleic acids, and small molecules such as ions and nucleotides can play a supporting role, as we shall see, but for the most part these molecules participate in signaling by changing the properties of proteins. Thus if we want to understand the changes that underlie cell signaling, we must look to the properties of proteins. In cell signaling, the most basic units of information are changes in the state of proteins. For example, a protein may have very different activities depending on whether or not it has a particular chemical modification (such as phosphorylation of a certain side chain). Perhaps the unmodi- fied protein is inactive (off) and the phosphorylated form is active (on). The state of this protein is changed by input from other proteins that can add the phosphate group or remove it. In this example, we can think of the enzymes that add phosphate (termed protein kinases) as “writers,” in that they add information in the form of post-translational marks, and the enzymes that remove those marks (protein phosphatases in this example) as “erasers” (Figure 1.6b). This kind of reversible change is essential to any kind of information-processing scheme. Furthermore, signaling pro- teins are often subject to regulation by multiple inputs, both positive and negative (Figure 1.6c). This allows the state or activity of the protein to depend in a relatively complex way on the specific combination of inputs at any particular time. There is a limited number of ways in which the state of proteins can change If information transfer requires change, and proteins lie at the heart of signaling, it follows that changes in the properties of proteins will provide the basis for cell signaling mechanisms. It turns out that the number of such changes that are used is really quite limited. These basic currencies of cell signaling are listed in Table 1.2—and a subset is illustrated in Figure 1.7—and each is described briefly below. The next five chapters of this book deal with each of these currencies in greater detail. Interactions between proteins, or between proteins and other biomol- ecules, can dramatically affect many aspects of their behavior, such as their activity and localization. The assembly or disassembly of multipro- tein complexes is frequently a key step in transmitting cellular signals. Protein–protein interactions and their role in signaling are the topic of Chapter 2. Table 1.2 The currencies of cell signaling Currencies of cell signaling Book chapter Protein–protein interactions Chapter 2 Conformation Chapter 3 enzymatic activity Chapter 3 Post-translational modification Chapter 4 subcellular localization; Concentration Chapter 5 small-molecule signaling mediators Chapter 6 binding/dissociation conformational change post-translational modification X localization Figure 1.7 Different signaling currencies. Different ways that the state of proteins can be changed are illustrated. In all cases, the on state (activation) is indicated by a pink halo. Proteins are not fixed in their three-dimensional shape or conformation; instead, they can often adopt multiple conformations that differ greatly in their activity. Protein conformation switches such as G proteins, which switch between active and inactive conformations depending on whether they are bound to GDP or GTP, are central to many signaling pathways. Many signaling proteins are enzymes, which catalyze specific reactions such as the chemical modification of other proteins, lipids, or other bio- molecules. The activation (or inactivation) of such enzymes by upstream signals can lead to widespread and massive downstream effects, due to the remarkable efficiency and specificity of enzymes as catalysts. Because changes in enzymatic activity are often tightly coupled to conformational changes, these two currencies and their interrelationship are discussed together in Chapter 3. Proteins are subject to a number of different types of chemical modifi- cations after they have been synthesized (these are collectively termed post-translational modifications). These modifications are performed by signaling enzymes, and include the addition or removal of small chemical groups such as phosphate, or of more substantial structures such as the small protein ubiquitin. Proteolysis (breaking the polypeptide backbone of the protein) is a particularly extreme form of post-translational modi- fication. Chapter 4 examines the post-translational modifications used in signaling, how they are regulated, and the downstream consequences of modification. The cell is not a well-mixed and homogeneous solution, so changes in the localization of proteins within the cell can affect their activities dramati- cally. For example, the relocalization of a protein from the cytosol to the nucleus allows it to interact with the genomic DNA, which is entirely restricted to the nucleus. Similarly, many important targets of signaling enzymes are confined to the plasma membrane, and thus recruitment of proteins to the membrane can be a critical step in propagating signals. The regulation of subcellular localization and the role of changes of locali- zation in signaling are discussed in Chapter 5. A number of important signaling pathways involve changes in the abun- dance of small molecules such as calcium ions or cyclic AMP (cAMP). These small-molecule signaling mediators, often termed second messen- gers, are created and destroyed by upstream synthetic and degradative enzymes, and they exert their effects by binding to and altering the activi- ties of downstream target proteins. Signaling that involves small-molecule mediators can have rather distinct properties due to the potential of these molecules to diffuse rapidly throughout the cell. Small signaling media- tors are the topic of Chapter 6. In addition to these changes in the state of proteins, signaling also depends on the protein abundance (concentration). The rates at which biological reactions occur and the steady-state levels of their products are proportional to the concentrations of the reactants; the higher the concen- tration, the more likely two components are to interact with each other. The balance between the rates of synthesis and degradation of various components sets their overall concentration in the cell. Thus, a change in the concentration of a particular protein can itself represent a way to store information. Of course local concentrations can be changed dramati- cally by changes in subcellular localization, even under conditions where the total amount of the molecule in the cell does not change. Most changes in state involve simultaneous changes in several different currencies Although there is a limited number of ways in which proteins can change to transmit information, a change in one of these currencies is often intimately linked to change in one or more of the others. For example, phosphorylation of an enzyme (a change in post-translational modifica- tion) may induce a conformational change in the modified enzyme, which in turn changes its catalytic activity (Figure 1.8). The switch from one state to the other (inactive to active) involves changes in multiple proper- ties that cannot easily be disentangled from each other. For the sake of simplicity, many of the following chapters discuss each type of change in isolation, but this way of organizing the book deliberately de-emphasizes the functional linkages and interrelationships among them. To illustrate this point, we will consider a rather typical example of a switch between states that involves many different changes in protein properties. Src family kinases are enzymes that catalyze the phospho- rylation of tyrosine side chains on target proteins in response to various extracellular signals. They play an important role in diverse physiological pathways, such as cell adhesion and lymphocyte activation. The namesake of the family is the Src oncogene, originally identified in a chicken tumor virus. Src was the first oncogene to be cloned and sequenced, and encodes the first tyrosine kinase to be identified, and so it holds an important place in the history of signal transduction research. The two predominant states for Src family kinases are a “closed” catalytically inactive state that is tightly folded together, and a more “open” catalytically active state (Figure 1.9). The inactive state is stabilized by phosphorylation of a negative regulatory tyrosine at the C-terminus, tyrosine 527 (Tyr527). In the active state, Tyr527 is unphosphorylated while a second site, tyrosine 416 (Tyr416), is phosphorylated. In this conformation, two protein-binding segments (the SH2 and SH3 domains) that had been involved in intramo- lecular interactions in the inactive state are now accessible to bind to other proteins. These domains are also important to localize the activated kinase to specific sites in the cell, for example focal adhesion complexes. phosphorylation conformational change in catalytic activity Figure 1.8 Different changes in protein state can be coupled. In this example, phosphorylation of a hypothetical protein is tightly coupled with conformational change and change in catalytic activity. INPUT signaling protein OUTPUT OFF (closed) protein binding, relocalization ON (open) SH3 SH2 INPUT pTyr527 phosphorylation pTyr416 P OUTPUT substrate phosphorylation pTyr527 domain protein binding, relocalization active kinase domain Let’s go through in a little more detail the sequence of events leading from the “off” state to the “on” state. We will assume the initial input is dephos- phorylation of Tyr527 by a phosphatase (a change in post-translational modification). The immediate consequence of dephosphorylation is to destabilize the closed conformation, which allows the SH2 and SH3 domains to dissociate from the catalytic domain (a change in conforma- tion). This conformational change increases the activity of the catalyt- ic domain (a change in enzymatic activity), which makes it more likely that the activating site, Tyr416, will become phosphorylated (a change in post-translational modification). In turn, Tyr416 phosphorylation stabi- lizes the active conformation of the catalytic domain. The SH2 and SH3 domains, released from their intramolecular interactions with the cata- lytic domain, are now free to bind to other cellular proteins (a change in protein–protein interactions). These interactions result in localization of Src to specific sites in the cell where its substrates are found (a change in subcellular localization). In this rather typical example, the switch from the inactive to active state involves the concerted change of at least six distinct yet interrelated prop- erties of the protein. Note also that such a regulatory scheme can accom- modate multiple inputs that could impact the final state, including the levels of kinases that phosphorylate Tyr527, of phosphatases that dephos- phorylate Tyr527, and of kinases and phosphatases that act on Tyr416, and the local concentrations of binding partners for the SH2 and SH3 domains. lInKIng sIgnalIng noDes Into PathWays anD netWoRKs In the previous section, we discussed how changes in the state of pro- teins provide the fundamental currencies of cellular signal processing. However, most signal transduction tasks do not involve a change in state of a single protein or node, but instead involve many different changes linked together in long chains (pathways) or interconnected networks. The increased complexity afforded by linking together individual signal- ing nodes is what permits more sophisticated and complex information processing. Information transfer involves linking different changes of state together In general, information transfer involves the conversion of one type of change into another type. We have already seen examples of this with- in a single signaling protein, as above, where dephosphorylation of the Figure 1.9 Inactive and active states of a Src family tyrosine kinase. In the inactive state (left), tyr527 is phosphorylated (ptyr527) and the protein is folded in a compact, “closed” conformation via intramolecular interactions of the sh2 and sh3 domains. upon activation (induced by dephosphorylation of tyr527), the protein adopts a more open conformation in which catalytic activity is high, tyr416 is phosphorylated, and sh2 and sh3 domains can bind to other proteins. Figure 1.10 Linking signaling nodes together. Change in one signaling node can be induced by input from upstream nodes, and its output can, in turn, induce changes in downstream nodes. In this way, multiple nodes can be linked together to form pathways and networks. transmission between signaling proteins regulatory site on Src led to conformational changes, which in turn led to increased catalytic activity. But this is true in an even broader sense, in that virtually all signaling inputs must be converted into another form, often many times, in order to generate the appropriate cellular response. It is this concept of conversion of one type of signal into another that led to the widespread use of the term “signal transduction” to describe the field of cell signaling. In addition to the conversion of different currencies within the same sig- naling protein, multiple proteins are often functionally linked together by their ability to change each other’s state (Figure 1.10). For example, activation of an enzyme is linked to the post-translational modification of its substrate; modification of a protein may be linked to its ability to bind to a second protein; relocalization of an enzyme may be linked to modifica- tion of a substrate that is present only in the new location; conformational change in one protein may be linked to its dissociation from a second pro- tein; and so on. These physical or functional linkages between proteins provide a way for the output of one signaling node to serve as the input for another. Much of the actual research in signal transduction has involved the discovery of such connections between signaling proteins, and the demonstration of their functional importance to particular cell behaviors. Multiple state changes are linked together to generate pathways and networks Links between individual signaling nodes are generally assembled together into much larger architectures—pathways and networks. The term pathway is typically used to describe a linear chain of interactions where the output of each node serves as the input for the next downstream node. A specific example was provided above in the case of the signaling from the EGF receptor through Ras, Raf, and the MAP kinase cascade (see Figure 1.4). Like a row of falling dominoes, a long series of changes of state are linked together so that the initial input (high concentrations of the activating ligand for the EGF receptor in this case) is tightly linked to the ultimate output (phosphorylation of substrates such as transcrip- tion factors by the MAP kinase Erk). Thus, a pathway defines the chain of interconnections that link a specific input to a specific output. In fact, in most cases, such a linear pathway is a gross oversimplification of the actual relationships between the individual proteins involved. This is because at each step, it is more common for each node to receive inputs from multiple nodes, and for its output to affect the activity of multiple nodes. We have already discussed above how the Src family kinases can receive multiple activating and inactivating inputs. The same example also illustrates how the output of one node can provide inputs into mul- tiple downstream nodes, as Src kinases can phosphorylate diverse sub- strates, serving to activate, inhibit, or otherwise modulate the activity of many different proteins in the cell. It is this branched network of functional interactions among signaling proteins that allows more complicated information processing beyond the simple cause-and-effect relationships implied by a linear pathway. These interrelationships can generate behaviors such as feedback regulation, integration, adaptation, and complex spatiotemporally controlled output patterns. In Chapter 11, we will discuss in much more detail how these systematic behaviors can be built by linking together simple signaling nodes in various ways. Cellular information-processing systems have a hierarchical architecture In this chapter, we have discussed cellular information processing on a variety of different scales—from changes in the state of individual signal- ing proteins, all the way up to communication between distant organs in a multicellular animal. While it may seem at first glance that these mecha- nisms of information transfer are very different at these different scales, if we consider more closely, we can discern a common logic that allows ever-more complex systems to be built out of relatively simple components and rules. This idea is illustrated in Figure 1.11. At the smallest scale, individual proteins act as molecular signaling devices, receiving inputs from the envi- ronment or upstream components and changing state in response, there- by generating an output signal. A Src family kinase is such a molecular signaling device. At a larger scale, a series of molecular devices are linked together into a network signaling device, in which the interrelationships can generate more complex and sophisticated information-processing behaviors. The EGF receptor–Ras–MAP-kinase pathway is an example of such a network signaling device. Finally, at the scale of the cell, the outputs of various network signaling devices are harnessed together into a larger network that regulates complex cellular behaviors such as prolif- eration, differentiation, and directed movement. molecular signaling device network signaling device cellular signaling device INPUTS OUTPUTS INPUTS OUTPUTS INPUTS OUTPUTS Figure 1.11 Signaling machines at different scales. Cell signaling involves information- processing devices that operate at several scales, from individual proteins, to networks, to whole cells. molecular signaling devices are linked together to make network signaling devices, which, in turn, are linked to regulate behaviors at the cellular level. these devices all share a common set of generic signal processing tasks that they must accomplish, by receiving and integrating inputs and converting these to outputs. We have organized most of this book around core molecular modules and mechanisms, and focused on explaining how they are commonly used to build generic information-processing systems within cells. This modular and conceptual treatment of signaling emphasizes how certain molecular components cannot be viewed as uniquely belonging to any one specific process; rather, in most cases, they have been incorporated in diverse proc- esses through evolution because they are well suited to perform a particu- lar class of information-processing tasks. Moreover, this view allows us to see the forest from the trees and to understand the logic of the higher- order architecture and organization of signaling systems. While this approach highlights the core fundamentals underlying cell sig- naling, it poses a danger that one might also lose touch with the broader physiological context in which a particular signaling mechanism operates. To balance this, we have tried to organize this book in a hierarchical way that matches the hierarchical organization of signaling systems: to focus many of the earlier chapters on fundamental molecular mechanisms for transmitting information, but to transition in later chapters to how these molecular systems are used in a hierarchical manner to build networks that solve larger cellular and physiological problems. summaRy Cell signaling encompasses the ability of cells to detect changes in the environment and respond to those changes—a fundamental property of living systems. Our current understanding of cell signaling has resulted from a convergence of research in a variety of fields including cancer biol- ogy, developmental biology, endocrinology, biochemistry, and genetics. It is now apparent that common strategies and cellular components are used over and over to solve various cell signaling tasks. To perform this informa- tion processing, cells use genetically encoded biomolecules, predominantly proteins. Switching between protein states is the most fundamental build- ing block for cell signaling mechanisms, and the number of ways that the state of proteins can be changed as a result of signaling inputs—the currencies of signaling—is quite limited. Such changes of state are often linked together, either within the same molecule (a molecular signaling device) or between different proteins (to generate signaling pathways and networks). By understanding how different signaling proteins are func- tionally linked together, we can understand the behavior of the signaling pathway or network as a whole. QuestIons Imagine that you are a primitive single-cell organism. What type of new signaling response could provide you with a potential evolution- ary advantage over the other competitor organisms in the primordial soup? When might the same signaling response provide a fitness disad- vantage? What general principles govern whether a new input–output response would be advantageous? What are some of the common signals and stresses that must be detect- ed by both single-cell organisms and individual cells that are part of multicellular organisms? What are some of the different signals that they must detect? What types of output response might be unique to cells from multicellular organisms? Multicellular organisms often use complex organ systems to sense and process external information. For example, an animal will use its vis- ual system and nervous system to detect many external stimuli. What ReFeRenCes types of inputs, decisions, and outputs must be made by the individual cells that make up the visual and nervous systems? Why are signaling proteins involved in cancer (such as oncogene prod- ucts) also often found to be involved in the process of development? What is the relationship between signaling and transcriptional regula- tion? What are the commonalities and differences between these two systems? What types of cellular response can only be generated by sig- naling systems? Does phosphorylation of a signaling protein by an upstream kinase represent a signaling input or output? Explain. How can the localization of a signaling protein be considered to be a currency for storing information? Describe how inputs can modify localization. Describe how localization can change downstream physi- ological outputs. What are the other molecular currencies that signaling proteins can use to store and transmit information? How can upstream inputs alter these currencies? And how can changes in these currencies be convert- ed into downstream responses? ReFeRenCes Alon U (2006) An Introduction to Systems Biology: Design Principles of Biological Circuits. Boca Raton, FL: Chapman & Hall/CRC. Cox AD & Der CJ (2010) Ras history: The saga continues. Small GTPases 1, 2–27. Koshland DE Jr (2002) Special essay. The seven pillars of life. Science 295, 2215–2216. Lim WA, Lee CM & Tang C (2013) Design principles of regulatory networks: searching for the molecular algo- rithms of the cell. Mol. Cell 49, 202–212. Margolis B & Skolnik EY (1994) Activation of Ras by receptor tyrosine kinases. J. Am. Soc. Nephrol. 5, 1288–1299. Martin GS (2001) The hunting of the Src. Nat. Rev. Mol. Cell Biol. 2, 467–475. Sternberg PW (2006) Pathway to RAS. Genetics 172, 727–731. Tyson JJ, Chen KC & Novak B (2003) Sniffers, buzz- ers, toggles and blinkers: dynamics of regulatory and signaling pathways in the cell. Curr. Opin. Cell Biol. 15, 221–231. Taylor & Francis Taylor & Francis Group https://taylorandfrancis.com Principles and Mechanisms of Protein Interactions The proteins that participate in signaling, like the components of any complex machine, must interact with each other in order to perform their function. Furthermore, changes in these interactions between signaling components can convey information. In this chapter, we will consider non- covalent physical interactions between signaling proteins and their bind- ing partners, or ligands. The cytosol and other biological fluids are very highly concentrated solu- tions of proteins and other components, in which molecules are continu- ously jostling and colliding with each other. In some cases, these collisions lead to binding—the relatively stable association of the two components. The likelihood that a particular interaction will occur at various concen- trations of components, and the dynamics of how rapidly such interactions form and dissociate, all play fundamental roles in defining the behavior of signaling mechanisms. Here we will examine the principles that govern the physical interactions of signaling molecules, and how these interac- tions are used to transmit information. Protein–protein interactions will dominate our discussion in this chapter, but it is important to realize that the interactions of proteins with other cell components are also important for signal transduction. In particular, the binding of proteins to specific lipids is important for localizing signal- ing proteins to defined membrane compartments or for modulating their activity. And, of course, the specific binding of proteins to nucleic acids underlies the regulation of processes such as DNA replication and the transcription and processing of mRNA. Most of the concepts that we will introduce in discussing protein–protein interactions also underlie other macromolecular binding interactions. (a) (b) Figure 2.1 Changes in binding can lead to relocation of proteins. (a) a transmembrane receptor (brown) resides on the surface of the cell, and its potential binding partner resides in the cytosol. In its inactivated state, the receptor cannot bind the cytosolic protein. (b) Upon activation, the receptor can now bind to its partner. Most of the previously cytosolic protein is now relocated to the membrane, in complex with activated receptor. Scaffolds and MAP kinase cas- cades are discussed in Chapter 3 ProPertIes of ProteIn–ProteIn InteractIons In this section, we will discuss the factors that determine whether or not two proteins will interact with each other in the cell. In particular we will focus on two critical parameters of binding interactions, affinity and spe- cificity, and address the quantitative description of binding interactions and their dynamic behavior. Changes in protein binding have both direct and indirect functional consequences A variety of signaling inputs can alter the binding of proteins with each other. These include changes in a protein’s abundance or its distribution throughout the cell, changes in post-translational modifications such as phosphorylation, and changes in protein shape or conformation. In turn, protein–protein interactions can dramatically affect the behavior of the binding partners involved; this is why changes in binding are capable of transmitting information during signaling. One of the most straightforward consequences of binding is the assembly of more complicated multifunctional structures from sim- pler component proteins. By combining the catalytic activity, binding activity, or other functions from several proteins, the resulting assemblies can perform more complex functions than can the individual components. For example, many different inputs can be integrated into one output pathway, or one input signal can be diverted down many different output pathways, or different steps in a process can be coordinated. Another common result of binding is to change the subcellular localization of one of the binding partners. For example, most cell-surface receptors undergo conformational changes or post-translational modification of their cytosolic domains upon activation. These changes create new binding sites for cytosolic proteins. Since the receptors are confined to the plasma mem- brane, by binding to the receptor, a previously soluble protein can be effec- tively localized to the membrane (Figure 2.1). Many enzymes, such as phospholipases and other lipid-modifying enzymes whose substrates reside in the membrane, are regulated by recruitment to the membrane in this way. Another important consequence of binding is to bring enzymes into contact with potential substrates. For example, many protein kinases bind stably to their substrates via specific docking sites or substrate-binding domains. Such binding may be important to ensure that only a very specific substrate is phosphorylated, and that this phosphorylation is highly efficient. Such interactions may also be mediated by specialized proteins termed scaffolds. Scaffolds bind multiple proteins, such as enzymes and their substrates, involved in a single process. For example, in the MAP kinase cascade, a conserved signaling pathway that couples upstream signals to the phos- phorylation of substrates such as transcription factors, a scaffold protein simultaneously binds three different protein kinases. These three kinases function in a cascade, in which they sequentially phosphorylate and thereby activate each other. The fact that they are all assembled into a single com- plex by the scaffold makes the overall reaction faster than if the three kinas- es needed to encounter each other separately through random collisions in solution. Furthermore, specificity and efficiency are increased because interaction with other competing substrates is prevented (Figure 2.2). The interaction of proteins with one another can also affect their biological activities quite directly. For an enzyme, binding of another protein can induce changes in its shape (allosteric changes) that increase or decrease its catalytic activity or other properties (Figure 2.3). This type of regulated conformational change will be considered in more detail in Chapter 3. Even in cases where binding does not result in dramatic changes in the con- formation of a protein, it may otherwise affect activity in a number of ways, for example by blocking access to substrates or to other binding partners. Protein binding can be mediated by broad interaction surfaces or by short, linear peptides The basis for protein–protein binding (or the binding of proteins to other macromolecules) is a specific fit, or complementarity, between the two interacting surfaces. The shape of the two surfaces largely determines this fit. Like pieces of a puzzle, two surfaces are more likely to bind if they have complementary shapes, with bumps and ridges on one surface fitting into holes and grooves on the other (Figure 2.4). Binding is also stabilized by hydrophobic and electrostatic interactions and hydrogen bonds between the interacting surfaces. As one would intuitively expect from simple rules of chemistry, hydrophobic patches on one surface gener- ally interact with hydrophobic patches on the other, negatively charged side chains are likely to interact with positively charged side chains (Figure 2.4d), and the formation of hydrogen bonds is favored. One way to describe protein–protein interactions is by their buried surface area. This is the region that is exposed to solvent in the uncom- plexed proteins, but is buried in the protein–protein interface in the complex (Figure 2.4c). Typical protein–protein interactions among signaling proteins have a buried surface area of ~ 1200–2000 Å2; in general, the greater the buried surface area, the stronger the interaction. In many cases, this binding surface is relatively hydrophobic, and thus binding to another protein is favored because it shields the hydrophobic region from the polar solvent. Individual amino acids do not contribute equally to binding; instead, some amino acids within the interface are absolutely critical for binding, while others can be altered by mutation without affecting binding. (a) Figure 2.2 Protein binding can make reactions more efficient and specific. (a) enzyme a and its two potential substrates B and c are free in solution; the reaction is relatively inefficient and targets both B and c. (b) a scaffold protein binds specifically to both enzyme a and substrate B, but not to substrate c. the enzymatic reaction is highly efficient and specific for substrate B. (b) inactive active Figure 2.3 Binding can directly alter the activity of proteins. (a) In its unbound state, a protein adopts an inactive conformation. Upon binding of a second protein (orange), changes in the conformation of the first protein result in its activation. (a) (b) Figure 2.4 The nature of protein-binding interfaces. (a) a schematic representation of a protein–protein interaction, illustrating the complementary shapes of the interacting surfaces (orange). top, a surface representation of human growth hormone receptor (green) bound to growth hormone (blue), based on the x-ray crystal structure. Below, the binding surfaces of the receptor and growth factor are shown by peeling away the hormone from the receptor surface and rotating it 180°. the buried surface area of the interface is shaded o r a n g e. (c) the surface electrostatic potential of the two binding surfaces (red, acidic; blue, basic; light gray, hydrophobic); the orientation is the same as in the lower part of panel (b). (a) (b) (c) (d) Modular binding domains are discussed in more detail in Chapter 10 Protein–protein interaction surfaces can be roughly divided into two categories. In the first, relatively broad surfaces of the two binding partners interact, with each binding surface composed of amino acids that may be widely separated in the linear sequence of the protein (see Figure 2.4). In the second, one of the binding surfaces consists of a rela- tively short, linear peptide sequence (usually four to eight amino acids in length) that fits into a corresponding groove on the surface of its bind- ing partner (Figure 2.5). Both types of interactions are seen in signaling proteins, and there are important differences in their physical and evolu- tionary properties. Peptide ligands are short and defined by only a few key amino acids, so such binding sites can be created or destroyed rather frequently by ran- dom mutation of protein-coding genes. By contrast, surface interactions are less likely to arise by chance, because many amino acids, spread over distant regions of different proteins, play a part in the binding. Such interaction pairs are likely to evolve relatively slowly to fulfill a specific physiological function. Binding between two surfaces, because of the extensive contact area, is potentially stronger than surface–peptide binding. Many of the protein interactions in signal transduction involve compact, structurally discrete modules or domains, whose sole known function is to confer on proteins containing them the ability to bind to specific types of ligands. Each type of modular protein-binding domain binds a character- istic linear peptide motif. Two specific examples are SH3 domains, most of which bind to proline-rich peptides that adopt a specific helical second- ary structure, and SH2 domains, which bind to tyrosine-phosphorylated peptides. Examples of lipid-binding domains are the PH domains, many of which bind to specific phosphoinositol-derived lipids. Signaling proteins often contain more than one of these binding modules, each conferring specific binding properties. The presence of modular domains is often obvious from the protein sequence, and because such domains have rela- tively predictable binding partners they provide important clues to the binding properties, and even the function, of the proteins that contain them. The presence of an SH2 domain in a protein, for example, strongly suggests that it functions in signaling pathways regulated by tyrosine phosphorylation. The affinity and specificity of an interaction determine how likely it is to occur in the cell Signaling proteins nearly all bind and interact with other molecules dur- ing the process of signal transduction. Thus it is critical to understand the likelihood that any particular signaling molecule will interact with Figure 2.5 A peptide–protein interaction. (a) a schematic representation of a short, linear peptide (pink) binding to a corresponding groove on the surface of its binding partner ( g r ee n ). for clarity the peptide is shown at the end of a protein; in reality most binding peptides are located at internal sites, in loops or relatively unstructured regions of proteins. (b) a surface representation of an sH2 domain (green) binding to a tyrosine-phosphorylated peptide (pink). the area ( y e ll o w ) on the sH2 domain surface that is buried by the bound peptide. the peptide is depicted in backbone format. (d) the electrostatic potential of the sH2 domain surface (light pink, acidic; blue, basic; light gray, hydrophobic). note the concentrated positive charge (blue) in the area that binds the negatively charged phosphotyrosine (ptyr; indicated by arrow). another under specific conditions. We shall now look at the qualitative and quantitative aspects of binding. First we will consider affinity and specificity, two important parameters of binding interactions. The affinity of an interaction is a measure of the intrinsic strength of the binding interaction. Simply put, if the affinity of the interaction between two species is high, then there is a higher probability of finding the molecules in complex at equilibrium. In contrast to affinity, the specificity of an interaction is not an absolute measure but a relative one—it reflects the relative affinity of a particular interaction (between protein P and ligand A) with respect to other pos- sible interactions (between protein P and all other molecules in the cell). Unlike affinity, which is an intrinsic property of two interacting mole- cules, specificity can only be defined in relation to a particular set of other potential interactions. Because the cellular environment in which signaling occurs is both extremely concentrated (roughly 20% protein by weight) and diverse (con- context 1: all proteins context 2: specific taining many thousands of distinct molecular species, varying in concen- tration over many orders of magnitude), any potential binding reaction is in constant competition with an enormous number of alternative interac- tions. Despite this competition, if an interaction is highly specific it can predominate because it has a higher affinity compared with those of com- peting interactions. The specificity of a particular interaction can vary, however, depending on the exact intracellular context. The relative nature of specificity is underscored by the fact that a single interaction between two molecules can be accurately described as either highly specific or nonspecific, depending on the context. For example, anti- bodies that bind to phosphotyrosine (Figure 2.6) are widely used rea- gents in signaling research. Such antibodies are considered to be highly specific for tyrosine-phosphorylated proteins, binding with much higher affinity to these targets than to the corresponding unphosphorylated pro- teins or the larger set of proteins that contain tyrosine. Yet in another con- text, these antibodies can be considered to be highly nonspecific, in that they bind with similar affinities to virtually all tyrosine-phosphorylated proteins, independent of the protein sequence context surrounding the phosphorylated tyrosine (Figure 2.6). It is possible to generate more spe- cific antibodies that bind a particular phosphorylated site in a protein; such antibodies have much higher affinity for that site compared to all other phosphotyrosine-containing sequences. Another example of relative specificity is provided by the lectins, which are proteins that bind to the sugar groups that make up the complex car- bohydrates found on cell-surface proteins. A lectin may be highly specific for a particular sugar group, failing to bind appreciably to other, closely related compounds. If the particular sugar recognized is widely distrib- uted on a large number of different proteins in the cell, however, the lec- tin could also be described as nonspecific, in the sense that it binds with similar affinity to many different protein targets. These examples show how specificity is a relative quantity that is dependent on the definition of other interactions that are considered as competitive. The strength of a binding interaction is defined by the dissociation constant ( K d) Biologically important binding reactions can vary widely in their strength, ranging from weak and transient interactions that may be dif- ficult to detect to extremely strong interactions that are nearly as stable as covalent bonds. Thus binding is not an all-or-none phenomenon, and all phosphotyrosine-containing proteins nonspecific Figure 2.6 Specificity is a relative quantity that depends on context. (a) an antibody (blue) that recognizes phosphotyrosine would be considered specific in its ability to bind the relatively rare phosphotyrosine residues (pink circles) among all proteins (context 1). (b) However, if one is trying to distinguish a specific phosphotyrosine- containing protein from all other phosphotyrosine-containing proteins (context 2), then the antibody would be considered nonspecific. Phosphospecific antibodies are discussed in Chapter 13 a quantitative measure of the strength of binding interactions is needed in order to estimate whether they are likely to occur in the cell. The most commonly used measure of affinity is the dissociation constant ( K d) of the interaction—the ratio of unbound and bound species observed at equilibrium. The definition of the dissociation constant is based on the law of mass action, which describes the relationship between the concentrations of reactants and products in a chemical reaction. In the case of a binding reaction between two molecules, the reactants are the two individual bind- ing partners, and the product is the bound complex. Thus for the reaction A + B ↔ AB, the rate of formation of AB is given by a rate constant (the on-rate or kon) times the concentration of the binding partners: kon[A][B]. Similarly, the rate of dissociation of the AB complex is given by another rate constant (the off-rate, or koff) times the concentration of the complex: koff[AB]. At equilibrium, the rate of formation of the complex equals the rate of dissociation of the complex, so kon[A][B] = koff[AB]. Rearranging, we obtain koff/kon = [A][B]/[AB]. The equilibrium constant koff/kon is defined as the dissociation constant, or Kd. K = k _off = [A][B] ^d [AB] Equation 2.1 The dissociation constant has units of concentration (moles per liter, or M) and is quite useful because it tells us what fraction will be complexed at a given concentration of individual components. Consider a situation where there is a relatively small amount of A, so B is in large excess. When half of the total amount of A is complexed with B, [A] = [AB]. Then, according to Equation 2.1, under these conditions Kd = [B]. Thus, the dissociation constant is the concentration of B at which half of A is free, and half is complexed with B. We can rearrange Equation 2.1 to obtain an equation that describes the fraction of the total amount of A that is complexed with B, or its fractional occupancy: Fractional occupancy of A = [AB] = [B] [A] + [AB] Kd + [B] Equation 2.2 1 0.5 Kd concentration of B If we plot the fractional occupancy of A as a function of [B] (again we assume that the total amount of A is low compared with that of B, so [B] is very nearly equal to the total concentration of B), we get a hyperbolic plot, often called a binding isotherm (Figure 2.7). It is obvious from this curve and Equation 2.2 that when the concentration of B is much higher than the Kd, almost all of A is complexed with B, while at con- centrations of B much lower than the Kd, almost all of A is free (uncom- plexed). Putting actual numbers into Equation 2.2, we see, for instance, that if [B] is nine times higher than the Kd, then 90% of A will be com- Figure 2.7 plexed with B; on the other hand, only 1% of A will be complexed when the concentration of B is 1/99 of the Kd. Another very useful way to think A binding isotherm. for a simple binding reaction a + B ↔ aB, the fractional occupancy of a is plotted as a function of the concentration of its binding partner B (under conditions where B is in excess, so unbound B is approximately equal to total B). the concentration of B where 50% of a is bound (fractional occupancy of a = 0.5) corresponds to the dissociation constant (Kd) for the binding reaction. about fractional occupancy and the dissociation constant is in terms of binding probability. For any individual molecule of A, when the concen- tration of B is equal to the Kd there is a 50% chance it will be bound to B; if the concentration of B is 1/99 the Kd, there is only a 1% chance that the molecule of A will be bound to B. The dissociation constants for some rep- resentative biological interactions are provided in Table 2.1. Note that because the dissociation constant is an equilibrium constant, it does not tell us the rates of complex formation or dissociation; the kinetics of bind- ing are discussed later in this chapter. 10−15 M avidin–biotin _{− 14} thrombin–hirudin (leech antico- agulant peptide) Methotrexate–dihydrofolate re- extremely high affinity protein–small molecule interaction Hirudin blocks blood clotting by binding and inhibiting thrombin Methotrexate is a small- Table 2.1 Physiological dissociation constants 10−11 M 10−9 M ductase (DHfr); platelet-derived growth factor (PDGf)–PDGf receptor restriction enzyme Ecori–Dna site; catalytic and regulatory subunits of protein kinase a (PKa) molecule drug that inhibits DHfr restriction enzymes cleave specific sites in Dna _{− 7} sH2 domain–phosphotyrosine (ptyr) site 10−6 M ca2+–ca2+-activated enzymes _{− 5} sH3 domain–binding site; atP–kinase _{− 4} Glutathione–glutathione S-transferase (Gst) courtesy of c.t. Walsh. Kinases use atP as a substrate in phosphotransfer reactions The dissociation constant is related to the binding energy of the interaction The dissociation constant also has a thermodynamic meaning, because the relative amounts of free components and of the complex present at equi- librium are directly related to differences in the free energy of the two states. Simply put, if the complex has lower free energy than the individ- ual components, formation of the complex will be favored at equilibrium; conversely, if the complex has higher free energy than the components, little complex will be present at equilibrium. For the binding reaction A + B ↔ AB, this thermodynamic relationship is given by the equation: ΔGo = –RTln [AB] [A][B] Equation 2.3 where ΔG° is the standard free energy for the binding reaction, R is the gas constant, and T is the temperature in degrees K. The term ([AB]/[A][B]) is the equilibrium constant (Keq) for the reaction. For a binding reaction this is termed the association constant (Ka) and, as can be seen from Equation 2.1, it is equal to 1/Kd. Thus we can write: ΔGo = –R TlnKa = –R Tln( 1 ) = R T lnKd Kd Equation 2.4 If we put in values for the gas constant and standard temperature, and convert from the natural logarithm, then ΔG° = 1364 logKd (in calories) or 5707 logKd (in joules). So for a biological interaction of moderate affinity (assume Kd = 10−8 M), then ΔG° is roughly −11,000 cal mol−1, or −11 kcal mol−1 (roughly −46 kJ mol−1). The fact that ΔG° is negative for the bind- ing reaction means that under standard conditions, binding is favored; at equilibrium, most of the components will exist in a complex. From Equation 2.4 we can also see that relatively small differences in the bind- ing energy will necessarily cause large differences in Kd. For example, the favorable free-energy contribution of a hydrogen bond in a protein is typi- cally 1–1.5 kcal mol−1 (~ 4–6 kJ mol−1), depending on context; thus adding or subtracting a single hydrogen bond to a binding interface between two proteins has the potential to change the Kd by more than tenfold. This fundamental relationship between binding energy and the dissociation constant is the basis of biological specificity, where quite subtle changes in the surface of a protein (changes in amino acid side chains or in overall conformation) can lead to enormous differences in the likelihood of two proteins binding to each other. Another important thermodynamic aspect of binding is that the free energy of binding, like any change in free energy, can be described in terms of two components: the change in the disorder or entropy (S) of the system, and the change in enthalpy (H), or the heat given off or absorbed upon binding. ΔG = ΔH – TΔS Equation 2.5 Figure 2.8 The quantitative definition of specificity. In this example, protein P interacts preferentially with ligand B compared with ligand a. Plots of the change in free energy as the binding reactions progress are shown for binding to a (pink) and B (blue); for both reactions, the free In this equation, ΔH is the change in enthalpy and ΔS is the change in the entropy of the system. Binding is favored (ΔG is negative) when it results in a decrease in enthalpy (heat is given off on binding) or an increase in entropy (the disorder of the system increases). In general, one might expect that entropy would decrease after binding, owing to the decreased disorder of the components. But it is the entropy of the entire system, including the solvent, that must be considered, and the effect of binding on the entropy of the many ordered water molecules that form a shell around the surface of the protein can be quite large. Thus, overall entropy can increase upon binding. In practice, some binding interactions are enthalpically driven, and others are entropically driven. How are the enthalpy and entropy of binding relevant to signal transduc- tion? Understanding the thermodynamic basis of binding is critical for the ability to predict the affinity of a biological interaction. As an example, computer algorithms can simulate the docking of a virtual library of com- pounds onto a protein surface, and predict which are likely to bind with high affinity; this can help to identify those compounds that may ulti- mately be developed into new pharmaceuticals. However, such in silico predictions are still quite challenging. As we have seen, even relatively small errors in calculated binding energies propagate into large errors in Kd. Furthermore, the overall binding energy is the sum of many distinct and counterbalancing changes in enthalpy and entropy, both of the inter- acting proteins and the surrounding solvent, many of which are difficult to quantify precisely on a theoretical basis. The free energy of binding also gives us a way to quantify the specifi- city of interactions. Figure 2.8 shows a free-energy diagram illustrat- ing the specificity of protein P for interaction with two competing ligand molecules, A and B. The difference in free energy between the binding of P to A and P to B is the free energy of specificity. Specificity can also be quantitatively described as the ratio of the dissociation constants for the two ligands (K /K ). However, the specificity described here only reflects energy of the complex is lower than the ^{B A} free energy of the unbound components, indicating that binding is favored at equilibrium. the difference between the free energy before and after binding (ΔG) is the free energy of binding. the free energy of specificity of protein P for the interaction with ligand B versus ligand a is equal to the difference in the free energy of binding of the two ligands under standard conditions. the preference of protein P for binding ligand B over ligand A and does not reflect the preference of protein P for binding ligand B over any other competing ligands. The dissociation constant is also related to rates of binding and dissociation The dissociation constant describes the fraction of components bound to each other at equilibrium. In biological systems, however, binding reactions very seldom achieve equilibrium. In fact, it is changes in protein binding over relatively short periods of time that are usually most important for transmitting information during signaling. How can we describe the rates at which multicomponent systems assemble and disassemble in the cell, and what is the relationship of these rates to the dissociation constant? As described before, for the bimolecular reaction A + B ↔ AB, the rate of formation of AB is given by a rate constant (on-rate or kon) times the con- centration of the binding partners, kon[A][B], while the rate of dissociation is given by another rate constant (off-rate or koff) times the concentration of the complex, koff[AB]. Let us consider each of these in a bit more detail. From the rate term kon[A][B] we see that the actual rate of formation of the complex (in moles per liter per second: M s−1) depends on the con- centration of each of the free components A and B. This makes intuitive sense, as binding can occur only when a molecule of A collides with a molecule of B, and the likelihood of such a collision is proportional to their concentrations. We can think of kon as a combination of the rate of diffu- sion of A and B (which governs the rate of collision) and the likelihood that two molecules will stick to each other once they collide. In this case, kon is a bimolecular rate constant and has units of M−1 s−1. There is a limit to how large kon can be, however: with rare exceptions, the rate of binding cannot be higher than the random rate of collision of A and B. For typical proteins in aqueous solution, the diffusion-limited rate of collision is on the order of 108–109 M−1 s−1. If kon approaches this rate, then a very high percentage of colliding molecules bind to each other, at least momentarily. The actual on-rates for biological binding reactions are usually consider- ably lower than this limit, typically 105–106 M−1 s−1. From the rate term koff[AB] we see that the rate of dissociation (again, in M s−1) is entirely dependent on the concentration of the AB complex. For any individual complex, the off-rate (koff) describes the probability of dis- sociation as a function of time. This is a unimolecular rate constant, with units of s−1. In fact, from koff we can calculate the half-life of the complex, or the amount of time it will take for half of the complex to dissociate (or, for an individual complex, the time at which there is a 50% likelihood it has dissociated). The general first-order rate equation describing such unimolecular reactions is kt = ln[initial amount]/[amount at time t]. When time t is equal to the half-life, then [initial amount]/[amount at time t] is equal to 2; thus under these conditions t = ln2/k, or 0.693/k. Half-life of complex = 0.693/koff Equation 2.6 Thus if koff for a particular reaction is 10−2 s−1 (in the range typically seen for physiological interactions), then the half-life of the complex is 0.693/10−2 seconds, or 69.3 seconds (about a minute). The dissociation constant is the ratio of two rate constants (koff/kon), and so the affinity of a binding reaction is inextricably linked to both the rate of add ligand Figure 2.9 remove ligand formation and the rate of dissociation of the complex. Thus, two different complexes can have the same dissociation constant and thus the same overall affinity, yet have entirely different kinetic behavior as a result of compensating differences in on-rate and off-rate. Consider a situation where both the on-rate and the off-rate are relatively high. In this case, complex formation is rapid even at low levels of the individual compo- nents, but the half-life of those complexes is very short. Compare this with a second reaction with the same dissociation constant, where both the on-rate and the off-rate are relatively low. Now complex formation is quite slow unless the concentrations of components are high, but the complexes are relatively stable once formed (Figure 2.9). Such considerations are Interactions with the same affinity can have different rates of binding and dissociation. the time-course of binding is shown for two interactions with the same dissociation constant (Kd), one with relatively fast on-rate and off-rate (pink line), and one with relatively slow on-rate and off-rate (green line). although the two reactions reach the same level of binding at equilibrium, the rates of binding and dissociation after ligand addition or removal are very different. crucial to the dynamic behavior of signaling mechanisms, as they govern how rapidly existing complexes can be remodeled and how quickly the system can respond to changes. We stated above that in most cases, the on-rate cannot be faster than the diffusion-limited rate of collision. This has profound implications for the half-life of biological complexes. Consider a relatively high-affinity inter- action, for example between a receptor and its ligand (Kd = 10−11 M). Using the definition of the dissociation constant, we can write: Kd = 10−11 = koff/kon. Since kon can be no greater than ~ 108 M−1 s−1, then koff must be less than 10−3 s−1. From Equation 2.6, we can then calculate that the half-life of the receptor–ligand complex must exceed 0.693/10−3 seconds, or 693 seconds (about 12 minutes). If a more typical value for kon is used (106 M−1 s−1), the calculated half-life is nearly a day. Thus a signal generated by binding of the ligand to the receptor will be quite stable and sustained, unless other mechanisms (such as degradation or other modification of the ligand or receptor) are provided to down-regulate it more rapidly. In the case of very high-affinity interactions, such as that between biotin and avidin (Kd = 10−15 M), the half-life of the complex must be many days, making such interactions essentially irreversible. This is consistent with the bio- logical role of avidin, which is found in chicken eggs. By binding and irre- versibly sequestering biotin, an essential microbial nutrient, avidin helps prevent the growth of bacteria in the egg. ProteIn InteractIons In tHeIr cellUlar and MolecUlar context The dissociation constant provides essential information about the intrin- sic properties of a protein–protein interaction and places physical limits on how it can behave in a cell. However, the likelihood that the interaction actually takes place in vivo is highly dependent on context. Taking the most obvious case, a particular interaction may be very strong in vitro, but if the two potential partner proteins are not expressed at the same time in the same cell, or in the same subcellular compartment, then the interaction will not occur. On the other hand, an interaction that appears relatively nonspecific in vitro may actually be more specific in vivo. For example, if a protein has several partners that all bind with the same affinity in vitro, but only one of these is coexpressed with the protein in a particular cell, then the in vivo binding specificity of that interaction will be high. Furthermore, our discussions of affinity so far have involved some assump- tions that may not be valid in biological systems. For example, simple binding equations assume that all components are well mixed and freely diffusing in solution, conditions that may not be met in the cell. We have also assumed that only one molecule of A binds to one molecule of B. In this section, we will explore a few situations where these assumptions are violated in biological systems, and how this can affect binding in the cell. The apparent dissociation constant can be strongly affected by the local cellular environment and other binding partners One way in which the apparent affinity of biological interactions can be significantly increased is through the presence of multiple binding sites on the two interaction partners. This is known as the avidity effect. The most prominent example of avidity, and where the effect was first noted, is in the binding of antibodies to ligands with multiple binding sites (known as polymeric ligands) such as bacterial cell walls, which consist of many identical subunits. Antibodies are dimeric, Y-shaped molecules possessing off-rate low (b) off-rate high (c) off-rate high two identical binding sites for an antigen (loosely defined as a molecule recognized by an antibody). Thus when an antibody binds an antigen that is present in many copies on a surface, both of its antigen-binding sites are likely to be bound to the surface simultaneously. The fact that the antibody is held onto the surface by two points of contact instead of one makes dis- sociation much less likely, because this would require that both sites disso- ciate at the same time. If one of the binding sites momentarily dissociates, the other is likely to remain bound, and since the concentration of the anti- gen on the surface is very high, the first site will rebind rapidly (remember, the rate of association is directly proportional to the concentration of the binding partner) (Figure 2.10a). The effective off-rate is therefore much slower than the off-rate for an individual binding site, and the apparent dissociation constant is correspondingly much lower. In the case of anti- bodies, the off-rate can be essentially zero, and binding is thus irreversible. Note that avidity depends on there being multiple binding sites on each partner. Thus an intact antibody (with two antigen-binding sites) binds to a polymeric ligand such as a bacterium much better than does a proteolyt- ic fragment of the antibody containing only a single binding site, termed an Fab fragment (Figure 2.10b). However, when an antibody recognizes a soluble, monomeric antigen instead of a surface with multiple binding sites, there is no avidity effect and an Fab would bind just as well as the intact immunoglobulin (Figure 2.10c). The above example illustrates how the effective concentration of a lig- and in the immediate vicinity of its binding partner, or its local concen- tration, influences binding. The local concentration of a ligand may be very different from its overall (or average) concentration in the cell. For example, in the case of an intramolecular binding interaction (between two parts of the same protein), the local concentration of the reactants is extremely high and thus binding can be favored even when the isolated interaction (intermolecular) is relatively weak. Local concentration also has an important role in facilitating binding to structures with a high density of binding sites, such as membranes or other surfaces. Under such conditions, it may be quite difficult for a molecule bound to such a surface to escape into solution; if it should dissociate, the very high local concen- tration of binding sites on the surface makes it likely that it will rebind before it can diffuse away (Figure 2.11). Ideal affinity and specificity depends on biological function and ligand concentrations Most binding interactions in signaling might be assumed to have high affinity and specificity, to ensure efficient and unambiguous transfer of information. However, this is not the case—binding interactions are observed that span a wide range of both affinities and specificities. What affinity and specificity would be most appropriate for a given interaction depends strongly on the function of the interaction and the endogenous concentrations of the partners. For example, many complexes are preas- sembled and need to be long-lived for their function; in such complexes, Figure 2.10 Avidity of antibody binding. (a) antibody molecules (blue) have two identical antigen-binding sites. When an antibody binds an antigen present in many copies on a surface, such as a bacterial cell wall, both antigen-binding sites are bound simultaneously. If one of the sites dissociates, the antibody is still held by the second site and the first site rapidly rebinds, and the off-rate is low. (b) If an antibody fragment containing a single antigen-binding site (an fab fragment) is bound to the surface, the off-rate is relatively high. (c) If the antibody binds the same antigen in a soluble, monomeric form, the off-rate is also relatively high. (b) Figure 2.11 Effect of surface density on binding. When there is a high local concentration of a binding site on a surface, a ligand dissociating from one site is likely to rebind immediately, and so dissociation of the ligand from the surface is a rare event. If the density of binding sites is low, however, rebinding is much less likely, and dissociation is favored. the affinities of the partners are likely to be high. In particular, it is important for the dissociation constant to be lower than the endogenous concentrations of the components so that binding will be close to satura- tion. However, there is a limit to how high an affinity is required—once below the endogenous ligand concentration, decreasing the dissociation constant by many orders of magnitude would provide little improvement in the degree of binding, and might be less likely to evolve. By contrast, if an interaction is dynamically regulated, as are many sig- naling interactions, then it is important that the dissociation constant be roughly equal to or slightly higher than the endogenous ligand concentra- tion. This weaker affinity (higher Kd) is important for two reasons. First, it allows small changes in either ligand concentration or other modulat- ing factors (for example, local concentration or allosteric changes) to lead to large changes in the fraction of ligand bound—that is, it allows the interaction to be regulated. Second, weaker affinities allow the interac- tion to be potentially more dynamic. As described above, the dissociation constant (Kd) of an interaction is defined by its off-rate (koff) divided by its on-rate (kon). Since the on-rate is essentially limited to the rate of diffu- sion (≈ 108–109 M−1 s−1), then the off-rate is limited by the affinity. A higher off-rate (shorter half-life) may be required for interactions that must be rapidly disassembled as part of their biological function. There are functional constraints on interaction affinities and specificities Because endogenous protein concentrations are so closely tied to binding affinities, it is useful to survey the range of individual protein concentra- tions observed in biological signaling systems. Overall, the concentrations of intracellular signaling proteins tend to be significantly higher than those of extracellular signaling proteins, such as hormones. As we shall see below, such concentrations constrain the range of observed affinities and specificities. Signaling interactions can be divided into broad classes on the basis of affinity and specificity (Figure 2.12). Interactions that have both high affinity and high specificity include hormone–receptor interactions; for example, human growth hormone (hGH) and its receptor interact with Kd = 3 × 10−10 M. Such receptors must recognize low concentrations of hormone in the bloodstream, and have affinities that allow binding in the range of the signaling concentrations of their ligands. For exam- ple, under basal conditions the concentration of hGH is <10−10 M, while during activation the concentrations reach ~ 10−9−10−8 M. Thus, the dis- sociation constant is closely matched with the change in concentration that must be physiologically detected—resulting in a significant signal- dependent change in the fraction of bound receptor. These types of inter- actions must also be specific enough to prevent cross-talk between related hormones and receptor sets. Interactions that have low affinity (here arbitrarily defined as interactions with Kd >10−7 M) but high specificity include the interactions of modular peptide-recognition domains, such as SH3 domains, with their partners. These interactions function in the intracellular environment, where protein concentrations are much higher. Low affinities are probably necessary to allow for subtle, dynamic regula- tion of transient signaling complex assembly and disassembly in response to changing input. High specificity, however, is still essential to achieve proper flow of information. An extreme example of an interaction that has high affinity but relatively low specificity is the interaction of major histocompatibility complex (MHC) (a) high low high 10–12 10–10 affinity low 10–6 10–4 mole/liter Figure 2.12 Range of biological affinities and concentrations. (a) the general range of affinities and specificities observed for a particular class of interactions. (b) the endogenous concentrations of these classes of proteins. overall, the Kd values of signaling interactions are matched to endogenous concentrations so that the extent of binding can be easily regulated by small environmental changes. hGH, (b) high dissociation constant (Kd) human growth hormone; MHc, major histocompatibility complex molecule; hsp70/clients, chaperone binding to newly synthesized client proteins during folding. low 10–12 10–10 10–6 10–4 mole/liter physiological concentrations molecules with peptide antigens. MHC molecules bind peptides derived from proteins degraded inside cells, and carry them to the cell surface where they can be surveyed by lymphocytes of the immune system. In this way, protein components of intracellular infectious agents such as viruses, or from bacteria that have been engulfed by phagocytes, can be detected and elicit an immune response. To ensure detection of microbial intruders, each MHC molecule must be able to bind a wide range of peptides, and this is achieved by combined recognition of general features of the peptide back- bone with recognition of specific amino acid side chains that vary between peptides recognized by any given MHC molecule. Despite the resulting promiscuity of this interaction, binding is quite tight—estimated Kd val- ues for such complexes are between 10−6 and 10−9 M. These unusual bind- ing properties are observed because the MHC molecule folds around the peptide as an integrated unit. The intimate engagement of MHC–peptide leads to a very slow off-rate—approximately 10−6 s−1, equivalent to a half- life of more than 100 hours. This extremely slow off-rate helps to ensure that the peptide–MHC complex remains at the cell surface long enough for sustained interaction with the antigen receptors of lymphocytes, a require- ment for immune activation. In contrast, the housekeeping role of chaperone proteins, which bind transiently to hydrophobic regions on newly synthesized proteins during folding to prevent their misaggregation, requires interactions that have low affinity and low specificity. Chaperone interactions must be nonspe- cific enough to allow binding to a wide range of ligands, and allow rapid disassociation. Chaperones like hsp70 bind a wide variety of hydropho- bic peptides with Kd values ranging from 10−7 to 10−5 M when bound to ADP. When ATP is bound, the affinity of hsp70 for the peptide is further reduced by between 10- and 100-fold. This enables ATP hydrolysis and exchange of ADP for ATP to drive cycles of peptide binding and release. Chaperones are present at high concentrations in the cell, ensuring that binding is favored despite their relatively low affinity for their ligands. Interaction affinity and specificity can be independently modulated One might intuitively assume that specificity and affinity are well cor- related: the tighter the binding of a protein to its correct partner, the less likely it is to cross-react with others. As suggested above, however, this is not always the case. In fact, affinity and specificity can be tuned independently via several different mechanisms. We will consider here the interaction of a protein with a series of related competing ligands. If the interaction of the protein with all its ligands is initially both rela- tively weak and nonspecific, there are, in principle, several ways in which improved affinity and specificity for one of the ligands can evolve. In a process of positive discrimination, the affinity of the protein for one of its ligands can be increased by the formation of additional favora- ble stereochemical interactions. Whether these new interactions increase specificity, however, depends on their nature. Favorable interactions that recognize chemical features common to all members of a compet- ing ligand family will increase the affinity, but not the specificity, for the given ligand. For example, many relatively nonspecific peptide-binding proteins, such as the MHC molecules described previously, recognize their peptide ligands partly through extensive hydrogen-bonding to the peptide backbone—which, unlike bonding to side chains, is not specific to a particular sequence of amino acids. Positive discrimination will gener- ally increase specificity only if the increase in affinity involves the forma- tion of specific side-chain interactions—elements that are unique to the correct ligand compared with its related competitors (Figure 2.13a–c). (a) no specificity receptor ligands A B Figure 2.13 Increasing specificity through positive and negative discrimination. Binding of a receptor to two possible ligands, a and B, is initially weak and nonspecific. (b) Modification of the receptor to increase complementarity to both ligands results in higher affinity but no increase in specificity. (c) Modification of the receptor to increase complementarity only to ligand B results in an increase in specificity for ligand B versus a. (d) Modification of the receptor to decrease complementarity to ligand a results in reduced affinity for ligand a and thus an increase in specificity for B versus a. (e) the receptor could also be altered to simultaneously decrease complementarity to ligand a and increase complementarity to ligand B, resulting in an even greater increase in specificity. free-energy diagrams on the right show how the changes alter the difference in free energy of binding between the two ligands (ΔGspecificity). (b) increased affinity no specificity (c) positive discrimination for B (d) negative discrimination against A (e) positive and negative discrimination Structural studies have shown that for most proteins that recognize spe- cific peptide ligands, the binding interface consists of a combination of specificity-determining contacts to specific side chains of the target, and contacts to generic features that are common to all peptides. Specificity can also be increased via a process of negative discrimination. In this case, affinity for related competitor ligands is decreased while the affinity for one ligand remains unchanged (Figure 2.13d–e). Nega- tive discrimination often involves the presence of interactions that would be unfavorable if the protein were to form a complex with the incorrect competing ligand. Where many closely related protein interactions occur in the same cell, such as those involving families of interaction domains, there is evidence that negative discrimination prevents incorrect cross- interactions in some cases. Negative discrimination can also be observed in the specific binding of SH3 domains to proline-containing peptide motifs (Figure 2.14). Pock- ets on the SH3 domain recognize proline via the imino acid backbone group. This interaction is not highly optimized, but is highly discrimina- tory against peptides lacking an imino group at these positions. Because proline is the only one of the 20 natural amino acids with an imino back- bone group (all others have an amide backbone group), the selection for proline by this binding site is absolute. This strong negative selectivity would not be observed if there were other imino acids among the constit- uents of natural proteins, illustrating how the effectiveness of negative discrimination as a mechanism for specificity is highly dependent on the nature and range of potentially competing ligands present in the cell. In most known interactions, affinity and specificity are tuned by both positive and negative discrimination. For example, charge–charge (salt bridge) interactions are observed at the interfaces in many protein– protein complexes. These not only provide a favorable energy for binding when they are complementary within the correct complex, but they often also repel related ligands that lack charge at the appropriate position, thus preventing the formation of incorrect complexes. The negative discrimination at such surfaces is strong because it is highly energetically unfavorable to bury a charged residue at an interface without a compen- sating interaction with an oppositely charged residue. Cooperativity involves the coupled binding of multiple ligands Many macromolecular complexes in cells are composed of multiple interact- ing partners, and the higher-order interplay between these partners can yield interesting and functionally important binding behaviors. One of these higher-order effects is cooperative binding, which is yet another mechanism by which the affinity and specificity of an interaction can be increased. Cooperativity refers to an interaction in which binding of one ligand enhances the binding of an additional ligand(s). In thermodynamic terms, cooperativity is observed when the free energy of two ligands binding simultaneously is different from the sum of the free energies of the two ligands binding individually. If the binding of one ligand increases the affinity for an additional ligand, then positive cooperativity is said to have occurred. Here we will focus only on positive cooperativity, although nega- tive cooperativity—when the binding of one ligand decreases the affinity for an additional ligand—does occur. An important consequence of positive cooperativity is that assembly of a complex occurs in more of an all-or-none fashion: the multiple ligands cooperate to assemble the complete complex, and intermediate states of assembly are poorly populated (Figure 2.15). (b) proline other 19 amino acids tolerated unfavorable Figure 2.14 Negative discrimination enables SH3 domains to specifically recognize proline. the binding pocket of sH3 domains tolerates an imino group (a), but poorly recognizes the standard amide group found in most peptide linkages (b). Because proline is the only amino acid with an imino backbone group, this is the only one of the 20 natural amino acids tolerated at this site and thus the binding of proline by the sH3 domain is highly specific. poorly populated intermediate states Figure 2.15 Positive cooperativity leads to all-or-none assembly. an example of a three-component assembly in which binding is cooperative. the binding of any pair of the three components shown on the left facilitates the binding of a third component. thus, intermediate two-component states are rare and the components are said to cooperate to favor assembly of the complete complex. Diverse molecular mechanisms underlie cooperativity Cooperativity can be produced by many different mechanisms, which differ in the source of the additional interaction energy (Figure 2.16). For example, if two ligands directly interact with one another when they are bound to the same receptor protein, they will enhance each other’s binding in a cooperative manner—the two binding events will be dependent, with the second binding event being more favorable than the first. Here, the source of the energy for cooperativity is the ligand– ligand interaction. Cooperativity is also observed if binding of one ligand leads to a conformational change in the receptor protein that in turn increases its affinity for a second ligand; for example, it alters the conformation of the second binding site. In this case, the extra energy of cooperativity comes from the reorganization of the receptor protein structure. Cooperativity can also occur when two interaction domains in a protein bind to two distinct sites on the same ligand. When one domain binds to the ligand, the increased effective concentration of the second binding site further enhances its binding; here, the covalent tethering of the two bind- ing sites reduces the entropic cost of the second binding event (avidity, described above, is an example of this type of cooperativity). A common example of cooperativity in intracellular signaling is when one protein is recruited to the membrane by a protein–lipid interaction, thereby increasing its local concentration (see Chapter 5, Figure 5.10). This coop- eratively increases its effective affinity for another membrane-localized partner. In addition, cooperativity can be observed if two binding sites on the protein are occluded by a single inhibitory element—binding of one ligand displaces the inhibitory element, thus apparently increasing affinity for the second ligand. Cooperative binding has a variety of functional consequences One outcome of cooperativity can be to increase the biological specifi- city of an interaction, restricting it to a particular time and place. For proteins participating in multiple cooperative interactions, for example, (b) (c) ligand–ligand interaction Figure 2.16 ligand-induced conformational change tethering/effective concentration Diverse mechanisms for cooperative binding of two ligands to one receptor. (a) the binding of the second ligand to the receptor is enhanced by ligand–ligand interactions. (b) conformational changes in the receptor induced by the binding of the first ligand increase the affinity for the second ligand. (c) If ligands are tethered, then binding of one ligand by the first binding site increases the local concentration at the second binding site, enhancing binding. the likelihood of interaction depends on the concentration of a specific set of ligands, rather than of a single ligand. The correct combination of ligands may only be present at a specific time or at a specific location in a cell. Cooperativity between multiple interaction domains within a protein often serves to increase specificity. For example, the intracellular pro- tein kinase ZAP-70 in the T lymphocytes of the immune system has two phosphotyrosine-binding SH2 domains (Figure 2.17). On its own, each domain has only modest specificity. When linked in the intact protein, however, they cooperate to recognize particular tandem arrangements of phosphotyrosine motifs known as immunoreceptor tyrosine-based activat- ing motifs (ITAMs), which are present in the cytoplasmic tails of impor- tant immune-system receptor proteins. When positive cooperativity involves multiple ligands of the same spe- cies, it is known as homotypic cooperativity; cooperativity involving two or more different ligands is referred to as heterotypic cooperativity. As well as making binding interactions more specific, homotypic cooperativ- ity can make them act more like a binary switch, which is either in the “on” or “off” state. In such cases, the binding curve is no longer a simple hyperbola, but assumes a sigmoidal shape (Figure 2.18). Within a nar- row critical range of ligand concentration, the protein shifts from a state in which almost all sites are unbound to one where almost all sites are bound; thus, at this threshold range of concentration, very small changes in concentration can have disproportionately large effects on fractional occupancy. By contrast, for single-site (noncooperative) binding, a greater than 80-fold increase in ligand concentration is required to increase the fractional occupancy from 10% to 90%. Because binding is often linked to protein activity, many signaling pathways use cooperativity to sculpt the response of the system so that activation of a protein only occurs at a precise concentration threshold of a ligand. Protein assemblies differ in their stability and homogeneity There is wide variation in the size, complexity, and uniformity of multipro- tein complexes involved in signaling. At one extreme of this spectrum are very stable structures that have a precise and invariant molecular archi- tecture and stoichiometry (that is, the numbers of the different types of subunits). Examples include ribosomes, proteasomes, and nuclear pores. The assembly of such complexes is often highly cooperative, and partially assembled structures are unstable. Such macromolecular machines execute important functions (such as making, destroying, or transport- ing proteins) in the cellular response to a signal, but their immutability makes them poorly suited for the business of sensing, integrating, and transmitting signals that are rapidly changing over time. The role of ZAP-70 in T cell signaling is discussed further in Chapter 12 individual domains weak affinity nonspecific linked domains high affinity and specificity for tandem phosphorylated motifs cooperativity: effective concentration and domain–domain interactions Figure 2.17 Cooperative recognition of a tandem phosphotyrosine motif by coupled SH2 domains. (a) Individual sH2 domains show weak affinity and specificity for a single phosphotyrosine motif. (b) In the protein kinase ZaP-70, linked sH2 domains show high affinity and specificity for phosphotyrosine motifs linked in tandem. this is the result of cooperativity due to interactions between the sH2 domains and also to ligand tethering and increased local concentration. Figure 2.18 The effect of positive cooperativity on binding curves. Binding of one molecule of ligand B (orange circle) to one subunit of protein a (brown) increases the affinity of the other subunits for ligand B. the p i n k solid line shows the sigmoidal binding curve characteristic of such positive homotypic cooperativity. note that within a critical range of concentration (shaded area), relatively small differences in the concentration of B have large effects on binding. for comparison, the b l u e dotted line shows a simple binding curve with no cooperativity. 1 0.5 concentration of B The protein complexes that perform more dynamic roles in signaling are often much less stable and homogeneous. Interactions among the compo- nents are likely to be relatively weak, and there can be many possible binding partners for a given binding site. Thus, many different possible combinations of interactions are possible. An important property of such structures is that their interactions are likely to change and reorganize relatively quickly. A good illustration is provided by the complexes induced by the stimulation of receptors coupled to tyrosine kinases (see Figure 4.11 in Chapter 4). Activation of such receptors induces the phosphorylation of many different sites on the receptor itself, which then serve as docking sites to recruit SH2-domain-containing proteins. Receptor phosphoryla- tion is relatively slow and inefficient, however, and each site is subject to dephosphorylation by phosphatases. Thus, individual receptor mol- ecules are likely to bear different and constantly changing combinations of phosphorylated sites. Furthermore, each site may bind to several dif- ferent SH2-containing proteins with similar affinity, so which one, if any, binds will depend on chance and the local availability of binding partners. This situation leads to a combinatorial explosion of possible states for the receptor; millions of distinct receptor complexes can be described for even a relatively small number of phosphorylation sites and binding partners. Such complexes, which can be termed dynamic molecular assemblies, differ greatly from more stable macromolecular machines such as the ribosome. This is an important concept to keep in mind when thinking about signal- ing pathways: although we often loosely refer to the signaling machinery, this analogy should not be taken too literally. The actual complexes that transmit signals can be difficult (if not impossible) to define precisely, and the contributions of a single interaction among many possible interactions can be difficult to tease out. What advantages might there be to signaling mechanisms based on such dynamic molecular assemblies? It may be that because such complexes exist in a wide range of possible states, depending on the specific combinations of binding partners, they are better able to respond to and transmit complex, finely graded signals. The relative insta- bility of such assemblies also allows them to be highly sensitive to rapid and subtle changes in the environment. And since they function in a combinato- rial fashion, a relatively limited number of components can participate in an almost infinite variety of pathways having different inputs and outputs. sUMMary Protein–protein interactions play a part in virtually all signaling events in the cell. Changes in these interactions, in turn, drive many other changes of importance for signal transmission, as we will see in coming chapters. The dissociation constant (Kd) of an interaction is an intrinsic, quanti- tative measure of the likelihood that the interaction will occur, whereas the specificity of an interaction is determined by the relative affinities of competing interactions. The affinity and specificity of physiological bind- ing interactions varies widely, but is tuned to the specific circumstances so that information can be processed reliably. QUestIons What is the concentration of a protein if it is present at one molecule per cell? Assume the volume of the cell is ~10–12 liter. If a cell is ~25% protein (mass/vol), then what is the approximate total number of protein molecules in a cell? Assume an average protein molecular mass of 100 kD. Questions You measure that there are approximately 10,000 copies of protein X in the cell. Again, assuming that the volume of a mammalian cell is ~10–12 liter, what is the approximate concentration of this protein when distributed throughout the whole cell? What happens to the concentration if all of protein X is translocated to the nucleus (use an estimated nuclear volume of ~10–13 liter)? Protein X described in Question 3 has a physiological binding partner Y in the nucleus that is present at a concentration of ~10–10 M. You observe a significant interaction between molecules X and Y (>50% of Y is bound to X) only after all of molecule X is concentrated in the nucleus (which we estimate is one-tenth the volume of the entire cell). Can you estimate the Kd for the interaction of X and Y? The observed extent of an interaction between two molecules in a liv- ing cell is often different from what might be expected from the simple Kd of the interaction measured in vitro. What factors might lead to a higher fraction of binding than expected? What factors might lead to a lower fraction of binding than expected? Proteins X and Y interact with each other, both in vivo and in vitro. You identify a disease-causing mutation in protein X that leads to dis- ruption of the X–Y interaction in vivo. Surprisingly, when you purify proteins X and Y, the in vitro binding affinity is unchanged by the mutation. What is a simple hypothesis for how the mutation in X might change the extent of binding to Y in vivo? A protein domain recognizes a peptide with the following sequence profile: RxxF/L/VxF (the residues F, L, and V at position 4 are recog- nized with equal affinity to each other; x denotes any amino acid is tolerated at that position). A homologous protein domain recognizes peptides with the profile: RxxVxF. Which domain has higher spe- cificity? Which domain will bind with higher affinity? Discuss your answers in terms of how positive or negative discrimination could be involved in these two interactions and how these mechanisms might correlate with affinities. Glutathione S-transferase (GST) has often been used as a “tag” for recombinant proteins, allowing the GST-tagged proteins to be purified away from other proteins using small beads coated with a high surface density of glutathione. Table 2–1 on page 27 shows that the affinity of the GST–glutathione interaction is relatively low (Kd ~ 10–4 M). What are the advantages and disadvantages of the GST–glutathione system as a purification tag, compared to one with much higher affinity, such as biotin–avidin? GST-tagged proteins can often be purified efficiently even when working at concentrations lower than 10–4 M, at which one might not expect to observe a high fraction of binding. What features of the GST interaction with glutathione-coated beads might account for this? The emergence of new interactions between signaling proteins is thought to be a key driver of the evolution of new signaling pathways and behaviors. Describe different mechanisms whereby a new pro- tein–protein interaction could evolve. Most protein–protein interactions involved in signaling are dynamic (that is, the kinetics of the interaction—on-rates and off-rates—are fast). Why might this be important? What other types of function might require protein interactions that are less dynamic? How are the dynamics of protein interactions related to their thermodynamic affinities? A number of proteomic databases are now available that enumerate all known protein interactions in a given system. For example, over 20,000 human protein–protein interactions have been identified in a recent dataset. Generally, these datasets provide little if any informa- tion about affinity. What are the limitations of such databases? Are there conditions where a predicted interaction might occur only rarely in a cell? When might a biologically important interaction not be pre- dicted by such databases? Scaffold proteins bind to multiple binding partners simultaneously. You discover a new scaffold protein that binds both molecules A and B, with no independent interaction between A and B. Signal transmission depends on the concentration of the trimolecular complex (scaffold plus A and B). How will signal output (proportional to the amount of the trimolecular complex) depend on the concentration of the scaffold pro- tein? What are the conditions under which signal output will be maxi- mal? How would you expect signal output to be affected if the scaffold is experimentally overexpressed to a level much higher than normal? references ProPertIes of ProteIn–ProteIn InteractIons Ajay & Murcko MA (1995) Computational methods to predict binding free energy in ligand-receptor complex- es. J. Med. Chem. 38, 4953–4967. Hammes GG (2000) Thermodynamics and Kinetics for the Biological Sciences. New York: John Wiley & Sons. Harrison SC (1996) Peptide–surface association: the case of PDZ and PTB domains. Cell 86, 341–343. Jones S & Thornton JM (1996) Principles of protein– protein interactions. Proc. Natl Acad. Sci. U.S.A. 93, 13–20. Kastritis PL & Bonvin AM (2012) On the binding affin- ity of macromolecular interactions: daring to ask why proteins interact. J. R. Soc. Interface 10, 20120835. Lo Conte L, Chothia C & Janin J (1999) The atomic structure of protein–protein recognition sites. J. Mol. Biol. 285, 2177–2198. Moreira IS, Fernandes PA & Ramos MJ (2007) Hot spots—a review of the protein-protein interface determi- nant amino-acid residues. Proteins 68, 803–812. Neduva V & Russell RB (2005) Linear motifs: evolution- ary interaction switches. FEBS Lett. 579:3342–3345. Nooren IM & Thornton JM (2003) Diversity of protein– protein interactions. EMBO J. 22, 3486–3492. Wells JA (1996) Binding in the growth hormone receptor complex. Proc. Natl Acad. Sci. U.S.A. 93, 1–6. Winzor DJ & Sawyer WH (1995) Quantitative Character- ization of Ligand Binding. New York: John Wiley & Sons. Wyman J & Gill SJ (1990) Binding and Linkage: Func- tional Chemistry of Biological Macromolecules. Mill Val- ley, CA: University Science Books. ProteIn InteractIons In tHeIr cellUlar and MolecUlar context Bhattacharyya RP, Reményi A, Yeh BJ & Lim WA (2006) Domains, motifs, and scaffolds: the role of modular inter- actions in the evolution and wiring of cell signaling cir- cuits. Annu. Rev. Biochem. 75, 655–680. Bukau B, Weissman J & Horwich A (2006) Molecular chaperones and protein quality control. Cell 125, 443–451. Capra EJ, Perchuk BS, Skerker JM & Laub MT (2012) Adaptive mutations that prevent crosstalk enable the expansion of paralogous signaling protein families. Cell 150, 222–232. Edwards LJ & Evavold BD (2011) T cell recognition of weak ligands: roles of signaling, receptor number, and affinity. Immunol. Res. 50, 39–48. Gibson TJ (2009) Cell regulation: determined to signal discrete cooperation. Trends Biochem. Sci. 34, 471–482. Hatada MH, Lu X, Laird ER et al. (1995) Molecular basis for interaction of the protein tyrosine kinase ZAP-70 with the T-cell receptor. Nature 377, 32–38. Mayer BJ, Blinov ML & Loew LM (2009) Molecular machines or pleiomorphic ensembles: signaling com- plexes revisited. J. Biol. 8, 81. Müller KM, Arndt KM & Plückthun A (1998) Model and simulation of multivalent binding to fixed ligands. Anal. Biochem. 261, 149–158. Nguyen JT, Turck CW, Cohen FE et al. (1998) Exploit- ing the basis of proline recognition by SH3 and WW references domains: design of N-substituted inhibitors. Science 282, 2088–2092. Pawson T & Nash P (2003) Assembly of cell regulatory systems through protein interaction domains. Science 300, 445–452. Sadegh-Nasseri S, Stern LJ, Wiley DC & Germain RN (1994) MHC class II function preserved by low-affinity peptide interactions preceding stable binding. Nature 370, 647–650. Szwajkajzer D & Carey J (1997) Molecular and biologi- cal constraints on ligand-binding affinity and specificity. Biopolymers 44, 181–198. Zarrinpar A, Park SH & Lim WA (2003) Optimization of specificity in a cellular protein interaction network by negative selection. Nature 426, 676–680. Taylor & Francis Taylor & Francis Group https://taylorandfrancis.com Signaling Enzymes and Their Allosteric Regulation Many of the key steps in cellular signal transduction are controlled by specific chemical reactions, ranging from covalent modifications like phos- phorylation to the generation of small diffusible mediators such as cAMP. In most cases, these chemical reactions are intrinsically slow, and thus specific enzymes are required for them to occur at reasonable rates. Cells use enzymes like protein kinases to catalyze phosphorylation reactions, and enzymes like adenylyl cyclase to catalyze the formation of cAMP from ATP. These enzymes are required for signaling processes to occur at the time scales (fractions of a second to minutes) necessary for cellular sens- ing and responses. In this chapter, we will discuss the basic mechanism of signaling enzymes, focusing on a few canonical examples: kinases and phosphatases, which control protein phosphorylation, and G proteins and the enzymes that regulate them. These represent two of the most important enzymatic control systems used throughout eukaryotic signaling pathways. Phos- phorylation is the protein modification most frequently used in signal- ing, while G proteins use conformational changes to transmit signaling information. In this chapter, we will review the detailed chemical mecha- nisms by which these common classes of enzymes catalyze their reactions. Other classes of signaling enzymes are discussed throughout the book, including those that create or destroy other types of post-translational modifications (Chapter 4), those that produce or destroy second messen- gers such as cAMP (Chapter 6), lipid-modifying enzymes (Chapter 7), and proteases (Chapter 9), though their detailed molecular mechanisms will not be covered. When discussing signaling enzymes, it is impossible to separate their cat- alytic mechanisms from their mechanisms of regulation. The central role of signaling enzymes is, after all, to relay information. Thus, their abso- lute catalytic efficiency is often less important than the capacity of their (a) INPUT kinase (b) INPUT GEF/GAP G protein conformational change catalytic activity phosphorylation catalytic activity conformational change effector binding Figure 3.1 Signaling proteins convert conformational changes into catalytic activity changes, and vice versa. (a) Kinases undergo conformational changes in response to diverse inputs, which in turn regulate the kinase catalytic function (phosphorylation). (b) G proteins are controlled by opposing enzymes— guanine nucleotide exchange factors (GEfs) and GTPase-activator proteins (GAPs)— which regulate the conformation of the G protein and, in turn, its interactions with downstream effector molecules. catalytic activity to be differentially regulated in response to inputs such as ligand binding or covalent modification (unlike, for example, a meta- bolic enzyme whose function might be to produce the maximal amount of a chemical product). A hallmark of signaling proteins, such as kinases and G proteins, is their ability to convert conformational changes into catalytic activity changes, or vice versa (Figure 3.1). The mechanisms by which these canonical classes of signaling proteins can couple conforma- tion change with catalytic activity are a major focus of this chapter. PRinciPlES of EnzymE cATAlySiS Enzymes are biological macromolecules that catalyze chemical reactions in living systems. They are remarkable for their ability to promote specific chemical reactions. They can greatly increase the rates of reactions that occur spontaneously but slowly. In addition, they can promote unfavorable reactions that would not occur spontaneously by coupling them to ener- getically more favorable reactions (such as ATP or GTP hydrolysis). Enzymes have a number of properties that make them useful for transmitting signals in the cell Signaling enzymes provide a flexible way to relay cellular information— they are capable of both receiving inputs and transmitting outputs. The output of a signaling enzyme is to catalyze a functional change in down- stream targets. A substrate that is phosphorylated by a kinase enzyme might undergo a change in conformation and activity, while a small-mole- cule mediator like cAMP that is produced by the enzyme adenylyl cyclase might change the function of a number of downstream effectors. Thus, the output from the enzyme is information that is stored and passed down- stream via the states of the downstream targets of the enzyme. In turn, the activity of these enzymes themselves is often subject to regu- lation by upstream inputs. Signaling enzymes often show low levels of activity under basal conditions, but large increases in activity upon regu- latory changes such as ligand binding (conversely, active enzymes can be inhibited by such inputs). Because signaling proteins act as relay switch- es, they are often selected not for optimum catalytic efficiency, but rather for the ability to be regulated. At the molecular level, many of these individual signal-transducing events involve some sort of change in the conformation in the enzyme (broadly speaking, a change in its three-dimensional arrangement or shape), which, in turn, is coupled to a change in its activity. Conformation- al changes that are induced by upstream inputs such as ligand binding or post-translational modification, and which result in changes in the pro- tein activity, are referred to as allosteric changes. This conformational PrinciPles of enzyme catalysis coupling between an upstream input and a change in protein activity provides a basic mechanism for the protein to propagate signals. In this chapter, we discuss the principles that drive allosteric changes, and show examples of how these changes can be used to propagate signals. Another common feature of signaling enzymes is that they often occur as complementary pairs. For example, a specific phosphorylation reac- tion might be catalyzed by a protein kinase, while the reversal of this modification—the removal of the phosphate group—is catalyzed by a pro- tein phosphatase. The kinase would essentially act as a “writer” element substrate “WRITER” kinase “ERASER” that drives the substrate to a new state, while the phosphatase acts as an “eraser” that returns the substrate to the original state (Figure 3.2). All information-storage and -transmission systems, be they natural cellular systems or human-made electronic ones, require some sort of mechanism like this to write and erase information in a controllable fashion. Enzymes have another useful property for transmitting information: the potential to amplify a signal. Because they are catalysts, enzymes remain unchanged while promoting their target reaction, and thus they are capa- ble of carrying out many rounds of the reaction. This catalytic behavior can result in signal amplification: a single activated enzyme molecule can, under the right conditions, generate many molecules of product. Enzymes use a variety of mechanisms to enhance the rate of chemical reactions Enzymes can typically increase the rate of a specific reaction by several orders of magnitude relative to the uncatalyzed reaction in water. In some cases, including the phosphoryl transfer reactions discussed below, rate enhancements can be as large as 1020-fold or more. Here, we will consider the ways that enzymes can achieve such remarkable rate enhancement and specificity. In any reaction, the reactants and the products have defined ground- state energies, and the conversion of reactants to products requires the molecules to pass over a transition state of much higher energy. Enzymes catalyze reactions by stabilizing the transition state relative to the ground state of the reactants (Figure 3.3). Lowering the transition-state energy reduces the free-energy barrier for the reaction and increases the prob- ability that reactants will pass over the barrier and be converted into the products. While enzymes can alter the relative free-energy difference between ground states and transition states, the thermodynamic equilib- rium between free reactants and products is unchanged. In other words, at infinitely long times, all chemical reactions will reach an equilibrium based on the free energies of the reactants and products; this equilib- rium is not affected by the presence of an enzyme. What is affected is how quickly this equilibrium is approached. As mentioned above, enzymes can accelerate reaction rates by many orders of magnitude. phosphatase Figure 3.2 Signaling enzymes act as “writers” and “erasers” of information- carrying marks. Phosphorylation of a protein acts as a mark that can control its activity. Kinases catalyze the writing of this mark (phosphorylation reaction), while opposing enzymes—phosphatases— catalyze the erasure of this mark (dephosphorylation). many other cellular signal-processing systems are also jointly controlled by analogous opposing regulatory enzymes. Figure 3.3 Enzymes catalyze chemical reactions by lowering the free energy of the transition-state barrier. free-energy diagram illustrating the effect of an enzyme on reaction energetics. An uncatalyzed reaction (blue dotted line) is slow because the conversion of reactants to products must normally pass through a high- energy transition state. An enzyme does not change the ground-state energies of the reactants and products, but increases the rate of conversion by reducing the free- energy barrier to the reaction, sometimes introducing a new reaction intermediate, as shown here by the pink line. This and other mechanisms by which enzymes alter the reaction energetics are discussed in the text. reaction coordinate There are several general mechanisms that enzymes can use to stabilize the transition state of a chemical reaction, all of which are observed in various signaling enzymes. First, enzymes can lower the transition-state energy by binding and orienting key reactive groups within a substrate (or two different substrates) such that they react with one another in a more favorable fashion. Second, enzymes can provide general acids and general bases to donate or accept protons that are transferred to and from the substrate during the reaction. The use of general acids, bases, and metal ions to activate attacking groups and stabilize charge devel- opment can be particularly important for nucleophilic displacement reactions (such as phosphoryl transfer reactions) that require activation of an attacking nucleophile and stabilization of a leaving group. Third, enzymes can provide a binding site that is more complementary to the electrostatic or geometric properties of the transition state than the ground state; in this case, some of the binding energy of the transition state is used to reduce the free-energy barrier for the reaction. This type of electrostatic or geometric catalysis can be accomplished by the precise positioning of active-site functional groups and cofactors such as metal ions. Finally, in some cases, enzymes alter the path or mechanism of a reaction, causing the reaction to proceed through a reaction intermediate that does not normally occur in the uncatalyzed reaction (see Figure 3.3). If the free- energy barriers are lower along this new reaction path, then the rate will be faster (this is akin to finding an alternative path through a mountain range, which, though less direct, requires less cumulative ascent). In the context of signaling enzymes, a typical example of this type of catalysis involves the use of enzyme functional groups as nucleophiles, leading to the formation of covalent intermediates. Several phosphatase enzymes that remove phosphate groups from protein residues form enzyme-bound covalent intermediates as part of their catalytic cycle. Methods to analyze the catalytic power of enzymes quantitatively (Michae- lis–Menten analysis) are discussed in Chapter 13. However, we will briefly introduce here two terms that are often used to describe enzymatic prop- erties. The catalytic rate constant (kcat) describes the maximum rate of reaction that can be achieved by the enzyme (that is, in the presence of saturating substrate concentration). The Michaelis–Menten constant (Km) is a measure of the affinity of the enzyme for its substrate. The Km is the substrate concentration at which half-maximal reaction velocity is achieved; if the Km is relatively low, the reaction will proceed efficiently, even at relatively low concentrations of substrate. Enzymes can drive reactions in one direction by energetic coupling Enzymes do not alter the thermodynamic equilibrium between reactants and products; rather, enzymes accelerate the approach to equilibrium. Many biologically important signaling reactions are thermodynamically unfavorable, however. That is, the free energy of the products is greater than that of the reactants, so at equilibrium reactants would predominate over product. For example, protein phosphorylation is thermodynamically unfavorable (for tyrosine phosphate) or approximately neutral (for serine/ threonine phosphates) relative to inorganic phosphate and the unphos- phorylated protein. To overcome this fundamental problem, signaling reactions are often coupled to highly energetically favorable reactions that drive the overall process in only one possible direction. For example, phosphorylation of protein targets is coupled with cleavage of the high- energy β–γ phosphodiester bond of ATP. Thus, a kinase reaction will be allosteric conformational changes essentially irreversible as long as the cell supplies a sufficiently high con- centration of ATP, which is normally the case. driven by hydrolysis of high-energy bond By the same logic, removal of phosphate from proteins cannot be accom- plished through the exact reverse reaction (that is, regeneration of ATP), because this would be thermodynamically unfavorable. Instead, dephos- phorylation proceeds through an alternative reaction: hydrolysis of the phosphoester linkage to the protein by water. This reaction produces inor- ganic phosphate and the free, unphosphorylated protein. This process is thermodynamically favorable for tyrosine phosphates and approximately ATP substrate kinase ADP P neutral for serine/threonine phosphates. Nevertheless, under physiologi- cal conditions, hydrolysis of serine/threonine phosphates is driven to com- phosphatase H20 pletion by the large excess of water relative to the concentration of protein (the molar concentration of water is ∼55 M). Thus both phosphorylation and dephosphorylation are energetically driven in one direction, forming a unidirectional reaction cycle (Figure 3.4), ultimately powered by the cell’s constant supply of ATP. The energy provided by ATP functions anal- ogously to the electric power that is required to drive an electronic circuit. Because uncatalyzed phosphorylation and dephosphorylation reactions are incredibly slow, the degree to which a substrate is phosphorylated will be determined by the kinetics of the opposing enzymatic phosphorylation and dephosphorylation reactions and not by the thermodynamic stability of one state over the other. An important consequence of this behavior is that the output of the system (the degree of substrate phosphoryla- tion) will essentially be a readout of the relative activities of upstream enzymes—a kinase and a phosphatase. These are ideal properties for a dynamic system that must transmit both positive and negative signals rapidly. We will see many times that signal transduction mechanisms exploit and manipulate reaction kinetics to transmit information. AlloSTERic confoRmATionAl chAnGES The ability of upstream signaling inputs to induce changes in protein con- formation, which in turn lead to changes in protein activity, is fundamen- tal to many signaling mechanisms. This coupling allows enzymes to act as nodes that can relay and process information. In this section, we will consider the basis for conformational change in proteins, and the different types of rearrangements that are commonly seen in signaling enzymes. Conformational flexibility of proteins enables allosteric control Signaling mechanisms take advantage of the intrinsic conformational flexibility of proteins, which provides a means by which regulatory inputs can modulate protein function. When looking at the crystal structure of a protein, it is easy to think of the molecule as a rigid, static structure. This impression is quite misleading, however. Although most proteins adopt a stably folded state, they are also highly dynamic, constantly sampling a range of slightly different conformational substates (Figure 3.5). This continuous jostling motion increases with temperature (it is sometimes driven by excess H ₂ 0 Figure 3.4 Phosphorylation/dephosphorylation reactions form a unidirectional, energetically driven cycle. Both the kinase-catalyzed phosphorylation reaction and the phosphatase-catalyzed dephosphorylation reaction are energetically favorable, allowing the reactions to form a unidirectional cycle that can alternate between the phosphorylated and dephosphorylated state. Energy provided by cellular ATP synthesis drives this cycle. (a) (b) Figure 3.5 A folded protein can exist in multiple conformational substates of slightly differing stabilities. in the cell, multiple conformational states of a protein exist, and their relative abundance at equilibrium is determined by their relative free energy. (a) conformation A (circle) predominates over conformation A′ (hexagon) or A″ (square). (b) conformation A has a lower relative free energy than either A′ or A″. note that all of these conformations have much lower free energy than the unfolded state. Figure 3.6 Inputs can change the relative stability of distinct conformational substates. (a) Binding of ligand (orange triangle) stabilizes conformation A′, lowering its free energy relative to conformations A and A″. (b) Phosphorylation makes conformation A″ the most stable conformation. Dotted lines indicate relative free energies in the absence of ligand binding or phosphorylation. ligand binding phosphorylation See Chapter 13 for a review of protein structure active-site conformation called thermal “breathing” of the structure), but occurs even at physio- logical temperature. Some of these conformational substates will be more stable than others (will have lower free energy), and thus will be adopted by a higher fraction of molecules in the population. Other, less stable sub- states will be more poorly populated. The relative free energies of these conformational substates, however, can be altered by the binding of other molecules (ligands) or by post-translational modifications. Thus, a previ- ously unstable conformational substate may be stabilized through events such as ligand binding or phosphorylation (Figure 3.6). An allosteric conformational change refers to a ligand- or modification- induced change in the secondary, tertiary, or quaternary structure of a protein. Figure 3.7 shows schematically how changes such as ligand binding can be associated with stabilization of distinct folded states of proteins. In this illustration, the protein exists in two conformational sub- states in the absence of ligand, one with low activity and another with high activity. In the case of an enzyme, the active-site residues might be in the correct position to promote catalysis in the active state, but not in the inactive state. In this example, the inactive conformation is intrinsically more stable; thus, in the absence of ligand, the protein has low activity. However, ligand binding alters the relative free energy of the two sub- states, leading to a relative stabilization of the active conformation over the inactive conformation. Thus, increasing ligand concentration has two energetically linked effects on the protein. First, it leads to an increase in the fraction of proteins with bound ligand. Second, because ligand bind- ing stabilizes the active state, it leads to an increase in the proportion of absence of ligand presence of ligand LOW activity stable HIGH activity unstable proteins with high activity. Thus, the protein behaves as a sensor that can be activated in response to increasing ligand concentration. There are a number of other specific ways in which signaling inputs can affect output by changing the stability of different protein conformations. For example, similar effects can be mediated by covalent modifications such as phosphorylation. Furthermore, such allosteric changes need not be activating—they can also inhibit or otherwise alter protein activity. For example, ligand binding can preferentially stabilize a conformation with lower activity, or one that binds to a different set of interaction partners. Many signaling proteins can respond to multiple allosteric inputs, some positive and some negative. In this way, a protein can integrate many dif- unstable stable Figure 3.7 The general principle of allosteric regulation. An input such as ligand binding can stabilize an otherwise unstable protein conformation. if this unstable conformation has higher activity, ligand binding increases activity by increasing the fraction of the population in the high-activity conformation. ferent environmental signals through changes in its conformation. Signaling proteins employ diverse classes of conformational rearrangements From a thermodynamic perspective, inputs such as ligand binding or phos- phorylation can modify the energies of different conformational substates by either disrupting interactions unique to one state, or introducing new, favorable interactions that stabilize another state. There are many diverse structural mechanisms for these types of conformational changes, some disorder/order transitions (secondary structure) tertiary structure transitions quaternary structure transitions (e.g. hinge-bending) oligomerization Figure 3.8 interdomain rearrangements reorganization of monomers within an oligomer Many kinds of conformational changes are observed in signaling proteins. Regions in the protein can undergo disorder-to-order transitions (or vice versa) leading to changes in secondary structure (illustrated here by formation of an α helix). Proteins could undergo changes in tertiary structure, such as hinge-bending between two subdomains, or the formation of new interdomain interactions. finally, proteins can undergo changes in quaternary structure, including changes in the molecule’s oligomeric state, or reorganization of the monomers within a preformed oligomer. Areas undergoing significant changes in conformation or interaction are highlighted in light pink. of which are summarized in Figure 3.8. Some changes involve transi- tions between order and disorder, whereby a region of a protein may be unstructured and flexible (adopting no defined secondary structure) in one substate, but structured in another. Indeed, many key regulatory proteins appear to have important intrinsically disordered regions that adopt a defined structure upon ligand binding or post-translational modification. Other changes involve tertiary rearrangements. These may involve rela- tive repositioning of atoms within a single folded unit (intradomain chang- es in tertiary structure). An example of such a change is a hinge-bending motion in an enzyme. In proteins composed of multiple domains, regulatory conformational changes might involve changes in the relative positions of structurally independent modular domains (interdomain changes in terti- ary structure). In proteins capable of oligomerization, regulatory changes may be mediated through quaternary changes including oligomerization or a change in the relative orientation of monomers within an oligomer. This type of quaternary structural change is observed in hemoglobin, a classic example of an allosterically regulated system. The binding of oxygen to one subunit in the hemoglobin tetramer induces a change in quaternary structure that increases the oxygen-binding affinity of the other subunits. PRoTEin PhoSPhoRylATion AS A REGulAToRy mEchAniSm The addition of the phosphate group to proteins by protein kinases and its removal by protein phosphatases is a frequently used mechanism to regulate protein activity. In this section, we discuss the properties of phos- phorylation that lend itself to this role, and the molecular consequences of phosphorylation on protein conformation. Phosphorylation can act as a regulatory mark Protein phosphorylation represents a key mechanism for altering the structure and function of target substrate proteins. Several reasons have been suggested to explain why phosphorylation is so commonly used for serine N H O threonine N biological regulation. First, phosphorylation reactions can be coupled O– to ATP hydrolysis to drive the reaction to completion. Second, although hydrolysis of ATP or of a phosphorylated residue may be highly ener- N getically favorable, these reactions are, by themselves, kinetically unfa- H O vorable—hydrolysis is extremely slow in the absence of a catalyst. Thus, these reactions provide excellent points for kinetic control via enzymes that are required to add and remove phosphate groups (kinases and phos- phatases). If, for example, ATP or a phosphorylated residue were kineti- cally less stable (that is, spontaneously hydrolyzed rapidly), then kinases O– and phosphatases would not be able to provide significant rate enhance- ment over the uncatalyzed reactions, and could serve little regulatory N function. This combination of thermodynamic instability but kinetic sta- H O tyrosine N H O H O OH O O– O– N H O bility may help explain why phosphate ester bonds drive so many biologi- cal processes. In eukaryotes, proteins are most commonly phosphorylated (marked) on the three hydroxylated amino acid residues: serine, threonine, and tyrosine (Figure 3.9). Phosphorylation on histidine and aspartate com- monly occurs in prokaryotes. Serine and threonine phosphorylation is controlled by serine/threonine kinases (Ser/Thr kinases), which write these marks, and by serine/threonine phosphatases (Ser/Thr phos- phatases), which erase these marks. Modification of serine and threonine is controlled by the same class of enzymes, because both amino acids Figure 3.9 The chemical structures of common protein phosphorylation modifications. in eukaryotes, the most common phosphorylated residues are serine and threonine. Tyrosine phosphorylation is less common, but also plays a central role in signaling in multicellular organisms. The phosphate group is highlighted in pink. At neutral ph, the phosphate modification is likely to introduce two new negative charges. See Chapter 4 for a more detailed discussion of different types of pro- tein post-translational modifications have short side chains that are stereochemically identical except for a single methyl group. By contrast, the bulky aromatic tyrosine side chain sterically requires a larger active site and therefore a distinct class of enzyme in most cases. Tyrosine phosphorylation is controlled by tyrosine kinases (Tyr kinases) and tyrosine phosphatases (Tyr phosphatases). However, there are a few kinases that can phosphorylate serine and tyro- sine residues, as well as dual-specificity phosphatases (DSPs) that have the flexibility to accommodate both phosphotyrosine and phosphoserine/ threonine substrates. Phosphorylation can either disrupt or induce protein structure Protein phosphorylation is a common mechanism for regulating protein structure and function. What explains the ability of phosphoryl modifi- cation to promote conformational changes? The phosphate group is rela- tively compact, but represents a major chemical change in the protein. A newly introduced phosphate moiety (pKa 6.7) is most likely to carry a double negative charge at neutral pH. The introduction of this large electrostatic perturbation (relative to the unmodified residue, which is uncharged) at specific sites in a protein can have dramatic conformational effects, both disrupting previously existing interactions and generating new ones. From structural analysis of proteins that are phosphorylated, we know that phosphorylation can alter both local structure (regions near the site of phosphorylation) and long-range (tertiary or quaternary) structure. At the local level, introduction of a new phosphate group on an amino acid can lead to dramatic steric or electrostatic effects. Such effects can disrupt an otherwise structured part of the protein (for example, an α helix) in the region bearing the phosphorylated residue (Figure 3.10a). Disrup- tion of local structure is observed if phosphorylation results in repulsion of another nearby negative charge, or disruption of a hydrogen bond made by the nonphosphorylated form of the side chain. The resulting conforma- tional changes could, for example, move active-site residues of an enzyme out of position, resulting in a loss of activity. (a) (b) local disruption P local ordering Figure 3.10 Protein phosphorylation can both disrupt and induce structure. introduction of a phosphate modification can sterically or electrostatically disrupt interactions previously made by the modified residue (or nearby residues). here, phosphorylation disrupts a hydrogen bond (pink dotted line) between an α helix and an adjacent strand of a β sheet. As a consequence, the α helix unfolds. A new phosphate group can also form new electrostatic or hydrogen-bonding interactions with other parts of the protein, inducing new structure. here, an introduced phosphate hydrogen-bonds with two positively charged residues, allowing a In other cases, introduction of a phosphate group can lead to the forma- tion of new local structures (Figure 3.10b). In such cases, a newly intro- duced phosphate group often participates in new interactions with nearby positively charged moieties. Structural analysis has revealed two common types of interactions between the phosphate group and the rest of the protein. First, a phosphate group can engage in hydrogen bonds with the main-chain amide groups at the N-terminus of an α helix (Figure 3.11a). Such interactions help to neutralize the intrinsic dipole moment of the α helix (its propensity to carry a positive charge at the N-terminus and a negative charge at the C-terminus). The second major type of interaction is between the phosphate group and the positively charged guanidinium group of one or more arginine side chains (Figure 3.11b). Such electro- static interactions can have the effect of inducing the formation or stabili- zation of local structure. In many cases, a new phosphoryl group will both disrupt preexisting interactions and induce new ones, leading to signifi- cantly different structures in the unphosphorylated and phosphorylated states. As detailed below, this type of phosphorylation-induced local order- ing is commonly observed in the activation loop of protein kinases, where only the phosphorylated loop adopts a conformation that is compatible with substrate binding and catalysis. Phosphorylation can also have longer-range effects on tertiary and qua- ternary structure (Figure 3.12). For example, phosphorylation can pre- vent the binding of a protein to a partner molecule or to another domain of the same protein. Addition of the negatively charged phosphate group to a binding surface may sterically and electrostatically block ligand or substrate binding, thus negatively regulating protein activity. Conversely, if the ligand is a negative regulator, then phosphorylation would result in an increase in protein activity. Phosphorylation can also promote new long-range intramolecular and intermolecular interactions. Signaling proteins often contain protein interaction domains, such as SH2 domains, that specifically recognize phosphorylated amino acid motifs. Interactions (b) previously unstructured region to fold into an α helix. See Chapter 10 for more on modu- lar interaction domains Figure 3.11 Common interactions made by phosphate groups in proteins. The doubly charged phosphate group is commonly observed to participate in several specific types of interactions. (a) it can O O O– O– + NH2 NH2 NH interact with two successive free amide groups at the n-terminus of an α helix. The helix has a dipole moment oriented with positive charge at the n-terminal end; interaction with phosphate neutralizes this positive charge. (b) Phosphate can interact with two nitrogens in the guanidinium group of an arginine side chain. long-range disruption Figure 3.12 Long-range effects of protein phosphorylation. (a) Phosphorylation at the interface of a tertiary or quaternary structural interaction can disrupt the interaction. Phosphorylation that creates a docking site for a phospho-recognition domain can result in the formation of a new tertiary or quaternary interaction. The illustration depicts tertiary (intramolecular) rearrangements. (b) long-range ordering with these domains will only occur when the target protein is phosphor- ylated. This type of long-range phosphorylation-dependent conformation- al change is a common method of functionally linking phosphorylation to enzyme activity, as detailed below for Src family kinases. PRoTEin KinASES Protein kinases are among the most numerous signaling enzymes in eukaryotes, playing a central role in many important signaling pathways. While protein kinase catalytic domains have a common overall structure and catalytic mechanism, these common elements are regulated in a vari- ety of ways to allow kinases to adapt to different modes of control and to phosphorylate different classes of substrates. Since kinases are a canoni- cal example of signaling enzymes, here we review the fundamental mech- anisms of kinase catalysis and the ways in which this catalytic activity can be regulated by allosteric inputs. The structure and catalytic mechanism of protein kinases are conserved Both Ser/Thr kinases and Tyr kinases share a common folded structure for their catalytic domains. The fold of this bilobed structure, shown in Figure 3.13, is similar to those of distantly related metabolic kinases that phosphorylate small-molecule substrates. The active site of the enzyme, including binding sites for ATP and substrate peptide, lies at the groove between the two lobes of the domain. The reaction that is catalyzed by kinases requires nucleophilic attack on the ATP γ phosphate by the hydroxyl group of a substrate amino acid (Figure 3.14a). Thus, the enzyme must be sterically compatible with both ATP and peptide substrates, and must provide properly positioned catalytic residues that increase the nucleophilicity of the attacking hydroxyl group, and stabilize the transition state and developing charge on the leaving group. Given the precise requirements for this class of reactions, it is not sur- prising that protein kinases have a number of highly conserved catalyt- ic residues, even across both Tyr kinases and Ser/Thr kinases. We will Figure 3.13 Structure of protein kinases. Schematic depiction of a protein kinase catalytic domain, showing the characteristic bilobed structure. The active site, which binds ATP and peptide substrates, lies between the n- and c-terminal lobes. X-ray crystal structure of the insulin receptor kinase domain in complex with ATP and substrate peptide (PDB 1iR3). colors correspond with the diagram in part (a); peptide substrate is indicated in orange. (a) ATP N-lobe C-lobe substrate binding pocket (b) C-helix in position catalytic loop activation loop Figure 3.14 The active site of protein kinases. Key catalytic residues are contributed by both the n- and c-lobes, including lys72 and Asp184, which help to coordinate the ATP, and Asp166 on the catalytic loop (orange), which acts as a general base to activate the substrate hydroxyl moiety. Proper positioning of these and other catalytic residues is dependent on tertiary interactions with key elements in the kinase structure, especially the c-helix (purple) in the n-lobe and the activation loop (green) in the c-lobe. in many kinases, the c-helix and activation loop are used as structural levers for controlling kinase activity. cyclin-dependent kinase (cDK) activation is regulated by proper positioning of the c-helix (purple) and the activation loop (green). Dramatic repositioning of the c-helix occurs upon cyclin binding, while proper positioning of the activation loop occurs upon activating phosphorylation of Thr197. Together, these conformational changes lead to proper assembly of the active site and clearance of the peptide-binding site. numbering of key residues is based to their positions in the protein kinase A (PKA) sequence. (b, Adapted from m. huse and J. Kuriyan, Cell 109:275–282, 2002. With permission from Elsevier.) Lys72 Thr 197 Glu91 cyclin activation loop in position inactive CDK CDK bound to cyclin phosphorylated CDK bound to cyclin describe these key conserved catalytic residues using the residue number- ing from cyclic AMP-activated protein kinase A (PKA), a Ser/Thr kinase that was one of the first protein kinases to be structurally characterized (Figure 3.14a); other individual kinases have diverse residue numbering (and different insertions and deletions within the catalytic domain). The N-terminal lobe contains residues that are primarily involved in ATP coordination. This includes the phosphate-binding loop (P-loop), which is a flexible glycine-rich segment. In addition, Lys72 plays a critical role in coordinating the negatively charged phosphate moieties of ATP. Many of the remaining critical catalytic residues are contained in the slightly larger C-terminal lobe. Asp166 lies within a segment known as the cata- lytic loop; this residue serves as the general base that abstracts the proton from the attacking substrate peptide hydroxyl group. Two other C-lobe residues, Asp184 and Asn171, play a critical role in ATP binding. See Chapter 12 for more on the cell cycle and cyclin-dependent kinases The activation loop and C-helix are conserved molecular levers that conformationally control kinase activity As signaling molecules, protein kinases typically function as switches that must exist in at least two conformations: an inactive conformation in which active-site residues are not aligned for catalysis; and an active con- formation, stabilized by the proper upstream inputs, in which the active- site residues are optimally positioned for catalysis. There are several core mechanisms that are used by most protein kinases to regulate catalytic activity. Most of these mechanisms rely on a conserved set of structural elements—the activation loop and the C-helix, highlighted in Figure 3.14. Because of their central position and interactions with many catalytic residues, these two elements essentially act as regulatory levers that can control kinase activity. Diverse regulatory inputs that shift these levers result in the movement of catalytic residues in or out of proper position. The activation loop is perhaps the most important regulatory segment in the protein kinase family. This loop is within the C-lobe and lies at the active site. The length, sequence, and conformation of the activation loop differ in individual kinases. However, in nearly all kinases, the activation loop is observed to undergo large conformational changes associated with distinct activity states (Figure 3.14b). The activation loop controls kinase activity through two coordinated mech- anisms. First, residues in the activation loop often participate in hydro- gen bonds with residues immediately adjacent to catalytic residues. This structural coupling means that the position of the activation loop can alter catalytic efficiency. Second, the activation loop, in some cases, can directly occlude the binding site for the peptide substrate, thus blocking enzyme activity. In such cases, activating stimuli move the loop out of the way, allow- ing substrate access to the active site. In most protein kinases, activation requires phosphorylation of residues within the activation loop. These phos- phorylation events usually alter the conformation of the loop, both exposing the peptide-binding site and properly positioning the catalytic residues. The C-helix is another regulatory lever within the kinase domain. This helix is in the N-lobe and contains a conserved residue, Glu91, which forms a hydrogen-bond network with the conserved catalytic residue, Lys72. Precise positioning of Lys72 is required for kinase activity and its interaction with Glu91 ensures it is properly positioned for ATP bind- ing. Thus, movement of the C-helix, and the subsequent movement of the Glu91–Lys72 pair, can dramatically alter the enzyme activity level. An example is provided by the cyclin-dependent kinases (CDKs), serine/ threonine kinases that control many important cellular events including the cell cycle. CDK regulation centers around the C-helix and activation loop. In the inactive CDK structure, the C-helix and the catalytic Lys72 residue are far out of optimal position. Activating inputs (phosphorylation and binding of cyclin, a regulatory subunit) induce a large translation and 90° rotation of the C-helix, resulting in the proper positioning of the Glu91–Lys72 pair (Figure 3.14b). Insulin receptor kinase activity is controlled via activation-loop phosphorylation Control of kinase activity by phosphorylation of the activation loop is illustrated in the example of the insulin receptor tyrosine kinase (IRK) (Figure 3.15). The insulin receptor binds to insulin, the key hormone in glucose homeostasis, and mediates the physiological responses to it in cells throughout the body. The IRK exists as a heterotetramer that con- tains two identical catalytic domains linked by disulfide bonds. Activation movement of activation loop Tyr 1158 Tyr 1162 Asp 1150 Asp 1150 Asp 1132 Arg 1155 pTyr 1158 Asp 1132 substrate peptide pTyr 1163 pTyr 1162 Tyr 1163 Arg 1164 of IRK requires phosphorylation on three resides in its activation loop: Tyr1158, Tyr1162, and Tyr1163. In the inactive, unphosphorylated state, Tyr1162 is hydrogen-bonded to the active-site Asp1132 residue located in the catalytic loop. This interaction positions the activation loop in a way that prevents both peptide and ATP binding. Phosphorylation of Tyr1162 destabilizes the autoinhibitory conformation of the activation loop, expos- ing the other tyrosines for phosphorylation. Phosphorylation of the loop then promotes new interactions between the phosphorylated tyrosines and Arg1155 and Arg1164, which configure the activation loop for cataly- sis and expose the catalytic cleft to substrates. Phosphorylation mediates long-range conformational regulation of Src family kinases The Src family kinases, a group of cytosolic tyrosine kinases that regulate processes such as cell adhesion and lymphocyte activation, illustrate how phosphorylation can induce long-range conformational changes that regu- late activity (Figure 3.16). In particular, phosphorylation of a tyrosine side chain in the C-terminal tail of Src kinases (Tyr527) inhibits kinase activity through large-scale rearrangements involving two regulatory domains. In addition to the kinase domain, Src family members have two protein inter- action modules: an SH2 and SH3 domain. The SH2 domain recognizes phosphotyrosine (pTyr) residues and can bind the pTyr on the C-terminal tail. This binding, in turn, promotes the interaction of the SH3 domain with another intramolecular recognition motif in the linker between the SH2 domain and the catalytic domain. These two intramolecular inter- actions together induce an inactive conformation of the kinase domain, essentially stabilizing a conformation that cannot efficiently phosphor- ylate substrates. Disruption of these autoinhibitory interactions, either by dephosphorylation of tyrosine in the tail or by binding of the SH2 or SH3 domains to competing external ligands, activates the enzyme. This type of regulation depends on the dynamic nature of protein confor- mation. In order for the kinase to be activated in response to signals, the Figure 3.15 Activation-loop phosphorylation stabilizes the active conformation of the insulin receptor tyrosine kinase (IRK). (a) in the inactive, unphosphorylated state of the insulin receptor kinase, Tyr1162, located in the activation loop (green), forms a hydrogen bond with the catalytic residue Asp1132 in the catalytic loop (orange). This interaction occludes the active site. (b) in the active form of the kinase, three tyrosine residues (Tyr1158, Tyr1162, and Tyr1163) in the activation loop are phosphorylated (phosphates are highlighted in red). Phosphorylation of Tyr1162 (pTyr1162) disrupts the inhibitory interaction between it and the catalytic residue Asp1132, and phosphorylated Tyr1162 and Tyr1163 then form salt bridges with Arg residues elsewhere in the protein. The overall effect of these changes is a rightward shift of the activation loop (green arrow), which opens up access to the active site for the peptide substrate (blue). note also that the n-terminal lobe of the kinase has rotated relative to the c-terminal lobe upon activation. ATP is not shown for clarity. (b) SH3 SH2 phosphorylated SH3 SH2 unphosphorylated inactive active Figure 3.16 Phosphorylation controls tertiary structure and activity of the Src family kinases. (a) Structure of the Src family tyrosine protein kinase hck in its inactive form. The catalytic domain is linked to two small protein-binding domains (orange)—an Sh3 domain and an Sh2 domain—that make intramolecular interactions that clamp the catalytic domain in its inactive conformation. These intramolecular interactions are disrupted upon activation, allowing the Sh2 and Sh3 domains to interact with other proteins. The c-helix is colored purple, activation loop green, and catalytic loop orange. Side chains of Tyr416 and pTyr527 are shown in stick format. (b) Schematic representation of the inactive and active conformations of a generic Src family kinase. Phosphorylation of the c-terminal tyrosine (Tyr527) stabilizes the inactive conformation through interaction with the Sh2 domain. The active conformation is also stabilized by phosphorylation of Tyr416 in the activation loop. note that the active conformation can also be stabilized by binding of the Sh2 or Sh3 domains to ligands in trans, even if the c-terminal site is phosphorylated. inactive conformation must be relatively unstable. For example, if we imagine that the SH2–pTyr527 interaction had very high affinity (was energetically very favorable), the two would almost never be dissociated; the pTyr527 site would not be exposed to phosphatases for dephospho- rylation, and the SH2 domain would not be free to bind to other phos- photyrosine sites. Activation would be infrequent and would take a very long time. For this reason, intramolecular regulatory interactions usu- ally have relatively low affinity, allowing for dynamic regulation. Full activation of the Src kinase also requires the subsequent autophos- phorylation of Tyr416 within the activation loop, similar to other kinases such as IRK discussed above. Phosphorylation of Tyr416 is generally per- formed by another Src family kinase in trans, for example after kinase aggregation or co-localization in response to signals. Thus, Src is actually controlled by phosphorylation in two ways: a negative regulatory phos- phorylation at the C-terminal tail is responsible for tertiary interactions that lock the kinase in an inactive conformation, while a positive regula- tory phosphorylation in the activation loop is required to locally order the active site for optimal activity. Multiple binding interactions regulate protein kinase substrate specificity Eukaryotic cells contain many different protein kinases (∼500 distinct genes in humans; ∼100 in yeast), raising the question of what determines which specific proteins will serve as substrates for each kinase. The sub- strate specificity of kinases is determined by a combination of factors (Figure 3.17). First, as for most enzymes, the active sites of kinases INPUT: upstream regulators OUTPUT: substrate/ effector active site docking interactions modular domains scaffolds/adaptors Figure 3.17 Mechanisms of substrate specificity in protein kinases. multiple mechanisms that contribute to substrate specificity are depicted, with specificity-determining regions highlighted in pink and substrates in dark brown. The active site of a kinase can recognize specific sequence motifs flanking the residue to be phosphorylated. Some kinases, particularly Ser/Thr kinases, have docking sites away from the active site that can recognize docking-motif peptides within substrates. Kinases may also contain modular binding domains that recruit target substrates. finally, kinases can use third- party accessory proteins, such as adaptors or scaffolds. These proteins bind the kinase, but also provide interaction sites for specific substrates. When bound to different scaffolds or adaptors, a kinase might display quite different substrate specificities, thereby dramatically increasing the functional flexibility of one kinase. (Adapted from R.P. Bhattacharyya et al., Annu. Rev. Biochem. 75:655–680, 2006. With permission of help determine specificity. For example, the substrate-binding pocket of Tyr kinases is considerably deeper than that of Ser/Thr kinases, which allows them to specifically accommodate the much larger tyrosine side chain. In addition, the active site often has adjacent pockets that show preferences for particular side chains at residues flanking the residue to be phosphorylated. CDKs, for example, almost exclusively phosphorylate serine or threonine residues that are immediately followed by a proline. However, unlike most canonical enzymes, kinases also use interactions that lie far outside of their active sites to determine substrate specificity. Many Ser/Thr kinases have docking sites on the kinase domain that are distant from the active site and which recognize specific peptide motifs. Kinase substrates often have a combinatorial structure with a docking motif and a substrate motif separated by a linker element; the presence of two specificity-determining sequences greatly increases the ability of the kinase to discriminate among different substrates. In some cases, docking sites can play a regulatory role. For example, several members of the AGC Ser/Thr kinase family have a docking site on the N-terminal lobe referred to as the PIF pocket or hydrophobic motif pocket. Peptide docking motifs that bind in the PIF pocket not only aid in determining substrate specifi- city, but they can also allosterically activate the catalytic domain of the kinase. Other domains that lie outside of the kinase domain can also help deter- mine substrate specificity; these are known as accessory domains. For example, the SH2 and SH3 domains of Src family kinases are important both in regulation of kinase activity and in substrate recognition. Pro- teins that contain SH2- and SH3-binding motifs can serve as optimal substrates because these interactions preferentially target the kinase to these substrates. And because the SH2 and SH3 domains normally inhibit kinase activity through intramolecular interactions, as discussed above, substrate binding also serves to activate the kinase. In this way, maxi- mal activity is attained only when the kinase is bound to an appropriate substrate, decreasing the likelihood of off-target activity that could have deleterious effects. Annual Reviews.) In some cases, accessory subunits that noncovalently associate with a kinase domain can help determine substrate specificity. For example, the cyclins that associate with CDKs have two major functions: as described earlier (see Figure 3.14b), they are allosteric activators that cause the kinase to adopt an activated conformation; but they also contain a dock- ing site, so that proteins containing a complementary docking motif are preferentially phosphorylated by CDKs (typically, conformational changes that activate the kinase lead to an increased kcat, while additional binding interactions lead to lower values of Km). CDKs can associate with differ- ent cyclin subunits at different stages of the cell cycle. Since each of these cyclin subunits recognizes distinct docking motifs, this provides a mecha- nism for CDKs to phosphorylate different sets of substrates at different times. Often the broad class of accessory proteins that help direct kinases and other signaling enzymes toward specific substrates are known as scaffold proteins, which are discussed in detail at the end of this chapter. As can be seen by these examples, accessory domains and subunits often play a role in regulating kinase activity and determining its resulting substrate specificity. It is worth noting that tyrosine kinases generally use diversification of accessory domains to achieve distinct specificity—these kinases tend to be modular, multidomain proteins. In contrast, serine/ threonine kinases tend to be smaller proteins that achieve distinct sub- strate specificity by associating with different accessory subunits. Histidine kinases and prokaryo- tic two-component systems are described in Chapter 4 Protein kinases can be divided into nine families The canonical protein kinases, for the most part, appear to have evolved in eukaryotes. Prokaryotes do have a distinct type of kinase, histidine kinase, which is not found in most eukaryotes. Genomic sequencing studies have revealed that prokaryotes also have a number of Protein Kinase-Like (PKL) genes, which appear to comprise an ancient, dis- tantly related family (which also includes some atypical eukaryotic kinases). The functions of prokaryotic kinase-like genes have not been well characterized. Within eukaryotes, the protein kinase family has expanded dramatically. In the human genome, there are ∼500 putative protein kinases, mak- ing them one of the largest families of enzymes in the genome. Of these, Ser/Thr kinases far outnumber the Tyr kinases: there are ∼400 Ser/Thr kinases and ∼90 Tyr kinases. Because of their greater number and their distribution among diverse eukaryotic species, Ser/Thr kinases are thought to have evolved considerably earlier than Tyr kinases, probably near the time of the emergence of eukaryotes. Tyrosine kinases are found almost exclusively in metazoans (multicel- lular animals). The exception to this rule is that they are also found in choanoflagellates, the closest unicellular organisms to metazoans. This observation has led to the proposal that tyrosine kinase signaling first emerged about one billion years ago, near the point at which multicellular organisms evolved. The added signaling capacity provided by this novel system may have played a key role in facilitating the emergence of mul- ticellular forms of life, with their increased needs for cell–cell signaling. In humans, the protein kinase family can be divided into nine distinct sub- families based on sequence homology. In addition to the tyrosine kinase (TK) family, there are eight families of Ser/Thr kinases (Figure 3.18). Presumably, the utility of the protein kinase as a fundamental regulatory device led to its expansion and diversification into this large number of distinct subfamilies. (a) Figure 3.18 Subfamilies of protein kinases. (a) classes of Ser/Thr kinases and Tyr kinases, with the approximate numbers in each class within the human genome indicated. canonical members of each class include protein kinase A (AGc), calmodulin-dependent protein kinase (cAmK), casein kinase 1 (cK1), cyclin-dependent kinase (cmGc), mAP/Erk kinase (STE), Raf (TKl), receptor guanylyl cyclase A (RGc), and Src (tyrosine kinase). Kinases classified in “other” include well-defined families, such as the Polo family, that cannot be grouped with one of the major classes. (b) Phylogenetic tree showing the relationship between human kinases. (b, Adapted from G. manning et al., Science 298:1912–1934, 2002. With permission from AAAS). PRoTEin PhoSPhATASES Protein phosphatases catalyze the removal of covalent phosphate modifi- cations on protein side chains. As mentioned above, this reaction function- ally opposes the phosphorylation reaction catalyzed by kinases, although it is not the chemical reverse reaction. Rather, the phosphatase reaction is a distinct reaction that cleaves a phosphorylated protein into inorganic phosphate and the free protein. Together, the kinase and phosphatase reactions form a unidirectional cycle in which both reactions are under kinetic control. As with kinases, protein phosphatases are classified based on the phos- phorylated amino acid substrates on which they act. Thus, there are two classes: the serine/threonine (Ser/Thr) phosphatases and tyrosine (Tyr) phosphatases. Serine/threonine phosphatases are metalloenzymes There are three major families of Ser/Thr phosphatases: the PPP, PPM, and FCP families (Figure 3.19). The PPP phosphatases share a core cata- lytic domain of 280 residues, and are found in all eukaryotes ranging from yeast to man. They are also found in bacteria and archaebacteria, indicat- ing an ancient origin from a common pre-eukaryotic ancestor. There are 13 PPP family members in humans and 12 in yeast. PPP phosphatases are metalloenzymes that bind a pair of metal ions (typically Fe3+ and either Zn2+ or Mn2+). This bimetal center contributes to catalysis in sev- eral ways: the metals coordinate and orient the substrate, they stabilize charge building in the transition state, and they activate a water molecule such that it can perform a nucleophilic attack on the substrate phosphate (Figure 3.20). PPP phosphatases show specificity for phosphoserine and phospho- threonine, but in some cases are capable of acting on phosphotyrosine. Typically, these enzymes show only modest sequence specificity for the protein in which the phospho-modification occurs. In most cases, PPP phosphatases are found in complexes with other proteins that determine which specific substrates are targeted. Each of these phosphatases can participate in several distinct complexes, allowing their localization to diverse areas in the cell and participation in a broad range of distinct pathways (Figure 3.21). A few PPP members such as PP1 and PP2B (also known as calcineurin) also have docking grooves on their catalytic subunits, which contribute to substrate specificity by recognizing specific peptide motifs in substrates. The second family of Ser/Thr phosphatases, the PPM phosphatases, includes PP2C and pyruvate dehydrogenase phosphatase. This class of enzymes is also present in eukaryotes, bacteria, and archaebacteria, sug- gesting a pre-eukaryotic origin. There are 10 PPM proteins in humans. Like the PPP family, the PPM phosphatases are metalloenzymes, but dif- fer from PPP family members in that they require Mg2+. The active site of the PPM phosphatases is structurally similar to that of PPP phosphatases but there is no sequence similarity, suggesting that these two families of phosphatases evolved independently (a case of evolutionary convergence). As with the PPP phosphatases, catalysis by PPM phosphatases occurs via activation of a water molecule that directly hydrolyzes the substrate phosphoester bond. Most PPM family members are monomeric, and thus their substrate preferences appear to be determined largely by accessory domains that occur within the same polypeptide chain as the catalytic domain. (a) protein phosphatases (~140) pSer/Thr phosphatases (28) pTyr phosphatases (107) PPP (13) PPM (10) FCP (5) Cys-based (103) Asp-based (4) (b) Figure 3.19 Subfamilies of protein phosphatases. (a) classes of protein phosphatases, with the approximate numbers in each class within the human genome indicated. Phosphoserine/threonine (pSer/Thr) phosphatases can be divided into three classes: PPP, PPm, and fcP. canonical members of each class include PP1 and calcineurin (PPP), PP2c (PPm), and fcP1 (fcP). Several phosphotyrosine (pTyr) phosphatases use an active-site aspartate (for example, Eya protein), but the majority use an active-site cysteine and can be divided into three classes. class i cys-based phosphatases can be further subdivided into the classical protein tyrosine phosphatases (PTPs), which only act on phosphotyrosine, and the dual-specificity phosphatases (DSPs), which can act on both phosphotyrosine and phosphoserine/ threonine. canonical class i PTPs include PTP1B (a classical PTP) and the mAP kinase phosphatases, which are DSPs. Some members of the DSP family (for example, PTEn) can act on nonprotein substrates such as phospholipids. The only class ii PTP in humans is lmPTP, and the three class iii PTPs are cdc25A, cdc25B, and cdc25c. (b) Phylogenetic trees showing the relationship between human phosphatases. The large dendrogram shows the large family of class i cys-based phosphatases, which encompasses both the protein tyrosine and dual-specificity phosphatases. The other unrooted trees show the other more distantly related families of phosphatases. (b, Adapted from S.c. Almo et al., J. Struct. Funct. Genomics 8:121–140, 2007. With permission from Springer Science and Business media.) His151 HN +NH substrate O His151 OH substrate HO P O H2N –O O– H2N Asp90 His92 HO ^– Fe3+ N –O O– 2 1 O– H O – O Zn2+ N Asn150 His281 H2O Asp90 His92 HO ^– O– Fe3+ H O N – Asn150 O His281 N N O Asp118 H N H His199 N O Asp118 H N H His199 Figure 3.20 Reaction mechanism of serine/ threonine phosphatases. Ser/Thr phosphatases have a bimetal active site, with the metal ions coordinating the phosphorylated substrate. The phosphoryl group is transferred to water in a single step with no covalent intermediate. Two alternative mechanisms have been proposed involving either (1) attack by a bridging hydroxide, or (2) attack by a terminal, metal-bound hydroxide. in either case, a his residue (his151) has been suggested to act as a general acid. Residue numbering is for human calcineurin (PP2B). Consistent with their similar mechanisms of action, both the PPP and PPM classes of Ser/Thr phosphatases are sensitive to the inhibitor oka- daic acid, which is an important tool in analysis of signaling pathways involving Ser/Thr phosphorylation. The third family of Ser/Thr phosphatases, the FCP phosphatases, was discovered relatively recently. These enzymes are dependent on Mg2+ for catalysis. They play a role in dephosphorylating the C-terminal domain of RNA polymerase, which plays an important role in transcriptional regula- tion. In addition, emerging evidence suggests that these phosphatases can regulate growth factor signaling pathways. Most tyrosine phosphatases utilize a catalytic cysteine residue Protein tyrosine phosphatases (PTPs) can be divided into several different families (see Figure 3.19). There are two major classes based on the identity of the key catalytic residue. The cysteine-based phosphatases are far more common than the aspartate-based phosphatases; for example, the human genome encodes 103 Cys-based phosphatases and just 4 Asp-based phos- phatases. The Cys-based phosphatases can be further subdivided into three (b) Figure 3.21 Serine/threonine phosphatases can form different holoenzyme complexes. (a) Regulatory proteins, including modulators (green) and inhibitors (pink), control Protein Phosphatase 1 (PP1) function. (b) Structures of PP1 holoenzyme complexes. PP1 (blue) is shown in surface rendering, with catalytic metal ions colored y e ll o w; modulatory or inhibitory subunits are depicted as ribbons. left: structure of PP1β in complex with the modulator myPT1 (green). myPT1 provides additional surfaces for substrate recognition and thus controls PP1 substrate specificity. center and right: PP1γ in absence (center) and presence (right) of inhibitor-2 (pink). inhibitor-2 blocks access to the active site. (Adapted from D.m. Virshup and S. Shenolikar, Mol. Cell 33:537–545, 2009. With permission from Elsevier.) O O HO Asp181 HO substrate O O Asp181 –O O Asp181 HO O Cys215 S– –O H P substrate O –O H ₂ O H Cys215 H –O H H H Cys215 S– –O OH –O H H N N Arg221 H N N Arg221 H N N Arg221 H +NH ₂ +NH ₂ +NH ₂ classes. By far the largest class of Tyr phosphatases in eukaryotes is the class I Cys-based PTPs (99 in humans), which includes the classical PTPs and the dual-specificity phosphatases (DSPs, also known as VH1-like phos- phatases). Class II Cys-based PTPs (of which there is only one example in humans) are of low molecular weight and related to bacterial arsenate reductase. Class III Cys-based PTPs include the cell-cycle regulatory phos- phatase Cdc25 (there are three Cdc25 enzymes in humans). The three classes of Cys-based tyrosine phosphatases are fairly diverse, but they share several common features (Figure 3.22). All have a cata- lytic motif comprising cysteine and arginine residues separated by five other amino acids (Cx₅R). The conserved cysteine acts as a nucleophile that attacks the substrate phosphotyrosine, forming a covalent phos- phocysteine intermediate and releasing the substrate tyrosine. In the second step of the reaction, a water molecule then hydrolyzes the phos- phocysteine intermediate to yield the free phosphate group. The con- served arginine residue in the catalytic motif functions to stabilize the transition state, by coordinating with the phosphoryl group throughout the reaction, while another conserved residue—an aspartic acid—acts as a general acid, donating a proton to the hydroxyl of the substrate tyro- sine (the leaving group). The aspartate then acts as a general base to activate water for nucleophilic attack in the second half-reaction. Other residues that are not absolutely conserved assist in the cysteine nucle- ophilic attack by lowering its pKa, thus favoring the more nucleophilic deprotonated form. Unlike Ser/Thr phosphatases, the Cys-dependent Tyr phosphatases are metal-independent and insensitive to the inhibitor okadaic acid. They are, however, sensitive to the inhibitor vanadate, which stereochemically mim- ics the pentacoordinate phosphate transition state. Phosphate normally has four coordinated oxygen atoms, but in the nucleophilic displacement of the dephosphorylation reaction, it must pass through a pentacoordi- nate transition state in which the attacking nucleophile and the leaving group are both present (Figure 3.23). Because these PTPs use a catalytic cysteine residue, they are, in some cases, subject to oxidative regulation. For example, when the phosphatase PTP1B is oxidized, the catalytic cysteine and a neighboring serine residue form a sulfenyl amide ring, which inactivates the enzyme. However, this moiety can be reversibly reduced to restore enzyme activity. This unusual mode of regulation is thought to allow reactive oxygen species and nitro- gen oxides to modulate phosphotyrosine signaling. As mentioned above, the class I Cys-based PTPs can be further subdi- vided into two distinct subclasses—the classical PTPs, which only act Figure 3.22 Reaction mechanism of tyrosine phosphatases. class i cys-based Tyr phosphatases, such as PTP1B or Shp2, utilize an active-site cysteine residue. The cysteine side chain acts as the initial nucleophile, resulting in release of the protein portion of the substrate and formation of a covalent, thiophosphate intermediate. in the second half-reaction, the phosphate is released after nucleophilic attack by water. in some PTPs, activity of the enzyme is regulated by reversible oxidation of the catalytic cysteine (in PTP1B, the catalytic cysteine and a neighboring serine can reversibly form a sulfenyl amide ring). (a) O S: P O O pentacoordinate transition state O O S O O O O S P HO O (b) O O O V O O vanadate Figure 3.23 Vanadate mimics the transition state of the dephosphorylation reaction. (a) in the catalytic reaction of cys-based tyrosine phosphatases, nucleophilic attack on the phosphorus atom by the active site cysteine proceeds through a pentacoordinate transition state. (b) This transition state is mimicked by the pentacoordinate structure of orthovanadate, which is a potent inhibitor of tyrosine phosphatases. note that the dephosphorylation reaction also involves a second pentacoordinate intermediate, when water attacks the covalent phosphoenzyme intermediate to release inorganic phosphate (see figure 3.22). on phosphotyrosine, and the dual-specificity phosphatases (DSPs). In humans, there are 38 classical PTPs and 61 DSPs. In many cases, the term PTP is used to refer only to the classical PTPs. The classical PTPs are highly selective for phosphotyrosine, most likely because of their deep active-site pocket, which only the longer phosphotyrosine residue can reach into. By contrast, dual-specificity phosphatases can, in some cases, hydrolyze phosphoserine and phosphothreonine. In fact, some members of the DSP family can also act on nonprotein substrates including phos- pholipids and RNA. See Chapter 10 for more on the modular architecture of signaling proteins Tyrosine phosphatases are regulated by modular domains while serine/threonine phosphatases often associate with regulatory accessory subunits Tyrosine phosphatases differ significantly from Ser/Thr phosphatases in their numbers and their protein architectures. There are around 100 Tyr phosphatases and a similar number of Tyr kinases in humans. This contrasts with the Ser/Thr phosphatases, which are vastly outnumbered by Ser/Thr kinases (∼30 Ser/Thr phosphatases versus ∼400 Ser/Thr kinases). Diversification of Tyr phosphatases occurs mostly with respect to pro- tein architecture—most Tyr phosphatases are complex modular proteins in which the catalytic phosphatase domain is one of several modular domains within the same polypeptide chain (Figure 3.24). Highly diverse combinations of accessory domains are linked to both classical PTP and DSP domains. There are even PTP domains that are found in transmem- brane proteins, in a class of proteins referred to as receptor PTPs. These diverse accessory domains are thought to determine the targeting of these proteins to particular subcellular sites and to specific substrates. The accessory domains can also participate in allosteric autoinhibitory interactions, thus regulating the phosphatase activity much as accessory domains regulate tyrosine kinase activity. This contrasts with the Ser/ Thr phosphatases, where individual phosphatase polypeptides general- ly participate in a number of alternative holoenzyme complexes. In this case, various accessory subunits in the complexes are thought to play key (a) SH2 SH2 PTP FERM PDZ PTP Fn PTP PTP TM Shp2 PTP1 CD45 Figure 3.24 Modular domain structure of tyrosine phosphatases. (a) Domain structures of three representative PTPs, showing multiple accessory domains in addition to the catalytic domain (blue). Tm, transmembrane segment. (b) X-ray crystal structure of Shp2 in the inactive state. Two Sh2 domains (yellow and green) play a dual role in regulating Shp2: they allosterically regulate the phosphatase domain (by occluding the active site in the inactive state), and they target the phosphatase to specific locations and substrates. The roles in functional diversification: they target the enzyme to specific sites, determine substrate preferences, and mediate regulation. G PRoTEin SiGnAlinG G proteins store and transmit information based on their conforma- tional state, and thus provide another canonical example of the tight coupling between protein activity and conformation. Serving as the basis for one of the most widely used and important signaling mecha- nisms in eukaryotes, G proteins play crucial roles in diverse signaling pathways including hormone signaling, cytoskeletal regulation, and nuclear import and export. G proteins are named for their ability to bind guanine nucleotides—both guanosine triphosphate (GTP) and guanos- ine diphosphate (GDP). The primary function of G proteins is to serve as conformational switches: they adopt significantly different conforma- tions depending on whether GTP or GDP is bound (Figure 3.25). In gen- eral, the GTP-bound state is the “active” conformation, which is capable of binding and modulating the activity of downstream effectors, while the GDP-bound state is the “inactive” conformation, with much lower affinity for these effectors. G proteins are conformational switches controlled by two opposing enzymes In the absence of regulatory inputs, cycling between the active and inac- tive states of G proteins is very slow. G proteins can hydrolyze bound GTP to GDP (thereby switching themselves from the active to inactive conformation), but they are very poor enzymes. In fact, the reaction half- time for GTP hydrolysis (the time at which there is a 50% chance that OUTPUTS active-site cysteine is shown in pink. Figure 3.25 G proteins are conformational switches whose state is controlled by two opposing enzymes. The GDP- bound conformation of the G protein is inactive, while the GTP-bound conformation is active and able to bind downstream effectors. Both nucleotide exchange and GTP hydrolysis are extremely slow in G proteins on their own. Activation (exchange of GDP for GTP) is speeded up by guanine nucleotide exchange factor (GEf) enzymes, while inactivation is promoted by GTPase- activator protein (GAP) enzymes. The activity of GEfs and GAPs is regulated by signaling inputs. Figure 3.26 The molecular basis of the G protein conformational change. (a) The GDP-bound conformation of a G protein is depicted schematically. (b) in the GTP- bound form, the γ-phosphate group on the bound nucleotide hydrogen-bonds with the main-chain atoms of conserved Thr and Gly residues, leading to conformational rearrangements of the switch i and ii regions of the protein. These rearrangements create an effector-binding site. The amino acid numbering corresponds to the small GTPase Ras. See figure 13.8 for a comparison of the X-ray crystal structures of the GDP- and GTP-bound forms of Ras. (Adapted from G. Petsko and D. Ringe, GTP will have been cleaved to GDP and Pi) ranges from several minutes to more than an hour for different G proteins. In addition, the release of bound GDP after hydrolysis is also very slow—both GDP and GTP bind G proteins with high affinity (Kd in the nM to pM range), and thus their dissociation rates (koff) are necessarily low. This system is very useful for signaling, however, because regulatory enzymes can overcome these kinetic barriers. The nucleotide-binding state of G proteins is controlled by two opposing enzymes: guanine nucleotide exchange factors (GEFs), which activate G proteins, and GTPase-activator proteins (GAPs), which inactivate them. GEFs acti- vate G proteins by catalyzing the release of GDP and the subsequent binding of GTP. GAPs inactivate G proteins by catalyzing the hydrolysis of bound GTP to GDP. Thus, these opposing enzymes form a kinetically controlled “writer/eraser” system analogous to kinase/phosphatase sys- tems described earlier in this chapter. G protein regulation is also similar to phosphorylation because the cycle of regulatory reactions, while under tight kinetic control, is thermody- namically favorable. Inactivation of a G protein is favorable because it is coupled to hydrolysis of the high-energy phosphodiester bond of GTP. Activation of a G protein, via rebinding of GTP, is favorable because the cell provides a constant excess of GTP over GDP (approximately tenfold excess), while the affinities and association rates for GTP or GDP are similar. Thus, the cell’s production of GTP provides the energy to drive G protein signaling, while GEFs and GAPs provide the kinetic controls that harness this energy for regulatory control. The presence of the GTP γ-phosphate determines the structure of G protein switch I and II regions The nucleotides GTP and GDP differ only by the presence of a terminal phosphate on GTP (the three phosphates on GTP are designated as α, β, and γ, starting from the nucleotide ring). Despite its small size, the highly charged γ-phosphate moiety plays a critical role in controlling the con- formation of G proteins (Figure 3.26; see also Figure 13.8). The guanine nucleotide binds in a conserved pocket. The γ-phosphate, when present, interacts with two loops known as switch I and switch II, forming hydrogen bonds with the main-chain atoms of two invariant Gly and Thr residues. These interactions cause significant structural rearrangements in switch I and II. Because these regions form a critical part of the binding site for downstream effectors, these conformational changes dramatically affect the ability of the G protein to bind such effectors. Thus, in a sense, (a) Protein Structure and function. oxford: oxford university Press, 2004.) switch II switch I the G protein functions to convert a single phosphate difference into a large conformational change. In its GTP-bound conformation, a particular G protein often is capable of binding to many different downstream effectors. Thus, its activation can induce a variety of downstream effects through its interaction with different effectors. Although the switch I and II regions are critical for binding to all effectors, the other residues on the surface of the G protein that contribute to the binding site may vary from effector to effector. For this reason, it has been possible to construct specific point mutants of G proteins that bind to some effectors, but not to others. Such mutants can be helpful in teasing out which effectors are important for a particular downstream effect of G protein activation. There are two major classes of signaling G proteins A typical eukaryotic cell contains well over 150 different G proteins, involved in diverse signaling pathways. These G proteins can be divided into several distinct superfamilies, including two families that play key roles in cell signaling. The first family is the small G proteins; these monomeric G proteins are often referred to as small GTPases. The second family is the heterotrimeric G proteins, which have α, β, and γ subu- nits. A third family of G proteins, which we will not discuss here, plays a central role in translation (this family includes the elongation factor EF-tu). All of the G protein superfamilies have at their core a 20 kD G domain, which binds the guanine nucleotide and can adopt alternative conforma- tions depending on whether GDP or GTP is bound. The small G proteins essentially consist of a single G domain, while the heterotrimeric G pro- teins contain a G domain in their Gα subunits (Figure 3.27). Below, we describe the mechanisms of regulation of both classes of G proteins. Subfamilies of small G proteins regulate diverse biological functions The small G proteins consist of a single 20–25 kD domain. The found- ing member of this family is Ras, a central regulator of cell proliferation and differentiation. Ras was first identified as an oncogene (a gene whose disregulated activity can lead to the uncontrolled growth characteristic (a) small G protein heterotrimeric G protein domain Figure 3.27 Structure of G proteins. X-ray crystal structures of (a) Ras, a small G protein, and (b) a heterotrimeric G protein complex. Bound GDP (orange) is shown in ball- and-stick format. The α subunit of the heterotrimeric G protein consists of a G domain (purple) homologous to small G proteins, and an additional helical domain (blue). The β (yellow) and γ (pink) subunits are tightly associated via a coiled-coil interaction. Table 3.1 Small G protein subfamilies: functions, downstream effectors, and upstream GEF and GAP domain K-ras, rap1a rhoa, cdc42, rac1 rab23, rab4a cell proliferation and differentiation cell shape and movement Vesicular traffick- ing and secretion rasgef (cdc25 homology domain)1 rhogef (Dbl ho- mology domain), DocK domain2 mss4 domain, sec2 domain, VsP9 domain rasgaP, rapgaP rhogaP rabgaP arf arf6, arl4 Vesicular traffick- ing and secretion arfgef (sec7 domain) arfgaP ran ran nuclear import rangef rangaP 1 RasGEf (cdc25 homology) domains are often associated with an n-terminal Ras exchanger motif (REm) domain 2 RhoGEf (Dbl homology) domains are often associated with a c-terminal pleckstrin homology (Ph) domain of cancer). There are approximately 150 small G proteins in humans, and these can be further subdivided into at least five major subfamilies: the Ras, Rho, Rab, Arf, and Ran families (Table 3.1). Each of these subfamilies is, in general, involved in regulating distinct cellular functions: Ras pro- teins regulate pathways involved in cell differentiation and proliferation; Rho proteins regulate the cytoskeleton and control cell shape and move- ment; Rab and Arf proteins regulate membrane vesicle-associated proc- esses including vesicle formation, trafficking, and secretion; Ran proteins control nuclear export and import, formation of the nuclear envelope, and mitotic spindle formation. Not only do G protein families differ in the particular set of downstream effector functions that they control, but even within these families there are further functional subdivisions. For example, the Rho family G pro- teins, in their active state, interact uniquely with key cytoskeletal regula- tory proteins. The approximately 25 human Rho subfamily members can be further divided, however, into the RhoA, Rac1, and Cdc42 subclasses. Each subclass is associated with distinct functions in cytoskeletal regula- tion. For example, Rac1-related G proteins are associated with formation of actin-based protrusive structures such as lamellipodia, while RhoA- related G proteins are associated with formation of actin–myosin contrac- tile structures (Figure 3.28). protrusion Many upstream receptors feed into a small set of common heterotrimeric G proteins Heterotrimeric G proteins contain three subunits—α, β, and γ (see Figure 3.27). The 50 kD Gα subunit contains the conserved 20–25 kD G domain that is homologous to the small G proteins. This domain binds GTP or GDP and regulates the conformational change of the protein. The Gα sub- unit also contains an unrelated helical domain. The Gβ and Gγ subunits do not themselves have any enzymatic activity. contraction Figure 3.28 Differential activation of the G proteins Rac and Rho contributes to directed cell motility. Rac activity (green) is concentrated at the leading edge of the cell, where it promotes actin-mediated protrusion. Rho activity (orange) is concentrated at the back of the cell, where it promotes actin–myosin-mediated contraction. activated GPCRs Figure 3.29 The activity cycle of heterotrimeric G proteins. in the basal state, the Gα subunit is bound to GDP and in complex with the Gβγ subunits. Activation of the G-protein-coupled receptor (GPcR) by ligand binding leads to conformational changes that induce its guanine nucleotide exchange factor (GEf) activity, resulting in exchange of GDP for GTP in the Gα subunit. The resulting conformational changes in the Gα subunit lead to its dissociation from the receptor and from the Gβγ subunits. The Gα and Gβγ subunits then are competent to bind downstream effectors. Reversion to the GDP-bound state leads to reassociation of the heterotrimeric complex. GAP, GTPase- activator protein; RGS, regulators of G protein signaling. When the Gα subunit is bound to GDP, it also associates with the Gβ and Gγ subunits to form the heterotrimer. However, when the Gα subunit is in the GTP-bound, or “active,” state, it dissociates from the Gβ and Gα subu- nits (which stay tightly associated to each other) (Figure 3.29). Dissocia- tion of the subunits occurs because the switch I and II regions in the Gα subunit—the regions that undergo the largest nucleotide-dependent con- formational shifts—form part of the heterotrimer binding interface. When dissociated, both the Gα and Gβγ subunits can bind to various downstream effectors (channels and enzymes like adenylyl cyclase), changing their activity and thus leading to biological effects. Heterotrimeric G proteins are activated by G-protein-coupled receptors (GPCRs). In humans, there are many hundreds of distinct G-protein-coupled receptors, making them the most highly represented class of signaling proteins. There are only 16 genes for Gα subunits in humans, which can be divid- ed into four main families: Gsα, Giα, Gq/11α, and G12/13α. Although these Gα subunits have a similar mechanism of activation, they have different effector-binding properties (Table 3.2). For example, the Gsα isoform is primarily responsible for activation of adenylyl cyclase and production of the signaling mediator cAMP. Thus, a large number of diverse upstream receptors feed into a small set of common G proteins. The mechanism by which downstream signaling is directed to specific functional outputs using this limited set of common G proteins is still unclear. However, there is growing evidence that mechanisms such as cell-specific expression of receptors and restricted subcellular localization mediated by scaffold proteins can help to limit and direct the downstream effectors that are targeted by specific receptor–G protein complexes. See Chapter 8 for more on GPCRs Table 3.2 Families of Gα subunits and their effectors Family Subtype Effectors gsα gs(s)α gs(L)α gs(Xl)α golfα adenylyl cyclases ↑ (gs,s(Xl)olfα) maxi K channel ↑ (gsα) src tyrosine kinases (c-src, hck) ↑ (gsα) gtPase of tubulin ↑ (gsα) gi/oα go1α go2α gi gzα gtα ggustα adenylyl cyclase ↓ (gi,o,zα) rap1gaPii-dependent erk/maP kinase activation ↑ (giα) ca2+ channels ↓ (gi,o,zα) K+ channels ↑ (gi,o,zα) gtPase of tubulin ↑ (giα) src tyrosine kinases (c-src, hck) ↑ (giα) rap1gaP ↑ (gzα) grin1-mediated activation of cdc42 ↑ (gi,o,zα) cgmP-PDe ↑ (gtα) ggustα: ? gq/11α gqα g11α g14α g15α Phospholipase cβ isoforms ↑ p63-rhogef ↑ (gq/11α) Bruton’s tyrosine kinase ↑ (gqα) K+ channels ↑ (gqα) tric ↑ (gqα) g12/13α gα₁₂ gα₁₃ Phospholipase D ↑ Phospholipase cε ↑ nhe-1 ↑ inos ↑ e-cadherin-mediated cell adhesion: ↑ p115rhogef ↑ PDz-rhogef ↑ leukemia-associated rhogef (larg) ↑ radixin ↑ Protein phosphatase 5 (PP5) ↑ aKaP110-mediated activation of PKa ↑ hsp90 ↑ Table modified from G. milligan and E. Kostenis, Br. J. Pharmacol. 147(Suppl 1): S46–55, 2006. up arrow indicates activation; down arrow indicates repression. REGulAToRy EnzymES foR G PRoTEin SiGnAlinG As was mentioned above, for nearly all G proteins the intrinsic rates of the nucleotide exchange reaction and the nucleotide hydrolysis reaction are extremely slow. In the cell, the actual rates of these reactions (which change in response to signals within seconds), are kinetically controlled by two opposing classes of enzymes: the guanine nucleotide exchange factors (GEFs) that accelerate the rate of nucleotide exchange, and the GTPase-activator proteins (GAPs) that accelerate the rate of GTP → GDP hydrolysis (see Figure 3.25). Thus, GEFs kinetically control the activation of the G protein, while GAPs kinetically control its deactivation. Upstream signaling inputs can therefore control the amount of active G protein by coordinately regulating the activity of either the GEFs or GAPs. There are diverse GEFs and GAPs, larger in number than the G pro- teins themselves. GEFs and GAPs are themselves highly regulated, and the diversity of these proteins provides a variety of ways to link diverse upstream signaling inputs to the control of a common set of G proteins. We shall see in the following sections the mechanisms by which GEFs and GAPs interact with G proteins and modulate their activity. G-protein-coupled receptors act as GEFs for heterotrimeric G proteins The activation of heterotrimeric G proteins is controlled by G-protein- coupled receptors (GPCRs). GPCRs have a common overall structure with seven membrane-spanning segments, but different GPCRs respond to a very wide range of extracellular inputs. Typically, when a GPCR detects its ligand, it undergoes a conformational change that allows the receptor to act as a GEF and catalyze the exchange of GDP for GTP on the Gα sub- GPCRs (GEFs) ~900 Figure 3.30 RGS proteins (GAPs) ~70 G α subunits ~16 unit (see Figure 3.29). Interestingly, the exchange mechanism requires the Gβ and Gγ subunits. This requirement maximizes the efficiency of G protein activation by preventing the receptor from targeting already- activated (GTP-bound) Gα subunits. Signaling mediated by heterotrimeric G proteins is extremely widespread. GPCRs are found in eukaryotes ranging from yeast to human and are the most numerous type of receptor in the human genome (∼900; nearly 5% of the total number of genes) (Figure 3.30). They transduce diverse signals, including those induced by light, odorants, hormones, lipids, and proteins. Rhodopsin, which is found in the rod and cone cells of the retina, is an example of a GPCR that responds to light. As discussed below, regulators of G protein signaling (RGS) proteins serve as GAPs to counteract the effects of GPCRs. Signal processing by heterotrimeric G proteins. Signal inputs are received by an enormous number of G-protein-coupled receptors (GPcRs) that act as guanine nucleotide exchange factors (GEfs), and by regulators of G protein signaling (RGS) proteins that act as GTPase-activator proteins (GAPs). These inputs are funneled through a relatively small number of Gα subunits. The total number of each class of protein in humans is provided. The visual signaling system is described in detail in Chapter 12 Distinct GEF and GAP domains regulate specific small G protein families The activity of small G proteins is linked to diverse upstream signals by GEF and GAP proteins that act in a coordinate manner to control the lev- els of the active, GTP-bound G protein. The number of GEF and GAP pro- teins exceeds the number of downstream G proteins. For example, there are ∼20 Rho family G proteins but ∼80 GEF and ∼70 GAP proteins that act on them. Presumably, the large number of GEF and GAP proteins func- tions as an adaptor layer that allows diverse upstream inputs to plug into common G-protein-mediated output responses (Figure 3.31). Like many signaling proteins that are involved in transmitting and processing information, these GEFs and GAPs are modular, multidomain proteins. Many of these domains, which differ widely in the individual examples, function as protein or lipid interaction domains that regulate or localize enzymatic function. However, all GEFs and GAPs contain one or more dedicated catalytic domains with intrinsic GEF or GAP activity. There are a relatively small number of types of GEF and GAP domains, with usually one type serving to regulate one subfamily of G proteins. For Figure 3.31 Signal processing by Rho family small G proteins. The activities of a relatively large number of RhoGEfs and RhoGAPs are regulated by signal inputs. These inputs modulate the activity of a somewhat smaller number of Rho family small G proteins. The total number of each class of protein in humans is provided. GEf, guanine nucleotide exchange factor; GAP, GTPase-activator protein. RhoGEFs ~80 Rho family G proteins ~20 RhoGAPs ~70 Figure 3.32 Representative domain architectures of RhoGEFs and RhoGAPs. The core catalytic domains (b l u e for GEf, o r a n g e for GAP) are combined with diverse regulatory domains that determine input control and localization (brown boxes). Some proteins contain both GEf and GAP activities, which may control signaling dynamics by coordinating both activation and inactivation of the G protein. GEf, guanine nucleotide exchange factor; GAP, GTPase-activator protein. RhoGEFs CH RhoGEF PH C1 SH3 SH3 SH3 PH RBD RhoGEF PH FERM RhoGEF PH PH RhoGAPs SH3 WW WW PH RhoGAP BAR PH RhoGAP SH3 RhoGAP/GEF See Chapter 10 for a discussion of modular domains and the architec- ture of signaling proteins Membrane recruitment as an acti- vation mechanism is discussed in more detail in Chapter 5 example, Rho family G proteins generally are activated by GEF proteins that have a Dbl homology (DH) catalytic domain or an unrelated DOCK domain. In contrast, Ras family G proteins are generally activated by GEF proteins containing a Cdc25 homology catalytic domain. Thus, GEFs and GAPs have distinct functional regions: a catalytic region that determines its specific output, and multiple regulatory domains that determine how and by what it is regulated (Figure 3.32). A summary of types of GEF and GAP domains, and what G proteins they act on, is shown in Table 3.1. GEF and GAP proteins show several forms of regulation (Figure 3.33). First, the protein or lipid interaction domains in the protein can medi- ate regulated changes in localization. Second, many of these proteins are autoinhibited by intramolecular interactions, in which other domains in the protein interact with the catalytic domain to inhibit its intrinsic activity. This type of modular regulation can lead to activation when the autoinhibitory interactions are disrupted by upstream inputs such as ligand binding or phosphorylation. Third, certain interactions with the catalytic domains can also allosterically increase the intrinsic GEF or GAP activity. These forms of regulation are observed for the GEF Sos, which, as part of its function, can activate the G protein Ras via its Cdc25 GEF domain (a) Figure 3.33 Three major mechanisms of GEF and GAP regulation. (a) GEf and GAP catalytic domains can be recruited to the sites of their target G protein on the membrane, (b) they can be activated by relief of autoinhibition, or (c) they can be allosterically activated. many GEf and GAP proteins exhibit multiple mechanisms of regulation. Activating ligands are indicated as orange circles and diamonds. (b) (c) (Sos also has a Dbl domain and can act as a RhoGEF as well). Sos can be activated by recruitment to the membrane where it can act on Ras (which is exclusively localized on the membrane). Recruitment is mediated by the adaptor protein Grb2, which binds to tyrosine-phosphorylated proteins on the membrane via its SH2 domain. Domains N- and C-terminal to the Cdc25 domain also act to autoinhibit Sos exchange activity, and these autoinhibitory interactions may be relieved by interaction with upstream ligands like Grb2. Finally, Sos exchange activity is also further stimu- lated by the active state of Ras, via an allosteric binding interaction. This allosteric activation is postulated to contribute to positive feedback in Ras activation. Another example of autoinhibitory regulation is the GEF Epac, which acts on the small G proteins Rap1 and Rap2. Epac is activated by cAMP, which relieves autoinhibition of the catalytic domain mediated by a cyclic nucleotide binding (CNB) domain. GTP GDP GEFs catalyze GDP/GTP exchange by deforming the nucleotide-binding pocket As mentioned above, most G proteins bind both GTP and GDP with high affinity and have half-times for dissociation in the range of minutes to hours. GEFs, in general, act by modifying the structure of the nucleotide- binding pocket, such that the affinity for the nucleotide is significantly decreased (Figure 3.34). It is important to note that the GEF does not favor dissociation of GDP over GTP, nor does it favor rebinding of GTP over GDP. Rather, the unidirectional nature of the G protein cycle (prefer- ential binding of GTP) is driven by the much higher concentration of GTP over GDP maintained in the cell. Several structures of small G proteins bound to GEFs have been solved; these show how various structurally unrelated catalytic GEF domains interact with G proteins (Figure 3.35). Despite their sequence and struc- tural divergence, all the GEF domains bind near or at the nucleotide-binding Figure 3.34 General mechanism of GEF action. Guanine nucleotide exchange factor (GEfs) act to pry apart the nucleotide-binding regions of the G protein, promoting the release of GDP. The GEf binds most tightly to the G protein when it is not bound to nucleotide (essentially the transition state for the nucleotide exchange reaction). Rab-Mss4 GEF Ras-Sos GEF Ran-RCC GEF Figure 3.35 Structures of small G proteins bound to their GEFs. X-ray crystal structures of three G protein–GEf complexes illustrate the variety of structures and mechanisms used by different GEfs to promote release of GDP. in all structures, the G protein ( b l u e ) is shown in the same orientation; GEfs are colored y e ll o w. Regions of the G protein that undergo significant conformational change upon GEf binding are colored in p i n k. GDP (orange ball-and-stick structure) is shown in the position it would occupy in the native G protein (uncomplexed with GEf). (Adapted from J.l. Bos, h. Rehmann and A. Wittinghofer, Cell 129:865–877, 2007. With permission from Elsevier.) (a) (b) Figure 3.36 General mechanism of GAP action. (a) GAPs bind above the nucleotide-binding pocket of the active G protein and promote GTP hydrolysis. (b) many GAPs, including those acting on Ras and Rho, orient a catalytic residue (Gln) in the G protein to polarize the water molecule that attacks the γ-phosphate of GTP. They may also insert an “arginine finger” (Arg) into the active site, which plays a key role in stabilizing the transition state of the substrate. (b, Adapted from J.l. Bos, h. Rehmann and A. Wittinghofer, Cell 129:865–877, 2007. With permission from Elsevier.) pocket of their substrate G proteins. In most cases, some element of the GEF domain acts to pry apart the switch I and II regions of the G protein from the rest of the structure, thus severely deforming the nucleotide- binding pocket. In these structures, the GEF sterically occludes the Mg2+- binding sites that are required for phosphate-group binding, as well as key residues in the switch I or II regions involved in nucleotide bind- ing. Thus, these diverse catalytic domains utilize a convergent molecular mechanism. GAPs order the catalytic machinery for hydrolysis As mentioned above, GAP activity can accelerate the hydrolysis of GTP by G proteins by several orders of magnitude. An efficient hydrolysis reac- tion, in principle, requires a water molecule that is properly oriented, polarized, and occluded from bulk solvent for optimal nucleophilic attack. It also requires stabilization of the negatively charged γ-phosphate in the transition state. In most G proteins, the catalytic residues that perform these functions are either improperly disordered or missing. In the small G protein Ras, the Gln61 residue that interacts with the nucleophilic water is well ordered in the Ras–RasGAP complex, but is disordered in free Ras (Figure 3.36). In addition, a key arginine in RasGAP, referred to as the arginine finger, is inserted into the active site, and appears to sta- bilize the transition state by neutralizing the γ-phosphate. This catalytic function is missing in isolated Ras, and is provided in trans by RasGAP. Interestingly, mutations at the Gln61 position of Ras are frequently found in human tumors. Such mutants are constitutively active because they are unable to efficiently hydrolyze GTP, even in the presence of GAPs. Although different GAPs are very diverse and have unrelated structures (Figure 3.37), they all use similar mechanisms of providing or ordering missing or disordered catalytic elements required for nucleotide hydroly- sis. In some cases, the residues that order the nucleophilic water are, in turn, ordered by the GAP interaction; in other cases, such residues are provided by the GAP itself. In nearly all cases, the arginine finger that stabilizes the transition state is provided in trans by the GAP. Ras-RasGAP Rho-RhoGAP Ran-RanGAP Figure 3.37 Structures of small G proteins bound to their GAPs. X-ray crystal structures of three small G protein–GAP complexes illustrate the structural diversity of GAPs. in all structures, the G protein (blue) is shown in the same orientation; GAPs are colored y e ll o w and GTP is o r a n g e. (Adapted from J.l. Bos, h. Rehmann, and A. Wittinghofer, Cell 129:865–877, 2007. With permission from Elsevier.) Regulators of G protein signaling (RGS) proteins act as GAPs for heterotrimeric G proteins Like small G proteins, Gα subunits have intrinsic but extremely slow GTPase activity. Thus, rapid deactivation of responses requires factors that can catalyze GTP hydrolysis. Regulators of G protein signaling (RGS) proteins function as GAPs, accelerating the rate of GTP hydroly- sis of the activated Gα subunits. More than 20 different RGS proteins have been identified. In several cases, it appears that heterotrimeric G proteins operate within multiprotein complexes containing specific GPCRs, RGS proteins, and downstream effectors, leading to tightly controlled activa- tion, downstream signaling, and deactivation. RGS proteins use a slightly different mechanism to that described above for small G protein GAPs. In this case, the arginine finger exists in the Gα subunits, provided by the helical domain that is linked to the Ras-like G domain. Thus, the heterotrimeric G proteins have a “built-in” arginine finger and RGS proteins are thought to play a role primarily in properly ordering the catalytic residues in the Gα subunit to optimize catalysis. Additional mechanisms are used to fine-tune the activity of G proteins Many G proteins have lipid groups attached at their C-termini that are required for membrane localization and function. This modification is used to provide an additional mechanism for controlling G protein activ- ity, especially for small G proteins of the Rho and Rab families. Guanine nucleotide dissociation inhibitors (GDIs) are proteins that can bind specific G proteins and shield their prenyl groups, thus maintaining the G protein in the cytoplasm. Thus, GDIs lock G proteins in their GDP- bound state and prevent their localization to the membrane. GDI dis- placement factors (GDFs) are enzymes that can promote the release of GDIs from G proteins, thus allowing their localization to the membrane and subsequent activation by GEFs. Thus, GDIs and GDFs provide an additional layer of regulation for prenylated G proteins. The activity of heterotrimeric G proteins is also modulated by proteins containing GoLoco motifs (GoLoco domains). These 19-residue motifs bind to specific Gα subunits, but only in the inactive GDP-bound state. Thus, proteins containing the GoLoco motif act as GDIs that can, in prin- ciple, inhibit G protein activation. Nonetheless, in some cases, GoLoco domains can play positive roles. For example, binding of GoLoco domains to inactivated Gα subunits can prolong activity of dissociated Gβγ subu- nits, because they competitively block reassociation of the heterotrimer. Recently, the GoLoco domain in the protein PINS (partner of inscruta- ble) has been found to modulate heterotrimeric G protein activity in a GPCR-independent manner as part of the machinery that positions the mitotic spindle and attaches microtubules to the cell cortex during cell division. SiGnAlinG EnzymE cAScADES So far we have examined in detail the properties of individual signaling enzymes, such as kinases, phosphatases, GEFs, and GAPs. But in the cell, signaling enzymes do not function individually, but rather are embedded within pathways and networks in which they function as one relay node within a much larger network. These higher-order signal transduction networks are constructed by functionally connecting the individual nodes. This often means that individual signaling enzymes are linked in series Lipid modification and the role of GDIs and GDFs are discussed in more detail in Chapter 5 into cascades, where the output of one enzyme directly or indirectly reg- ulates the activity of the next enzyme. Below, we discuss conserved kinase and G protein cascades and how they are used in signaling. The role of kinase aggregation in signaling across the plasma mem- brane is discussed in Chapter 8 The three-tiered MAP kinase cascade forms a signaling module in all eukaryotes As discussed earlier, many protein kinases require phosphorylation on their activation loops to become active. Thus, it is inherently easy for kinases to be regulated in series, as their activation usually requires the action of some upstream kinase. As a result, it is fairly common to find conserved pathways with kinases acting in series in a cascade. (A number of other kinases are activated by phosphorylation by a different molecule of the same type; for example, after kinase dimerization or aggregation.) Among the most prevalent kinase cascades in all eukaryotes are the mitogen-activated protein (MAP) kinase cascades. This is a pathway module composed of three kinases that act in series, and it is utilized in a remarkably wide variety of cellular responses (Figure 3.38a). The most upstream member is referred to as a MAP kinase kinase kinase (MAPKKK). Activation of a MAPKKK by upstream inputs allows it to phosphorylate and, in turn, activate a MAP kinase kinase (MAPKK) on two Ser/Thr residues in its activation loop. The MAPKK, in turn, activates the most downstream member of the module, MAP kinase (MAPK), by phosphorylating it on its activation loop on specific Ser/Thr and Tyr residues (note that MAPKKs are one of the few members of the Ser/Thr kinase family that can catalyze a Tyr phosphorylation reaction). When activated, MAPKs phosphorylate a large number of downstream targets. In many cases, this involves translocation to the nucleus, where phospho- rylation of specific transcription factors results in changes in gene expres- sion patterns. These three kinases form what is essentially an obligate cascade, because in most cases the only significant substrates for MAPKKKs are MAPKKs, and the only significant substrates for MAPKKs are MAPKs. Different MAPKKKs can be regulated by a variety of specific inputs, and differ- ent MAPKs can phosphorylate many different targets (Figure 3.38b). Thus, because both the input and output of the module are flexible, these three-tiered cascades can be used to mediate a host of different response (a) INPUT (b) INPUTS Figure 3.38 MAP kinase (MAPK) cascades. mAPK cascades consist of three kinases that successively phosphorylate and activate one another. The mAP kinase kinase kinase (mAPKKK), when activated by inputs, can in turn phosphorylate and activate the mAP kinase kinase (mAPKK). The activated mAPKK can phosphorylate the mAPK on two positions in its activation loop, leading to activation of the mAPK. mAPK modules are often activated by a large number of different inputs, and can result in activation of specific programs that involve a large and diverse set of mAPK substrates. OUTPUT OUTPUTS MAPKKK MAPKK MAPK mitogenic signals stress inputs or cytokines Figure 3.39 Examples of different mammalian MAPK modules. The three major families of mAPK cascades in mammals are illustrated. The Erk mAPK family controls cell proliferation programs, while the JnK and p38 families control stress responses (and are often referred to as stress-activated protein kinases or SAPKs). proliferation stress responses behaviors. Cells often contain a number of individual proteins that belong to each of the three kinase families, thus further flexibility results from using different combinations of individual MAPK cascade components. In principle, the phosphatases that dephosphorylate MAPKKKs, MAP- KKs, and MAPKs could be important points of regulation. Relatively lit- tle is known, however, about these reverse reactions and which specific enzymes are involved. In mammalian cells, the canonical MAPK cascade involves the MAPKKK Raf (which is recruited to the membrane and activated by the small G protein Ras), the MAPKK MEK (MAPK/Erk kinase), and the MAPK Erk (Figure 3.39). Some of the most important substrates of Erk are mitogen- ic transcription factors, such as Fos and Jun. The Raf–MEK–Erk module is used to mediate cell responses to various mitogenic signals (explain- ing the origin of the name “MAP kinase”). This module is also used in immune-cell activation, in developmental pathways, and in many other contexts. However, there are a number of other related MAPK modules in mammalian cells, which involve distinct kinase family members and which are used to mediate other types of behaviors including responses to stress or cytokines. These include pathways that activate the JNK (c-Jun N-terminal kinase) and p38 families of MAP kinases (also referred to as stress-activated protein kinases or SAPKs). The number of MAPK cascade components found in different eukaryo- tic organisms is shown in Figure 3.40a. Even a simple eukaryote such as budding yeast contains several distinct MAPK cascades, all sharing the same three-tiered structure, but varying with respect to the specific kinases involved, and the physiological inputs and outputs that are linked by the cascade (Figure 3.40b). There is a particularly large expansion of MAPK components in plants. Scaffold proteins often organize MAPK cascades The large number of related MAPK cascade components, and the poten- tial cross-talk that could occur between these components, raises the issue of how specific signaling responses can be generated in the cell. The very same kinase can play a role in two distinct pathways, in which case the problem of pathway specificity is particularly acute. In many cases, such specificity is accomplished through the use of scaf- fold proteins—proteins that interact with multiple proteins within a pathway, organizing them into a single complex. Not surprisingly, MAPK (b) MAPKKK MAPKK MAPK human H. sap 16 9 14 yeast S. cer 4 3 4 plant thal 60 20 10 MAPKKK MAPKK MAPK pheromone starvation high osmolarity Figure 3.40 mating ilamentous growth osmoresponse Alternative linkages in MAPK cascades. (a) Different species vary in the number of individual mAPK cascade components. in principle, there are many ways in which these components can be linked. H. sap, Homo sapiens; S. cer, Saccharomyces cerevisiae; A. thal, Arabidopsis thaliana. (b) in budding yeast, there are at least three physiologically distinct mAPK cascades, which respond to mating pheromone (mating pathway), nitrogen starvation, and high osmolarity stress. Activation of each cascade results in a different output. however, the three cascades utilize shared components, including the mAPKKK Ste11 and the mAPKK Ste7. in some cases, scaffold proteins organize specific components to provide specificity (see figure 3.41). Figure 3.41 Scaffold proteins physically organize MAPK cascades. The KSR (kinase suppressor of Ras) and JiP (JnK interacting protein) proteins organize the mammalian Erk (proliferation) and JnK (stress response) pathways, respectively. The Ste5 and Pbs2 proteins organize the yeast fus3 (mating) and hog1 (osmotic stress) pathways. cascades were one of the first pathways shown to be organized by scaf- folds. Some of the scaffold proteins that are found in yeast and mamma- lian MAPK pathways are shown in Figure 3.41. Scaffolds are thought to modulate MAPK signaling in multiple ways (Figure 3.42). Most simply, scaffold proteins can promote efficient signal- ing between component proteins in a pathway by increasing their proxim- ity to each other. They are also thought to insulate the components from cross-talk with incorrect but potentially competing partner proteins, by sequestering them into distinct complexes that may be localized in dis- tinct regions of the cell. For example, in yeast, the MAPK pathway that responds to mating pheromone is organized by the scaffold protein Ste5, while the MAPK pathway that responds to high salt (osmotic stress) is organized by the protein Pbs2, which serves as both the scaffold and the MAPKK in this pathway (see Figure 3.41). These scaffolds are thought to control information input and output of the common pathway component, the MAPKKK Ste11. The population of Ste11 that is associated with the Ste5 scaffold is thought to play a role only in mating pheromone signaling, while the population of Ste11 that is associated with the Pbs2 scaffold is thought to play a role only in osmolarity signaling. Thus, Ste11 that is human scaffold proteins yeast scaffold proteins MAPKKK MAPKK MAPK (b) (c) Figure 3.42 Mechanisms by which scaffold proteins control information flow. Scaffolds can increase the effective concentration of two interacting enzymes for one another through tethering and enhanced proximity. (b) Scaffolds can also proximity insulation allosteric modulation activated by one pathway will not cross-activate the other pathway, which is crucial for proper behavior and survival of the cell. There is also evidence that scaffold proteins can control kinase activity allosterically. For example, the mammalian scaffold protein KSR appears to allosterically activate the MAPKKK Raf. Thus, Raf associated with the scaffold is more active than Raf that is not associated with the scaffold. The yeast mating scaffold protein, Ste5, has been shown to allosterically modulate the mating MAPK Fus3 so that only when associated with the scaffold does it become a good substrate for a coassociated Ste7 MAPKK enzyme. In essence, Fus3 has evolved to be a “locked,” inactivatable kinase, but the Ste5 scaffold acts as a specific “key” allowing it to be activated by Ste7. This behavior is thought to be important for signaling specificity, because the yeast starvation response is mediated by “free” Ste7 that is not associated with a scaffold. This lock-and-key control of the mating MAPK Fus3 prevents its misactivation by Ste7 in response to starvation. G protein activity can also be regulated by signaling cascades Although enzymatic cascades are often associated with kinases, cascades composed of other classes of signaling enzymes can also be found. Cas- cades involving G protein regulatory enzymes have been found that con- trol cell morphology and intracellular trafficking. For example, endocytic trafficking pathways are characterized by sequential progression of an endosome through different states, each associated with a distinct Rab small G protein. In some cases, an upstream-activated Rab G protein is found to recruit and/or activate a GEF that then activates a downstream Rab, essentially forming a G-protein-mediated cascade (Figure 3.43a). serve to insulate a protein from alternative partners and substrates. (c) finally, scaffolds can serve as allosteric regulators for pathway members. for example, a pathway protein may be inactive (or inactivatable) unless it is associated with the scaffold, thus preventing its ability to signal when associated with alternative partners. Figure 3.43 G protein signaling cascades. During vesicle trafficking, activation of one G protein (Rab1) leads to the activation of a second G protein (Rab2). in this case, activated Rab1 binds to and activates the Rab2 GEf in a cascade-like manner. (a) INPUT (b) GEF G protein PAK kinase (effector) G protein cascade scaffold protein OUTPUT in addition, activated Rab1 can also lead to a negative feedback loop in which it also recruits and activates the GAP for Rab1. This leads to the inactivation of Rab1. Together, these linkages lead to the temporal sequence of events in which a Rab1-marked vesicle is converted in an all- or-none fashion into a Rab2-marked vesicle. G protein signaling can also be controlled by scaffold proteins. The yeast protein Bem1 organizes a G protein cascade that controls cell polarity. Bem1 can bind the GEf cdc24, its substrate (the G protein cdc42), and the cdc42 effector Ste20 (PAK kinase), thus leading to efficient signaling through the pathway. moreover, binding of Bem1 to sites of activated cdc42 is thought to provide positive feedback by localizing cdc24, and thereby activating more cdc42 at these sites. Recent studies have also shown that, in some cases, an activated Rab G protein will also recruit a GAP that will inactivate the G protein upstream of it in the cascade. This forward GEF cascade interwoven with a back- ward GAP cascade is thought to restrict the overlap between the two Rab proteins, leading to a sharp and irreversible transition from one step to the next along the trafficking pathway. Scaffold proteins have also been identified that mediate small G protein signaling. For example, the yeast protein Bem1 is thought to act as a scaf- fold that binds the GEF enzyme Cdc24, its target (the G protein Cdc42), and the protein kinase Ste20, a downstream effector of activated Cdc42 (Figure 3.43b). SummARy Enzymes have a remarkable ability to catalyze specific biological reac- tions. They greatly increase the rate of thermodynamically favorable reactions, and can promote thermodynamically unfavorable reactions by coupling them to more favorable ones. The activity of signaling enzymes is frequently regulated by allosteric changes in conformation induced by upstream signaling inputs, which allows these enzymes to function as relay nodes to transmit information. The most common inputs that induce conformational change are ligand binding and post-translational modification. Canonical signaling enzymes, such as kinases, are opti- mized to allow coupling between conformational changes and catalytic function. Many cell signaling processes are regulated by opposing enzyme “writ- ers” and “erasers” that switch substrates between active and inactive states. For example, kinases and phosphatases control the addition and removal of phosphate modifications. Phosphorylation is a particularly useful class of information-carrying mark because both the phosphor- ylation and dephosphorylation reactions are thermodynamically favora- ble, but kinetically slow. Thus, the distribution between phosphorylated and dephosphorylated states can be kinetically controlled by the appro- priate enzymes. Another important class of enzymatic “writers” and “erasers” in cell signaling are the GEFs and GAPs, which activate and deactivate G proteins. G proteins can bind either GTP or GDP, but are only active in the GTP-bound state. GEF enzymes activate G proteins by catalyzing the exchange of bound GDP for GTP, while GAP proteins inactivate G proteins by accelerating their intrinsically slow GTPase activity. Signaling enzymes are often organized into cascades in which an upstream enzyme activates a downstream enzyme in series. Such cascades often form evolutionarily conserved modules, such as the three-tiered MAPK cascades used by all eukaryotes to transmit signals. These cascades are often organized by scaffold proteins, and they can be flexibly configured to control the flow of cellular signals in different ways. QuESTionS What aspects of enzyme catalysis are described by the kinetic param- eters kcat, Km, and kcat/Km? The isolated catalytic domains of many sig- naling enzymes often have Km values that are high relative to the substrate concentrations at which they operate in the cell. Why might this be, and how might this lend itself to particular mechanisms for regulating overall catalytic activity? Questions Compare and contrast some of the selective pressures on signaling enzymes with those acting on enzymes involved in metabolism, in par- ticular enzymes involved in high-flux metabolic reactions. If you were to design a new post-translational modification system for signal transmission, what key features would you include? Treatment of a cell with growth factor X leads to tyrosine phosphoryla- tion of a number of target proteins, consistent with a model in which a tyrosine kinase is activated by growth factor stimulation. Interest- ingly, treatment of the cell with vanadate (an inhibitor of phosphoty- rosine phosphatases) also leads to the accumulation of the same set of tyrosine-phosphorylated species. Can you explain why this increase in target protein phosphorylation is observed with vanadate? What are the implications of your model for how kinases and phosphatases operate as a regulatory system, and the speed at which signals will be turned on and off by this system? You are given a purified preparation of protein kinase and test its abil- ity to phosphorylate a peptide substrate at various concentrations of enzyme. You notice that at low concentrations of enzyme, the rate of substrate phosphorylation is initially low but increases substantially over time. However, if you use a higher concentration of kinase, this apparent lag is not seen. How can you explain these results, and how would you test your hypothesis? In mammalian cells, the hydroxyl amino acids (serine, threonine, and tyrosine) are the major sites of regulatory phosphorylation. In bac- teria, histidine phosphorylation predominates in signaling. The free energy of hydrolysis (ΔG) of phosphohistidine is larger than for the phosphorylated hydroxyl amino acids, and the rate of spontaneous hydrolysis is much faster. What differences between eukaryotic and bacterial cells could account for the selective pressure for eukaryotic cells to use phosphorylation of the hydroxyl amino acids instead of histidine for their signaling pathways? Summarize the general mechanisms by which diverse regulatory inputs that act on kinase domains can increase the enzyme’s ability to phosphorylate substrates. Summarize how the gain of a single phosphate group in the GTP- bound state of a G protein (versus the GDP-bound state) can lead to changes in the interactions of the G protein. G proteins are almost always active (and transmit a downstream sig- nal) in their GTP-bound forms and inactive when bound to GDP. In principle, however, it would be possible for the GDP-bound form to be active and the GTP-bound form to be inactive. How might you design a signaling system where the GDP-bound form is active in transmitting a signal? Explain how receptor binding would lead to activation of the G protein, and conversely how the G protein would eventually be down- regulated. In this system, what kind of G protein mutants would be con- stitutively active (that is, active in the absence of receptor stimulation)? Some of the most common mutations associated with cancer are found in the small G protein Ras. Oncogenic mutants of Ras lead to consti- tutive signaling. Suggest several possible mechanisms for why these forms of Ras show constitutive signaling. Conversely, mutations of Ras have been isolated that have dominant negative activity (that is, they block signaling by the normal, endogenous G protein). Suggest possible mechanisms for how these mutants block signaling. REfEREncES PRinciPlES of EnzymE cATAlySiS Fersht A (1999) Structure and Mechanism in Protein Science. New York: WH Freeman and Company. Hunter T (2012) Why nature chose phosphate to modify proteins. Philos. Trans. R. Soc. Lond. B Biol. Sci. 367, 2513–2516. Knowles JR (1980) Enzyme-catalyzed phosphoryl trans- fer reactions. Annu. Rev. Biochem. 49, 877–919. Lassila JK, Zalatan JG & Herschlag D (2011) Biological phosphoryl-transfer reactions: understand- ing mechanism and catalysis. Annu. Rev. Biochem. 80, 669–702. Walsh CT (1978) Enzymatic Reaction Mechanisms. New York: WH Freeman and Company. Westheimer FH (1987) Why nature chose phosphates. Science 235, 1173–1178. Wolfenden R (2006) Degrees of difficulty of water-con- suming reactions in the absence of enzymes. Chem. Rev. 106, 3379–3396. AlloSTERic confoRmATionAl chAnGES Cui Q & Karplus M (2008) Allostery and cooperativity revisited. Protein Sci. 17, 1295–1307. Dyson HJ & Wright PE (2005) Intrinsically unstruc- tured proteins and their functions. Nat. Rev. Mol. Cell Biol. 6, 197–208. Kern D & Zuiderweg ER (2003) The role of dynamics in allosteric regulation. Curr. Opin. Struct. Biol. 13, 748–757. Lim WA (2002) The modular logic of signaling pro- teins: building allosteric switches from simple binding domains. Curr. Opin. Struct. Biol. 12,61–68. PRoTEin PhoSPhoRylATion AS A REGulAToRy mEchAniSm Johnson LN & Lewis RJ (2001) Structural basis for con- trol by phosphorylation. Chem. Rev. 101, 2209–2242. Westheimer FH (1987) Why nature chose phosphates. Science 235, 1173-1178. PRoTEin KinASES Hubbard SR (1997) Crystal structure of the activated insulin receptor tyrosine kinase in complex with peptide substrate and ATP analog. EMBO J. 16, 5572–5581. Huse M & Kuriyan J (2002) The conformational plastic- ity of protein kinases. Cell 109, 275–282. Johnson LN, Noble ME & Owen DJ (1996) Active and inactive protein kinases: structural basis for regulation. Cell 85, 149–158. Kannan N, Taylor S, Zhai Y et al. (2007) Structure and func- tional diversity of the microbial kinome. PLoS Biol. 5: e17. Manning G, Whyte DB, Martinez R et al. (2002) The pro- tein kinase complement of the human genome. Science 298, 1912–1934. Meharena HS, Chang P, Keshwani MM et al. (2013) Deciphering the structural basis of eukaryotic protein kinase regulation. PLoS Biol. 11: e1001680. Reményi A, Good MC & Lim WA (2006) Docking interac- tions in protein kinase and phosphatase networks. Curr. Opin. Struct. Biol. 16, 676–685. Taylor SS, Zhang P, Steichen JM et al. (2013) PKA: les- sons learned after twenty years. Biochim. Biophys. Acta 1834, 1271–1278. PRoTEin PhoSPhATASES Almo SC, Bonanno JB, Sauder JM et al. (2007) Struc- tural genomics of protein phosphatases. J. Struct. Funct. Genomics 8, 121–140. Alonso A, Sasin J, Bottini N et al. (2004) Protein tyrosine phosphatases in the human genome. Cell 117, 699–711. Barford D (1995) Protein phosphatases. Curr. Opin. Struct. Biol. 5, 728–734. Cohen PT (2004) Overview of protein serine/threonine phosphatases. In Protein Phosphatases (Arino J, ed.), pp. 1–20. Berlin: Springer. Fauman EB & Saper MA (1996) Structure and function of the protein tyrosine phosphatases. Trends Biochem. Sci. 21, 413–417. Guan KL & Dixon JE (1991) Evidence for protein- tyrosine-phosphatase catalysis proceeding via a cysteine-phosphate intermediate. J. Biol. Chem. 266, 17026–17030. Karisch R, Fernandez M, Taylor P et al. (2011) Global proteomic assessment of the classical protein-tyrosine phosphatome and “Redoxome”. Cell 146, 826–840. Kennelly PJ (2001) Protein phosphatases—A phyloge- netic perspective. Chem. Rev. 101, 2291–2312. Tonks NK (2006) Protein tyrosine phosphatases: from genes, to function, to disease. Nat. Rev. Mol. Cell Biol. 7, 833–846. G PRoTEin SiGnAlinG Hamm HE & Gilchrist A (1996) Heterotrimeric G pro- teins. Curr. Opin. Cell Biol. 8, 189–196. references Heo WD & Meyer T (2003) Switch-of-function mutants based on morphology classification of Ras superfamily small GTPases. Cell 113, 315–328. Milligan G & Kostenis E (2006) Heterotrimeric G-pro- teins: a short history. Br. J. Pharmacol. 147(Suppl 1), S46–S55. Sprang SR (1997) G protein mechanisms: insights from structural analysis. Annu. Rev. Biochem. 66, 639–678. Venkatakrishnan AJ, Deupi X, Lebon G et al. (2013) Molecular signatures of G-protein-coupled receptors. Nature 494, 185–194. Vetter IR & Wittinghofer A (2001) The guanine nucle- otide-binding switch in three dimensions. Science 294, 1299–1304. REGulAToRy EnzymES foR G PRoTEin SiGnAlinG Bos JL, Rehmann H & Wittinghofer A (2007) GEFs and GAPs: critical elements in the control of small G pro- teins. Cell 129, 865–877. Ross EM & Wilkie TM (2000) GTPase-activating pro- teins for heterotrimeric G proteins: regulators of G pro- tein signaling (RGS) and RGS-like proteins. Annu. Rev. Biochem. 69, 795–827. Rossman KL, Der CJ & Sondek J (2005) GEF means go: turning on RHO GTPases with guanine nucleotide- exchange factors. Nat. Rev. Mol. Cell Biol. 6, 167–180. Schmidt A & Hall A (2002) Guanine nucleotide exchange factors for Rho GTPases: turning on the switch. Genes Dev. 16, 1587–1609. SiGnAlinG EnzymE cAScADES Bhattacharyya RP, Reményi A, Yeh BJ & Lim WA (2006) Domains, motifs, and scaffolds: the role of modular inter- actions in the evolution and wiring of cell signaling cir- cuits. Annu. Rev. Biochem. 75, 655–680. Bose I, Irazoqui JE, Moskow JJ et al. (2001) Assembly of scaffold-mediated complexes containing Cdc42p, the exchange factor Cdc24p, and the effector Cla4p required for cell cycle-regulated phosphorylation of Cdc24p. J. Biol. Chem. 276, 7176–7186. Ferrell JE Jr (1996) Tripping the switch fantastic: how a protein kinase cascade can convert graded inputs into switch-like outputs. Trends Biochem. Sci. 21, 460–466. Good M, Zalatan JG & Lim WA (2011) Scaffold proteins: hubs for controlling the flow of cellular information. Science 332, 680–686. Johnson GL & Lapadat R (2002) Mitogen-activated pro- tein kinase pathways mediated by ERK, JNK, and p38 protein kinases. Science 298, 1911–1912. Rivera-Molina FE, Novick PJ. (2009) A Rab GAP cas- cade defines the boundary between two Rab GTPases on the secretory pathway. Proc Natl Acad Sci USA 106, 14408-13. Taylor & Francis Taylor & Francis Group https://taylorandfrancis.com role of Post-Translational Modifications in signaling Signal transduction requires change in some component of the cell in response to an incoming signal. One of the most commonly used mechanisms for changing the properties of proteins is the covalent modification of their structure. Such changes, collectively termed post- translational modifications (PTMs), range from the addition of small chemical groups as in phosphorylation and methylation, to more substantial additions such as lipid groups or entire polypeptides, or even cleavage of the peptide backbone through proteolysis. These post- translational modifications are catalyzed by specific enzymes and, in many cases, are reversed through the action of other enzymes, so change can be rapid, specific, and tightly regulated. The functional consequence of post-translational modification is to change the activity of the modi- fied protein—its binding or enzymatic activity, localization, or confor- mation. The variety of post-translational modifications possible, and the large number of different sites on a protein that can be modified, expand enormously the possible states for each protein. Given the rapid- ity, variety, and regulatory potential of post-translational modifications, it is no surprise that they are centrally important for virtually all signal transduction mechanisms. THE logic of PosT-TranslaTional rEgulaTion Post-translational modifications enable the regulation of cellular proc- esses by providing a way to change protein properties rapidly. In the absence of post-translational modifications, any major changes in the protein complement of the cell would require new protein syn- thesis, which is energetically costly to the cell. Moreover, the synthe- sis of new protein is a relatively slow process. Furthermore, without Methods to detect and characterize post-translational modifications are discussed in Chapter 13 Figure 4.1 Modification of proteins with small functional groups. structures of commonly modified amino acid side chains and their modified forms are depicted, along with the enzymes that add and remove each modification. Modifications are highlighted in pink. post-translational modification, the structural diversity of proteins would be limited to what could be built from the 20 standard amino acids. Thus, post-translational modification greatly expands the diversity and dynamic behavior of the proteins encoded by the genome (~20,000 protein-encoding genes in humans). Below, we briefly introduce the most common post-translational modifications used in signaling. Proteins can be covalently modified by the addition of simple functional groups A variety of small chemical groups can be transferred enzymatically to the side chains of proteins (Figure 4.1). Such groups can change the surface charge distribution, hydrophobicity, hydrogen-bonding capacity, or conformation of the proteins they modify. One of the most prevalent of these modifications is phosphorylation, the transfer of the terminal phosphate group from ATP to proteins, most commonly to the hydroxyl groups of serine, threonine, or tyrosine side chains. The enzymes that per- form the transfer are protein kinases, and the enzymes that reverse it by removing phosphate are protein phosphatases. NH + NH lysine acetyl transferase lysine N deacetylase H O serine/threonine kinase N N H serine/threonine H phosphatase N H lysine methyl transferase N -acetyl lysine CH3 NH + serine phosphoserine serine/threonine kinase O– lysine 2 lysine demethylase N H O N serine/threonine N O phosphatase H O N -methyl lysine threonine phosphothreonine NH2 CH3 NH OH Tyrosine kinase O O– O– HN NH + arginine HN NH + N Tyrosine N H O phosphatase H O methyl transferase N N H O H O tyrosine phosphotyrosine HO arginine O O– glutamate N -methyl arginine CH3 prolyl hydroxylase O O methyl transferase N N H O glutamate H demethylase proline 4-OH-proline glutamate O -methyl glutamate N -acetylation involves the transfer of the acetyl group from acetyl CoA to the terminal ε-amino group of lysine side chains of target proteins. This modification is often seen on the N-terminal tails of histones, the major protein component of the nucleosomes that package genomic DNA into chromatin; thus acetylation can regulate chromatin structure and gene expression. The reaction is catalyzed by histone acetyl transferase (HAT) and can be reversed by the action of histone deacetylase (HDAC). This nomenclature reflects the prominence of histones as acetylation targets, but given the importance of this modification to a variety of other proteins, the terms lysine acetyl transferase (KAT) and lysine deacetylase (KDAC) are more accurate. The simple single-carbon methyl group can also be transferred to proteins. The targets of N -methylation are the amino groups of lysine and arginine side chains and, like acetylation, this is a common modification of histone tails. N-methyl transferases transfer the methyl group from S-adenosyl methionine (SAM) to the protein; these transferases can be divided into lysine methyl transferases (KMTs) and protein arginine methyl transferases (PRMTs). The sequential addition of multiple methyl groups to the same lysine nitrogen can generate di- and trimethylation. Unlike many other post-translational modifications, N-methylation is very stable, and relatively few lysine demethylases (KDMs) and no arginine demethylases catalyzing the reverse reactions are known. In prokaryotes, O -methylation of glutamate side-chain oxygens is important for transmembrane signaling, for example, in bacterial chemotaxis. Finally, the hydroxyl group can be transferred to several amino acid side chains; proline hydroxylation is important for signaling in a number of contexts, such as the cellular response to hypoxia. Proteins can also be covalently modified by the addition of sugars, lipids, and even proteins Larger organic compounds (carbohydrates and lipids) can also be transferred to proteins. Virtually all proteins exposed to the extracel- lular environment—secreted proteins and the extracellular portions of transmembrane proteins, as well as those residing in the lumen of the endoplasmic reticulum (ER) or Golgi apparatus—are modified by the addition of complex, branched carbohydrate chains. This proc- ess, termed glycosylation, can occur either on the hydroxyl groups of serine or threonine (O-glycosylation) or the amino group of asparagine (N-glycosylation). The enzymes that sequentially add sugar groups and trim the resulting polysaccharides to their final form are confined to the lumenal face of the ER and Golgi. Although glycosylation can pro- foundly affect the folding, trafficking, and ultimately the function of proteins, there are few examples in which changes in glycosylation store or transmit a signal, so these modifications will not be further discussed here. However, it has lately been appreciated that a very different type of glycosylation can play an active role in signaling: the addition of single N-acetyl glucosamine (GlcNAc) sugar groups to serine and threonine residues of cytosolic and nuclear proteins (sometimes termed GlcNAcylation) (Figure 4.2a). Lipids are another important class of molecules that can be added to pro- teins, particularly simple fatty acids such as myristate and palmitate, as well as more complex prenyl groups such as farnesyl and geranylgeranyl. Given the very hydrophobic nature of these lipids, their addition often restricts a target protein to cellular membranes, though there are some examples of lipid-modified proteins that reside in the cytosol (for exam- ple, when the lipid group can be sheltered in a hydrophobic pocket on Figure 4.2 Glycosylation and proline isomerization. (a) addition of a single N-acetyl glucosamine group (glcnac, (a) HO N O-GlcNAc transferase OH HO O HO O NH highlighted in pink) to a serine residue. (b) Proline cis-trans isomerization. Peptidyl H O-GlcNAc hydrolase HN O prolyl cis-trans isomerase (PPiase) catalyzes rotation about the peptide bond (pink) n-terminal to proline. The amino acids n-terminal and c-terminal to proline are depicted in blue and green, respectively. (b) serine GlcNAc-serine O N PPIase N N O N R O R R N trans cis Lipid modifications and their role in protein localization are discussed in greater detail in Chapter 5 the protein surface). In many cases, lipids are added co-translationally (during translation or very soon thereafter) and are stable, and there- fore are not well suited for transmitting signals. Some lipid modifications, however, particularly S -palmitoylation (the addition of the palmitic acid group to the sulfur of cysteine side chains), are more dynamic and may play a more active role in signaling. Entire protein chains can also be covalently linked to another protein. Ubiquitin is a 76-residue protein that is enzymatically added via its C-terminus to lysine side chains in a target protein, generating an iso- peptide bond linking the two. In some cases, long chains of ubiquitin can be generated by sequential addition. Ubiquitylation is an important and widely used signal, often used to target proteins for proteolytic degrada- tion. Several other ubiquitin-like proteins (UBLs), such as SUMO and Nedd8, are also transferred to target proteins by similar mechanisms (termed sumoylation and neddylation). This type of modification can be considered the post-translational addition of a modular protein domain to an existing protein. The cleavage of the protein backbone itself, termed proteolysis, can be con- sidered the most drastic of all post-translational modifications. Proteolytic cleavage of proteins can destroy their activity, or can generate biologically active molecules from larger, inert precursors. The enzymes that mediate proteolysis are termed proteases. The unique role of proteolysis in cellular signaling is the subject of Chapter 9. On the other extreme is the relatively subtle conformational modification of proline residues in protein, termed prolyl cis-trans isomerization. This is not strictly speaking a post-translational modification, as there is no change in covalent structure of the protein; instead, it is a switch in the conformation of the proline residue (rotation around the peptide bond) from the cis to the trans conformation (Figure 4.2b). Spontaneously, this reaction proceeds very slowly, but it can be speeded up greatly through the action of peptidyl prolyl cis-trans isomerase (PPIase). This conforma- tional switch plays a role in some signaling mechanisms. Post-translational modifications can alter protein structure, localization, and stability Of course, for post-translational modifications to be able to transmit infor- mation, they must in some way impact the biological activity of their host proteins. They do this by changing the physical structure of the protein (a) polar serine kinase phosphatase negative charge Figure 4.3 Chemical effects of protein modification. (a) serine and phosphoserine and (b) lysine and N-acetyl lysine are depicted in stick format, with the molecular surface superimposed. Both the phosphate and acetyl groups greatly increase the bulk of the side chain. Electrostatic potential is also indicated on the surface (red = negative, blue = positive potential). Phosphorylation introduces (b) serine phosphoserine strong negative charge, while acetylation decreases positive charge compared to the unmodified side chains. lysine N- acetyl transferase lysine deacetylase lysine N -acetyl lysine (generally the shape, charge, and hydrophobicity of the surface) which, in turn, affects how that protein behaves and interacts with other molecules in the cell. For example, phosphorylation of a hydroxyl amino acid such as serine converts a relatively small, uncharged, moderately hydrophilic side chain into a much bulkier one carrying a strong negative charge. Adding the acetyl group to the amino nitrogen of lysine side chains also increases bulk, but also partially quenches the strong positive charge of the unmodified amino group (Figure 4.3). Such changes at the local level can have profound effects on the more global level of protein function. Some of these effects are introduced below. Post-translational modifications are often tightly coupled to changes in protein conformation (see, for example, Figure 4.4). Such changes may be in the conformation of the polypeptide backbone itself, particularly when the modified residues exist in relatively unstructured loops. Alter- natively, post-translational modification may alter the intramolecular arrangement of the different folded domains of the protein. The result- ing conformational changes again can have diverse and profound second- ary effects on protein activity, such as stimulating or inhibiting catalytic activity, altering the ability to bind to other proteins, or enabling further post-translational modifications by unmasking or burying potential modi- fication sites. Another important consequence of post-translational modification is to alter protein–protein interactions. Modification can either increase or decrease the affinity of a protein for one or more binding partners (that is, can create or destroy a binding site) by virtue of its effects on the charge distribution, hydrogen-bonding possibilities, and shape of the binding change conformation, activity promote protein binding Figure 4.4 Diverse effects of protein modification. (a) Post-translational modification (pink circle) of a substrate protein (light brown) results in conformational changes and activation of the substrate. (b) Modification creates a binding site for another protein (orange). (c) Modification of the substrate prevents the binding of a protein (blue) that binds to the unmodified substrate. (d) Modification results in a change in subcellular localization of the substrate from the cytosol to the nucleus. (e) Modification leads to proteolytic degradation of the substrate. – prevent protein binding change subcellular localization cytosol + M – M ^M nucleus change proteolytic stability – proteolysis Notch signaling is described in more detail in Chapter 5 surface (Figure 4.4a,b). Indeed, as discussed in the next section, the cou- pling of post-translational modification to changes in protein–protein interaction is a very common theme in signaling. In this way, changes in the activity of enzymes that mediate post-translational modifications are converted into changes in the physical interactions between proteins. Post-translational modification can also lead to changes in the inter- action of proteins with other cellular components, for example, nucleic acids or membrane lipids. Changes in protein interaction often lead, in turn, to secondary effects, such as changes in subcellular localization (for example, if the protein is cytosolic and its binding partner is an integral membrane protein) or further changes in post-translational modification (as when a binding partner is itself an enzyme that can modify the first protein). Changes in subcellular localization can also be a direct consequence of post-translational modifications. For example, lipid modification directs many proteins to stably interact with cellular membranes. There are also several examples where transmembrane proteins are proteolytically processed in response to signals, releasing from the membrane an active fragment that can diffuse away and exert its effects elsewhere in the cell, as in the case of Notch signaling (Notch is a transmembrane recep- tor that, upon binding to an extracellular ligand, is cleaved to release an intracellular fragment that translocates to the nucleus and regulates gene transcription). Furthermore, post-translational modification can alter the dynamics of the shuttling of proteins between different sub- cellular compartments (termed protein trafficking). There are many cases where the phosphorylation of nuclear localization or nuclear export sequences of a protein either enhances or blocks its interaction with the normal nuclear import or export machinery, thus altering the distribu- tion of the protein between the nucleus and cytosol. In the same vein, ubiquitylation of plasma membrane proteins often serves as a signal for their internalization (and possible degradation) through the process of endocytosis. Finally, post-translational modifications can alter the proteolytic stabil- ity, and therefore the expression level, of a protein. Some phosphoryla- tion events can target a protein for proteolytic degradation, whereas other modifications can specifically stabilize a protein. As we will learn later in this chapter, phosphorylation is often coupled to a second modification, ubiquitylation, which leads to proteolytic degradation of the ubiquitylated target. Post-translational control machinery often works as part of “writer/eraser/reader” systems A common theme among these diverse post-translational modifications is that they are often controlled by “writer/eraser/reader” systems. In Chapters 1 and 3, we discussed how enzymes that can mark proteins by post-translational modifications can be termed “writers”, while enzymes that can remove these post-translational marks can be termed “erasers”. In some cases, the post-translational modifications directly alter protein structure and activity. However, in many cases, the physical changes in the marked proteins are interpreted indirectly by cytosolic “readers”— proteins that contain modular domains that bind only to proteins contain- ing the appropriate post-translational mark. Examples of such systems are shown in Figure 4.5. It is important to note that in such systems, it is changes in the overall level of post-translational marking that lead to downstream signaling. We are naturally biased to focus on the affirmative act of protein marking by writers, but the removal of preexisting marks by erasers may be at least as important in some contexts. Such modular writer/eraser/reader systems lend themselves to the rapid diversification and adaptation of the signaling machinery in the course of evolution. This is because they provide a ready means to link together two previously unlinked proteins into a new signaling pathway, simply by generating a new site of post-translational modification by point muta- tion, or by adding a modular binding domain through recombination. All of the post-translational modifications described in this chapter are catalyzed by specific enzymes (some covalent modifications, such as the oxidation of certain amino acid side chains, can occur spontaneously, but these are generally not exploited in signal transduction mechanisms). Thus, the state of post-translational modification of a protein necessarily depends on the activity and abundance of the enzymes that perform the modification and those that can remove it. Whether or not a particular site is modified depends on the local concentration of the modifying enzymes, their overall state of activation, and their activity toward that specific site. WRITER: tyrosine kinase READER: SH2 domain OUTPUT ERASER: tyrosine phosphatase OUTPUT: receptor signaling WRITER: histone acetyl transferase READER: bromo domain OUTPUT ERASER: histone deacetylase OUTPUT: increased transcription WRITER: ubiquitin ligase READER: UIM domain OUTPUT ERASER: deubiquitinase OUTPUT: DNA damage response Figure 4.5 Three writer/eraser/reader systems. (a) in signaling by tyrosine kinase growth factor receptors, tyrosine phosphorylation of the activated receptor creates binding sites for effector proteins containing sH2 domains. (b) acetylation of histones by histone acetyl transferase creates binding sites for chromatin modifying factors with bromo domains. (c) ubiquitylation of proteins associated with damaged Dna creates binding sites for signaling and repair complexes containing uiM domains. For a particular reaction to occur (a writer adding a mark, or an eraser removing one), all three requirements must be satisfied: little if any modi- fication will occur if there is no enzyme in the vicinity, if the enzyme is not catalytically active, or if the active enzyme cannot efficiently modify the particular site. Because each of these parameters may be modified by upstream signals, however, post-translational modifications can respond to environmental changes and thus transmit information. Figure 4.6 Multiplicity of modification sites and modification types greatly increases the number of possible states for a protein. a number of different scenarios are presented; proteins are depicted as vertical lines and modifications as colored circles. (a) a single modification is possible (two possible states: modified and unmodified) at three possible sites. (b) The number of possible sites is increased to five. Two different modifications are possible at the same site (e.g., lysine acetylation and lysine methylation), generating three possible states per site. (d) Three different modifications are possible at each site (e.g., acetylation, methylation, and ubiquitylation), generating four possible states per site. Post-translational modifications allow very rapid signaling and transmission of spatial information Because post-translational modification is mediated by enzymes, it can be very rapid and lead to massive amplification of a signal. Enzymes can be extremely efficient catalysts, with a single enzyme molecule being capa- ble of modifying many substrate molecules per second. Thus, only a few activated enzyme molecules can exert an enormous effect in a very short period of time. These properties are useful for signaling mechanisms that need to operate on relatively short time scales (seconds to minutes). By contrast, changes in transcription leading to synthesis of new proteins typically require much longer times to exert their effects (minutes to hours). The only signaling mechanisms with the potential for more rapid response are those driven by changes in membrane potential and ion flow, used extensively in the nervous system where speed is essential. The enzymes responsible for post-translational modification are often activated in specific places in the cell and they and their products have finite diffusion rates, and so such systems can be used to transmit spatial information. These systems can be used to detect where an input signal originates, and can control resulting spatial responses, including complex changes in cellular organization or morphology. This type of information cannot be propagated by transcriptional regulatory systems. inTErPlay BETwEEn PosT-TranslaTional MoDificaTions Many proteins can be modified at many different sites and in a vari- ety of ways. An important consequence of this diversity of modification is to greatly increase the number of potential molecular states of a pro- tein (Figure 4.6). To take a simple example, if a protein has ten possible phosphorylation sites, and each site can be phosphorylated or dephospho- rylated independently, 210 (1024) distinct phosphorylated states are pos- sible. When other potential modifications are considered, the number of (a) (b) (c) (d) 3 sites 1 modification type/site 5 sites 1 modification type/site 3 sites modification types/site sites 3 modification types/site 23 = 8 possible states 25 = 32 possible states 33 = 27 possible states 43 = 64 possible states possible states for a single protein can become virtually infinite. In prin- ciple, each of these different states can have a distinct biological activity. Thus, the genomic coding potential is expanded enormously by the abil- ity of each protein to be modified in various ways after translation. This combinatorial complexity also creates problems in characterizing the properties of proteins, because in most cases it is not easy to physically separate each of the modified states for analysis, so one can only assay the average properties of the ensemble of modified states. In this section, we will briefly consider some of the ways that individual post-translational modifications can interact with each other to regulate protein activities. We will then delve more deeply into a specific example, p53, which uses a host of post-translational modifications to integrate sig- nals from throughout the cell and thereby regulate cell-cycle progression and programmed cell death. A post-translational modification can promote or antagonize other modifications There are a number of examples where one post-translational modifica- tion can directly affect either the likelihood or nature of additional modi- fications. Such interrelationships can be important in allowing proteins to integrate and process multiple upstream signals and perform logical operations. Since many post-translational modifications target the same amino acid residues, a single residue can be modified in different, mutually exclusive ways. Competition for sites by different modifying enzymes can result in a switch between two (or more) different states of the modified protein. Lysine residues are commonly used in this way; the terminal amino group can be subject to acetylation, methylation, ubiquitylation, and modifica- tion by other ubiquitin-like groups (Figure 4.7). Similarly, the same ser- ine and threonine residues can be subject to either phosphorylation or to N-GlcNAcylation. An example of the latter is provided by the transcrip- tion factor Myc, an important mediator of cell growth and proliferation. The balance between phosphorylation and N-GlcNAcylation of certain serine residues can be regulated by upstream signals, and the different (a) lysine lysine acetyl transferase lysine demethylase lysine deacetylase Lys methyl transferase ubiquitin ligase deubiquitinase (b) O-GlcNAc hydrolase O-GlcNAc transferase S/T kinase S/T phosphatase Figure 4.7 Switching between distinct states of post-translational modification. lysine residues can be acetylated, methylated, or ubiquitylated. switching between each state requires removal of the first modification before addition of the second. (b) serine residues can be either phosphorylated [by a serine/threonine (s/T) kinase] or glcnacylated. (b) READER: F-box of SCF WRITER: cyclin-dependent kinase WRITER: SCF ubiquitin ligase READER: UBD of proteasome F-box ubiquitin ligase READS: WRITES: P SCF P UU OUTPUT SCF: a combined reader/writer ERASER: phosphatase ERASER: deubiquitinase OUTPUT: proteolysis Figure 4.8 Linked writer/eraser/reader systems. During the cell cycle, proteins phosphorylated by cyclin-dependent kinases are targeted for destruction in the proteasome. (a) The scf ubiquitin ligase complex acts as both a reader (of phosphorylation) and a writer (of ubiquitylation). (b) Phosphorylation of a substrate by cyclin-dependent kinase leads to binding of scf via its f-box domain. scf polyubiquitylates the substrate, which is then recognized by ubiquitin-binding domains (uBDs) of the proteasome. Proteasome binding results in the proteolytic destruction of the protein. Cyclin-dependent kinases and cell-cycle control are discussed in more detail in Chapter 12 Figure 4.9 Regulation of histone methylation by phosphorylation. under basal conditions, the lysine demethylase lsD1 binds to the n-terminal tail of histone H3 and removes methyl groups from lys4. upon androgen stimulation, protein kinase c-β₁ (PKcβ₁) phosphorylates Thr6, preventing the binding of lsD1 and leading to increased methylation of lys4. modified forms have very different activities and different functional consequences. Another way in which modifications can affect each other is when one type of modification is required for subsequent modifications of another type. An excellent example of this is provided by cyclin-dependent kinases (CDKs), which regulate cell-cycle transitions. CDKs phosphorylate specific protein targets on serine and threonine residues. These phosphorylated sites then serve as binding sites for a large ubiquitin ligase complex which polyubiquitylates the phosphorylated protein, thereby targeting it for destruction by the proteasome (Figure 4.8). In this pathway, three distinct post-translational modifications (phos- phorylation, ubiquitylation, proteolyis) are functionally linked—one modification promotes the next by making the modified protein a bet- ter substrate for the next modifying enzyme. In effect, multiple writer/ eraser/reader systems are linked together. The ultimate result, in the case of cyclin-dependent kinases, is to convert the reversible activation of a kinase into the essentially irreversible destruction of the substrates of that kinase. Different modifications can also be mutually antagonistic—the presence of one modification can prevent other modifications in the same protein or associated proteins. One type of modification may block interaction with other modifiers by decreasing the affinity of their interaction, or modifica- tion may make a protein a poorer substrate for the second enzyme. This is seen in the case of modification of histone tails, which regulate chro- matin structure and thus the ability of DNA to be transcribed (discussed in more detail below). For example, a threonine residue in histone H3 is phosphorylated by a protein kinase (specifically, protein kinase C-β₁) after androgen hormone stimulation. This post-translational modification blocks the ability of a lysine demethylase, LSD1, to remove methyl groups from a nearby lysine residue. The resulting increase in histone methyla- tion at this position is important for the induction of transcription by the hormone (Figure 4.9). phosphorylation by PKCβ₁ histone H3 p53 is tightly regulated by a wide variety of post-translational modifications A particularly well-characterized illustration of the diversity of post- translational modifications is provided by p53, a master regulator of cel- lular responses to a wide range of environmental stresses such as DNA damage. Depending on the specific stress and cellular context, p53 can induce momentary cell-cycle arrest while the cell attempts to repair any damage, or permanent cell-cycle arrest and apoptosis (a form of programmed cell death). Thus, p53 prevents cells from replicating inap- propriately and passing on damaged genomic DNA, earning it the nick- name “guardian of the genome.” It is the most commonly mutated gene in human cancers, underscoring its central importance as a check on inap- propriate cell proliferation (such genes that antagonize cell proliferation and survival pathway and promote tumari genesis when mutated are termed tumor suppressors). Not surprisingly, given the importance of its task and the potentially fate- ful consequences of its activation, p53 is very tightly regulated by a wide variety of inputs. The primary way in which environmental conditions communicate to p53 is through a vast array of post-translational modifi- cations, which include phosphorylation, acetylation, methylation, ubiquit- ylation, sumoylation, and neddylation (Figure 4.10a). Roughly 10% of the amino acids of p53 have been shown to be subject to at least one form of post-translational modification, and many can be modified in more than one way. Further regulation of p53 occurs through its interactions with over a hundred binding partners and by its subcellular localization but, as we shall see, these too are primarily controlled by post-translational modifications. Apoptosis is discussed in more detail in Chapter 9 Figure 4.10 Regulation of p53 by post- translational modification. (a) The domain structure of p53 is depicted to scale, along with known sites of post- translational modification. for p53 domains: Ta, transcriptional transactivation; Pro, proline-rich domain; nls, nuclear localization signal; Tet, tetramerization domain; nEs, nuclear export signal; reg, regulatory domain. (b) Mechanism of activation of p53. Top: in its basal, unactivated state, p53 is bound to Mdm2, which polyubiquitylates it and targets it for destruction by the proteasome. cell stresses such as Dna damage lead to phosphorylation of p53, loss of Mdm2 binding, and recruitment of acetyl transferases such as p300/cBP. The resulting acetylation leads to recruitment of transcriptional activators. p53 is stabilized, tetramerizes, and binds specific Dna sequences to induce transcription of p53-responsive genes. (a, adapted from K.a. Boehme and c. Blattner, Crit. Rev. Biochem. Mol. Biol. 44:367–392, 2009. with permission from informa Healthcare.) (a) TA Pro DNA binding NLS Tet NES Reg phosphorylation acetylation methylation ubiquitin SUMO Nedd (b) cell stresses p53 phosphorylation proteasomal dedradation basal conditions activators activated conditions The level and activity of p53 are regulated by ubiquitylation and acetylation When active, p53 binds to specific DNA sites and induces the transcrip- tion of genes that regulate the cell cycle, DNA repair, apoptosis, and other activities. In unstressed cells, however, the levels of p53 are very low. This is mostly due to the polyubiquitylation of p53 on a number of C-terminal lysine residues by a ubiquitin ligase, Mdm2 (Figure 4.10b). As we will discuss below and in Chapter 9, polyubiquitylated proteins are rapidly degraded by a specialized proteolytic structure, the protea- some. Also associated with p53 are deubiquitinases like HAUSP, which can remove ubiquitin chains and thereby stabilize p53. In addition to polyubiquitylation, p53 can be monoubiquitylated at low levels of Mdm2. This modification leads to the export of p53 to the cytosol, where it can regulate apoptosis and another form of cellular self-destruction, autophagy. Clearly the extent of ubiquitylation is critical for regulating the stabil- ity of p53 and thus determining its overall concentration in the cell. The ubiquitylation of p53 is, in turn, regulated by phosphorylation. A variety of protein kinases, such as ATM, ATR, and Chk1/2 (which are activated by DNA damage), can phosphorylate p53 and render it less susceptible to polyubiquitylation by Mdm2 and other ubiquitin ligases. This is an exam- ple of one type of post-translational modification (phosphorylation) nega- tively regulating another modification (ubiquitylation). The acetylation of a number of lysine residues on p53 dramatically enhances the recruitment of transcriptional coactivators, thereby stimulating transcription of specific p53-dependent promoters. The extent of acetylation is thought to determine, at least in part, which specific promoters are activated by p53 (that is, those that promote cell-cycle arrest, or those that promote apoptosis). The primary acetyl transferase for p53 is termed the p300/CBP complex, though others exist. Importantly, the most prominent sites of acetylation are the same lysine residues that are polyubiquitylated by Mdm2, so their modification (ubiquitylation versus acetylation) serves as a switch to toggle between two distinct states: unstable and transcriptionally inactive versus stable and transcriptionally active. This is an example of two alternative modifications of the same residues resulting in very different functional outputs. Additional modifications further fine-tune p53 activity In addition to phosphorylation, ubiquitylation, and acetylation, other post-translational modifications contribute to p53 regulation (see Figure 4.10a). Many of these target the same lysine residues that are subject to ubiquitylation and acetylation, thus adding to the rich com- plexity of p53 regulation. For example, methylation of lysine and arginine residues of p53 can either enhance or suppress transcriptional activities on particular promoters, most likely through regulating interactions with other specific transcriptional coactivators. Transfer of the ubiquitin- like proteins SUMO and Nedd8 to lysine residues can also regulate the transcriptional activity of p53, its subcellular localization, or both. Clearly, the number of lysines on p53 subject to modification, and the diverse modifications that are possible at each, have the potential to generate an enormous number of different activity states for this protein. An additional layer of complexity is provided by a dense network of regula- tory relationships connecting various p53-modifying enzymes. For exam- ple, activation of p53 leads to increased expression of Mdm2 which, in turn, polyubiquitylates p53 and down-regulates its activity (an example of a negative feedback loop). In a second example, ubiquitin ligases and acetyl transferases not only compete to modify the same lysine residues on p53, but can also modify each other and thereby either inhibit or stim- ulate their respective activities. The rich diversity of post-translational modifications of p53 greatly expands the ways in which its abundance, activity, and localization can be regulated. This not only enables p53 to respond to and integrate diverse cellular inputs, but also allows it to have diverse output activities that can be finely tuned to the needs of the system. ProTEin PHosPHorylaTion In eukaryotic cells, protein phosphorylation is a very common modifica- tion. It is estimated that more than a third of all human gene products are phosphorylated, and this fraction is almost certain to grow as more sen- sitive methods are used to detect phosphorylation. The hydroxyl amino acids (serine, threonine, and tyrosine) are by far the most common targets of phosphorylation in eukaryotes. In Chapter 3, we discussed some of the properties of the phosphate group that make it particularly useful for sig- naling: in short, phosphorylation allows the cell to use a readily available raw material (ATP) to induce stable and significant alterations in protein structure and function. Although the phosphoester linkage is relatively stable to spontaneous hydrolysis, in the cell its reactions can be rapidly catalyzed through the opposing actions of protein phosphatases and pro- tein kinases. The importance of phosphorylation in regulating signaling pathways is suggested by the increasing importance of kinase inhibitors in the treat- ment of human disease. Since all protein kinases use ATP as a substrate of the phosphoryl transfer reaction, ATP analogs (small molecules that mimic ATP, but cannot be used for phosphate transfer) are useful as pro- tein kinase inhibitors. Many of these compounds are now used clinically as drugs, for example, to inhibit kinases that cause cancer. Phosphorylation is often coupled with protein interactions In addition to its direct effects on protein structure, phosphorylation has another important role in signaling: it can dramatically affect the interaction of a protein with other proteins in the cell. As already noted, both serine/threonine and tyrosine phosphorylation lie at the heart of writer/eraser/reader systems, in which proteins containing modular phosphorylation-specific binding domains “read” changes in phospho- rylation by binding specifically to certain proteins only after they are phosphorylated. This type of system was first appreciated in signaling by receptors with tyrosine kinase activity. When cells were stimulated with ligands for such receptors, in many cases the most abundantly phosphorylated substrate was found to be the receptor itself. This rather puzzling observation raised the question of how the signal was transmitted, in the absence of significant phosphorylation of downstream substrates. The discovery of a modular domain that binds specifically to peptides in the tyrosine- phosphorylated state (the SH2 domain) provided a solution to the puzzle: autophosphorylation of the receptor led to the recruitment of SH2- containing proteins from the cytosol to the receptor on the membrane. This change in localization brought SH2-containing enzymes into close proximity with their substrates on the membrane, thereby increasing their activity (Figure 4.11a). How phosphorylation can allos- terically alter protein structure is discussed in Chapter 3 Signaling by receptor tyrosine kinases is discussed in more detail in Chapter 8 (a) ligand (b) ligand Figure 4.11 Receptor tyrosine kinases create phospho-binding sites. (a) Binding of ligand induces dimerization of the receptor, activation of the catalytic domain, and tyrosine autophosphorylation on a number of sites. These sites serve to recruit effector proteins with phosphotyrosine-binding domains such as sH2 domains. The identity of the effectors bound depends on the sites that are phosphorylated and the effectors expressed in the cell. (b) some receptor tyrosine kinases do not autophosphorylate extensively, but instead phosphorylate scaffold proteins, which then serve to recruit downstream effectors. a number of common effectors are depicted: src, src family nonreceptor tyrosine kinase; Pi3K, phosphatidylinositol 3-kinase; Plcγ, phospholipase cγ; shp2, tyrosine phosphatase shp2. gaP is a ras gTPase activator protein, grb2 is an adaptor protein, and sos is a ras guanine nucleotide exchange factor. The SH2 domain is the founding member of the family of phosphospecific binding modules, of which more than ten are now known. The SH2 domain is the major binding partner for tyrosine-phosphorylated sites, though a few other phosphotyrosine-binding domains have been identified. In the case of phosphoserine and phosphothreonine, a larger number of modu- lar binding domains are known, likely reflecting the earlier evolutionary origin and higher overall prevalence of serine/threonine phosphorylation compared to tyrosine phosphorylation. Phosphospecific binding domains have a positively charged pocket that interacts directly with the phosphate group, and adjacent surfaces that interact with amino acids surrounding the phosphorylated site. Rough- ly half of the binding energy is provided by the phosphate group, and half by the surrounding residues. In this way, at normal physiological concentrations of binding domain and substrate, only those substrates that are phosphorylated and which also contain favorable residues in the vicinity can bind to a significant extent; the affinity of the domain for the unphosphorylated site, or to phosphate group alone, is too weak to support binding. In the case of tyrosine phosphorylation, the receptor itself is often phos- phorylated on many different sites, each of which is recognized by a subset of SH2-domain-containing proteins, providing the potential for assembly of a large, multicomponent signaling structure nucleated by the phosphorylated receptor (Figure 4.11a). In other cases, the receptor phos- phorylates an intracellular scaffold protein on many sites, which can then recruit many distinct effectors (Figure 4.11b). For example, in signaling by the insulin receptor, the scaffold IRS1 (insulin receptor substrate 1) is phosphorylated by the activated receptor on around ten different tyro- sine residues, and serves as a platform for assembly of a downstream signaling complex consisting of proteins with phosphotyrosine-binding domains. IRS1 itself contains a PTB domain, another phosphotyrosine- binding domain, which presumably helps recruit the scaffold to the acti- vated (autophosphorylated) receptor. The domains that bind phosphoserine/phosphothreonine, and the con- sequences of this binding, are much more varied than in the case of phosphotyrosine. A number of domain families including FHA, WW, BRCT, Polo-box, MH2, and WD40 have members that can bind phosphothreonine- or phosphoserine-containing motifs. There is also a family of small proteins, termed the 14-3-3 proteins, which specifically bind to proteins phosphorylated on serine or threonine. Unlike other modular phosphoprotein-binding domains, 14-3-3 proteins do not contain any other functional domains, though they do form homo- or heterodimers. 14-3-3 dimers either interact with multiple phosphorylated sites on the same phosphoprotein, or with two different phosphoproteins. They reg- ulate activity upon binding either by sterically blocking access to other binding partners, or by inducing conformational changes. Kinases and phosphatases vary in their substrate specificity In every cell, there are many hundreds of thousands of different potential phosphorylation sites (serine, threonine, or tyrosine residues) located on the surface of proteins. This poses a problem, as the specificity of signal output may depend on the phosphorylation of one or a few sites within this vast excess of irrelevant sites. As we have seen, kinases and phosphatases select the substrates they act upon at a number of levels (see Figure 3.17). For example, the catalytic cleft itself can discriminate between different potential phosphorylation sites depending on the amino acids flanking the potential phosphorylation site. Kinases vary consider- ably in their intrinsic specificity, however, and some (for example, most tyrosine kinases) are quite promiscuous in the variety of peptides that can be efficiently phosphorylated. A second level of specificity is provided by contacts with the substrate outside of the catalytic cleft, including other regions of the kinase itself, or proteins associated with the kinase. Substrate specificity can be further enhanced by association with scaffold proteins. Scaffolds bind a kinase (a) Figure 4.12 specific phosphoprotein- binding domain nonspecific phosphatase (b) kinase specificity binding specificity likelihood a phosphosite will be bound Protein binding domains can contribute to apparent kinase specificity. (a) a variety of proteins phosphorylated on a variety of different sites is depicted. under conditions of high nonspecific phosphatase activity, most sites will be rapidly dephosphorylated. sites that are bound by specific phosphoprotein-binding domains, however, may be protected from phosphatase activity and thus come to predominate in the population. (b) The likelihood that a site will be bound by a phosphoprotein-binding domain (and thus be protected from dephosphorylation) depends both on kinase specificity (the likelihood that the site will be phosphorylated) and the specificity of the phosphoprotein-binding domain (the likelihood that a site, if phosphorylated, will bind). See Chapter 3 for a more extensive discussion of scaffolds (or other enzyme) and its potential substrates, aligned like peas in a pod; thus substrates are presented at high concentration and at optimal orientation for phosphorylation by the kinase. Scaffolds can also anchor proteins to specific subcellular locations, thereby altering the potential substrates that might be encountered by the kinase. Phosphospecific binding modules such as 14-3-3 proteins and SH2 domains can also contribute, albeit indirectly, to specificity in the cell. Not only do such proteins play a critical role as readers of the presence or absence of phosphorylation, but they have the potential to protect sites from dephosphorylation. The constitutive rates of protein dephosphor- ylation in the cell can be very high, so that any sites that are not pro- tected in some way will rapidly be dephosphorylated. In such a setting, a relatively nonspecific protein kinase might phosphorylate many sites, most of which are irrelevant, and only those sites that successfully bind to reader proteins will be protected and thus persist (Figure 4.12). Thus the functional relationship between readers, writers, and erasers may allow some of the burden for substrate specificity to be borne by the bind- ing proteins. Multiple phosphorylation of proteins can arise by different mechanisms When more than one site is phosphorylated on the same protein, there are a number of ways in which these phosphorylation events can depend on each other. For instance, phosphorylation by one kinase can make a protein a better (or worse) substrate for a second kinase that phosphor- ylates different sites. This is seen in glycogen synthase kinase 3 (GSK-3), which is a key serine/threonine kinase in a number of signaling path- ways. Typically, GSK-3 only recognizes and efficiently phosphorylates substrates after they have first been primed by phosphorylation by another kinase, such as CK1 or CK2 (Figure 4.13a). Crystal structures revealed that the substrate-binding cleft of GSK-3 contains a positively charged pocket that recognizes the phosphate group of the primed sub- strate, promoting substrate binding and subsequent phosphorylation of the second site. When a substrate can be phosphorylated at multiple sites by the same kinase, this can occur either by a distributive or processive priming processive phosphorylation distributive phosphorylation Figure 4.13 Diverse modes of multiple phosphorylation. (a) Priming: a kinase phosphorylates a substrate efficiently only after the substrate has been previously phosphorylated on a different site, often by a different kinase, creating a structure that fits a binding pocket on the kinase. (b) Processive phosphorylation: a kinase binds to a substrate and phosphorylates multiple sites on that substrate before dissociating. (c) Distributive phosphorylation: a kinase phosphorylates only one site before dissociating from the substrate. Multiple phosphorylation requires multiple rounds of binding, catalysis, and dissociation. mechanism (Figure 4.13b,c). In distributive phosphorylation, each site is phosphorylated independently—for each site, the kinase binds, transfers phosphate to the substrate, and then dissociates. By contrast, in proces- sive phosphorylation, the kinase remains associated with the substrate and phosphorylates multiple sites sequentially. In this latter case, the unphosphorylated and highly phosphorylated states are much more prev- alent than states of phosphorylation at just one or a few sites. Priming, distributive phosphorylation, and processive phosphorylation all can lead to phosphorylation of multiple sites, but differ significantly in how the final state of phosphorylation depends on the concentration and activity of the kinases responsible. For example, requiring multiple distributive modifications for an enzyme to be active is one way to generate switchlike activation. Histidine and other amino acids can be phosphorylated, especially in prokaryotes In multicellular organisms, phosphorylation of serine, threonine, and tyrosine accounts for almost all protein phosphorylation. However, other amino acid side chains can also be phosphorylated, including histidine, Switchlike activation and ultrasen- sitivity are described in Chapter 11 Figure 4.14 Two-component signaling systems. a histidine kinase (HK) responds to input signals by becoming activated and autophosphorylating on a histidine residue. This phosphate is then transferred to an aspartate on a response regulator (rr) protein, inducing conformational changes leading to signal output. Many rr proteins bind to genomic Dna and regulate transcription upon phosphorylation. INPUT arginine, and aspartate, and such modifications are important for signal- ing in prokaryotes and some eukaryotes. Both histidine and aspartate phosphorylation are key elements of a com- mon and highly conserved prokaryotic signaling mechanism—the two- component system. In bacteria, two-component systems are the most common means of transducing information from outside the cell to the interior. In their most typical form, they consist of two proteins: a his- tidine kinase and a response regulator (RR). The histidine kinase transfers the γ-phosphate of ATP to one of its own histidine residues (an autophosphorylation). This phosphate is then rapidly transferred to the carboxyl group of an aspartate side chain on the RR. Phosphorylation induces conformational changes in the RR domain that lead to down- stream effects (Figure 4.14). In many bacterial two-component systems, the phosphorylated RR is a transcriptional activator that binds DNA and regulates gene transcription. A particularly well-studied two-component system regulates bacterial chemotaxis (the ability of bacteria to swim toward or away from envi- ronmental cues). In this case, chemotactic receptors on the cell surface are associated with a histidine kinase (CheA). When the receptors bind a noxious chemical (chemorepellant), CheA transfers its phosphate to a RR (CheY). Phosphorylated CheY binds to the flagellar motor and regulates its activity by changing the direction of flagellar rotation which, in turn, changes the direction of swimming by inducing tumbling (see Chapter 11; Figure 11.23). Several aspects of the two-component system make it distinct from the phosphorylation of hydroxyl amino acids more common in eukaryotes. The biochemical nature of the phosphate linkages are different from the phosphoester linkage in hydroxyl amino acids—a phosphoramidate in the case of histidine phosphorylation, an acyl phosphate in the case of aspar- tate phosphorylation (Figure 4.15). These two linkages are very high- energy bonds, which are kinetically much more susceptible to hydrolysis than the phosphoesters under physiological conditions. In practice, this means that phosphohistidine and phosphoaspartate are much more tran- sient, with half-lives of minutes in most cases. In addition to the rapid –O O– N ⁺ O– N H N N N –O O– O N Figure 4.15 Phosphohistidine and phosphoaspartate. structures of histidine, aspartate, and their phosphorylated derivatives are shown. attack of phosphohistidine by aspartate during the phosphotransfer reaction H O H O H O H O is indicated by the gray dotted arrow. histidine phosphohistidine aspartate phosphoaspartate Phosphate groups are highlighted in pink. spontaneous hydrolysis, both the histidine kinase and RR domains have intrinsic phosphatase activity for the RR domain, ensuring that any sig- nal is relatively transient. Most bacterial cells contain a relatively large set of histidine kinases linked to receptors, and RRs linked to effector domains. This raises the question of whether histidine kinase and RR domains couple in exclusive pairs, or whether they have overlapping sets of interactions. A study of the bacterium Caulobacter crescentus, which encodes 62 histidine kinases and 44 RRs, found that most histidine kinases were quite specific in transferring phosphate to only one or a few RRs at physi- ological concentrations and time scales. Greater promiscuity was seen in in vitro experiments done at high concentrations or for an extended time. Thus, specificity is ensured and unwanted cross-talk is avoided at the kinetic level; only a small fraction of possible interactions occur rapidly enough at the relatively low kinase and RR concentrations seen in the cell for phosphotransfer to occur before dephosphorylation of the histidine kinase. Two-component systems and histidine phosphorylation are also present in eukaryotes Two-component systems have also been found in plants, slime molds, and fungi, with the RRs feeding into typical eukaryotic signaling pathways such as those involving MAP kinases or cAMP. However, two-component systems are not found in multicellular animals (metazoans). The enzymes responsible for histidine autophosphorylation, phosphotransfer to the RR, and dephosphorylation bear no resemblance to the kinases and phos- phatases that modify hydroxyl amino acids in eukaryotes, so this mode of signaling is apparently evolutionarily distinct. One hypothesis for why two-component systems were lost in metazoans (and replaced with kinases and phosphatases that modify hydroxyl amino acids) is that the lability of phosphohistidine and phosphoaspartate made it difficult to transmit signals reliably in larger cells. In a very small bacterial cell, the distance between the cell membrane and the target (for example, chro- mosomal DNA) is short, so spontaneous dephosphorylation is unlikely before the signal can be transmitted. These distances become consider- ably larger in most eukaryotic cells, giving more time for the signal to be lost by dephosphorylation. The more stable phosphorylation of hydroxyl amino acids allows signals to be more precisely controlled over longer time scales and distances. Despite this, histidine phosphorylation has been observed in mass spec- trometric analysis of metazoan proteins, and may be regulated in some specialized instances. For example, it has been shown that the KCa3.1 K+ channel is activated by histidine phosphorylation of its C-terminus, and that this modification is important for T cell activation. Phosphorylation is mediated by NDPK-B, a member of the nucleotide diphosphate kinase K+ Figure 4.16 Regulation of a mammalian K+ channel by histidine phosphorylation. The potassium (K+) channel Kca3.1 is closed in the basal state. upon histidine phosphorylation by the kinase nDPK-B, the channel opens, allowing K+ to exit the cell. Dephosphorylation and channel closing is mediated by the phosphatase PHPT-1. Figure 4.17 The ubiquitylation machinery. The successive action of E1, E2, and E3 enzymes results in the transfer of ubiquitin (orange triangle) to substrate protein (light brown). family, and phosphate removal is mediated by a protein histidine phos- phatase (PHPT-1) (Figure 4.16). It is not clear whether this is merely the first example of a more extensive class of physiologically important and regulated histidine phosphorylation events, or whether it is an interest- ing one-off solution to a very specific biochemical problem. In general, the study of histidine phosphorylation and other transient phosphorylated species is hampered by their lability under typical conditions of cell lysis and analysis. aDDiTion of uBiquiTin anD rElaTED ProTEins Entire globular proteins can also be covalently added to proteins, result- ing in major changes to the structure and activity of the modified target. The addition of the small, 76-residue protein ubiquitin or its relatives to target proteins has profound effects on the biological activity of the modified protein. For example, the addition of long chains of ubiquitin (polyubiquitylation) often tags proteins for destruction by a specialized protein-degrading complex, the proteasome. By contrast, addition of sin- gle ubiquitin units (monoubiquitylation) or shorter polyubiquitin chains is used to target proteins for endocytosis, or to mediate specific protein– protein interactions. Ubiquitin and polyubiquitin chains are recognized by ubiquitin-binding domains (UBDs) on other proteins. Specialized enzymes mediate the addition and removal of ubiquitin The addition of ubiquitin to a substrate involves three distinct proteins working in series (Figure 4.17). First, an E1 ubiquitin activating enzyme uses the energy of ATP hydrolysis to covalently attach the C- terminus of ubiquitin to a cysteine residue on the E1 protein. Second, the activated ubiquitin is transferred to a cysteine on an E2 ubiquitin con- jugating enzyme. Finally, the ubiquitin is transferred to an amino group (generally a lysine side chain) on the substrate protein with the help of an E3 ubiquitin ligase. In vertebrates, there are two E1 enzymes, ~50 E2 enzymes, and many hundred E3 ligases. It is the E3 ligase, and to a lesser extent the E2 enzyme, that determines which substrates will be ubiquitylated and also the nature of the linkage of the resulting ubiquitin chains. A similar set of enzymes is used to transfer other ubiquitin-like (UBL) peptides—such as SUMO, Nedd8, and ISG15—to their substrates. The UBL peptides have the same overall structure and fold as ubiquitin, though they direct interaction with distinct binding partners and thus confer distinct biological activities to the proteins they modify. There is considerable variety in how the ubiquitin subunits are linked together in polyubiquitin chains (Figure 4.18). For proteins targeted for degradation, the C-terminal glycine of each ubiquitin is generally cou- pled to Lys48 of the preceding subunit, whereas linkage via Lys63 is usu- ally used to tag proteins for other fates such as endocytosis. Linkage to ATP ADP + + + + addiTion of uBiquiTin and rElaTEd ProTEins the N-terminus (generating head-to-tail or linear chains) or to the other five lysines of ubiquitin also occurs. Generally, polyubiquitylation results in long, unbranched chains connected by the same linkage, but mixed linkages and branched structures are also possible given the number of possible attachment points. Ubiquitin is thus rather unique among post- translational modifications used for signaling in the structural diversity of possible modifications that can be generated from a single molecular building block. Deubiquitinases (DUBs) represent the flip side of the ubiquitin-conjugating machinery—proteins that can erase the post-translational marks, poten- tially saving the marked proteins from destruction or reversing other ubiquitin-mediated activities. Approximately 80 DUBs are known in the human genome. These enzymes are, in fact, a specialized type of protease, catalyzing cleavage of the isopeptide bond between a lysine amino group and the C-terminus of ubiquitin. Different DUBs can be specific for dif- ferent substrates and different types of linkages in polyubiquitin chains. Furthermore, they are subject to regulation by post-translational modifi- cation and regulated protein–protein interactions. E3 ubiquitin ligases determine which proteins will be ubiquitylated Consistent with their pivotal role in determining what substrates will be modified, the E3 ligases exhibit a wide variety of structures and bind- ing specificities. These proteins essentially function as adaptors, bring- ing the E2 and its activated ubiquitin into close proximity to the lysine to be modified on the target protein. Most E3 ligases fall into two major classes: the RING group and those containing HECT domains. The RING E3 ligases are by far the largest group, with more than 600 examples in humans. These proteins interact with the E2–ubiquitin complex via the zinc-coordinating RING finger motif, and facilitate ubiquitin transfer to substrate proteins. Generally, it is the E2 enzymes and not the RING E3 ligases that specify the type of linkage (for example, Lys48 versus Lys63). Thus, different combinations of E2 enzymes and RING E3 ligases can mediate transfer of polyubiquitin chains of different types to a huge vari- ety of distinct substrates (Figure 4.19). The HECT domain class of E3 ligases is defined by the relatively large HECT domain, which recognizes the E2–ubiquitin complex and catalyzes ubiquitin transfer first to itself, then to substrate proteins. With respect to signaling, the critical issues for all E3 ligases are how they choose substrates and how their activity is regulated. Many E3 ligases contain well-characterized modular protein-binding domains such as WW, WD40, and SH2 domains, which mediate interaction with specific substrate proteins. Other E3 ligases interact indirectly with substrates through additional E3-associated adaptors. In some cases, one of the pri- mary substrates of ubiquitylation is the E3 ligase itself; autoubiquityla- tion can be promoted by E3 dimerization or oligomerization. This is very analogous to the autophosphorylation of receptors with kinase activity upon ligand-induced dimerization or clustering. In many cases, either the catalytic or binding activity can be regulated by other post-translational modifications, either of the substrate or the E3 ligase itself. Phosphorylation of a potential substrate often increases its affinity for the E3 ligase; for example, the E3 ligase Cbl, which contains an SH2 domain, interacts specifically with substrates when they are tyro- sine phosphorylated. In another example, the large E3 ligase complexes that regulate the cell cycle bind specifically to targets phosphorylated by (b) Figure 4.18 Ubiquitin structure. (a) Three- dimensional structure of human ubiquitin, with the most common sites of linkage (c-terminus, n-terminus, lys48, lys63) indicated. (b) Major forms of polyubiquitin. linkage of the c-terminus to lys48 (K48) or to lys63 (K63) generates polyubiquitin chains with different structures (orange). The polyubiquitylated protein is indicated in light brown. Aggregation-induced receptor activation is discussed in Chapter 8 E3 ligases that regulate the cell cycle are discussed in Chapter 9 ~50 E2s ~500 E3s linkage K48 K63 HTT substrate recognition cyclin-dependent kinases. In this way, a relatively transient modification (phosphorylation) can be converted to a signal that may doom a protein to degradation. The importance of E3 ligases in signaling processes is indi- cated by examples in which their mutation is implicated in diseases such as cancer. Ubiquitin-binding domains read ubiquitin-mediated signals in diverse cellular activities For the most part, the functional consequences of ubiquitylation are medi- ated by proteins that bind directly to the ubiquitin units of the modified protein. Ubiquitin binds specifically to a number of structurally dis- tinct families of binding domains, collectively termed ubiquitin-binding domains (UBDs). Most UBDs bind to the same small hydrophobic patch on the surface of ubiquitin surrounding Ile41, though different UBDs engage other residues surrounding this patch as well (see Figure 10.17a in Chapter 10). In many cases, multiple UBDs, either on the same protein or on associated proteins, engage multiple ubiquitin units on a modified protein, thereby increasing the affinity and specificity of the interaction through avidity and/or cooperative binding. Furthermore, some UBDs recognize specific ubiquitin–ubiquitin linkages, either by recognizing the interface between the two ubiquitin units, or through geometric con- straints. In this way, proteins containing UBDs can discriminate between monoubiquitylated and polyubiquitylated proteins, and polyubiquitylated Figure 4.19 Combinations of E2 and E3 enzymes can modify many different substrates with different polyubiquitin linkages. E2 conjugating enzymes (50 genes in humans) generally determine the linkage of polyubiquitin chains; for example, lys48 (K48), lys63 (K63), or head-to-tail (HTT). E3 ligases generally specify the substrates to be modified. combinations of different E2 and E3 enzymes generate a wide diversity of possible substrates and linkages. proteins with different linkages (for example, Lys48 versus Lys63). Below, we provide a few examples of how different ubiquitin modifications are sensed in a variety of signaling contexts. Proteins on the cell surface that are to be degraded are internalized in vesicles (endocytosis), then sent to the lysosome, a membrane-enclosed intracellular compartment where the protein and lipid components of the vesicle are broken down by digestive enzymes. Membrane proteins are tagged for destruction either by multiple monoubiquitylation or by Lys63- linked polyubiquitylation. Lysosomal targeting is mediated by a series of multiprotein complexes termed the ESCRT machinery (see Figure 8.28 in Chapter 8). Each ESCRT complex has distinct UBDs that interact with ubiquitylated cargo proteins. In particular, the ESCRT-0 complex, which first binds to internalized ubiquitylated targets, plays a critical role in identifying and concentrating those proteins that will be sent to the lysosome. This complex contains multiple UBDs, and in some cases (for example, the UIM domain of Hrs) a single ubiquitin-binding domain can bind two ubiquitin subunits simultaneously (Figure 4.20a). The multi- plicity of ubiquitin-binding sites in the ESCRT-0 complex, each of which alone has rather low affinity, promotes the cooperative binding of targets that contain multiple ubiquitins. Another area where ubiquitylation plays a key role is in the cellular response to DNA damage. In the case of double-strand breaks—a par- ticularly troublesome form of DNA damage because it can lead to the loss of chromosome fragments if not repaired before cell division—a key early event is the recruitment of an E2/E3 complex (Ubc13/RNF8) that adds Lys63-linked polyubiquitin chains to histones in the vicinity of the break. This signal is then amplified by the recruitment of a large effec- tor complex that includes the tumor suppressor protein BRCA1, itself a ubiquitin E3 ligase. Recruitment of this complex is mediated by the UBDs of an adaptor protein, Rap80. Rap80 contains two ubiquitin-binding UIMs that work together to recognize Lys63-linked diubiquitin. The linker region between the two UIMs positions them in such a way that they can productively interact only when the linkage is via Lys63—the Figure 4.20 Recognition of different polyubiquitin linkages by ubiquitin- binding domains. (a) The uiM motif of Hrs can bind simultaneously to two monoubiquitin molecules. The two ubiquitin molecules bind on opposite sides of the uiM and with a similar binding mode. (b) structure of lys63-linked diubiquitin bound to receptor-associated protein rap80 (coordinates provided by s. fukai, university of Tokyo, Japan). (c) The uBan domain in nEMo binds two linear (head- to-tail) diubiquitins. (adapted from i. Dikic, s. wakatsuki and K.J. walters, Nat. Rev. Mol. Cell. Biol. 10:659–671, 2009. with permission from Macmillan Publishers ltd.) distance between two ubiquitins is too short if the linkage is via Lys48 (Figure 4.20b). Head-to-tail (linear) polyubiquitin chains play an important role in the NF-κB signaling pathway. NF-κB is a transcription factor whose activa- tion is critical for innate and adaptive immune responses. A key step in this pathway is the activation of IKK, a multisubunit serine/threonine kinase. IKK activation is controlled by a regulatory subunit, NEMO which contains a UBD (the UBAN domain) that is highly specific for lin- ear polyubiquitin chains. Upstream signals lead to the addition of linear polyubiquitin chains to NEMO by LUBAC (linear ubiquitin chain assem- bly complex); this is likely to lead to conformational changes in the IKK complex, as the UBAN domain of one NEMO molecule interacts in trans with the linear polyubiquitin chains of another. The UBAN domain con- tacts both ubiquitins in a linear diubiquitin chain, but productive contacts are only possible when the two are connected by a head-to-tail linkage (Figure 4.20c). The NF-κB pathway is interesting in that it exploits three distinct types of polyubiquitin chains: Lys63 for activation of upstream kinases, Lys48 for targeting inhibitory subunits for degradation, and head-to-tail for activating IKK. HisTonE acETylaTion anD METHylaTion Chromosomal DNA is packaged into chromatin through its association with histone proteins. The overall structure of chromatin is regulated by the post-translational modification of histones and other proteins that interact with them. Chromatin structure has a direct impact on virtually all activities of the genome, including transcription, replication, repair, genomic imprinting, and chromosome segregation. Many cell signaling events directly or indirectly affect chromatin by changing its post-translational marks which, in many cases, provide binding sites for modification-specific readers. In this section, we will focus on the The NF-κB pathway is described in Chapter 9 remodeling of chromatin by post-translational modifications such as methylation and acetylation, particularly in transcriptional regulation. (a) H2A-N H3-N (b) H4-N H4-N H4 H3 H2A H2A-N H2B H3-N H2B-N H2B-N H1 Chromatin structure is regulated by post-translational modification of histones and associated proteins The basic unit of chromatin is the nucleosome, consisting of eight his- tone subunits arranged in a disclike structure, around which is wrapped ~147 base pairs of DNA. A typical nucleosome contains two molecules each of histone 2A (H2A), histone 2B (H2B), histone 3 (H3), and histone 4 (H4) (Figure 4.21a). The simplest higher-order chromatin structure is an extended conformation where individual nucleosomes are arranged like beads on a string; a linker histone, histone 1 (H1), associates with adja- cent nucleosomes and with the DNA that connects them (Figure 4.21b). Higher-order conformations result from interactions between nucleo- somes, leading to packaging of the chromatin into more dense and compact fibers. In general, the chromatin of genes that are actively transcribed or accessible for transcriptional activation is in a relatively loosely packed form termed euchromatin, while less transcriptionally active regions are in a more tightly packed and inert form termed heterochromatin. During mitosis and meiosis, when chromosomes need to be bundled into an easily handled form for nuclear division, the chromatin condenses dramatically compared to its state in interphase cells. Activities that use the genomic DNA as a template, such as transcription, require at least temporary loosening of the higher-order interactions of chromatin; without transient histone dissociation, transcription factors and RNA polymerase would be unable to access the DNA for binding. The primary means of modulating chromatin structure is by post-translational modification of the histones. The bulk of each histone is a globular domain that contacts DNA and other histones, and is mostly inaccessi- ble for modification. However, each histone has a fairly long, positively charged N-terminal tail that projects out from the nucleosome core and thus is accessible for modification by a variety of enzymes. In particular, a number of lysine residues in the histone tails can be either N-methylated, N-acetylated, ubiquitylated, or sumoylated. Arginine residues are subject to N-methylation and deimination, serine and threonine residues may be phosphorylated, and prolines are subject to cis-trans isomerization. The discovery of these modifications, along with proteins that seemed to bind specifically to different modified sites on histones, led to the pro- posal that specific patterns of modifications determine the activity state of the chromatin. While this concept has been useful in focusing research and thought, the situation is much more complicated than a simple “one modification, one output” model. As discussed above for p53, it is now clear Figure 4.21 Nucleosome structure. (a) arrangement of histone subunits in a standard nucleosome is depicted schematically. There are two subunits each of histones H2a, H2B, H3, and H4, around which ~147 nucleotides of genomic Dna are wound (brown). The n-terminal tails of the histones protrude from the nucleosome and are accessible for post-translational modification. (b) Multiple nucleosomes are arrayed along the Dna in chromatin fibers. Histone H1 (green) bridges two adjacent nucleosomes. that the post-translational modification of histones is highly complex, with many marks being present simultaneously and in a host of different possible combinations. Furthermore, some histone-binding proteins are themselves modified, providing additional variation. Histone modifica- tions are also highly dynamic: many of the marks turn over very rapidly, and thus are in dynamic equilibrium and subject to change over short time scales. Chromatin modification is intimately associated with gene transcription. Transcription is the ultimate end point of many signal transduction path- ways that exert their long-term effects through modulating the expres- sion of specific gene products. For transcription to occur, RNA polymerase (RNA Pol II in the case of protein-coding genes) must bind to the promoter region of the template DNA. This is generally facilitated by a variety of transcription factors, which bind DNA and facilitate binding of the polymerase. Histones must dissociate from the site of transcription ini- tiation to allow polymerase binding. Once polymerization of the RNA message has initiated, histones must be transiently stripped from the template to allow passage of the polymerase; generally, histones rapidly rebind after polymerase transit to prevent adventitious initiation of tran- scription at internal sites. Thus, transcription requires an elaborate and highly dynamic set of changes to chromatin. Two writer/eraser/reader systems are based on protein methylation and acetylation The lysine acetylation of histone tails is generally associated with actively transcribed chromatin. Some of the effects of histone acetylation are intrinsic; that is, they directly regulate interactions among nucleo- somes. For example, acetylation of Lys16 of histone H4 (abbreviated as H4K16Ac) prevents the compaction of nucleosome arrays by preventing contacts between H4 subunits on one nucleosome and H2B subunits on an adjacent nucleosome, most likely because of the loss of positive charge on lysine when it is acetylated. However, most effects of acetylation are extrinsic, due to the specific binding to lysine-acetylated sites of “reader” proteins with small modular protein domains such as bromo domains. Consistent with the connection between transcription and acetylation, a number of transcriptional activators and proteins that associate with transcriptional start sites have histone acetyl transferase (HAT) activity. Conversely, the activity that removes acetyl groups, histone deacetylase (HDAC), is frequently found in transcriptional co-repressor complexes. Both HAT and HDAC activities are most often found in large multifunc- tional proteins or protein complexes that also contain one or more bromo domains. In this way, both the addition and removal of acetyl groups to histone tails is processive: the complex remains stably associated with acetylated chromatin, allowing multiple sites in the vicinity to be modi- fied. The rate of turnover of acetyl groups on histones is quite high, with half-lives of several minutes, though turnover is slower for a minority of sites. This likely reflects a relatively low basal level of acetylation in actively transcribed regions, with transient increases associated with the loading and passage of RNA polymerase. The other major class of histone modification is N-methylation of lysine and arginine residues, by lysine methyl transferases (KMTs) and protein arginine methyl transferases (PRMTs), respectively. Either one, two, or three methyl groups can be added to the ε-nitrogen of lysine side chains; in the case of arginine, one or two methyl groups can be added, with the two either symmetric (one methyl group per amino group) or asymmetric (two methyl groups on one amino group) (Figure 4.22). Protein methylation was once thought to be essentially irreversible, but a number of lysine demethylases (KDMs) have now been identified that readily remove methyl groups from lysine. No arginine demethylases have yet been found, but arginine methylation can be counteracted in the cell by deimination of the side chain, converting arginine to citrul- line (Figure 4.22c). The readers for protein methylation marks represent a rather diverse group of binding domains with varying degrees of specificity and affinity. Domains that bind specifically to methylated lysines include the “royal family” domains (tudor, chromo, and MBT), WD40 domains, and the PHD finger (see Figure 10.16 in Chapter 10). Domains that have been shown to bind specifically to methylated arginines include a subset of tudor and Figure 4.22 Diverse modes of protein methylation. Modifications are highlighted in pink. (a) Mono-, di-, and trimethylated lysine. (b) Monomethylated and cis and trans dimethylated arginine. (c) The deimination of monomethyl arginine by peptidyl arginine deiminase yields N-methyl (a) H N+ N H CH3 H3C N H NH+ CH3 H3C N H CH3 CH N+ citrulline. unmethylated arginine can similarly be deiminated to generate citrulline, which cannot be methylated by arginine methyl transferases. N-methyl lysine N,N-dimethyl lysine N,N,N-trimethyl lysine (b) H3C NH CH3 CH3 NH H3C N CH3 HN NH+ HN NH + HN NH + N N N H H H trans N,N-dimethyl arginine N-methyl arginine cis N,N-dimethyl arginine (c) CH3 HN peptidyl CH3 HN HN NH + arginine deiminase HN O N N H H N-methyl arginine N-methyl citrulline BRCT domains. As is the case for enzymes that modify protein acetyla- tion, the enzymes that add or remove methyl groups from histones are often part of larger complexes that contain one or more reader domains as well, leading to a complex interplay between existing marks and the addition or removal of new marks. Lysine methylation of histones is associated both with activation and repression of transcription. For example, trimethylation of histone H3 at Lys4 (H3K4me3) is enriched on actively transcribed genes, while the tri- methylation at Lys9 of the same protein (H3K9me3) is enriched on tran- scriptionally silenced genes. Even the very same mark can have widely divergent effects, depending on the specific context of associated proteins and other marks. In contrast, arginine methylation is most always associ- ated with transcriptional activation. Chromatin modification in transcription is dynamic and leads to highly cooperative interactions In order for a gene to be transcribed, signals must lead to the recruit- ment of RNA polymerase to the promoter of the gene. Most often this is accomplished by increased binding of transcriptional activators near the promoter of the gene. This may be due to increased transport of the transcriptional activators to the nucleus, increased affinity for their bind- ing sites on DNA, increased binding to transcriptional coactivators, or increased synthesis; but in some way their activity and/or concentration in the nucleus must be increased as a result of the signal. These tran- scriptional activators, in turn, recruit a number of other proteins, includ- ing chromatin modifying complexes and RNA Pol II itself. Alternatively, some signals lead to the inactivation of transcriptional repressors already bound to the DNA, thereby relieving repression and allowing the recruit- ment of factors that promote transcription. Transcriptional activators such as p300/CBP or PCAF/Gcn5 generally associate with or contain HAT activities that acetylate chromatin in the promoter region. Acetylated histones, in turn, recruit a variety of factors that help promote transcription, including general transcription factors and chromatin remodeling factors such as the SWI/SNF complexes, which use the energy of ATP to loosen the association of histones with DNA near the site of transcription initiation. Transcriptional activators also recruit the RNA Pol II holoenzyme, which then binds to the template DNA and is poised to initiate transcription (Figure 4.23a). RNA polymerase itself becomes phosphorylated in the course of tran- scription, and this post-translational modification leads to important changes in protein interactions—yet another writer/eraser/reader sys- tem that intersects with the systems based on histone methylation and acetylation. RNA Pol II contains a long C-terminal domain (CTD) com- posed of a number of heptameric repeats, with the consensus sequence YSPTSPS. The serine, threonine, and tyrosine residues are subject to phosphorylation, and the prolines to cis-trans isomerization. One of the (b) (c) HAT RNA Ac Ac Ac capping Pol lI Set1 P Ac Ac Me Me Figure 4.23 Dynamic chromatin modifications during transcription. (a) Transcriptional activators recruit histone acetyl transferases (HaT), leading to acetylation of histones in the vicinity of the site of transcription initiation. in turn, acetylation recruits aTP-dependent chromatin remodeling complexes and promotes binding of rna polymerase ii (Pol ii). (b) as transcription is initiated, the c-terminal tail repeats of Pol ii are phosphorylated on ser5, recruiting rna capping factors and the set1 methyl transferase, which adds methyl groups to lys4 of histone H3. nascent rna transcript is depicted in orange. (c) as transcription proceeds, the Pol ii c-terminal repeats are phosphorylated on ser2, leading to recruitment of termination factors and the set2 methyl transferase, which adds methyl groups to lys36 on histone H3. This recruits histone deacetylases (HDac), which remove acetyl groups from nearby histones. key events of transcription initiation is the phosphorylation of the CTD on the fifth residue of the heptameric repeat (Ser5) by a cyclin-dependent kinase, Cdk7, which is a component of the general transcription factor TFIIH. Ser5 phosphorylation of the CTD repeats has at least two important con- sequences. First, it decreases the affinity of the polymerase for general factors bound to the promoter, such as the Mediator complex, allowing polymerase to escape the promoter and begin transcript elongation. Sec- ond, it provides specific binding sites for a variety of factors, including RNA capping enzymes that process the 5′ end of the RNA transcript. The phosphorylated CTD also binds the Set1 lysine methyl transferase, which leads to increased histone H3K4 methylation. H3K4 methylated sites, in turn, recruit more complexes that promote open and active chroma- tin structure (HATs, KMTs, and ATP-dependent remodeling complexes) (Figure 4.23b). Once the polymerase clears the promoter and is actively transcribing the gene, another cyclin-dependent kinase (a component of the P-TEFb tran- scription elongation complex) phosphorylates the CTD repeats on Ser2. This, in turn, recruits a second KMT (Set2), which methylates another site on the tail of histone 3, Lys36 (H3K36Me). One consequence of H3K36Me is the recruitment of HDACs, which remove acetyl groups from histones in the protein-coding region of the gene. This is thought to be important to reset the histones within the coding region to a basal, closed state after the passage of the polymerase, helping prevent inappropriate polymerase recruitment and transcription initiation from internal sites within the gene. Finally, the Ser2-phosphorylated RNA Pol II CTD helps recruit termination and polyadenylation factors needed to terminate the new transcript and process its 3′ end (Figure 4.23c). Specific phosphatases are also recruited that target the CTD and reset the polymerase back to its basal, unphosphorylated state. The above description is highly simplified, and is only meant to provide an overview of the complex and highly coordinated events of transcription. For example, the effects of DNA methylation and the phosphorylation, ubiquitylation, sumoylation, and proline isomerization of histones and associated proteins are not considered. But even at this relatively sim- ple level of description, it is apparent that writer/eraser/reader systems provide a flexible way to control complex and dynamic cell behaviors. A sequential, linked series of post-translational modifications, each creating binding sites for new factors that generate additional modifications (or remove previous ones), is a powerful and recurrent theme in signaling. suMMary Post-translational modification provides a rapid and efficient way to change the activity of proteins. The addition and removal of these modi- fications are catalyzed by the activity of specific enzymes. The large number of possible modification sites, along with the diversity of pos- sible modifications, enormously expands the number of potential protein states beyond what can be encoded by the genome. Modifications can directly affect protein activity, but in many cases the modifications are “read” by proteins that specifically bind to the modified sites. Phospho- rylation of the hydroxyl amino acids (serine, threonine, tyrosine) is the most widespread post-translational modification in signaling in metazo- ans. The addition of ubiquitin and its relatives is also widely used, and affords considerable diversity in the length of the chains that are added quEsTions and how the subunits are linked together. The structure and activity of chromatin is dynamically modulated by the post-translational modifica- tion of histones and histone-associated proteins, especially by acetylation and methylation. quEsTions In analyzing how a cell responds to stress, you discover several induced phosphorylation events on a membrane-associated protein, X. Phospho- rylation of protein X on Thr122 induces its interaction with protein Y. Describe possible alternative mechanisms by which this modification might induce binding to protein Y. Propose experiments to distinguish between these possibilities. Phosphorylation of the protein X from Question 1 on Ser54 disrupts its interaction with protein Z. Describe possible mechanisms by which this modification might disrupt binding to protein Z. How might you distinguish between these possibilities? Sometimes, phosphorylation of a protein on a specific residue leads to the degradation of that protein. Describe the general mechanisms by which degradation can be mediated by phosphorylation, and the inter- vening steps between phosphorylation and degradation. Figure 4–13 shows several different mechanisms by which a protein could be phosphorylated on multiple distinct sites. One mechanism is priming, where one kinase phosphorylates the target, creating a bind- ing site for a second kinase. Recruitment of the second kinase subse- quently leads to phosphorylation on a second (or additional) site. What kinds of kinetic responses might result from a protein that is controlled by a priming mechanism? Describe additional features of the signaling system that would be necessary in your model. How might regulation of the activity of enzymes that mediate protein post-translational modifications (writers and erasers) generate sig- nals that are highly localized and which transmit spatial information? What elements might limit the ability of enzymes to generate localized signals? What elements might enhance their ability to generate local- ized signals? Efforts are underway to map the full complement of protein post-trans- lational modifications in various cells under different conditions. Is it realistic to expect that such a task will ever be completed? What are the major challenges in such an effort? Is has been suggested that some experimentally detected post-trans- lational modifications of specific sites might have no function, or may even be detrimental to the cell. Under what conditions might this be the case? A number of enzymes that modify proteins contain a reader domain that binds to the modified substrate. For example, nonreceptor tyro- sine kinases have SH2 domains that bind to tyrosine-phosphorylated peptides; and histone acetyl transferases have bromo domains that bind acetylated lysines. What effect will these reader domains have on modification by such enzymes, and under what conditions would such an arrangement be useful? How might you engineer a system, based on a ubiquitin E3 ligase, to degrade a particular protein of choice in the cell? rEfErEncEs PosT-TranslaTional MoDificaTions anD THEir EffEcTs Seet BT, Dikic I, Zhou MM & Pawson T (2006) Read- ing protein modifications with interaction domains. Nat. Rev. Mol. Cell Biol. 7, 473–483. Walsh CT (2006) Posttranslational Modification of Pro- teins: Expanding Nature’s Inventory. Englewood, CO: Roberts and Co. Publishers. inTErPlay BETwEEn PosT-TranslaTional MoDificaTions Boehme KA & Blattner C (2009) Regulation of p53— insights into a complex process. Crit. Rev. Biochem. Mol. Biol. 44, 367–392. Butkinaree C, Park K & Hart GW (2010) O-linked beta- N-acetylglucosamine (O-GlcNAc): Extensive crosstalk with phosphorylation to regulate signaling and tran- scription in response to nutrients and stress. Biochim. Biophys. Acta 1800, 96–106. Kruse JP & Gu W (2009) Modes of p53 regulation. Cell 137, 609–622. Lothrop AP, Torres MP & Fuchs SM (2013) Deciphering post-translational modification codes. FEBS Lett. 587, 1247–1257. Metzger E, Imhof A, Patel D et al. (2010) Phosphoryla- tion of histone H3T6 by PKCbeta(I) controls demethyla- tion at histone H3K4. Nature 464, 792–796. Prabakaran S, Lippens G, Steen H & Gunawardena J (2012) Post-translational modification: nature’s escape from genetic imprisonment and the basis for dynamic information encoding. Wiley Interdiscip. Rev. Syst. Biol. Med. 4, 565–583. Yang XJ (2005) Multisite protein modification and intramolecular signaling. Oncogene 24, 1653–1662. ProTEin PHosPHorylaTion Gao R & Stock AM (2009) Biological insights from struc- tures of two-component proteins. Annu. Rev. Microbiol. 63, 133–154. Jin J & Pawson T (2012) Modular evolution of phospho- rylation-based signalling systems. Philos. Trans. R. Soc. Lond. B Biol. Sci. 367, 2540–2555. Johnson LN & Lewis RJ (2001) Structural basis for con- trol by phosphorylation. Chem. Rev. 101, 2209–2242. Podgornaia AI & Laub MT (2013) Determinants of spe- cificity in two-component signal transduction. Curr. Opin. Microbiol. 16, 156–162. Srivastava S, Zhdanova O, Di L et al. (2008) Protein his- tidine phosphatase 1 negatively regulates CD4 T cells by inhibiting the K+ channel KCa3.1. Proc. Natl Acad. Sci. U.S.A. 105, 14442–14446. Ubersax JA & Ferrell JE Jr (2007) Mechanisms of spe- cificity in protein phosphorylation. Nat. Rev. Mol. Cell Biol. 8, 530–541. aDDiTion of uBiquiTin anD rElaTED ProTEins Dikic I, Wakatsuki S & Walters KJ (2009) Ubiquitin- binding domains – from structures to functions. Nat. Rev. Mol. Cell Biol. 10, 659–671. Kerscher O, Felberbaum R & Hochstrasser M (2006) Modification of proteins by ubiquitin and ubiquitin-like proteins. Annu. Rev. Cell Dev. Biol. 22, 159–180. Komander D (2009) The emerging complexity of protein ubiquitination. Biochem. Soc. Trans. 37, 937–953. Metzger MB, Hristova VA & Weissman AM (2012) HECT and RING finger families of E3 ubiquitin ligases at a glance. J. Cell Sci. 125, 531–537. Searle MS, Garner TP, Strachan J et al. (2012) Structur- al insights into specificity and diversity in mechanisms of ubiquitin recognition by ubiquitin-binding domains. Biochem. Soc. Trans. 40, 404–408. HisTonE acETylaTion anD METHylaTion Barth TK & Imhof A (2010) Fast signals and slow marks: the dynamics of histone modifications. Trends Biochem. Sci. 35, 618–626. Berger SL (2007) The complex language of chromatin regulation during transcription. Nature 447, 407–412. Campos EI & Reinberg D (2009) Histones: annotating chromatin. Annu. Rev. Genet. 43, 559–599. Kouzarides T (2007) Chromatin modifications and their function. Cell 128, 693–705. Zentner GE & Henikoff S (2013) Regulation of nucleo- some dynamics by histone modifications. Nat. Struct. Mol. Biol. 20, 259–266. subcellular Localization of signaling Molecules One of the defining properties of a cell is the nonuniform distribution of its constituents. The most basic division is between the aqueous cytosol and the water-insoluble lipid environment of the plasma membrane that surrounds it. The plasma membrane serves a dual role both as a physi- cal barrier separating the cell’s contents from the environment, and as a dynamic interface mediating the selective passage of information and material to and from the interior of the cell. In eukaryotes, internal mem- branes serve to further compartmentalize key cellular functions such as transcription, replication, and energy production. Even within the same intracellular compartment, many components are unevenly distributed, and this lack of homogeneity is essential for many of the cell’s activities. Nearly all cells display some functionally important asymmetry, or polarity. For example, a migrating cell must protrude at the front and de-adhere at the back. In another example, epithelial cells have clearly distinct apical and basal faces. These polarity differences are due to an asymmetric distribution of specific molecular components. Here we will address the more general questions of how cell signaling mecha- nisms take advantage of the nonuniform distribution of cell components to transmit information, and how signals can modulate the distribution of those components. LocaLization as a signaLing currency One way of describing the uneven distribution of cell components is in terms of differences in their local concentration across the cell. Local concentration is the concentration of a component at a specific location, irrespective of the overall amount present elsewhere in the cell. Although there may be only a few molecules of some component in the entire cell, if they are all co-localized within a very small volume then their local plasma membrane Figure 5.1 membrane domain/pole organelle concentration at that spot will be very high. Local concentration is par- ticularly important for cell physiology because most macromolecules can- not freely diffuse throughout the cell; their movements in the crowded intracellular environment are highly constrained by physical barriers and by their interactions with other cellular components. The cell is not an idealized, well-mixed solution in equilibrium, and so very different things will occur at different sites due to differences in the local concentrations of components that are found there. Subcellular localization plays a critical role in signaling because it directly affects what reactions can occur. For both enzymatic and binding reac- tions, the actual rate of reaction is proportional to the concentration of Subcellular locations important in signaling. the biological activity of a signaling protein can be very different depending on whether it is localized to the cytosol, nucleus, or membranes. Membranes can be further divided into different functional domains, including bulk plasma membrane, specialized regions of the plasma membrane, or the membranes surrounding organelles or intracellular vesicles. trafficking of proteins or other components between all of these different compartments can transmit signaling information. each of the reactants. Obviously, if an enzyme is localized in the interior matrix of the mitochondrion and a potential substrate is found only in the nucleus, the two proteins will never encounter each other and the enzyme cannot act upon that substrate. In effect, the concentration of these two proteins toward each other is zero, no matter what their overall (average) concentrations might be per cell. On the other hand, the high local con- centrations that result when two components are co-localized can drive reactions very efficiently, both thermodynamically and kinetically. Changes in subcellular localization can transmit information Signaling mechanisms exploit the translocation of molecules among a number of distinct cellular compartments. As shown in Figure 5.1, mole- cules can be partitioned to the cytosol, nucleus, or membrane. Even within the membrane, molecules can have functionally distinct sublocalizations— bulk plasma membrane is distinct from membrane domains enriched in particular lipid or protein components, or from the membranes associated with physical structures such as the primary cilium, cell–cell junctions, or cell poles, for example. In this chapter, we will focus on two of the most common cellular translocations involved in signaling—the movement of proteins to and from the nucleus, or to and from cell membranes. The rationale behind regulated nuclear localization is obvious: genomic DNA and other chromatin constituents such as histones are found exclu- sively in the nucleus. Access to this compartment is therefore absolutely essential for the activity of proteins that act on chromatin, such as tran- scription factors. As we will see below, the cell has elaborate mechanisms that regulate nuclear import and export, providing ample opportunity for regulation in signaling. The rationale for changes in membrane localization is similar, as many signaling proteins and their substrates are exclusively found on mem- branes. These include transmembrane proteins and proteins contain- ing covalently attached lipid groups, as well as lipids themselves, which are frequent targets of modification in signaling. Although the plasma membrane surrounds the cytosol, and thus is available to interact with cytosolic components, a molecule is much more likely to encounter a mem- brane-bound partner if it is localized to the membrane itself than if it is free to diffuse in the cytosol. This is because the effective volume within which the molecule can roam is much smaller when it is confined to the membrane, so its local concentration with respect to membrane compo- nents is correspondingly higher. The magnitude of this effect can be illustrated by a hypothetical example. If we assume a spherical cell 20 µm in diameter, from simple geometry we can calculate its volume and surface area (roughly 4000 µm3 and 1200 µm2, respectively). But what is the effective volume for a protein that is con- fined to the plasma membrane? We can estimate this by calculating the volume of cytosol within 5 nm of the membrane (this distance is roughly the diameter of an average protein). In this example, the volume of cytosol inhabited by a membrane-bound protein is then (1200 µm2) × (0.005 µm) = 6 µm3 (Figure 5.2). Therefore a protein would be concentrated by a fac- tor of almost 700-fold (from 4000 µm3 to 6 µm3) when it is confined to the membrane, compared to when it is uniformly distributed throughout the cell. If the protein is an enzyme whose substrate is localized to the mem- brane, it is therefore much more likely to encounter and interact produc- tively with that substrate. Although the spherical cell used as an example is obviously idealized, comparable effects are seen when actual values for cell volume and sur- face area are used. For example, in one experiment, mouse fibroblasts spreading on a tissue-culture dish were found to have an average cytosolic volume of 16,000 µm3 and surface area of 8400 µm2, and relocation of a cytosolic protein to the membrane in this case would lead to an almost 400-fold increase in local concentration. Subcellular localization can be regulated by a variety of mechanisms The mechanisms that control the constitutive targeting of proteins to spe- cific subcellular compartments are a major focus of research in cell biology, and will not be discussed in great detail here. Broadly speaking, however, short peptide sequences termed sorting signals or targeting signals are often sufficient to target a protein bearing that sequence to a particular compartment or organelle. Such targeting sequences often work by pro- moting association with specialized proteins, termed trafficking proteins, that deliver them to their proper cellular address. Of more interest for cell signaling, however, are the ways in which the localization of proteins and other signaling molecules can be dynami- cally regulated in response to signals. As discussed in Chapter 4, post- translational modification is one way that signaling can directly or indirectly affect protein localization. For example, phosphorylation can destroy or create binding sites for partners involved in the transport or tethering of a protein to a specific site. Post-translational modifications can also have more direct effects, as in the addition of hydrophobic lipid (a) (b) Figure 5.2 20 μm groups that promote membrane association, or in proteolytic cleavage that physically releases one part of a protein from another that tethers it to a specific subcellular localization. Another widely used mechanism involves small modular domains that bind to specific membrane lipids such as phosphoinositides (phosphatidylinositol-derived lipids which may be further phosphorylated at specific positions on the inositol head group) whose levels are regulated during signaling. The membrane locali- zation of a protein bearing such a domain can be regulated by localized changes in the concentration of the cognate lipid. controL of nucLear LocaLization The movement of macromolecules such as proteins between the nucleus and cytosol is tightly controlled. The two compartments are separated from each other by the double lipid bilayer of the nuclear envelope, which is perforated by a relatively small number of aqueous pores. The nuclear pore complex is a large, multiprotein complex that serves as a selective gateway regulating the passage of macromolecules into and out of the nucleus. Relatively small macromolecules, up to a size of 40–60 kD, can passively diffuse through the nuclear pore and thus rapidly equilibrate between the two compartments. Any macromolecule larger than this must Effect of membrane localization on local concentration. (a) a protein (pink) is uniformly distributed throughout the cell. its local concentration in the vicinity of a membrane-associated binding partner (blue) is relatively low. (b) the protein is localized to the plasma membrane. this results in an increase in local concentration in the vicinity of its binding partner of nearly 700-fold (see text for details). be actively ferried through the nuclear pore by specialized transport pro- teins in a process that requires energy. Because this process is highly regulated, it provides ample scope for modulation during signaling. Short, modular peptide motifs direct nuclear import and export The efficient nuclear import or export of proteins is mediated by short (often less than ten residues) amino acid sequences in the protein itself. A nuclear localization signal (NLS) is sufficient to direct a protein containing it to enter the nucleus, while a nuclear export signal (NES) is sufficient to direct nuclear export. As will be discussed below, these sequences bind to specific transporter proteins, which negotiate passage of the protein through the nuclear pore complex. These targeting sequenc- es are modular, in the sense that they can be moved to different sites on the protein, or even to different proteins, without losing their ability to direct import or export. Although some NLS and NES motifs can be recognized from their primary sequence (for example, one class of NLS is lysine-rich and highly basic, while an NES often contains a characteristic leucine-rich motif), the precise sequence and/or conformational require- ments are not fully understood. All proteins are synthesized in the cytosol, so the existence of nuclear export signals implies that some proteins can be both imported into the nucleus and exported back out into the cytosol. Indeed, many proteins can be shown to contain multiple functional NLS and NES motifs. The localization of such proteins is thus in dynamic equilibrium as they shut- tle between the cytosol and nucleus. This has important implications for signal transduction. First, changes in the rate of import and/or export can rapidly affect the overall distribution of a protein between the nucleus and the cytosol. Second, the constant cycling between the two compart- ments allows such a protein to continuously sample conditions in both the cytosol and the nucleus and respond rapidly to changes in conditions in both compartments. G proteins and their regulation are described in Chapter 3 Nuclear transport is controlled by shuttle proteins and the G protein Ran The binding of specific transport proteins to the NLS or NES of a protein mediates its nuclear import or export. The most prominent class of such transport proteins is termed the karyopherins, which can be divided into the importins and exportins depending on whether they function primarily to import cargo to the nucleus or export it out to the cytosol. In mammalian cells, many different exportins and importins are expressed, each recognizing different classes of cargo proteins. The vectorial transport of proteins bound to karyopherins is control- led by a nuclear–cytoplasmic gradient of the GTP-binding protein Ran (Figure 5.3). Like all G proteins, Ran can exist in GTP-bound and GDP- bound states, which differ in their binding partners. Rcc1 is a guanine nucleotide exchange factor (GEF) that promotes release of GDP and bind- ing of GTP to Ran. It is found exclusively in the nucleus. By contrast, GTPase-activator proteins (GAPs) for Ran, which promote cleavage of GTP bound to Ran into GDP plus phosphate, are exclusively cytosolic. Because of this asymmetric distribution of Ran regulators, there is a much higher concentration of Ran-GTP in the nucleus than in the cytosol. The binding of exportins to their cargo is stimulated by the binding of Ran-GTP. Tri- molecular exportin/cargo/Ran-GTP complexes pass through the nuclear pore complex, whereupon cargo is released in the cytosol once the GTP Figure 5.3 Nuclear import and export are regulated by karyopherins and Ran. the nucleus contains high levels of ran-gtP, while ran-gDP predominates in the cytosol. Proteins containing nuclear localization signal (nLs) sequences bind importins (green) in the cytosol, pass through the nuclear pore, and are released in the nucleus by the binding of ran-gtP to the importin. Proteins containing a nuclear export signal (nes) bind exportins (blue) and ran-gtP in the nucleus, are exported, and are released in the cytosol when gtP bound to ran is hydrolyzed. bound to Ran is hydrolyzed. Conversely, importins bind to their cargo in the cytosol in the absence of Ran-GTP, and release it in the nucleus upon Ran-GTP binding. Note that each import–export cycle results in the cleav- age of one molecule of GTP. It is the energy provided by GTP hydrolysis that drives the transport of proteins against their concentration gradient, which is thermodynamically unfavorable. Several examples of how nucle- ar import can be regulated to mediate diverse cell signaling responses are described below. Phosphorylation of transcription factor Pho4 regulates nuclear import and export An elegant example of how the balance of nuclear import and export can be exploited in signaling is provided by the yeast transcription factor Pho4. Pho4 regulates the transcription of genes needed for growth under conditions of low phosphate. Therefore, when cells experience phosphate starvation, Pho4 must localize to the nucleus so it can bind to its DNA tar- gets. When phosphate is abundant, however, Pho4 activity is not needed and the protein is found predominantly in the cytosol. This regulation of subcellular localization of Pho4 by phosphate has been shown to depend on the activity of Pho80/85, a nuclear cyclin-dependent protein kinase. Pho80/85 is active under normal conditions when phosphate is abundant, and inactivated upon phosphate starvation. How does Pho80/85 regulate the subcellular localization of Pho4? It turns out that the phosphorylation of Pho4 directly affects both its import and export (Figure 5.4). First, phosphorylation of two sites within an NES motif promotes its binding to an exportin, leading to much more efficient nuclear export of Pho4 when the kinase is active. Second, phosphoryla- tion of a site within the NLS abolishes the binding of an importin, thus preventing import of Pho4 into the nucleus. Finally, phosphorylation of a third region of Pho4 eliminates binding to another transcription factor, Pho2; in the absence of Pho2 binding, Pho4 cannot bind strongly to its specific binding sites on the promoters of phosphate-responsive genes and thus activate their transcription. Upon inhibition of Pho80/85 under con- ditions of phosphate starvation, a combination of dephosphorylation and (a) (b) export exportin transcription phosphate HIGH: Pho80/85 kinase active P P P P NES Pho4 DNA binding/ OFF import transcrption transcriptional changes phosphate LOW: Pho80/85 kinase inactive ON Pho4 Figure 5.4 Regulation of Pho4 by phosphorylation. (a) the subcellular localization of the transcription factor Pho4 (indicated in pink) changes from cytosolic under normal conditions to nuclear when cells are starved of phosphate. (b) Pho4 contains a nuclear export signal (nes), a nuclear localization signal (nLs), and a region that binds the transcriptional coactivator Pho2. under normal conditions (top), Pho4 is phosphorylated by the Pho80/85 kinase at multiple sites. as a result, it binds exportin but not importin or Pho2. Pho4 is exported to the cytosol and cannot activate transcription. under phosphate starvation (bottom), Pho80/85 is inactivated, and unphosphorylated Pho4 binds to importin and to Pho2, but not to exportin. Pho4 is now localized to the nucleus and can activate transcription. new synthesis leads to a rapid increase in the pool of unphosphorylated, nuclear Pho4 and thus induction of the transcription of Pho4-dependent genes. Phosphorylation of Pho4 dramatically alters the nuclear/cytoplasmic equilibrium in three distinct ways: increasing nuclear export, decreasing nuclear import, and decreasing binding to transcriptional cofactors and DNA in the nucleus. A very important feature of this mechanism is that Pho4 activity is cooperatively regulated at multiple steps, making the response sharper and more switchlike than it would be if only one step was regulated. The use of phosphorylation to regulate the partitioning of proteins between nucleus and cytoplasm is quite common in signaling, and addi- tional examples will be discussed later. One frequently used mechanism involves the phosphorylation of serine and threonine residues that create binding sites for 14-3-3 proteins. These small proteins bind specifically to phosphorylated sites on their partners. In some cases, 14-3-3 binding serves to sterically block the binding of importins or of DNA, thereby excluding the bound protein from the nucleus or inhibiting its activity. For example, this mechanism is used to regulate the activity of the FOXO family of transcription factors, and also the Cdc25 family of protein phos- phatases that regulate cell-cycle progression (see Figure 10.12). Nuclear import of STATs is regulated by phosphorylation and conformational change In Chapter 3, we discussed a number of examples where conformational change was functionally linked to changes in enzymatic or binding activ- ity. Changes in subcellular localization can also be coupled to conforma- tional changes, as illustrated by the STAT (signal transducer and activator of transcription) proteins. These transcription factors func- tion in a common signaling module termed the JAK–STAT pathway, which relays signals from a variety of cytokines and hormones such as growth hormone, erythropoietin, and interferon. The seven closely related members of the STAT family each contain a DNA-binding domain, an SH2 domain that recognizes phosphotyro- sine, and a conserved tyrosine phosphorylation site. In unstimulated cells, the STAT proteins reside in the cytosol in a latent, inactive state (Figure 5.5). Stimulation of cell-surface cytokine receptors by their ligands leads to activation of JAK-family tyrosine kinases, which then phosphorylate STAT proteins on specific tyrosine residues. Phos- phorylation of STATs causes a conformational change that facilitates Figure 5.5 Nuclear localization of activated STAT. in unstimulated cells, stat is located in the cytosol and its nuclear localization signal (nLs) and Dna- binding domains are not functional. upon phosphorylation by JaK-family kinases, stats dimerize through interactions between sH2 domains and phosphotyrosine. conformational changes expose nLs and Dna-binding domains that mediate nuclear localization and binding to target sequences in Dna, respectively. formation of active homodimers or heterodimers with other phospho- rylated STAT molecules. This dimerization is mediated by mutual interaction of the SH2 domain and tyrosine-phosphorylated site on one partner with the corresponding tyrosine-phosphorylated site and SH2 domain of the other. This bivalent interaction is particularly stable due to the avidity effect. The active STAT dimer translocates to the nucleus where it binds spe- cific DNA sequences to stimulate gene transcription. In this case, both the nuclear localization signal (which interacts with importins) and the DNA- binding domain are not functional in the unphosphorylated form of STAT, and are only fully assembled and active upon conformational change and dimerization. This dual control of activity, at the levels of rate of nuclear import and binding to specific DNA targets, again makes the response to activation more switchlike. Localization of MAP kinases is regulated by association with nuclear and cytosolic binding partners Many extracellular signals lead to the activation of serine/threonine kinases of the MAP kinase family. These kinases are activated by phos- phorylation by upstream kinases (MAP kinase kinases, or MAPKKs). Substrates for MAP kinases are found in various subcellular localizations, but perhaps the best characterized of these substrates are nuclear tran- scription factors. To phosphorylate these nuclear substrates, however, the activated MAP kinase must itself be localized to the nucleus. When the subcellular localization of a MAP kinase such as Erk2 is tracked following activation, often a dramatic relocation from the cytosol to the nucleus is seen, corresponding with the appearance of the phosphorylated, active kinase (see, for example, Figure 13.15). The precise mechanism of this translocation is not yet completely resolved. Erk2 itself has no appar- ent NLS or NES sequences, and a number of experimental approaches have indicated that translocation does not, in most cases, require energy. Indeed, MAP kinases are small enough (∼40 kD) to enter the nucleus through pas- sive diffusion, and it is likely that this mechanism is sufficient to explain the observed rates of translocation. What then is responsible for the change in distribution upon activation? This is thought to be largely the result of differences in the binding partners for the unphosphorylated (inactive) and phosphorylated (active) forms of the MAP kinase. In the case of Erk2, the inactive form binds tightly to Mek, its upstream MAPKK, which is localized almost entirely in the cytosol. Phosphorylated Erk2, by contrast, does not bind strongly to Mek, and thus is free to diffuse into the nucleus, where it presumably preferentially binds to nuclear proteins (Figure 5.6). This relatively straightforward picture is almost certainly an oversim- plification of the actual mechanisms controlling MAP kinase subcellular localization. For example, in some situations, activated Erk can be found in the cytosol where a number of important substrates are localized. There is also some evidence for energy-dependent import and export of Avidity is described in Chapter 2 MAP kinase cascades are dis- cussed in Chapter 3 Figure 5.6 Dynamic localization of MAP kinases. the MaP kinase erk2 binds to different partners in its inactive and activated states. inactive erk binds tightly to its upstream activator Mek, which is predominantly found in the cytosol. upon activation by Mek, erk dissociates from Mek. erk is then free to diffuse passively into the nucleus, where it associates with nuclear proteins. nucleus Figure 5.7 Localization of Notch is regulated by proteolysis. notch is a transmembrane receptor. When engaged by its ligand Delta on an adjacent cell, notch is cleaved within the juxtamembrane region, releasing the notch intracellular domain (nicD). a nuclear localization signal (nLs) within the nicD facilitates its import to the nucleus, where it associates with cofactors and Dna to regulate transcription. General properties of membranes and their role in signaling are described in Chapter 7 MAP kinases, perhaps in complex with proteins that contain bona fide NLS or NES motifs. Some studies have also suggested that activation- induced dimerization of the MAP kinase is an important prerequisite for nuclear import. It is also worth noting that, once again, activation is inti- mately coupled to change in subcellular localization, thus increasing the apparent difference between the inactive and active states for nuclear substrates such as transcription factors. Notch nuclear localization is regulated by proteolytic cleavage The Notch signaling pathway is commonly used during development to specify cell fate and to regulate morphogenesis. It provides a mecha- nism for clusters of adjacent cells to communicate with each other and coordinate their activities. The receptors of the Notch family are trans- membrane proteins containing an intracellular domain linked to an extracellular domain that interacts with specific ligands, such as Delta, on adjacent cells. Ligand engagement leads to a series of proteolytic cleav- ages of Notch, the most important consequence of which is the release of the intracellular domain from the membrane. This domain (termed the Notch intracellular domain, or NICD) then translocates to the nucleus where it interacts with a DNA-binding protein and other cofactors to reg- ulate gene transcription (Figure 5.7). The NICD contains NLS motifs that presumably interact with importins and mediate nuclear import once the domain is freed from the plasma membrane. The fact that nuclear localization is the result of proteolytic processing confers several interesting properties to this signaling pathway. Most obviously, the release of the active NICD is irreversible; thus, this system is not well suited to respond to rapid, repeated changes in stimulus levels over time. Also, signal amplification is not possible, as one activated recep- tor generates just a single activated transcription factor. This pathway is also remarkable for its directness, as it does not involve any interven- ing molecules between receptor engagement and nuclear activity; thus, Notch signaling is less likely than other signaling pathways to respond to diverse upstream stimuli or lead to many different downstream effects. controL of MeMbrane LocaLization Cellular membranes, and in particular the plasma membrane, are the site of many signaling reactions. In the case of the plasma membrane, this is in large part due to the presence of transmembrane receptors convey- ing information from the outside environment to the cytosol. Membranes also offer a variety of lipids that can attract protein-binding partners and provide substrates for lipid-modifying enzymes. Furthermore, the two- dimensional surface of the membrane serves to restrict diffusion and increase the efficiency of interactions between membrane-associated pro- teins compared to those in the cytosol. Signaling can involve both proteins that are constitutively localized to the membrane, such as most receptors, and proteins that are conditionally localized to the membrane. Below we discuss both classes of protein and the diverse mechanisms that are used to target such proteins. Proteins can span the membrane or be associated with it peripherally The different ways in which proteins can be associated with membranes are illustrated in Figure 5.8. Some proteins exist as integral membrane proteins that actually span the lipid bilayer, while peripheral membrane proteins can associate with membranes by virtue of covalent lipid modifi- cation, through specialized domains that bind directly to membranes, or by indirect binding to other membrane-bound proteins. Integral membrane proteins such as transmembrane receptors are insert- ed into the membrane during the process of translation on the endoplas- mic reticulum, and are ultimately transported to their target membranes by vesicular transport. Different portions of such proteins face the extra- cellular (or lumenal) or cytosolic faces of the membrane. Because of the enormous energetic cost of dragging bulky hydrophilic protein domains through the lipid membrane after their initial synthesis, integral mem- brane proteins cannot be liberated from the membrane other than through proteolytic cleavage (as in the case of Notch, as discussed above). Proteins can be covalently modified with lipids after translation Some proteins are anchored in the membrane through the covalent addi- tion of bulky hydrophobic lipid groups. Such post-translational lipid modi- fications provide more scope for regulation than is possible for integral membrane proteins, because the lipid addition and removal steps can potentially be controlled by signals. Furthermore, the energetic cost of extracting the lipid group from the membrane is not prohibitive, so lipid- modified proteins can, in some cases, shuttle between the membrane and cytosolic compartments. Several common lipid modifications will be dis- cussed below, and their properties are summarized in Table 5.1. The glycosylphosphatidylinositol (GPI) anchor involves the addition of a modified phosphatidylinositol lipid group to the C-terminus of a pro- tein within the lumen of the endoplasmic reticulum (Figure 5.9). GPI- modified proteins then proceed through the normal secretory pathway to the outer leaflet of the plasma membrane. Because they are anchored to the membrane via the phospholipid head group, they can be released from the cell surface through the action of extracellular phospholipases. This cleavage is, however, irreversible. N -myristoylation involves the addition of the myristoyl group, a 14-carbon fatty acid, to glycine at the N-terminus of a cytosolic protein just after translation, following removal of the initiator methionine (Figure 5.9). This modification is thought to be essentially irreversible. A large number of signaling proteins are N-myristoylated, including hetero- trimeric G protein α subunits and other G proteins, nonreceptor tyrosine kinases, and the catalytic subunit of protein kinase A (PKA). The mem- brane association mediated by myristoylation is relatively weak, and thus some myristoylated proteins can partition between the membrane and cytosol depending on protein conformation, interactions with other pro- teins, and other post-translational modifications that are present. Thus, myristoylation can serve as one component of a regulatable membrane- switching mechanism. Indeed, in some proteins, such as the catalytic (a) (b) (c) (d) Figure 5.8 Mechanisms for localizing proteins to membranes. Mechanisms include: an integral membrane protein, covalent lipid modification, (c) association via a membrane-binding domain, and (d) association by binding to another membrane protein. Glycosylphosphatidylinositol (GPi) anchor C-terminus Strong no Table 5.1 Properties of common lipid modifications of proteins N-myristoylation n-terminus Weak no S-acylation internal cysteine Weak Yes Prenylation C-terminus Weak no subunit of PKA, the myristoyl group is buried in a hydrophobic groove on the protein surface and thus does not lead to membrane association. However, in many cases, myristoylation is sufficient to target a protein to the membrane. Thus, engineered myristoylation sites are often used experimentally to test the consequences of forced membrane localization of a protein of interest. S -acylation involves the addition of an acyl group, most often the 16-carbon fatty acid palmitate, to cysteine residues of a protein O (Figure 5.9). Unlike most other lipid modifications, S-acylation is read- ily reversible in the cell, thus proteins can cycle to and from the mem- O O brane upon addition or removal of lipid. As in the case of myristoylation, S-acylation leads to relatively weak membrane association, thus it often cooperates with other mechanisms of membrane association to mediate tighter membrane binding. For example, some Src family nonreceptor tyrosine kinases are thought to be transiently directed to the plasma membrane by N-myristoylation, then to become more stably associated after addition of palmitate groups by membrane-associated protein S-acyl transferases. The addition of relatively bulky isoprenoid lipids, such as farnesyl or geranylgeranyl groups, to the C-terminus of a protein is known as pre- nylation (Figure 5.9). The lipid is added to the cysteine within a so- called CAAX box motif (where A is an aliphatic amino acid and X is any amino acid), and this is followed by proteolytic removal of the AAX and methylation of the terminal carboxyl group. Like other lipid modi- fications, prenylation can be sufficient to direct membrane association, but stable binding often requires additional interactions. One of the most important classes of prenylated proteins includes various G pro- teins such as the Ras family of oncogenic small GTPases, whose function requires their membrane association. Small molecules that block pre- nylation, such as the statin class of anticholesterol drugs (which inhibit the rate-limiting step in isoprenoid biosynthesis), are being investigated for their ability to inhibit the function of prenylated proteins such as Ras family G proteins. N-myristoylation S-acetylation prenylation Figure 5.9 Protein lipid modifications. the structures and linkages of common lipid modifications are depicted. the outer (extracellular) leaflet of the plasma membrane is oriented toward the top of the figure. site of attachment to protein (green) is also indicated for each type of modification. gPi, glycosylphosphatidylinositol. Lipid kinases and lipid phosphatases are discussed in Chapter 7 Modular lipid-binding domains are important for regulated association of proteins with membranes Many signaling proteins can associate with membranes via small, modular lipid-binding domains. These domains include such exam- ples as the PH, FYVE, and PX domains. In many cases, a lipid-binding domain is highly specific for binding to a particular lipid head group. For example, different examples of the PH domain bind with relatively high affinity to specific phosphoinositides. Membrane association by such domains is therefore controlled by the amount and local concen- tration of these lipids which, in turn, are controlled by lipid kinases and lipid phosphatases whose activity can be regulated in signaling. The affinity of such domains for their lipid targets is often relatively low, so other factors (for example, interaction with other membrane-associated proteins) may be necessary for stable binding (Figure 5.10a). In this way, stable membrane association of a protein can be made to depend cooperatively on multiple inputs, allowing for integration of multiple signals and the generation of more switchlike, all-or-none changes in localization. A second class of membrane-binding motifs includes various positively charged protein surfaces that interact electrostatically with membranes, which generally carry a net negative charge due to the phosphate groups of phospholipids. These protein surfaces include positively charged amphipathic helical segments. Other types of positively charged motifs are somewhat specific for particular highly phosphorylated lipids, such as phosphatidylinositol 4,5-bisphosphate [PI(4,5)P ₂ ]. PI(4,5)P₂ is a relatively rare membrane lipid that plays a disproportionately large role in signaling through its ability to bind cytosolic proteins, and its ability to generate soluble second messengers after cleavage. Such positively charged motifs have interesting potential regulatory prop- erties. For example, they have the potential to be highly multivalent (to bind to multiple phosphate groups on the membrane); because of this, binding to the membrane surface can be highly cooperative and thus vary dramati- cally with relatively small differences in the local concentration of the lipid binding partners, such as phosphoinositides (Figure 5.10b). Furthermore, because the interaction of such motifs with membranes is dependent on their positive charge, their activity can be abolished by post-translational modifications, such as phosphorylation, that increase negative charge. This mechanism is used, for example, to regulate the membrane localization of the MAP kinase scaffold protein Ste5 during the cell cycle in yeast. In this case, phosphorylation of a positively charged membrane-localization motif by a cyclin-dependent kinase, which is activated in the G₁ phase of the cell cycle, abolishes Ste5 membrane localization. Some lipid-modified proteins can reversibly associate with membranes If the membrane binding of proteins containing bulky lipid groups is to be regulated in signaling, there must be a mechanism to shield the hydrophobic lipid group from the hydrophilic environment of the cytosol when it is not embedded in the membrane. One such mechanism involves allosteric changes in the three-dimensional structure of the protein itself. For example, some isoforms of the Abl nonreceptor tyrosine kinase are modified by N-myristoylation. In the inactive conformation of the kinase, the myristoyl group folds up into a hydrophobic groove on the surface of the protein and helps maintain and stabilize the inactive conforma- tion. Activation of Abl is associated with a concerted conformational change that eliminates the myristoyl binding pocket and thus exposes the free myristoyl group, which can then insert into the membrane. By this means, the kinase activity of Abl can be functionally coupled to mem- brane association. Lipid groups can also be shielded by association with specific binding proteins. This is best understood in the case of prenylated small G proteins, for example the Rho and Rab families. In the case of the Rab family, a protein called RabGDI (Rab guanine nucleotide dissociation inhibitor) binds to membrane-associated Rab proteins and can extract them from the membrane by shielding their prenyl groups in a hydropho- bic pocket (Figures 5.11 and 5.12). The RabGDI–Rab complex is soluble, and can then be transported to new sites in the cell, where interaction with target membranes leads to dissociation of the GDI and insertion of the prenyl group of the G protein into the lipid bilayer. This process may be facilitated by so-called GDI displacement factors (GDFs) on the target membrane, which interact with the GDI and stimulate release of the bound G protein. In some cases, guanine nucleotide exchange factors (GEFs) may also coordinate with the GDI in coupling membrane binding to the activation state of the G protein. RhoGDI (Rho guanine nucleotide dissociation inhibitor) plays a similar role in regulating the membrane association and guanine nucleotide binding of Rho family G proteins such as Rac and Cdc42. (b) Figure 5.10 Cooperative binding to membranes. (a) cooperative binding via multiple independent interactions. binding via a membrane-binding domain (left) or via association with another membrane protein (middle) is relatively weak in both cases, and most of the protein will therefore remain in the cytosol. When both interactions are present, however, binding is much stronger and most of the protein is associated with the membrane. (b) this binding domain has multiple sites for binding a specific membrane phospholipid (pink). When the local density of the lipid is low (left), binding is weak and most of the protein is cytosolic. When local density of the lipid is high (right), cooperative binding leads to very strong binding to the membrane. Figure 5.11 The RabGDI cycle. rab guanine nucleotide dissociation inhibitor (rabgDi) binds to gDP-bound rab ( p u r p l e ) and extracts it from the membrane. the soluble rabgDi–rab complex can then move through the cytosol to a new target membrane. Dissociation of rabgDi, which can be facilitated by gDi displacement factors (gDfs), allows insertion of rab into the target membrane. guanine nucleotide exchange factors (gefs) on the target membrane can activate rab by inducing release of gDP and binding of gtP. gaP, gtPase-activator protein; Pi, inorganic phosphate. (a) geranylgeranyl geranylgeranyl (b) geranylgeranyl 1 GDF Coupling effector protein activation to membrane recruitment is a common theme in signaling Activation of plasma membrane receptors often leads to the membrane recruitment of effector molecules which, in turn, become activated and transmit downstream signals. Below, we discuss the example of the Akt kinase, just one of many protein kinases activated as a direct consequence of regulated membrane recruitment. In such cases, the enzyme is acti- vated by the presence of activating enzymes or cofactors that are localized on the membrane. In other cases, membrane localization of an enzyme may not increase its state of activity, but its output is increased (that is, the number of reac- tions it can catalyze per unit time) simply by bringing it into close prox- imity with its substrates. The most striking example of this is activation of the small G protein Ras by receptor tyrosine kinases. This is a crucial step in signaling pathways that control proliferation and differentiation. The enzyme that actually activates Ras is a GEF called Sos. While Ras is tightly associated with membranes by virtue of covalent lipid modifica- tions, its activator Sos is cytosolic in unstimulated cells, and therefore only rarely encounters its membrane-bound substrate. Upon receptor activa- tion, however, Sos is recruited to the membrane where Ras is found, thus greatly increasing the efficiency of Ras activation by Sos (Figure 5.13). geranylgeranyl 2 Figure 5.12 GDI Akt kinase is regulated by membrane recruitment and phosphorylation The Akt family of serine/threonine kinases (also called the PKB family) Structure of Rab–RabGDI complex. ribbon diagram of the x-ray crystal structure of the yeast rab family g protein ypt1 ( y e ll o w ) complexed with rabgDi (blue). ypt1 is modified by addition of two geranylgeranyl lipid groups, indicated in pink and orange. (b) space-filling model showing the two ypt1 geranylgeranyl groups (pink and orange) fitting into hydrophobic grooves on the surface of gDi (white). (adapted from o. Pylypenko et al., EMBO J. 25:13–23, 2006. With permission from Macmillan Publishers Ltd.) Phosphoinositide kinases and phosphatases will be discussed in more detail in Chapter 7 provides a good example of activation by membrane recruitment. Akt plays a central role in transmitting signals that regulate cell growth, prolifera- tion, and survival. Its activation depends on phosphoinositides. Specifi- cally, Akt is dependent on phosphatidylinositol 3,4,5-trisphosphate [PI(3,4,5)P ₃ ] which is generated by a lipid kinase, phosphatidylinosi- tol 3-kinase (PI3K). PI3Ks can be activated by upstream signals by sev- eral distinct mechanisms, which themselves involve recruitment to the membrane through protein–protein interactions, coupled with conforma- tional changes. As we have seen before, membrane recruitment increases the availability of substrate for an enzyme, such as PI3K, that targets membrane components. The net result of PI3K activation is a high local concentration of its product PI(3,4,5)P₃. It is this increased density of PI(3,4,5)P₃ that triggers activation of Akt. Like many kinases, Akt is activated by phosphorylation on several regu- latory sites. At least one of these critical activating phosphorylations is performed by a second protein kinase, PDK1. Both Akt and PDK1 contain ligand a PH domain (a modular lipid-binding domain) that binds specifically to PI(3,4,5)P₃. Thus, the activating enzyme (PDK1) and its substrate (Akt) are both recruited to the same patch of membrane, where PDK1 then phosphorylates and activates Akt (Figure 5.14). Conformational changes upon membrane binding may also facilitate phosphorylation of Akt by PDK1. Akt phosphorylation induces further conformational changes that decrease its affinity for PDK1 and membranes, freeing Akt to dif- fuse throughout the cytosol and nucleus to phosphorylate targets in these compartments. Thus, the membrane plays a transient, albeit critical, role in Akt activation. MoDuLation of signaLing by MeMbrane trafficKing Figure 5.13 Activation of Ras. (a) in unstimulated cells, the ras activator sos is associated with the adaptor protein grb2 in the cytosol; its concentration at the membrane is relatively low. (b) a receptor tyrosine kinase is activated by binding its ligand and autophosphorylates, creating binding sites for the sH2 domain of grb2. thus, grb2 and sos are recruited to the membrane. the concentration of sos at the membrane is now high, and ras is activated. Cellular membranes are highly dynamic. Vesicles containing lipids and membrane proteins continuously shuttle from the endoplasmic reticulum and Golgi apparatus toward the cell surface. At the same time, plasma membrane components are internalized and sorted for different fates, such as destruction in lysosomes or recycling back to the surface. Thus, the many signaling proteins that are intimately associated with mem- branes can also be vectorially transported through these mechanisms. The resulting movement of signaling proteins between membrane com- partments with distinct subcellular locations, physical properties, and constituents, can profoundly influence their behavior, either positively or negatively. In this section, we will consider a few specific examples of where the transport of membrane proteins is important for signal output. (a) PIP₂ PIP₂ PIP₂ Proteins can be internalized by a variety of mechanisms Cell-surface proteins such as receptors are internalized by endocytosis, a process whereby a portion of the membrane invaginates and ultimately (b) PIP₃ PI3K activation PIP₃ PIP₃ Figure 5.14 Activation of Akt. (a) in unstimulated cells, both PDK1 and akt are cytosolic and inactive. PiP₂, phosphatidylinositol 4,5-bisphosphate. activation of phosphatidylinositol 3-kinase (P13K) by upstream signals leads to high local concentrations of the lipid phosphatidylinositol 3,4,5-trisphosphate (PiP₃) on the membrane. both PDK1 and akt bind to PiP₃ via their PH domains. once recruited to the membrane, PDK1 phosphorylates akt, promoting its activation. activated akt can dissociate from the membrane to phosphorylate substrates in other subcellular locations. Phosphorylation at a second site by a different kinase (the mtorc2 complex) is also required for full activation of akt. Figure 5.15 Mechanisms of endocytosis. clathrin-mediated endocytosis. caveolin-mediated endocytosis. Macropinocytosis. after endocytosis, vesicles are sorted and either recycled to the cell surface or directed to lysosomes for degradation. (c) macropinocytosis pinches off into the cytosol, forming a vesicle termed an endosome that is no longer physically connected to the outside environment. Receptors and other membrane components can be internalized by a variety of specific mechanisms (Figure 5.15). Perhaps the best charac- terized of these endocytic pathways is clathrin-mediated endocytosis. This involves the cooperative assembly of the protein clathrin into a hol- low framework or coat around a patch of membrane, which ultimately pinches off. In the case of receptors, this process is often promoted by their ubiquitylation which, in turn, is mediated by ubiquitin ligases that are recruited upon receptor activation. Once free of the plasma membrane, the vesicle sheds its clathrin coat and the vesicle (now termed an early endosome) and its contents are targeted to their next destination. By con- trast, caveolin-mediated endocytosis occurs through the invagination of patches of membrane, termed caveolae, enriched in the lipid cholesterol and the protein caveolin. Finally, macropinocytosis involves large-scale rearrangements of the actin cytoskeleton (often termed circular dorsal ruffles), leading to engulfment of a large area of membrane, its associated proteins, and extracellular fluid. For a given protein or lipid molecule, the kinetics of its internalization and its ultimate fate (that is, recycling to the cell surface versus destruction in the lysosome) are highly dependent on the specific mechanism of internalization. Receptor down-regulation is discussed in Chapter 8 Internalization of receptors can modulate signal transduction The most straightforward way in which membrane transport affects sig- naling is through the internalization of cell-surface receptors. Once inter- nalized, receptors no longer have access to extracellular ligands, and may be targeted for destruction in the lysosome, and so this is often used as a mechanism to down-regulate receptor signaling after stimulation. There are also examples where receptors are stored in intracellular vesicles and transported to the cell surface in response to signaling inputs. This mecha- nism is used to up-regulate the number of receptors for the neurotrans- mitter AMPA in the dendritic spines of neurons when they are exposed to certain types of stimuli, which is thought to be important for learning and memory. Any ligand already bound to its receptor is likely to remain bound for some time after internalization, due to the generally slow off-rate for receptor–ligand interactions and the small volume of the lumen of the vesicle, which ensures that the local concentration of ligand remains quite high even if it dissociates from the receptor. However, the pH in the lumen of an endocytic vesicle generally decreases over time en route to the lyso- some, and some ligand–receptor interactions are sensitive to this acidifi- cation. For example, among the closely related ligands and receptors of the epidermal growth factor (EGF)/EGF receptor family, differences in the acid stability of binding likely lead to differences in the time that internal- ized receptor–ligand complexes remain active. This, in turn, affects the fraction of receptors that are fated for degradation versus recycling back to the surface. There is considerable evidence that activated receptors, such as the EGF receptor, continue to signal after internalization. This makes sense, as the intracellular portion of the receptor that transmits downstream sig- nals remains exposed to the cytosol, whether the receptor is on the cell surface or an endosome. In some cases, however, the vesicle environment restricts the output signal compared to the plasma membrane. One way in which this can occur is in the availability of substrates. To provide a specific example, tyrosine kinase receptors such as the EGF receptor transmit signals in part by recruiting enzymes such as phospholipase C (PLC) and PI3K that target membrane lipids. The substrate for both of these enzymes is PI(4,5)P₂. It is easy to see how, in a small vesicle, recep- tor-associated PLC or PI3K activity would rapidly deplete any available PI(4,5)P₂. By contrast, on the cell surface, a much larger pool of substrate is available, and thus the potential magnitude of the downstream signal is much greater. TGFβ signaling output depends on the mechanism of receptor internalization The transforming growth factor β (TGFβ) receptor provides an excellent example of how the mode of endocytic transport can profoundly affect signal output. TGFβ family receptors possess intrinsic serine/threonine kinase activity. Ligand binding induces the heterodimerization and acti- vation of the receptors, which then phosphorylate and activate the down- stream effectors SMAD2 or SMAD3. This process is facilitated by the scaffold protein SARA. Phosphorylated SMAD2 and SMAD3 then dissoci- ate from the receptor and bind another subunit, SMAD4, and this complex translocates to the nucleus where it induces the transcription of specific target genes. TGFβ receptors are constitutively internalized by two distinct mecha- nisms, one of which promotes and one of which inhibits signal output (Figure 5.16). The early endosomes resulting from clathrin-mediated endocytosis promote activation. This is because these vesicles are high- ly enriched in the SARA protein, which promotes SMAD2 and SMAD3 phosphorylation and downstream signaling. This enrichment of SARA is likely due to high levels of the lipid phosphatidylinositol 3-phosphate [PI(3)P] in these vesicles, and the fact that SARA has a FYVE domain that specifically binds PI(3)P. Thus, clathrin-mediated endocytosis promotes TGFβ signaling, and blocking this pathway inhibits signal output. On the other hand, caveolin-dependent endocytosis of TGFβ receptors inhibits sig- naling. The vesicles resulting from this process are enriched for the “inhibi- tory SMAD” SMAD7, which recruits the ubiquitin ligases SMURF1 and SMURF2. SMURF-mediated ubiquitylation of the receptor targets it for degradation. Thus, the dynamic balance between the clathrin-dependent TGFβ signaling is discussed in greater detail in Chapter 8 Figure 5.16 Two pathways for internalization of TGFβ receptors. transforming growth factor β (tgfβ) receptors are internalized via clathrin-mediated endocytosis or caveolin- mediated endocytosis. in the clathrin- mediated pathway (left), downstream signaling is promoted by recruitment of sara and sMaD2 to phosphatidylinositol phosphate [Pi(3)P]-rich early endosomes. in the caveolin pathway (right), signaling is suppressed by recruitment of sMaD7 and the sMurf ubiquitin ligase, leading to receptor ubiquitylation and degradation. and caveolin-dependent endocytic pathways determines the ultimate strength of the downstream signal. Retrograde signaling allows effects distant from the site of ligand binding While it is clear that internalized receptors have the potential to gener- ate signals, in the case of neurons there is good evidence that receptor internalization is actually required for what is termed retrograde sign- aling. Neurotrophins are a family of ligands that engage receptor tyro- sine kinases, and transmit cell growth and survival signals in neurons. Because nerve processes can be very long (more than a meter long in the case of humans), transmission to the cell body of a signal generated by engagement of neurotrophin receptors at the end of an axon would be very slow and uncertain if it relied solely on the passive diffusion of protein intermediates or other signaling molecules. To circumvent this problem, it is thought that internalized vesicles containing the activated receptors, along with associated downstream effector proteins, are actively trans- ported along microtubules to the cell body by the molecular motor dynein (Figure 5.17). Such vesicles have been termed “signaling endosomes.” Ras isoforms in distinct subcellular locations have different signaling outputs Ras provides another example where the specific membrane localization of a signaling protein appears to affect its signal output. As we have already seen, Ras family small GTPases act as switches to regulate processes such as cell proliferation and differentiation. There are several isoforms of Ras in mammals that differ in their membrane-localization signals. While all forms are farnesylated at the C-terminus (see above), this lipid modifi- cation is not sufficient for stable membrane association, which requires a second signal. In the case of the H-Ras and N-Ras isoforms, this is provided by S-acylation with one or two palmitate groups, respectively, while the major K-Ras isoform contains a stretch of basic amino acid resi- dues that promotes plasma membrane localization through interaction with negatively charged membrane lipids. Palmitoylation of Ras occurs in the Golgi apparatus, and is followed by vesicular transport to the plasma membrane. Thus, N-Ras and H-Ras are thought to cycle between the Golgi (a) (b) (c) Figure 5.17 Retrograde signaling by neurotrophins. (a) neurotrophins activate their receptors at the end of neuronal processes, which can be a great distance from the cell body that must respond to the signal. (b) transmission of the signal would be slow and inefficient if effectors (light blue) activated by the receptor had to travel to the cell body via passive diffusion. (c) endocytosed receptor and associated effectors are actively transported to the cell body by the motor protein dynein. slow, inefficient fast, efficient and plasma membranes, with the rate of cycling and relative distribution among them determined by the rates of palmitoylation and depalmitoyla- tion. K-Ras does not require palmitoylation and thus is highly enriched in plasma membranes, though this can be regulated by the phosphorylation of residues flanking the polybasic site. Studies have shown that the plasma membrane and Golgi pools of Ras are activated and deactivated with different kinetics following stimula- tion, most likely because the specific GEFs and GAPs that can activate and inactivate Ras are differentially distributed in these compartments. Experiments in which activated Ras isoforms are artificially targeted to either the Golgi or endoplasmic reticulum showed that the downstream signaling outputs are distinct, presumably because of the availability of different Ras effectors in these compartments. In the fission yeast Schizosaccharomyces pombe, there is a single Ras protein that normally localizes to both the endoplasmic reticulum and plasma membrane. In this organism, Ras regulates both cell morphology and the mating response, and genetic studies have shown that these two pathways employ different upstream GEFs and downstream effectors. In the absence of wild-type Ras, expression of a Ras mutant localized exclu- sively to the endoplasmic reticulum rescues the morphology defects but not the mating defect (Figure 5.18). Conversely, a Ras mutant exclusively localized to the plasma membrane rescues the mating defect but not the morphology defect. These experiments clearly highlight the dependence of signal output on subcellular localization. Figure 5.18 Correlation of subcellular localization of Ras with signaling output. in wild-type fission yeast cells, ras localizes to both the plasma membrane (PM) and endomembranes (er). the ability of ras mutants with more restricted localization to rescue defects in mating and cell morphology is indicated in the table on the left by plus and minus symbols. on the right are photomicrographs of cells expressing the engineered ras mutants (fused to cyan fluorescent protein). Mutant 1 localizes exclusively to the plasma membrane, while mutant 2 localizes exclusively to endomembranes. (adapted from b. onken, H. Wiener, M.r. Philips and e.c. chang, Proc. Natl Acad. Sci. USA 103:9045–9050, 2006. With permission from national academy of sciences, usa.) mutant 1 mutant 2 suMMary The activity of a signaling protein is highly dependent on its location within the cell. Many signaling proteins undergo regulated changes in localization as a means to transmit information. Signaling proteins that control transcription often undergo regulated changes in nuclear locali- zation. Another key location for coordination of signaling is at the cell membrane. Many signaling proteins are localized to the plasma mem- brane, either constitutively or in a signal-regulated manner. Coordinated localization of proteins at the membrane often controls when and if sig- naling partner proteins can effectively communicate with one another. Membranes are dynamically redistributed throughout the cell by vesicu- lar transport. Signaling proteins associated with the membrane, such as receptors, can be internalized into endosomes and other organelles, which can be used to further modulate signal output. Questions Explain the basic principles by which changes in the subcellular locali- zation of signaling proteins can be used as a mechanism to transmit information. In your studies of how a cell responds to stimulation by a specific extra- cellular growth factor, you discover that a downstream protein trans- locates from the nucleus to the cytoplasm upon stimulation. How is it possible for a protein localized mostly in the nucleus to respond to signaling changes in the cytosol or plasma membrane? What are the different mechanisms by which phosphorylation can reg- ulate the nuclear–cytoplasmic localization of a protein? In what ways can internalization of a membrane-bound receptor affect signaling pathways involving that receptor, and the dynamics of the response? How could the signaling properties of a protein differ depending on its subcellular localization? Small G proteins play a central role in determining subcellular local- ization, not only for nuclear import/export (Ran), but also in vesicle transport and trafficking (Rab family). Which aspects of G proteins and their regulation make them well suited to this role? references controL of nucLear LocaLization Aaronson DS & Horvath CM (2002) A road map for those who don’t know JAK-STAT. Science 296, 1653–1655. Bridges D & Moorhead GBG (2005) 14-3-3 proteins: a number of functions for a numbered protein. Sci. STKE 2005(296), re10. Burack WR & Shaw AS (2005) Live cell imaging of ERK and MEK. J. Biol. Chem. 280, 3832–3837. Chatterjee-Kishore M, van den Akker F & Stark GR (2000) Association of STATs with relatives and friends. Trends Cell Biol. 10, 106–111. Ebisuya M, Kondoh K & Nishida E (2005) The dura- tion, magnitude and compartmentalization of ERK MAP kinase activity: mechanisms for providing signaling spe- cificity. J. Cell Sci. 118, 2997–3002. Fortini ME (2009) Notch signaling: the core path- way and its posttranslational regulation. Dev. Cell 16, 633–647. Komeili A & O’Shea EK (1999) Roles of phosphorylation sites in regulating activity of the transcription factor Pho4. Science 284, 977–980. Ranganathan A, Yazicioglu MN & Cobb MH (2006) The nuclear localization of ERK2 occurs by mechanisms both independent of and dependent on energy. J. Biol. Chem. 281, 15645–15652. referenCeS Riddick G & Macara IG (2005) A systems analysis of importin-α-β mediated nuclear protein import. J. Cell Biol. 168, 1027–1038. Stark GR & Darnell JE Jr (2012) The JAK-STAT path- way at twenty. Immunity 36:503–514. Stewart M (2007) Molecular mechanism of the nuclear protein import cycle. Nat. Rev. Mol. Cell Biol. 8, 195–208. Weis K (2003) Regulating access to the genome: nucleo- cytoplasmic transport throughout the cell cycle. Cell 112, 441–451. controL of MeMbrane LocaLization Behnia R & Munro S (2005) Organelle identity and the signposts for membrane traffic. Nature 438, 597–604. Calleja V, Alcor D, Laguerre M et al. (2007) Intramolecu- lar and intermolecular interactions of protein kinase B define its activation in vivo. PLoS Biol 5, e95. Carlton JG & Cullen PJ (2005) Coincidence detection in phosphoinositide signaling. Trends Cell Biol. 15, 540–547. Cho W (2006) Building signaling complexes at the mem- brane. Sci. STKE (2006)321:pe7. Dransart E, Olofsson B & Cherfils J (2005) RhoGDIs revisited: novel roles in Rho regulation. Traffic 6, 957–966. Fivaz M & Meyer T (2003) Specific localization and tim- ing in neuronal signal transduction mediated by pro- tein-lipid interactions. Neuron 40, 319–330. Hantschel O, Nagar B, Guettler S et al. (2003) A myr- istoyl/phosphotyrosine switch regulates c-Abl. Cell 112, 845–857. Magee T & Seabra MC (2005) Fatty acylation and pre- nylation of proteins: what’s hot in fat. Curr. Opin. Cell Biol. 17, 190–196. Pylypenko O, Rak A, Durek T et al. (2006) Structure of doubly prenylated Ypt1:GDI complex and the mecha- nism of GDI-mediated Rab recycling. EMBO J. 25, 13–23. Strickfaden SC, Winters MJ, Ben-Ari G et al. (2007) A mechanism for cell-cycle regulation of MAP kinase signaling in a yeast differentiation pathway. Cell 128, 519–531. MoDuLation of signaLing by MeMbrane trafficKing Di Guglielmo GM, Le Roy C, Goodfellow AF & Wrana JL (2003) Distinct endocytic pathways regulate TGF-β receptor signalling and turnover. Nat. Cell Biol. 5, 410–421. Eisenberg S & Henis YI (2008) Interactions of Ras pro- teins with the plasma membrane and their roles in sign- aling. Cell. Signal. 20, 31–39. Haugh JM (2002) Localization of receptor-mediated sig- nal transduction pathways: the inside story. Mol. Interv. 2, 292–307. Le Roy C & Wrana JL (2005) Clathrin- and non-clathrin- mediated endocytic regulation of cell signalling. Nat. Rev. Mol. Cell Biol. 6, 112–126. Onken B, Wiener H, Philips MR & Chang EC (2006) Compartmentalized signaling of Ras in fission yeast. Proc. Natl Acad. Sci. U.S.A. 103, 9045–9050. Rocks O, Peyker A & Bastiaens PI (2006) Spatio-tempo- ral segregation of Ras signals: one ship, three anchors, many harbors. Curr. Opin. Cell Biol. 18, 351–357. Zweifel LS, Kuruvilla R & Ginty DD (2005) Functions and mechanisms of retrograde neurotrophin signalling. Nat. Rev. Neurosci. 6, 615–625. Taylor & Francis Taylor & Francis Group https://taylorandfrancis.com second messengers: small signaling mediators Much of the information transmitted within cells is carried by large mac- romolecules, such as proteins or nucleic acids. As we have seen in previous chapters, incoming signals can alter proteins in a number of ways—they can induce conformational changes, changes in complex formation, or post-translational modifications. However, information can also be car- ried by much smaller and simpler molecules; these include Ca2+, various lipid-derived mediators, and the cyclic nucleotides cAMP and cGMP. In this chapter, we discuss these small signaling mediators and their spe- cial properties. ProPerties of small signaling mediators When signaling inputs cause a change in the concentration of a small sig- naling mediator, that change is detected by downstream effector proteins that bind the mediator. Mediator binding leads to conformational changes in the effectors and ultimately to changes in their activity. Thus, informa- tion is conveyed by the concentration and distribution of these media- tors, and how they change over time. These small signaling mediators are often referred to as second messengers, a term that derives largely from their historical discovery as signals that were produced downstream of hormone stimulation, where the hormone itself was considered the first messenger. Signaling by small-molecule mediators differs in several respects from more typical signaling mechanisms based only on changes in proteins. Changes in the concentration of the mediators can be quite rapid and can result in an enormous amplification of an input signal. And since most small signaling mediators are highly diffusible, their effects can spread rapidly throughout the cell. Figure 6.1 INPUT Small signaling mediators are controlled by an interplay of their production and elimination For a small signaling mediator to be an effective information-carrying molecule, its concentration must differ significantly under unstimul- ated and stimulated conditions. The steady-state concentration of these mediators is determined by a balance between the production of the mediator and its elimination. Usually, stimulation of the system by an input leads to an increase in the rate of synthesis of the mediator (for example, the allosteric activation of the enzyme that synthesizes the mediator), but stimulation can also involve a decrease in the breakdown or elimination of the mediator (Figure 6.1). This situation is analogous to the opposing activities of “writers” and “erasers” in the case of post- translational modifications of proteins. To provide a specific example, the mediator cAMP is produced from ATP by the enzyme adenylyl cyclase, and is eliminated by the enzyme cAMP phosphodiesterase. Many signal- ing pathways involve stimulation of adenylyl cyclase by heterotrimeric G proteins, leading to a large, transient increase in cAMP concentration in the cell. Unlike other mediators, the concentration of Ca2+ is not regulated by its synthesis and degradation. Instead, the concentration of Ca2+ in the cytoplasm is regulated by the opposing activities of pumps that remove it, and channels that allow it to flow back in. In the basal state, Ca2+ concentrations in the cytoplasm are kept low by Ca2+-pumps. These use the energy of ATP to actively transport Ca2+ against its concentration gradient either out of the cell, or into the endoplasmic reticulum, which serves as an intracellular reservoir of Ca2+. Upon signaling, Ca2+ chan- Small signaling mediators are controlled by a balance of synthesis and degradation. the concentration and distribution of signaling mediators are coordinately controlled by opposing activities—the enzyme that synthesizes the mediator, and the enzyme that degrades or removes it. in the case of Ca2+, levels are controlled in a similar fashion by the opposing activities of channels and pumps. small signaling mediators, because they are highly diffusible, can act at greater distances than macromolecular signaling molecules. nels open and the ions rapidly enter the cell, flowing down the concen- tration gradient, leading to a rapid increase in Ca2+ concentration in the cytoplasm. Upon cessation of the signal, channels close and pumps once again restore the basal concentration gradient. Thus, the opposing actions of channels and pumps play an equivalent role to synthesis and degradation by enzymes. In all of these cases, the proteins that increase and decrease mediator concentrations function in a coordinated fashion in many different sig- naling contexts. A variety of different upstream inputs can be linked to production of a particular mediator. Similarly, as discussed below, a single mediator can activate a wide range of downstream effectors. Small signaling mediators exert their effects by binding downstream effectors Small signaling mediators such as cAMP or Ca2+ exert their biological effects by binding to a wide range of downstream enzymes and channels that, in response, are either allosterically activated or repressed. In some cases, the mediator exerts its effects indirectly by binding to regulatory subunits, as in the case of protein kinase A (cAMP binds to an inhibitory subunit that releases the active kinase subunit). Lipid mediators can also bind and regulate effector proteins by recruiting them to the membrane. Often, proteins that respond to the same mediator will contain evolution- arily related modular domains or motifs that transduce signals from that mediator, such as cAMP- or Ca2+-binding modules. These binding modules are functionally analogous to “reader” domains that recognize post-trans- lational modifications, as discussed in Chapter 4. The fact that the small signaling mediators exert their effects solely through binding to downstream effector proteins distinguishes them from other ions and small molecules that participate in signaling. For ProPerties of small signaling mediators example, in excitable cells such as neurons and muscle, the regulated opening and closing of channels for Na+, K+, and other ions leads to very rapid changes in membrane potential that can be amplified and propa- gated. These changes in voltage across the membrane provide the basis of neuronal signaling, but are not considered further here. Note that in this case, the ions primarily exert their influence by carrying electric charge, not by binding and allosterically changing the activity of effector proteins. For effector proteins to properly read and respond to the mediator signal, it is critical that their dissociation constants for binding (and activation/ repression) be properly tuned so that binding is low at the basal steady- state concentration of mediator, but high under activated steady-state concentrations. If, for example, the dissociation constant of an effector was lower than the basal steady-state concentration of a mediator (affinity was too high), then most of the effector would already be bound to media- tor before stimulation; a further increase in mediator concentration upon signaling would have little or no impact. Small signaling mediators can lead to fast, distant, and amplified signal transmission The rate of diffusion in solution is inversely proportional to the radius of the diffusing molecule. Because signaling mediators are relatively small, they can, in principle, rapidly diffuse throughout a cell to transmit information. Signaling through such mediators can occur on the milli- second time scale and can rapidly cover distances of up to an entire cell. Figure 6.2 shows the relative sizes of a macromolecular signaling protein (a protein kinase catalytic domain; radius ~50 Å) and the small signaling mediators cAMP (radius ~5 Å) and Ca2+ (radius ~1 Å). Also shown are the relative concentration profiles for each molecule at a given time after dif- fusion from an equal concentration point source. Molecules in the smaller size range have the potential to spread farther and faster than macro- molecules. Some mediators, such as the gas nitric oxide, can even diffuse through membranes and mediate communication from cell to cell. The actual distance that small-mediator signals propagate, however, is also determined by the lifetime of the molecule. For example, nitric oxide is very short-lived, and therefore only propagates within a few cells. Simi- larly, in the case of Ca2+, the high ambient concentration of Ca2+-binding proteins results in the rapid sequestration of Ca2+ in the cytosol. But as discussed in the next section, this interplay between production, diffusion, The tuning of affinity to the physi­ ological concentration range of ligands is discussed in Chapter 2 Figure 6.2 Relative sizes and diffusion of protein and small-molecule signaling mediators. (a) relative sizes of a typical signaling protein [the kinase domain of protein kinase a (PKa)] and two small signaling mediators, camP and Ca2+. (b) theoretical distribution of molecules from a point source after a fixed time of diffusion. diffusion rates are calculated using the radii given in part (a) using the stokes– einstein equation, d = kT/(6πηr), where k is Boltzmann’s constant, T is the absolute temperature, η is solvent viscosity, and r is the radius. as can be seen, small-molecule mediators can diffuse further and faster than macromolecular ones. (a) kinase domain cAMP calcium (b) radius ~ 50 Å radius ~ 5 Å radius ~ 1 Å 200 100 0 100 200 distance from origin (μm) Figure 6.3 The rise and fall of small signaling mediator concentration in response to the beginning and end of an upstream stimulus. (a) Control over the steady-state level of a small-molecule signaling mediator is analogous to control over the steady-state level of water flowing in to and out of a bucket. an increase in steady-state level can be achieved by either increasing the flow rate of water into the bucket or by decreasing the flow rate out of the bucket (or by a combination of both mechanisms). (b) a transient rise in mediator concentration can be induced by an increase in synthesis. restoration of mediator concentration to a low, basal level after removal of the stimulus is dependent on a high basal degradation rate. (c) a similar increase in mediator concentration can also be induced by a transient decrease in degradation rate. and destruction/sequestration can lead to signaling outputs with complex spatial and temporal behaviors. Small signaling mediators also have the potential to greatly amplify sig- nals. In the case of mediators that are generated by enzymes, the acti- vation of only a small number of synthetic enzymes can result in the production of an enormous number of mediator molecules. Similarly, even the momentary opening of a small number of membrane channels will allow the passage of many Ca2+ ions. In either case, if the initial con- centration of the mediator is low it will rapidly and massively increase. It is the potential for enormous amplification combined with the poten- tial for rapid spread of the resulting signal throughout the cell that sets signaling by small-molecule mediators apart from other signaling mechanisms. Small signaling mediators can generate complex temporal and spatial patterns Because the levels of small signaling mediators are controlled by oppos- ing production and elimination activities, coordinated regulation of both can yield complex and diverse signaling dynamics. Figure 6.3a compares signaling mediator levels to the fill level of a vessel with water flowing in and out. The steady-state level of water can be changed either by tuning the IN flow rate or the OUT flow rate (or both). Similarly, change in the levels of second messengers can be coordinately controlled both by their production and their destruction. Figure 6.3b shows a schematic time profile of change in mediator concen- tration in response to a transient stimulus that increases the synthesis (a) (b) synthesis rate degradation rate mediator concentration (c) time detection threshold synthesis rate decrease flow out degradation rate mediator concentration time detection threshold rate. Efficient restoration of basal mediator levels, after the input stimu- lus ends, is dependent on a sufficiently rapid basal rate of mediator deg- radation. Transient increases in mediator level could also be induced by distinct complementary mechanisms: an input stimulus could decrease the degradation rate (Figure 6.3c), or coordinately increase mediator syn- thesis rate and decrease degradation rate. Small signaling mediators can also be used to generate more complex spatiotemporal patterns. For example, highly local activation of syn- thetic enzymes, but widespread distribution of degradation enzymes (or sequestrating binding proteins, in the case of Ca2+), can lead to sharp gradients of mediator. Temporal patterns, such as rapid adaptation or oscillation, can also be produced by an interplay between the synthesis and elimination enzymes. An example is provided later in this chapter of highly complex wave patterns that can be generated in the case of Ca2+ signaling. Classes of small signaling mediators The various classes of small signaling mediators are listed in Table 6.1, along with the enzymes or proteins involved in their production and elim- ination, and regulatory inputs and targets. In this section, we will focus in detail on the mechanism of action and regulation of a few prominent mediators: cyclic nucleotides and the lipid-derived signaling mediators inositol trisphosphate (IP₃) and diacylglycerol (DAG). The following sec- tion will focus on the special properties of signaling mediated by Ca2+. Table 6.1 Classes of commonly used small- molecule signaling mediators* camP adenylyl cyclase camP intracellular (long range) gPCrs Kinases, channels, cgmP guanylyl cyclase cgmP intracellular (long range) nitric oxide, rhodopsin Kinases, channels inositol trisphosphate (iP3) diacylglycerol (dag) Phospholipase C (PlC) Phospholipase C (PlC) dephosphorylation intracellular (long Phosphorylation membrane rtKs, gPCrs Ca2+ channels rtKs, gPCrs Kinases Phosphoinositides (e.g., PiP₃) Pi kinases Pi phosphatases membrane Kinases, many PH domains eicosanoids (e.g., prostaglandins) Cyclooxygenase, prostaglandin synthases dehydrogenase, reductase extracellular Pla₂ gPCrs Ceramide sphingomyelinase Ceramidase membrane tnf and il-2 receptors Kinases/ phosphatases sphingosine phosphate Ceramidase/ sphingosine kinase sphingosine phosphatase extracellular tnf and il-2 gPCrs Ca2+ Ca2+ channels (release from er stores) Ca2+-atPase pumps, Ca2+-sequestering proteins intracellular (0.1–1 μm) Channels, iP3 Channels, kinases, enzymes, cyto- skeletal proteins (calmodulin) nitric oxide (no) nitric oxide synthase reaction with o₂, heme binding Ca2+/calmodulin, phosphorylation cgmP (guanylyl cyclase) * this table shows selected examples from the classes, and is not an exhaustive list. gPCrs, g-protein-coupled receptors; gefs, guanine nucleotide exchange factors; rtKs, receptor tyrosine kinases; PiP₃, phosphatidylinositol 3,4,5-trisphosphate; Pi, phosphatidylinositol; Pla₂, phospholipase a₂; tnf, tumor necrosis factor; il-2, interleukin-2; er, endoplasmic reticulum. NO signaling will be discussed in more detail in Chapter 8 Figure 6.4 Synthesis and degradation of cyclic nucleotides. the structures of the signaling mediators camP and cgmP are shown, along with their precursors and inactive products. Small signaling mediators have a wide range of physical properties While small signaling mediators share the defining property of rapid and widespread effects, their actual physical properties are quite variable. Typical mediators such as cyclic nucleotides and Ca2+ are strictly water- soluble, and thus cannot pass through the lipid bilayer. By contrast, some lipid-derived mediators are quite hydrophobic and therefore remain con- fined to membranes. These can only diffuse laterally in two dimensions within the plane of the membrane. Some other lipid-derived mediators are somewhat water-soluble, and can thus partition both to the mem- brane and to the cytosol or extracellular space, allowing them to transit from one cell to another. The small free-radical gas molecule nitric oxide (NO) is an interesting special case that easily traverses cell membranes by virtue of its small size and uncharged nature. NO is produced by the enzyme nitric oxide synthase (NOS), which converts L-arginine into citrulline and NO. NOS is activated by Ca2+/calmodulin. Once produced, NO is able to diffuse across the cell membrane to adjacent cells. Its effects are highly transient and local because NO is intrinsically very unstable and reacts spontaneously with oxygen or heme. The major target of NO is the enzyme guanylyl cyclase, which produces cGMP (itself another signaling mediator). NO plays an important role in smooth muscle relaxation and vessel dilation, thereby regulating blood flow and pressure. The cyclic nucleotides cAMP and cGMP are produced by cyclase enzymes and destroyed by phosphodiesterases The cyclic nucleotides cAMP and cGMP are used as signaling media- tors by a wide variety of organisms ranging from bacteria to vertebrates. Both cAMP and cGMP are synthesized—from the high-energy precursors ATP and GTP, respectively—by cyclase enzymes (Figure 6.4). Vertebrate adenylyl cyclases are large transmembrane proteins that contain evolu- tionarily conserved cytoplasmic cyclase catalytic domains. These enzymes are allosterically regulated by heterotrimeric G proteins (Figure 6.5). O ^– O– O P precursor H inactive product H O O– P O O O– N NH2 N O ^– O– N NH2 N P N O CH ₂ O N P N O CH ₂ O N H H H H H H H H H H OH OH OH OH O ^– O– O P ATP H2H synthetic enzyme degradative enzyme 5'-AMP O O– P O O O– P N O guanylyl N cyclase N cGMP phosphodiesterase O– O– H2H N O N O CH ₂ O N N O CH ₂ O N H H H H H H H H OH OH H H OH OH GTP 5'-GMP INPUT Figure 6.5 Signaling via the small signaling mediator cAMP. the enzyme that synthesizes camP, adenylyl cyclase, is activated by heterotrimeric g proteins. camP is degraded by the enzyme phosphodiesterase. most of the effects of camP are mediated through regulation of protein kinase a (PKa), and ePaC, a guanine nucleotide exchange factor (gef) for the small g proteins rap1 and rap2. Binding of camP to the regulatory (r) subunit of PKa causes dissociation of the catalytic (C) subunit. Binding of camP to ePaC relieves autoinhibition of the gef activity of this protein. gPCr, g-protein- coupled receptor. Certain Gα subunits activate adenylyl cyclase, whereas others can inhibit it. Vertebrates have two classes of guanylyl cyclases, including a soluble form that is activated by the upstream signaling mediator nitric oxide (NO), and transmembrane (receptor) forms that can be regulated by vari- ous upstream ligands. cAMP and cGMP are degraded by phosphodiesterase enzymes (PDEs), which convert the active molecules to the inactive forms 5′-AMP and 5′-GMP, respectively. Although most signaling pathways raise the steady-state level of cyclic nucleotides by activating their production (via cyclases), the phosphodiesterases can also be regulated. Ca2+/ calmodulin can activate cAMP phosphodiesterase, while Giα can activate cGMP phosphodiesterase, leading in both cases to a decrease in cyclic nucleotide levels. Phosphodiesterases can also be an important drug target (Figure 6.6). The erectile dysfunction drug sildenafil (Viagra®) is an inhibitor of cGMP phosphodiesterase. In vascular smooth muscle cells, blocking the destruction of cGMP (combined with low basal cyclase activity) leads to a buildup of high steady-state levels of cGMP. This cGMP kinase Figure 6.6 cGMP-gated ion channels causes smooth muscle relaxation and blood vessel dilation, resulting in increased blood flow to the penis. Cyclic nucleotides regulate diverse cellular activities The primary targets of cAMP in vertebrates are protein kinase A (PKA) and EPAC, a regulator of small G proteins (Figure 6.5). Both share related cAMP-binding modules, also known as cyclic nucleotide binding domains, which are found in species ranging from bacteria to humans. The details of activation of PKA by cAMP are discussed in the next section. EPAC is a guanine nucleotide exchange factor (GEF) that activates Rap1 and Rap2, two small G proteins that play a key role in regulating cellular Signaling via the small signaling mediator cGMP. one of the enzymes that synthesizes cgmP, soluble guanylyl cyclase, is activated by nitric oxide (no). cgmP is degraded by the enzyme phosphodiesterase. cgmP regulates various effectors, including kinases and ion channels, stimulating vascular smooth muscle relaxation. the drug sildenafil (Viagra®) functions by inhibiting the cgmP phosphodiesterase, leading to a buildup of cgmP. The visual signal transduction system is described in Chapter 12 Cooperative binding is explained in Chapter 2 adhesion interactions with the extracellular matrix. EPAC has a cata- lytic GEF domain that is regulated by several other domains, including cyclic nucleotide binding domains that are related to those found in the PKA regulatory subunit (EPAC1 has one such domain, EPAC2 has two regulatory domains). These domains are involved in autoinhibition of the catalytic GEF domain, which is relieved by cAMP binding. cGMP has several downstream targets, including the family of cGMP- dependent protein kinases (cGK, also known as protein kinase G or PKG) and cGMP-regulated ion channels (Figure 6.6). In addition, this mediator appears to regulate some cGMP phosphodiesterases, the enzymes that break down cGMP. This represents a form of feedback regulation that may be important for controlling signaling dynamics. In general, the effectors of cGMP are involved in inducing vascular smooth muscle relaxation and increased blood flow, and also play a key role in vertebrate vision. The regulatory (R) subunit of protein kinase A is a conformational sensor of cAMP binding Protein kinase A (PKA), or cyclic AMP-dependent protein kinase, is one of the best-understood examples of an enzyme whose activity is directly regulated by a small signaling mediator. As its name suggests, the ability of PKA to phosphorylate substrates is entirely dependent on the presence of cAMP. Thus, the role of PKA is to convert changes in intracellular cAMP concentration into changes in the phosphorylation of serine and threo- nine residues in substrate proteins. PKA has two subunits: a catalytic (C) subunit and a regulatory (R) subunit that suppresses its activity when bound (Figure 6.7a). The kinase activity of the catalytic subunit is tightly controlled by its association with the regulatory subunit, which acts as a conformational switch controlled by cAMP binding. The ability of the regulatory unit to inhibit catalytic activity is largely due to its pseudosub- strate-like inhibitory segment, which blocks the active site when it is teth- ered to the catalytic subunit (Figure 6.7b). The R subunit normally exists as a dimer, so the actual inactive complex is an R2C2 heterotetramer. In the enzyme’s inactive state (when levels of cAMP are low, typically below 10–8 M), the catalytic and regulatory subunits are tightly associ- ated—the Kd is less than 10–9 M—such that essentially every catalytic subunit is complexed with a regulatory subunit. In this complex, the regu- latory subunit sterically blocks substrates from binding to the enzyme active site of the catalytic subunit. Thus, when cAMP levels are low, the vast majority of PKA is in an inactive state, sequestered through stable interaction between the two subunits. Each regulatory subunit has two distinct binding sites for cAMP, which exhibit positive cooperativity—binding of cAMP to one site increases the binding affinity of the other site, resulting in a sigmoidal dependence of activity on the concentration of cAMP. Thus, as cAMP concentrations rise and approach the Kd for binding to the R subunit (10–8–10–7 M), relatively small differences in cAMP levels lead to large differences in binding. Once two molecules of cAMP are bound to each regulatory subunit, a confor- mational change alters its binding interface with the catalytic subunit, decreasing its affinity and leading to dissociation of the catalytic subunit into its active form. The dissociated catalytic subunit can now phosphor- ylate substrates. Although the regulation of PKA appears straightforward, in reality the on–off switch is part of a regulatory apparatus that is considerably more complex and subtle. For example, there are actually several distinct R (a) (b) inhibitory segment C subunit R subunit cAMP 2 active C subunits inhibitory segment catalytic (C) subunit CNB-A PKA holoenzyme (R2C2 tetramer) 2 free R subunits regulatory (R) subunit CNB-B subunits with slightly different biological properties. Furthermore, each of the R subunits interacts with specific scaffold proteins, termed A-kinase anchoring proteins (AKAPs), that tether the inactive complexes to spe- cific subcellular localizations and are associated with distinct potential substrates (discussed below). Thus, the active catalytic subunit is only released in specific places in the cell where it is in close proximity to rel- evant substrates. We can see that even this relatively simple example actually involves a variety of coupled changes in protein conformation, protein–protein interactions, and subcellular localization, all induced by binding of a small signaling mediator. Some small signaling mediators are derived from membrane lipids Another common class of mediators is derived from the lipids that com- prise cell membranes. The membrane lipid phosphatidylinositol and its derivatives give rise to three classes of intermediates. Cleavage of phos- phatidylinositol 4,5-bisphosphate (PIP₂) by the enzyme phospholipase C yields a water-soluble head group, inositol 1,4,5-trisphosphate (IP ₃ ), and diacylglycerol (DAG), which is highly hydrophobic and remains embed- ded in the membrane (Figure 6.8). Both of these products are active sign- aling mediators. Alternatively, modification of phosphatidylinositol by lipid kinases and phosphatases can give rise to a range of phosphoinositide spe- cies in which the head group is differentially phosphorylated. An important example is the mediator phosphatidylinositol 3,4,5-trisphosphate (PIP₃). These species can activate various effector enzymes, or can be recognized by lipid-recognition domains that result in conditional membrane recruitment. Eicosanoids, such as the prostaglandins, are critical inflammatory media- tors that are synthesized from fatty acids such as arachidonic acid, which are derived from membrane lipids. These mediators, unlike most others, are sufficiently soluble in both lipids and water that they can leave the cell in which they are produced. They exert their biological effects by acti- vating G-protein-coupled receptors (GPCRs) on neighboring cells, leading to a range of effects on inflammation, blood clotting, and vascular tone and premeability. The third common class of lipid-derived mediators is those pro- duced from sphingomyelin. Receptor-mediated activation of the lipase Figure 6.7 Regulation of PKA by cAMP binding. in the absence of camP, PKa exists as a complex in which the catalytic (C) subunit is held in an inactive state by association with the regulatory (r) subunit. an inhibitory segment from the r subunit blocks the active site of the C subunit. When camP levels increase, the r subunit binds camP and dissociates from the C subunit, which is now active and can phosphorylate substrates. the inactive holoenzyme exists as a heterotetrameric complex containing two r and two C subunits. (b) structure of a single regulatory and catalytic PKa heterodimer. CnB-a and CnB-B, cyclic nucleotide-binding domains a and B. (b, adapted from P. Zhang et al., Science 335:712–716, 2012.) Figure 6.8 Hydrolysis of PIP ₂ by phospholipase C yields the signaling mediators IP3 and DAG. iP3 is soluble and diffuses rapidly through the cytoplasm, whereas dag remains in the membrane. –O P O O O 5 OH phospholipase C O– –O P O– –O P O O O 1 OH HO 4 O O 5 1 OH 4 –O P O O– OH HO O PIP –O P O IP O– 2 phosphatidylinositol 4,5-bisphosphate 3 inositol 1,4,5- trisphosphate The role of inositol lipids, eicosa­ noids, and sphingomyelin in signaling are discussed in greater detail in Chapter 7 sphingomyelinase cleaves off the phosphocholine head group from sphin- gomyelin, producing the membrane-restricted signaling mediator cera- mide, which can activate specific kinases and phosphatases. Alternative enzymatic cleavage and phosphorylation of sphingomyelin can produce the soluble mediators sphingosine and sphingosine 1-phosphate, which can also leave the cell of production. PLC generates two signaling mediators, IP ₃ and DAG As mentioned above, IP₃ and DAG are products of the hydrolysis of PIP₂ by phospholipase C (PLC). The β isoform of this enzyme (PLC-β) is acti- vated by heterotrimeric G proteins, whereas the γ isoform (PLC-γ) is acti- vated by tyrosine kinase signaling pathways. After cleavage, IP₃ can be inactivated by enzymatic dephosphorylation, while DAG can be further modified by phosphorylation to generate phosphatidic acid, which itself can mediate signals or can be further enzymatically processed into other bioactive lipids. The major effect of IP3 is to regulate intracellular Ca2+ levels. IP3 can diffuse through the cytoplasm to the endoplasmic reticulum, where it binds and activates the endoplasmic reticulum-associated IP₃ receptor (also termed the IP₃-gated Ca2+ channel). Activation opens these chan- nels, leading to the release of Ca2+ stores into the cytoplasm. As discussed below, this causes a rapid and sharp increase in intracellular calcium con- centration that can be exploited for signaling. Activation of protein kinase C is regulated by IP ₃ and DAG The protein kinase C (PKC) family of serine/threonine kinases illus- trates how enzyme activity and membrane association can be tightly reg- ulated in time and space by small signaling mediators. The PKC family consists of at least 10 members in mammals, all of which contain a C-termi- nal kinase domain, which is normally inhibited by an N-terminal segment that binds to the catalytic site and blocks substrate binding. In addition, the different family members contain various regulatory domains, termed C1 and C2 domains that allow them to respond to changes in the levels of intracellular Ca2+ and/or DAG. For typical PKC isoforms, the simul- taneous binding of Ca2+ and DAG induces a conformational change that relieves inhibition and activates the catalytic domain. In unstimulated cells, conventional PKC resides in the cytosol in its autoinhibited conformation. Upon activation of PLC, however, the result- ing increased membrane density of DAG leads to recruitment of PKC to the membrane via its tandem C1 domains. Binding of Ca2+ to the C2 domain is also required for high-affinity binding to the membrane. Conformational changes induced by binding of DAG and Ca2+ are then sufficient to cata- lytically activate the enzyme (Figure 6.9). Thus, membrane recruitment of PKC is intimately linked to its catalytic activity. Phorbol esters are organic compounds that mimic the structure of DAG and thereby promote the activation of PKC in vivo. They were first isolated as the active ingre- dient in croton oil, which promotes tumorigenesis when applied to the skin of experimental animals. Phorbol esters are used experimentally to manipulate the activity of PKC in cells. Since binding to neither DAG nor Ca2+ is sufficient to recruit and fully activate conventional PKC, this mechanism serves to integrate signals from DAG and Ca2+, the local concentrations of which are likely to have very different spatial and temporal patterns. Other PKC isoforms differ in their dependence on small signaling mediators. So-called “novel” isoforms are dependent on DAG only, while “atypical” isoforms require neither cal- cium nor DAG, and are regulated by other mechanisms. Thus, different PKC isoforms represent a diverse palette of related kinases that can be activated in distinct kinetic and subcellular patterns. CalCium signaling Calcium ions are among the most important intracellular signaling medi- ators. Calcium signaling depends on membrane Ca2+-pumps that gener- ate and maintain a gradient of Ca2+ across cell membranes. Normally, Figure 6.9 Activation of protein kinase C (PKC) by IP ₃ and DAG. activation of heterotrimeric g proteins leads to activation of phospholipase C (PlC), which cleaves phosphatidylinositol 4,5-bisphosphate (PiP₂) to form diacylglycerol (dag) and inositol trisphosphate (iP3). iP3 diffuses through the cytoplasm to activate iP3-gated Ca2+ channels in the endoplasmic reticulum, releasing stored Ca2+ (green circles) into the cytoplasm. Ca2+ and dag function cooperatively to activate PKC, which phosphorylates downstream targets. PlC can also be activated by tyrosine kinases. gPCr, g-protein-coupled receptor. the cytosol has very low concentrations of calcium (~10–7 M) compared to the concentration outside the cell, or in intracellular stores such as the endoplasmic reticulum, where it is approximately 10–3 M. Because of this 10,000-fold concentration gradient, the influx of Ca2+ ions from the environment or release from intracellular stores causes a very rapid and dramatic increase in cytoplasmic calcium concentration, which is used in various ways for signal transduction. Gated ion channels and their regu­ lation are discussed in more detail in Chapter 8 Activation of Ca2+ channels is a common means of regulation In signaling, changes in intracellular Ca2+ levels are initiated by chan- nel opening (Figure 6.10). In the plasma membrane, there are diverse Ca2+ channels that allow Ca2+ to pass through the membrane into the cytosol. Most of these are “gated”; this means that they only open to allow passage of ions under certain conditions. Some Ca2+ channels are gated by the binding of specific ligands (ligand-gated channels), while others can be gated by environmental changes such as tempera- ture, pH, or voltage across the membrane (for example, voltage-gat- ed channels). These different channels allow diverse signals to be converted into the common currency of increased Ca2+ concentration. Small molecules that can block Ca2+ channels are an important class of drugs, and have also been useful experimental tools in studying Ca2+- dependent signaling. Figure 6.10 Use of Ca2+ as an intracellular signaling mediator. Cytosolic Ca2+ concentrations are kept low (~10–7 m) by Ca2+-atPase pumps which drive Ca2+ into the extracellular environment and the endoplasmic reticulum. Cytosolic Ca2+ concentrations rise rapidly when Ca2+ channels open. these include ligand- or voltage-gated channels in the plasma membrane, or iP₃-gated channels in the membrane of the endoplasmic reticulum. the resulting increase in Ca2+ concentration directly activates some effectors, and activates others via the Ca2+-binding protein calmodulin. normal levels of Ca2+ are restored by the Ca2+-atPase pumps. Another physiologically important site of Ca2+ storage is the endoplasmic reticulum (or the analogous sarcoplasmic reticulum in muscle cells). The endoplasmic reticulum membranes are rich in IP3-gated Ca2+ channels (IP₃ receptors). Local increases in the level of IP₃ (caused by PLC-medi- ated cleavage of PIP₂—see above) will activate these receptors, leading to Ca2+ release from the endoplasmic reticulum into the cytoplasm. The IP₃ receptors are also regulated by Ca2+ itself. Intermediate levels of Ca2+ can lead to further activation of the channels, leading to positive feedback. High levels of Ca2+, however, inhibit the channel, leading to negative feedback. These unusual channel properties lead to unique spatiotem- poral modes of Ca2+ signaling, such as Ca2+ waves, which are discussed below. Ca2+ influx is rapid and local Ca2+ can diffuse rapidly. However, the actions of intracellular Ca2+ tend to be highly transient and local. First, Ca2+ levels are rapidly restored by the action of Ca2+-ATPase pumps. Second, because there is an extremely high concentration of Ca2+-binding proteins in the cytoplasm, free Ca2+ has a very short half-life before it is bound to protein and thus, at least temporarily, sequestered. These Ca2+-binding proteins act to buffer Ca2+ increases. These transient and spatially localized increases in Ca2+, how- ever, can be functionally important. We know a great deal about Ca2+ dynamics in living cells because of bio- physical tools that allow visualization of changes in intracellular Ca2+ concentrations. One common approach takes advantage of specialized fluorescent dyes, such as Fura-2, whose spectral properties depend on Ca2+ concentration. More recently, genetically encoded biosensors for Ca2+ concentration have been developed. In one example, two fluorescent pro- teins were fused to calmodulin, so that Ca2+ binding changes the distance between the two fluorescent proteins, leading to changes in their spectral properties (fluorescence resonance energy transfer, or FRET). The use of genetically encoded biosensors avoids the need to introduce a chemical dye into the cell, facilitating the visualization of Ca2+ changes, for exam- ple, in individual neurons in the brain. Intracellular Ca2+ exerts its effects by binding to downstream effector pro- teins. There is a large class of proteins that bind directly to Ca2+. These include kinases (for example, PKC), Ca2+-gated channels, cytoskeletal pro- teins, and synaptic-vesicle proteins (for example, synaptotagmin). In the synapses of neurons, Ca2+ binding to synaptotagmin plays a central role in stimulating the fusion of synaptic vesicles with the plasma membrane and the release of their contents. Many Ca2+-binding proteins have a con- served structural motif, known as the EF hand, which has a conserved set of acidic residues positioned to chelate Ca2+. Many downstream effec- tors, however, do not directly bind Ca2+, but instead are regulated by the common Ca2+ regulatory protein, calmodulin. Calmodulin is a conformational sensor of intracellular calcium levels The protein calmodulin (CaM) is an example of a calcium-binding sensor—upon binding calcium, it undergoes conformational changes that allow it to regulate the activity of a range of downstream effector proteins. Such a mechanism is useful for coupling a single input signal (change in Ca2+ concentration) to widespread and rapid changes in many different cellular activities, including enzyme action and the opening and closing of membrane channels. Biosensors are described in more detail in Chapter 13 Figure 6.11 Calcium binding triggers changes in calmodulin (CaM) structure and binding. atomic structure (bottom) and diagrammatic representation (top) of Cam in different states. (a) apo–Cam (not bound to calcium). asterisks indicate the positions of occluded peptide-interaction surfaces. (b) the Ca2+-bound state, with four Ca2+ ions (green circles) bound to each Cam molecule. the Ca2+-bound state has an altered conformation and the peptide-interaction surface is now exposed. (c) Ca2+–Cam bound to a peptide target ( o r a n g e ). the altered surface of the Ca2+-bound state allows Cam to bind with increased affinity to specific peptide targets on the surfaces of proteins. note that the view in panel (c) is rotated relative to that in panels (a) and (b). (a) unbound (apo–CaM) calcium bound (Ca2+–CaM) Ca2+–CaM bound to a peptide Calmodulin is a compact protein containing four EF hand calcium-bind- ing sites. The affinity of calcium for the EF hand sites in calmodulin ranges from 5 × 10–7 to 5 × 10–6 M, ideally suited for sensing increases in calcium above the resting intracellular level (~10–7 M). The conformation of CaM changes dramatically upon calcium binding (Figure 6.11). These conformational changes expose a new peptide-binding surface, which allows CaM to interact favorably with a wide range of binding partners that include protein kinases and phosphatases, adenylyl cyclases, tran- scriptional regulators, and membrane channels and pumps. Figure 6.11c shows how Ca2+–CaM can wrap around a recognized peptide, using the newly exposed binding surface. In the most familiar cases, unliganded CaM (termed apo–CaM) cannot bind the downstream target, but calcium- bound CaM (Ca2+–CaM) binds with high affinity. This mechanism operates to regulate Ca2+/CaM-dependent protein kinases (CaM-Ks). Binding of Ca2+–CaM to the CaM-K induces activating conformational changes in its catalytic domain, allowing the enzyme to phosphorylate substrates on serine and threonine residues. The modes of CaM regulation can be dif- ferent for different effectors. For example, in some cases, binding of CaM to the effector leads to its inhibition, not activation. Moreover, there are a few targets that bind constitutively to apo–CaM but are then released by calcium binding. Signaling can lead to propagating Ca2+ waves As described above, IP₃ receptors show an unusual Ca2+-dependent func- tion—they are activated by intermediate levels of intracellular Ca2+ but inhibited by higher levels of Ca2+. As a result, upon local stimulation by a pulse of IP₃, the channels in one region of the cell will open and lead to a local influx of Ca2+. Ca2+ will diffuse a short distance and will activate nearby IP₃ receptors, or the related ryanodine receptor, which is also a Ca2+-gated Ca2+ channel. This positive feedback effect can lead to a propa- gating wave of increasing Ca2+ (Figure 6.12). These Ca2+ waves can prop- agate further than individual Ca2+ ions could diffuse, given that the high ambient concentration of Ca2+-binding proteins in the cell limits the dif- fusion of free ions. As the waves propagate, once local Ca2+ concentration (a) (b) (c) calcium concentration positive feedback negative feedback Figure 6.12 Intracellular Ca2+ signaling can show complex dynamics such as propagating waves. (a) the typical iP3-gated Ca2+ channel responds to Ca2+ as well as iP₃. the channels show a bell- shaped dependence of activity on Ca2+ concentration. (b) When channels are first activated by iP3, the initial rise in Ca2+ leads to a local positive feedback loop that spatially propagates channel opening and further Ca2+ influx. However, once Ca2+ concentration becomes further elevated, it has an inhibitory effect on the iP₃-gated channels, resulting in a slower negative feedback loop. this complex regulation can lead to propagating waves of Ca2+. (c) a calcium wave in a rabbit urethral muscle cell, visualized by a calcium-sensitive dye (yellow color corresponds to high Ca2+ concentration). images were taken at a rate of 30 frames per second. some downstream signaling systems, such as Cam-Kinase ii, can read the frequency of such waves as a measure of initial input. (c, Courtesy of m. Hollywood, smooth muscle research Centre, dundalk institute of technology.) increases beyond a certain point, the IP₃ receptors will be inhibited, limit- ing further Ca2+ influx. Thus, as these waves pass, the channels will shut, allowing pumps to restore low Ca2+ levels (creating troughs that follow the crests of the wave). Together, these phenomena can lead to traveling waves of Ca2+, where the intensity of input is encoded in the frequency of the waves. Certain downstream effectors, such as CaM-Ks, are thought to be able to detect and respond to differences in Ca2+-wave frequencies, providing a novel way of encoding signaling information. There is now evidence that other small signaling mediators such as cAMP may also exhibit waves and oscillatory behavior. sPeCifiCity and regulation A limited number of small signaling mediators are used to mediate many diverse biological responses. Thus, a key question is how signaling by these seemingly ubiquitous mediators can yield specific responses in cells. This is an important conceptual challenge, since a single mediator can be generated by many different upstream inputs, and potentially targets a diverse set of downstream effectors (Figure 6.13). A further complication is that many of the small signaling mediators that we have discussed can directly or indirectly impact each other. These specific INPUT/OUTPUT multiple inputs multiple effectors Figure 6.13 Small signaling mediators are generated by diverse signaling inputs and generate diverse signaling outputs. a key question is how these mediators can yield specific input/ output linkages (an example is highlighted in yellow). pathway interactions mean that it can be difficult to disentangle the effect of one signaling mediator from the effects of others. We have seen how IP₃, DAG, and Ca2+ all cooperate to regulate PKC activity. A number of other examples can be found, such as when elevated Ca2+ leads to NO synthesis in vascular endothelial cells, inducing cGMP synthesis in nearby vascular smooth muscle. In this section, we discuss how scaffold proteins provide one mechanism to impart spatial and temporal specificity to small-mole- cule signaling. Scaffold proteins can increase input and output specificity of small-molecule signaling Scaffold proteins help to increase the specificity of small-mediator signal- ing by organizing upstream and downstream molecules (such as recep- tors and effectors, respectively) into localized complexes. For example, cAMP signaling specificity can be enhanced by a class of proteins known as A-kinase anchoring proteins (AKAPs) (Figure 6.14a). AKAPs con- tain multiple protein-binding sites, including one for protein kinase A (PKA), the major immediate downstream target for cAMP. AKAPs have an amphipathic α helix that binds selectively to a site on the regulatory subunit of PKA. As discussed above, at basal levels of cAMP, the R subunit of PKA stably associates with the catalytic (C) subunit, which is thereby maintained in an inactive conformation. The affinity of AKAPs for the R subunit of PKA is very high (Kd of ~10–8 M), thus inactive PKA is effectively tethered to the AKAPs. Individual AKAPs also possess targeting motifs that associate with spe- cific subcellular compartments, such as the plasma membrane, the mito- chondrion, or the centrosome. Thus, although PKA by itself lacks an intrinsic signal for subcellular localization, subsets of PKA are directed to specific sites within the cell by their association with AKAPs, and are held there in an inactive state in readiness for a cAMP signal. A rise in the local concentration of cAMP will trigger binding of the AKAP-associated R subunit and release of the catalytic subunit of PKA in an active conforma- tion, which will preferentially phosphorylate substrates in the vicinity of the AKAP scaffold (Figure 6.14b). AKAPs have additional protein-binding sites (interaction motifs) that help refine the specificity with which cAMP production and PKA activ- ity are coupled to specific upstream receptors and downstream targets. For example, some AKAPs contain targeting motifs that can localize the AKAP complex to the vicinity of a particular upstream GPCR. When this specific GPCR is activated and induces activation of nearby adeny- lyl cyclase molecules, the localized increase in cAMP will more rapidly activate the population of PKA associated with the AKAP scaffold. Thus, cAMP production and PKA activation in the complex are linked to only one of many possible upstream receptors. AKAPs can also control pathway output downstream from PKA—they often have binding sites for specific substrates of PKA, such as ion chan- nels. An AKAP scaffold can thereby link specific upstream receptors to specific downstream effectors. AKAP scaffold proteins can also regulate dynamics of cAMP signaling AKAP scaffolds can also be used to limit the timing of signaling. For exam- ple, certain AKAP proteins also contain binding sites for a phosphodi- esterase (PDE), the enzyme that degrades cAMP and therefore shuts off signaling. This PDE can have two functions that are critical for PKA INPUT (b) (c) cAMP synthesis rate cAMP degradation rate cAMP concentration Figure 6.14 time AKAP scaffolding proteins can enhance input/output specificity of cAMP-mediated signaling. (a) in the unstimulated condition, specific a-kinase anchoring proteins (aKaPs) are associated with specific g-protein-coupled receptor (gPCr) complexes. aKaPs contain targeting sites for specific upstream gPCrs ( p i n k ), binding sites for protein kinase a (PKa; b r o w n), and binding sites for specific downstream effectors (PKa substrates, orange). in addition, some aKaPs contain binding sites for phosphodiesterase (Pde; g r ee n). for clarity, only one heterodimer of the r2C2 PKa tetramer is shown. (b) activation of the receptor and adenylyl cyclase in the vicinity of the aKaP leads to activation of PKa and dissociation of the catalytic subunit, linking activation of specific receptors to phosphorylation of specific PKa outputs. in addition, phosphorylation and activation of Pde by PKa leads to degradation of camP. this generates a negative feedback loop that can control the duration of camP-mediated signaling. (c) schematic of the effect of aKaP on rates of camP synthesis and degradation. aKaP-mediated activation of Pde leads to more rapid signal termination than would be the case in the absence of aKaP (dotted line). activity. In the absence of external signals, the PDE has a basal activity that suppresses the level of cAMP in the vicinity of the AKAP, and this, in turn, ensures that PKA is not spuriously activated in unstimulated cells. However, as the level of cAMP rises, it overwhelms the degradative capac- ity of the PDE, and PKA is activated to phosphorylate local substrates. In the absence of any further activity, the PDE will gradually degrade the cAMP, leading to inactivation of PKA. One form of AKAP-associated PDE, however, is itself a PKA substrate that is stimulated upon phosphorylation. The activation of PKA induced by cAMP therefore stimulates the PDE to more efficiently degrade the local pool of cAMP. As a consequence of this negative feedback loop, there is a more rapid degradation of cAMP and inactivation of PKA (Figure 6.14c). The AKAP scaffold therefore tunes the kinetics of the response to a spe- cific cAMP response signal by binding an inducible inhibitor. Some AKAPs also bind to kinases other than PKA, such as protein kinase C (PKC), and also to phosphatases such as calcineurin. It is easy to see how very complex response dynamics can result from the localized activation of multiple enzymes that can potentially reinforce or coun- teract each other’s activities. Because these AKAP-mediated effects are largely restricted to the immediate vicinity of the scaffold, the same con- centration of cAMP can have very different outputs at different spots in the cell. summary A variety of small molecules serve as signaling intermediates. The levels of these small signaling mediators are determined by their relative rates of production and elimination. They exert their effects on downstream effec- tor proteins by binding and inducing conformational changes that alter the activity of the effectors. Small signaling mediators enable relatively weak signaling inputs to induce rapid, widespread, and highly amplified effects in cells. Major classes of small signaling mediators include Ca2+, the cyclic nucleotides cAMP and cGMP, and lipid-derived mediators such as DAG and IP₃. Many different inputs regulate the same limited number of mediators, which in turn regulate a wide variety of outputs. However, scaffold pro- teins can provide spatial and temporal specificity by physically assem- bling particular components into localized complexes. Questions What physical and chemical properties distinguish small signaling mediators from larger signaling proteins? What kinds of functional dif- ferences might result from these properties? In Chapters 3 and 4, we discuss the concept of enzymes that gener- ate and remove post-translational modifications (writers and erasers). By controlling the extent of a modification, these enzymes coordinately control the flow of signaling information. For Ca2+ signaling—where the concentration of cytoplasmic Ca2+ is used to transmit informa- tion—which molecules play the roles of writers and erasers? What is the source of the energy driving Ca2+ signaling? Under resting conditions, the basal rate of degradation of a small sig- naling mediator, such as cAMP, is relatively high. Why might this be important for cAMP-dependent signaling? Describe two different ways that, in principle, an incoming signal could result in an increase in cAMP levels in a cell. How are waves of Ca2+ generated in cells? How might oscillating waves provide an added dimension for encoding information? Why is specificity a particularly acute problem for small signaling mediators such as cAMP? How do scaffold proteins mitigate this issue? referenCes referenCes Classes of small signaling mediators Bos JL (2006) Epac proteins: multi-purpose cAMP tar- gets. Trends Biochem. Sci. 31, 680–686. Kim C, Xuong NH & Taylor SS (2005) Crystal structure of a complex between the catalytic and regulatory (RIα) subunits of PKA. Science 307, 690–696. Rehmann H, Prakash B, Wolf E et al. (2003) Structure and regulation of the cAMP-binding domains of Epac2. Nat. Struct. Biol. 10, 26–32. Taylor SS, Zhang P, Steichen JM et al. (2013) PKA: les- sons learned after twenty years. Biochim. Biophys. Acta 1834, 1271–1278. Zhang P, Smith-Nguyen EV, Keshwani MM et al. (2012) Structure and allostery of the PKA RIIβ tetrameric holoenzyme. Science 335, 712–716. CalCium signaling Chin D & Means AR (2000) Calmodulin: a prototypical calcium sensor. Trends Cell Biol. 10, 322–328. Clapham DE (2007) Calcium signaling. Cell 131, 1047–1058. sPeCifiCity and regulation Greenwald EC & Saucerman JJ (2011) Bigger, better, faster: principles and models of AKAP anchoring protein signaling. J. Cardiovasc. Pharmacol. 58, 462–469. Welch EJ, Jones BW & Scott JD (2010) Networking with AKAPs: context-dependent regulation of anchored enzymes. Mol. Interv. 10, 86–97. Wong W & Scott JD (2004) AKAP signalling complexes: focal points in space and time. Nat. Rev. Mol. Cell Biol. 5, 959–970. Taylor & Francis Taylor & Francis Group https://taylorandfrancis.com Membranes, lipids, and enzymes That Modify Them The water-impermeable lipid membrane that separates the cytosol from the outside environment is one of the defining features of all living cells. In eukaryotes, lipid membranes also partition the cytosol into different compartments and organelles, and membrane vesicles are used to shuttle proteins and other components between different sites in the cell. In addi- tion to their important structural roles, however, membranes and their lipid building blocks also play a very active role in cell signaling. They provide the raw materials for generating intracellular signaling media- tors such as diacylglycerol (DAG) and inositol trisphosphate (IP₃), and intercellular signaling molecules such as prostaglandins. They also play an important role in providing regulated binding sites for cytosolic pro- teins with lipid-binding domains. Because membrane lipids are arranged in two-dimensional sheets, the biochemical reactions in which they par- ticipate have different properties from more familiar reactions in aqueous solution, where all components can freely diffuse in three dimensions. In this chapter, we will examine the special properties of lipids and mem- branes, and the enzymes that modify them, in the context of their roles in signal transduction. Biological MeMBranes and Their ProPerTies The properties that allow a biological membrane to serve as an effective barrier between the cell’s contents and the environment are its imper- meability to water and other hydrophilic compounds, and its ability to spontaneously organize into self-sealing sheets and vesicles. These special properties arise from the biophysical nature of its molecular components. The fundamental driving force is the high thermodynamic cost of placing hydrophobic chemical groups (such as aliphatic hydrocarbons, which lack Figure 7.1 Organization and structure of the lipid bilayer. (a) Phospholipids spontaneously organize into bilayers, with the polar head groups of the lipids oriented toward the aqueous solution, and the fatty acid tails inside. in addition to large sheets (top), phospholipids can also organize into closed vesicles or liposomes (lower right). in some cases, particularly when the head group is bulky relative to the hydrophobic tail, amphipathic lipids can also organize into micelles (lower left). (b) a computer- generated dynamic simulation of the position of 100 phosphatidylcholine molecules arranged in a lipid bilayer in (a) lipid bilayer (b) aqueous solution. in this representation, the lipid head groups are red, the fatty acid tails are yellow, and water molecules are blue. (adapted from B. alberts et al., Molecular Biology of the cell, 5th ed. garland science, 2008; and s.W. chiu et al., Biophys. J. 69:1230–1245, 1995. With permission from elsevier.) more electronegative atoms such as oxygen or nitrogen) into a polar aque- ous environment, and of placing polar or charged chemical groups in a hydrophobic environment. Of course, this is why oil and vinegar separate in a bottle of salad dressing—the hydrophobic oil and the hydrophillic vinegar segregate away from each other into two distinct phases, mini- mizing the energetically unfavorable interface between the two. Biological membranes consist of molecules in which relatively long hydro- phobic tails are linked to polar head groups. Such molecules, containing both hydrophilic and hydrophobic portions, are said to be amphipathic. In aqueous environments, their lowest energy configuration is one in which the hydrophobic chains are arranged in the interior, shielded from the solvent, while the polar head groups are arrayed on the surface facing the solvent. While some amphipathic molecules readily form micelles— small, spherical structures with the hydrophobic portions sequestered in the interior—for the polar lipids that make up biological membranes, the most favorable arrangement is in a lipid bilayer (Figure 7.1). This con- sists of two layers of polar lipids, arranged in a sheet with the hydrophobic chains oriented inward and the polar head groups on the surface of the sheet facing the aqueous environment. As discussed below, lipid bilayers spontaneously seal into spherical vesicles or liposomes. Biological membranes consist of a variety of polar lipids The most abundant lipids in biological membranes are the phospholip- ids, or more properly the glycerophospholipids (Figure 7.2). All glyc- erophospholipids are built on a three-carbon glycerol backbone. Two of the hydroxyl groups of the glycerol are attached through an ester link- age to fatty acids, which consist of a carboxylic acid group linked to a long, hydrophobic aliphatic chain, generally 14–20 carbons in length. The third glycerol hydroxyl is linked to a phosphate group, which is highly polar and carries a negative charge at physiological pH. The phosphate itself is generally linked to other chemical groups, the most common being choline, serine, inositol, or ethanolamine; the resulting glycerophospholi- pids are termed phosphatidylcholine (PC), phosphatidylserine (PS), phos- phatidylinositol (PI), and phosphatidylethanolamine (PE) (Figure 7.3). Cis-double bonds may be present in the aliphatic chains of the fatty acids (most typically in the fatty acid in the middle, or sn-2, position of the glyc- erol backbone), which generate kinks in the chain because of the planar nature of the double bonds (Figure 7.2). As we will see below, such dou- ble bonds affect the biophysical properties of the membrane. Chains with double bonds are said to be unsaturated (that is, the carbons have fewer than the maximum number of hydrogen atoms associated with them). It is important to note that the specific fatty acids found on phospholipids (a) polar (b) choline CH2 N+(CH3)3 CH2 (hydrophilic) head group phosphate O O– O CH2 O CH CH2 O glycerol C O C O nonpolar (hydrophobic) tails CH2 CH2 CH2 CH2 CH2 CH2 CH2 CH2 CH2 CH2 CH2 CH2 CH2 CH2 CH2 CH2 CH CH cis-double bond CH2 CH2 CH2 CH2 CH2 CH2 CH CH2 CH2 CH2 CH2 CH2 CH3 CH2 CH2 CH3 can be quite variable, both in terms of the length of the chain and the posi- tion and number of double bonds, even when the head group (and thus the generic chemical name of the lipid) is the same. Another major component of biological membranes is the sphingolipids. These resemble phospholipids in that they have similar polar head groups, but the glycerol backbone and fatty acid tails are replaced by ceramide. Ceramide itself has two long aliphatic chains, derived by joining sphingo- sine (a sphingoid base) via an amide linkage to a fatty acid. For example, sphingomyelin (Figure 7.3) has a phosphorylcholine head group attached to ceramide. Since the aliphatic chains of sphingolipids tend to be long and saturated, these lipids are generally taller and narrower in shape than the glycerophospholipids, and thus can pack together more tightly. The last major components of eukaryotic membranes are the sterols, predominantly cholesterol in mammals. Cholesterol is unlike the other major membrane lipids in that it is a rigid, planar, polycyclic compound that is relatively nonpolar (Figure 7.4). Through its interaction with the aliphatic chains of other membrane lipids, it exerts potent effects on the biophysical properties of the membrane, such as its fluidity (see below). Structural properties of membrane lipids favor the formation of bilayers Because of the amphipathic nature and roughly cylindrical shape of most membrane lipids, their most energetically favorable configuration in aqueous solutions is in a bilayer, with the polar head groups oriented out- ward and the fatty acid chains sandwiched between (see Figure 7.1). (By contrast, micelles are generally favored for amphipathic molecules with relatively large polar heads and small hydrophobic tails.) The lipid bilayer consists of two back-to-back layers, or leaflets. Because it is energetically very costly to expose the hydrophobic fatty acid chains to the aqueous Figure 7.2 The structure of a glycerophospholipid. a glycerophospholipid is shown in (a) graphical, (b) chemical structure, and space-filling representations. note the kink in the fatty acid chain introduced by the planar cis-double bond. adapted from B. alberts et al., Molecular Biology of the cell, 5th ed. garland science, 2008.) (a) (b) (c) (d) NH ₃ CHOH CHOH CHOH (e) CH ₂ CH ₂ O O O O H COO CH ₂ O O O O CH ₂ CH ₂ O O O O CHOH CHOH CH O O O O CH ₂ CH ₂ O O O OH O CH ₂ CH CH ₂ CH ₂ CH CH ₂ CH ₂ CH CH ₂ CH ₂ CH CH ₂ CH CH CH ₂ O O O C O O O O C O O O O NH O C O O phosphatidylethanolamine phosphatidylserine phosphatidylcholine phosphatidylinositol sphingomyelin Figure 7.3 The chemical structures of the most common membrane phospholipids. (adapted from B. alberts et al., Molecular Biology of the cell, 5th ed. garland science, 2008.) Figure 7.4 The structure of cholesterol. cholesterol is shown in (a) chemical structure, (b) graphical, and (c) space- filling representations. (d) The interaction of cholesterol with phospholipids in a lipid bilayer, with the rigid ring structure aligned with the fatty acid chains of the phospholipids. (adapted from B. alberts et al., Molecular Biology of the cell, 5th ed. garland science, 2008.) environment, any rips or tears in the membrane tend to spontaneously reseal, a highly useful property for a barrier. Similarly, relatively small sheets will spontaneously form closed, spherical vesicles. Although indi- vidual lipids in a membrane experience considerable thermal motion, the hydrophobic core of the intact membrane presents a very effective barrier to water and to other hydrophilic compounds (see Figure 7.1b). However, a few signaling molecules—such as the small, uncharged gas nitric oxide, and hydrophobic organic molecules such as steroid hormones—are suf- ficiently small and lipid-soluble that they can pass through the hydropho- bic core relatively freely. The composition of the membrane determines its physical properties The inner and outer leaflets of the cell membranes differ in their lipid com- position. This is because most lipids cannot spontaneously “flip” between the two leaflets, due to the energetic cost of dragging the polar head group through the hydrophobic core of the membrane. For these lipids, transport (a) OH CH3 CH ₃ CH3 CH CH2 CH2 rigid steroid ring structure (b) polar head group nonpolar (c) (d) 3 2 1 polar head groups cholesterol- stiffened region more fluid CH2 CH CH3 CH3 hydrocarbon tail 0 region liquid-disordered solid gel liquid-ordered (raft) Figure 7.5 Different organization states of a lipid bilayer. (a) When many of the fatty acid chains are unsaturated, the chains are not densely packed and the bilayer adopts a liquid-disordered phase, with high lateral mobility. (b) When most fatty acid chains are saturated, they align closely with each unsaturated hydrocarbon chains with cis-double bonds saturated hydrocarbon chains saturated hydrocarbon chains, cholesterol enriched other and the membrane adopts a solid gel phase, with limited lateral mobility. if cholesterol is present, a membrane can adopt a liquid-ordered phase that is both highly ordered yet relatively fluid. (adapted between leaflets requires specialized enzymes and energy in the form of ATP. In general, the outer (extracellular) leaflet of the plasma membrane contains sphingomyelin and glycosphingolipids and is enriched for PC, while the inner leaflet is enriched for PE. The phospholipids PS and PI, which carry a net negative charge, are exclusively found on the inner leaf- let, consistent with the slight negative charge on the inner face of the membrane (the resting potential). The fact that PS is exclusively found on the inner leaflet is actually used by cells as a signal: the presence of PS on the outside of cells undergoing apoptosis, or programmed cell death, is an “Eat Me” signal that induces surrounding cells to engulf the apoptotic cell fragments. The physical properties of the membrane are highly dependent on its component lipids and, in particular, the aliphatic chains of those lipids. Just as butter and other familiar fats undergo temperature-dependent phase transitions from a solid gel to a more highly disordered liquid, the aliphatic chains of membrane lipids can undergo similar phase transi- tions. In general, the more tightly packed and ordered the aliphatic chains, the less fluid the membrane will be (Figure 7.5). The presence of double bonds, which introduce kinks into the aliphatic chains, makes it more difficult for them to pack together tightly; thus, increasing the fraction of unsaturated fatty acids increases membrane fluidity (the industrial proc- ess of hydrogenation removes these double bonds in liquid vegetable oils to make shortening or margarine that gels at room temperature). In addi- tion, the presence of relatively large amounts of membrane-associated protein (up to 50% by weight) and of rigid, nonpolar sterols can have dra- matic effects on fluidity. Membrane fluidity is relevant to signaling for several reasons. First, the amount of fluidity directly affects the behavior of proteins embed- ded in the membrane and of lipids that act as signaling intermediates (such as DAG) by affecting their rates of diffusion. Diffusion is much faster in a fluid membrane than in regions that are more ordered and gel-like. More important, since different phases can exist in the same membrane, this raises the possibility that different membrane domains will have different lipid and protein constituents as well as different physical properties. For example, the long, straight alkyl chains of sphingomyelin interact favorably with cholesterol, generating local lipid domains (lipid rafts) that are highly ordered, yet allow a high degree of lateral mobility (termed the liquid-ordered phase) (see Figure 7.5). Such structures have been difficult to visualize in living cells, implying that they are likely to be small and transient in nature. However, such lipid microdomains can potentially play an important role in signaling by concentrating key components and thereby increasing the efficiency of reactions, and by segregating components away from competing interactors or inhibitors. from B. alberts et al., Molecular Biology of the cell, 5th ed. garland science, 2008.) Apoptosis is discussed in Chapter 9 There are fundamental differences between biochemistry in solution and on the membrane The fact that membranes are two-dimensional has implications for the reactions that occur there. Intuitively, we can see that if two reactants are confined to the same membrane, they are more likely to interact than if they are free to wander about in solution. This is a simple con- sequence of less physical space to explore. However, the actual rate of diffusion of a molecule in the plane of the membrane is considerably slower than that of the same molecule in solution, so the absolute rate that two molecules encounter each other in solution versus on the mem- brane may not be all that different. However, those interactions are likely to be much more efficient in two dimensions (fewer competing interactions to wade through, and in many instances the interacting molecules are pre-oriented by their anchorage points to the membrane) (Figure 7.6). Another consideration, touched on above, is the likelihood that different domains of the membrane have quite different compositions and physical properties, and that migration between these domains may be impeded to various extents. Experimental evidence from single-molecule track- ing and fluorescence recovery after photobleaching (FRAP) suggests that membrane lipids and proteins that are embedded in the membrane diffuse significantly more slowly in actual biological membranes than in artificial membranes consisting of pure lipids. Furthermore, meas- urements of diffusion on very short time scales suggest that membrane (b) Figure 7.6 Membrane association can make molecular interactions more efficient. (a) Two interacting proteins (one green, one blue) bind productively only in a certain precise orientation (top). if molecules are free in solution, they can rotate around all three axes; only a small fraction of collisions will be correctly oriented to allow binding (middle). however, if the two molecules are confined to the membrane by an anchor (bottom), they can only rotate around one axis; thus a relatively large percentage of collisions will be oriented to allow binding. (b) For molecules in solution, a three-dimensional volume must be explored (top), while if molecules are confined to a membrane, only a two-dimensional plane must be explored, making interaction more likely (middle). in actual biological membranes, lateral diffusion is thought to be limited to relatively small areas separated by diffusion barriers (bottom), further increasing the likelihood of interaction between molecules. components can rapidly diffuse within small domains (several microns in size), but move much more slowly between adjacent domains. These slower diffusion rates may be due to the physical constraints of a high density of membrane proteins. It also seems that some cytoskeletal structures may act to corral membrane proteins, impeding their ability to freely diffuse over the entire surface of the cell membrane. Distinct membrane domains that differ in fluidity could also contribute to this anomalous diffusion. Thus, signaling proteins that are strongly asso- ciated with membranes through transmembrane domains or via lipid anchors are likely to interact with their near neighbors in a much less transient way than if they were cytosolic. Another unusual aspect of membrane signaling is the diversity of bio- physical properties of the molecules involved. The signaling compounds that are generated by modification of membrane lipids can be either lipid- soluble, thus constrained to the membrane from whence they came, or water-soluble, and thus free to diffuse into the cytosol or extracellular fluid. In some cases, one reaction generates both types of molecules, as when phospholipase C cleaves phosphatidylinositol 4,5-bisphosphate [PI(4,5)P₂] to generate DAG, which is confined to the membrane, and IP₃, which is water-soluble. DAG does not contain a polar head group and so can flip relatively easily from one leaflet of the membrane to the other. Membrane-localized signaling mediators such as DAG work primarily by recruiting and, in some cases, activating cytosolic proteins. Other lipid signaling mediators such as sphingosine 1-phosphate (S1P) and lysophos- phatidic acid (LPA) can partition into both the lipid and soluble phases. Thus, the downstream effects of signals involving membrane lipids may be exerted throughout the cell, or in specific membrane compartments, or even to distant cells, depending on the specific nature of the signaling intermediary that is generated. A final distinctive feature of signaling that involves membrane lipids is the rapid metabolism of these compounds from one bioactive species to another. This interconversion can make it difficult to say with confidence which specific lipid species is responsible for a particular output. In other words, if the levels of a single bioactive lipid are increased experimen- tally, this increase will rapidly alter the levels of a slew of breakdown products and further metabolites that can be derived from it, each of which may have different (but possibly overlapping) biological activities. This presents considerable difficulties in attempting to tease out specific cause-and-effect relationships in lipid signaling pathways. liPid-ModiFying enzyMes Used in signaling Signaling by membrane lipids involves their breakdown or chemical modification in response to input stimuli. These reactions are catalyzed by enzymes, most prominently phospholipases, lipid kinases, and lipid phosphatases. In this section, we will briefly outline the general proper- ties of these enzymes. We will then explore their specific roles in a number of important signal transduction pathways in greater detail in the follow- ing section. Cleavage of membrane lipids by phospholipases generates a variety of bioactive products The enzymes that break down phospholipids are termed phospholipases. They are classified on the basis of the bonds that they cleave. The major choline and PA OH O O CH ₂ O CH ₂ OH CH ₂ CH O O O C CH ₂ O phospholipase D CH ₃ CH ₃ + N CH ₃ CH ₂ CH ₃ phosphocholine and DAG OH CH ₂ O CH ₃ + N CH ₃ CH ₂ CH CH ₂ O P O O CH CH CH phospholipase C CH ₂ CH ₂ O O O C O C O 2 2 O O CH3 O P O OH C O C CH ₃ + N CH ₃ phospholipase A ₂ CH ₂ CH ₂ O O P O O CH ₂ CH CH ₂ O OH OH C O C O lyso-PC and fatty acid Figure 7.7 The sites of cleavage and the reaction products generated by phospholipase D (PLD), phospholipase C (PLC), and phospholipase A2 (PLA2). The substrate phospholipid in this example is phosphatidylcholine. Pa, phosphatidic acid; dag, diacylglycerol; lyso-Pc, lysophosphatidylcholine. phospholipases that participate in signaling are phospholipase A₂, phos- pholipase C, and phospholipase D (Figure 7.7). Phospholipase A ₂ (PLA ₂ ) cleaves the fatty acid chain from the sn-2 (middle) position of the glycerol backbone of glycerophospholipids, gener- ating a free fatty acid and a lysophospholipid containing a single fatty acid chain. Both of these products can generate bioactive components. In many cases, the fatty acid at the sn-2 position is arachidonic acid (AA). Arachidonic acid may itself activate some signaling proteins, and also serves as the basic building block of the eicosanoids, a large family of intercellular signaling lipids that includes the prostaglandins and leu- kotrienes, as discussed later in this chapter. Eicosanoids can leave the cell of origin and signal to nearby cells by binding to specific G-protein- coupled receptors (GPCRs). Lysophospholipid, the second product, is also sufficiently water-soluble that it can leave the membrane and signal to adjacent cells. Lysophosphatidic acid (LPA), the product of PLA₂ action on phosphatidic acid, can potently signal via a GPCR to regulate diverse activities such as proliferation, differentiation, and cell morphology. PLA₂ is a major component of many insect and snake venoms, presumably exerting its toxic effects by disrupting the biophysical integrity of mem- branes in the victim. Phospholipase C (PLC) cleaves the phosphorylated head group from the phospholipid, leaving the uncharged DAG behind. In the case of PI- PLC, which is specific for phosphatidylinositol, the released phosphoinosi- tol head group (such as IP₃) can diffuse rapidly in the cytosol and act as a soluble signaling intermediate, while DAG, which remains in the mem- brane, acts as an activator of proteins such as protein kinase C (PKC). Finally, phospholipase D (PLD) cleaves the unphosphorylated head group from the phospholipid, leaving behind phosphatidic acid (PA). PA not only serves as a substrate for PLA₂ to generate LPA, but itself plays an important role in activation of the mTOR (mechanistic target of rapamycin) kinase complex, a key regulator of metabolic and stress sign- aling and cell growth, as described below. A variety of lipid kinases and phosphatases are involved in signaling The phosphorylation and dephosphorylation of membrane lipids, particu- larly those derived from PI (phosphoinositides), are also commonly regu- lated in signaling to generate a variety of bioactive compounds. As might be expected, lipid kinases (which add phosphate) and lipid phosphatases (which remove it) both play a role in changing the local concentration of various lipid isoforms in the course of signaling. The activity of these pro- teins, in turn, is regulated by their abundance, by their post-translational modification, and, in particular, by their subcellular localization. Conse- quently, these enzymes frequently possess modular lipid- and protein- binding domains, which allow them to respond to changes in the local protein and lipid environment. One broad class of lipid kinases phosphorylates uncharged lipid signal- ing mediators, converting them to more highly polar products that also can act as signaling mediators. For example, DAG is converted to PA by DAG kinases. Thus, DAG kinases mediate a switch in local activity from one class of effectors (those activated by DAG binding) to another (those that are regulated by binding to PA). Similarly, sphingosine is converted by sphingosine kinase to S1P, which can leave the membrane and acti- vate GPCRs. The reverse reactions (dephosphorylation of PA and S1P) are mediated by a family of related integral membrane enzymes, the lipid phosphate phosphatases. The six-carbon inositol sugar head group of PI provides ample scope for the action of lipid kinases and phosphatases (Figure 7.8). Hydroxyl groups at positions 2 through 6 of the ring are available for phosphoryla- tion (the oxygen at position 1 is linked by a phosphodiester bond to the glycerol backbone). As will be described in more detail below, phosphor- ylation at positions 3, 4, and 5 is most associated with signaling, creating specific membrane binding sites for modular lipid-binding domains and providing precursors for signaling mediators generated by the action of PLC. In general, different kinases and phosphatases, each of which can have distinct subcellular localization and modes of regulation, are used to modify each position of the inositol ring. This enables very precise spatio- temporal control over the phosphorylation state of the phosphoinositides. Activation of PKC is discussed in Chapter 6 (a) PI PI(3,4,5)P ₃ O O O –O O OH HO O O O O O O O O O O O Figure 7.8 The structure of phosphoinositol lipids. (a) structure of phosphatidylinositol (Pi) and phosphatidylinositol 3,4,5-trisphosphate [Pi(3,4,5)P₃]. numbering of carbons of the inositol ring is indicated in orange. (b) space-filling models of the head groups of Pi and the major phosphoinositides recognized by lipid-binding domains. Akt activation is discussed in Chapter 5 exaMPles oF Major liPid signaling PaThWays We have outlined above how membranes and their lipid constituents can participate in signaling in many ways, both direct and indirect. In this section, we will consider in greater depth several examples where the modification of membrane lipids plays a particularly direct role in the process of signal processing and transmission. Phosphoinositides can serve as membrane binding sites and as a source of signaling mediators Phosphoinositides provide a compelling illustration of the two major roles of membrane lipids in signaling: as a source of diffusible signaling media- tors, and as specific membrane-localized binding sites that can recruit and activate signaling proteins. The lipid PI(4,5)P₂ is at the nexus of these two different functions. Cleav- age of PI(4,5)P₂ by PI-PLC generates DAG and IP₃ (see Figure 6.8), each of which goes on to activate downstream targets (such as PKC and endo- plasmic reticulum calcium channels, respectively). However, PI(4,5)P₂ itself can directly bind to and recruit to the plasma membrane a variety of signaling proteins, particularly those that regulate localized actin polym- erization. PI(4,5)P₂ can also be further modified by phosphorylation or dephosphorylation to generate distinct phosphoinositides that can bind and recruit other targets. In particular, PI(4,5)P₂ can be phosphorylated by phosphatidylinositol 3-kinase (PI3K) to generate phosphatidylinositol 3,4,5-trisphosphate [PI(3,4,5)P₃], a pivotal regulator of cell survival and growth, and many other activities. For example, PI(3,4,5)P₃ stimulates downstream signaling by recruiting and activating effectors such as the Akt kinase. PI(3,4,5)P₃ has another important property: unlike its precur- sor PI(4,5)P₂, it cannot be cleaved by PI-PLC. Thus, activation of PI3K has the effect of shutting down signaling from PI(4,5)P₂, whereas PLC activa- tion, by destroying the substrate for PI3K, has the effect of shutting down Box 7.1 In signaling, the activity of many lipid-modifying enzymes is regulated, in turn, by other lipids and the enzymes that produce or modify those lipids. These complex regulatory arrangements can have interesting consequences, such as promoting the formation of discrete membrane domains with different lipid composition, or promoting the precise switching from one type of signal to another. Three examples mentioned in the text are illustrated in Figure 7.9. PI3K and PI-PLC can both be activated by similar stimuli (for example, receptor tyrosine kinase activation), but the two consume the same substrate, PI(4,5)P₂. Thus, in situations where substrate is limiting, whichever enzyme is activated first at a particular site will signal transiently (until it exhausts the substrate) while preventing activation of the latecomer (Figure 7.9a). Phosphatidylinositol 5-kinase (PI5K) and PLD co- localize, and each makes a product that activates the other enzyme. Thus the activity of each reinforces the other, leading to high local concentrations of their products. This is an example of positive feedback (Figure 7.9b). The small G protein Rheb activates both mTOR and PLD. PLD makes a product, PA, which is also necessary for mTOR activity. This is an example of coherent feed-forward (Figure 7.9c). The requirement for PLD could function to limit mTOR activity to specific locations (where PLD is present), or to delay activation of mTOR fol- lowing Rheb activation (until PLD can generate sufficient PA). Such relationships and their depiction (network architecture) are discussed in chapter 11. (b) PI(4,5)P ₂ PI(4,5)P ₂ (c) Rheb PI-PLC PI3K PI5K PLD PLD mTOR PI(4,5)P2 PA PA Figure 7.9 Lipid-modifying enzymes involved in (a) mutual inhibition, (b) positive feedback, and (c) coherent feed- forward regulation. Pi(4,5)P₂, phosphatidylinositol 4,5-bisphosphate; Pi-Plc, phosphatidylinositol phospholipase c; Pi3K, phosphatidylinositol 3-kinase; Pi5K, phosphatidylinositol 5-kinase; Pld, phospholipase d; Pa, phosphatidic acid. PI3K signaling. This mutual inhibition presumably facilitates the precise spatiotemporal regulation of signal output within the cell (Box 7.1). Two well-understood ways in which PI-PLC can be activated are by GPCRs and by tyrosine kinases. Heterotrimeric G proteins activate the PLC-β family of PLCs. Generally, this is accomplished by binding of the PLC to Gα_q subunits, but the βγ subunits may participate in activating some PLC-β isotypes (see Figure 6.9). Examples of G-protein-mediated PLC activation include platelets activated in response to the thrombin receptor, and Drosophila retinal cells, where G protein activation is medi- ated by visual rhodopsin. By contrast, the PLC-γ family is activated by receptors with intrinsic or associated tyrosine kinase activity, such as mitogenic growth fac- tor receptors or B and T cell receptors. PLC-γ contains SH2 domains, which mediate recruitment and binding to the tyrosine-phosphorylated receptor or associated proteins. This interaction has two consequences: tethering PLC to the membrane where its substrate resides, and pro- moting phosphorylation of PLC by associated kinases. Tyrosine phos- phorylation is important for full catalytic activity of the enzyme. The primary downstream consequences of PLC activation include release of intracellular Ca2+ and activation of PKC, though biological effects have also been attributed to local decreases in PI(4,5)P₂ density, or to increases in fatty acids or PA due to further enzymatic modification of the DAG generated by PLC. Figure 7.10 Phosphoinositide metabolism in signaling. Major phosphoinositides involved in signaling are shown, and examples of enzymes that generate or destroy them. not all possible reactions indicated by arrows have been shown to be biologically significant. PiP, phosphatidylinositol phosphate; Pi3K, phosphatidylinositol 3-kinase; PiP5K, phosphatidylinositol(4)P 5-kinase; Pi-Plc, phosphatidylinositol-specific phospholipase c; dag, diacylglycerol; iP₃, inositol trisphosphate. Phosphoinositide species provide a set of membrane binding signals The other major arm of phosphatidylinositol signaling is the selective binding, recruitment, and activation of effectors to various phosphor- ylated forms of PI. This provides another example of the “writer/eraser/ reader” theme introduced in earlier chapters. In this case, the writers are the PI kinases, the erasers are the PIP phosphatases, and the read- ers are lipid-binding domains on signaling proteins. A number of modu- lar lipid-binding domains (including the PH, PX, and FERM domains) interact with specific phosphoinositides. Other proteins bind to phospho- inositides through less-well-defined basic (positively charged) motifs. The availability of five potential phosphorylation sites on the inositol ring (see Figure 7.8) means a great deal of information can be encoded in one small molecule. In principle, 25 or 32 different phosphorylation states are pos- sible, but in practice only the 3, 4, and 5 positions are used in cells, gen- erating eight combinations. Furthermore, the high density of phosphate groups on some phosphoinositides, and their associated negative charge, means that electrostatic attraction can provide considerable binding energy for interacting with positively charged protein surfaces. The major phosphoinositides that participate in signaling, along with examples of their effectors and the kinases and phosphatases that generate them, are shown in Figure 7.10. Different phosphoinositides vary widely in their subcellular distribution, and this plays an important cell-biological role by providing a kind of Vps34 EFFECTORS EFFECTORS EFFECTORS EFFECTORS (actin regulation) EFFECTORS (PKC) PIP5K PI-PLC PTEN synaptojanin SHIP PI3K PTEN EFFECTORS (Akt) relative abundance of lipids high low (a) PI(4,5)P ₂ PI(4)P PI(3)P Figure 7.11 Subcellular localization of phosphoinositides. subcellular localization of (a) Pi(4,5)P₂, (b) Pi(4)P, and (c) Pi(3)P, as revealed by the binding of green fluorescent protein (gFP)-tagged lipid-binding domains specific for each lipid. lipid-binding domains used are indicated beneath each panel. (From g. di Paolo and P. de camilli, Nature 443:651–657, 2006. With permission from Macmillan Publishers ltd.) GFP-PH(PLC-δ1) GFP-PH(FAPP1) GFP-2xFYVE(Hrs) “identity tag” for different types of membranes. For example, PI(4,5)P₂ is almost exclusively found on the plasma membrane, while phosphati- dylinositol 3-phosphate [PI(3)P] is found on endosomes and PI(4)P on the Golgi (Figure 7.11). This is largely due to the subcellular distri- bution of the enzymes that generate the different phosphoinositides (PI kinases and PIP phosphatases). Key regulatory enzymes include PI(4)P 5-kinases (PIP5Ks), which dominate local generation of PI(4,5)P₂, and PI 5-phosphatases, such as synaptojanin. Synaptojanin is associated with the endocytic machinery and degrades PI(4,5)P₂ on plasma membrane- derived vesicles after endocytosis, a key step for their proper targeting and recycling. Importantly, phosphoinositide signaling is intimately intercon- nected with signaling by small G proteins, both in regulating membrane trafficking and in signaling. Not only are many of the guanine nucleotide exchange factors (GEFs) and GTPase-activator proteins (GAPs) for small G proteins regulated by phosphoinositide binding, but the G proteins and phosphoinositides often act cooperatively to bind to and activate down- stream factors. There are several different PI3K isoforms activated by distinct upstream stimuli. In terms of signaling, the best understood of these is the het- erodimeric class IA subtype, for which the localization and activity of the catalytic subunit is regulated by a regulatory subunit containing SH2 domains. As in the case of PLC-γ described above, the SH2 domains medi- ate binding to activated tyrosine kinases, thereby coupling activation of the PI3K to activation of tyrosine kinases. Class IB subtypes possess a similar heterodimeric organization, but in this case respond to G protein βγ subunits. All class I PI3Ks use PI(4,5)P₂ as their preferred substrate, thus their predominant product is PI(3,4,5)P₃ (see Figure 7.10). By con- trast, the class III PI3K, Vps34, prefers PI as a substrate and thus gener- ates predominantly PI(3)P. Vps34 is found primarily in endosomes and the Golgi, and its major role is to regulate vesicle trafficking. PI(3,4,5)P₃ generated by class I PI3Ks can be dephosphorylated by several distinct phosphatases. PTEN specifically dephosphorylates the 3 position of the inositol ring, essentially counteracting the effects of the PI3K. Since PI3K and its product generally promote cell growth, proliferation, and survival, it is not surprising that PTEN is a strong tumor suppressor, and that loss of PTEN activity is often seen in tumors. Another family of phos- phatases that degrades PI(3,4,5)P₃ is the SHIP family, which specifically removes the phosphate from the 5 position to generate PI(3,4)P₂. This lipid has a distinct but overlapping set of effectors compared with PI(3,4,5)P₃. Recruitment of SHIP to activated T cell receptors, for example, is thought to be important for a temporal shift from one class of phosphoinositide- binding effectors to another. Phospholipase D generates the important signaling mediator, phosphatidic acid (PA) Phospholipase D cleaves the most abundant membrane phospholipid, PC, into phosphatidic acid (PA) and choline (see Figure 7.7). While free choline has no known signaling functions, PA affects a myriad of cellular events, both through the direct binding and recruitment of effector proteins, as well as through the biophysical effects of PA and its immediate metabo- lites on the lipid bilayer. There are two highly related PLD isoforms in vertebrates, PLD₁ and PLD₂, both of which require phosphoinositides such as PI(4,5)P₂ for activity. PLD₁ is also regulated by PKC, and by small G proteins including Rheb, RalA, and members of the Rho and Arf fami- lies. Less is known about the regulation of PLD₂. Activation of PLD likely involves the combination of membrane recruitment and allosteric activa- tion, also seen for many other lipid-metabolizing enzymes. One important physiological role for PLD is to cooperate with PI(4,5) P₂ and Rho family G proteins to regulate local organization of the actin cytoskeleton. This is accomplished in part through the physical interac- tion of PLD with PIP5Ks, which are activated by PA. Since both PLD₁ and PLD₂ are dependent on the product of PIP5K [PI(4,5)P₂], this is an example of a mutually reinforcing positive feedback loop that may help generate discrete, localized membrane domains with distinct signaling properties (see Box 7.1). Another role for PLD is in regulating the intracellular trafficking of mem- brane vesicles. PLD is a major effector of the Arf family of small G pro- teins, which function to regulate the transport and targeting of vesicles. The localized activation of PLD at sites of vesicle fusion and fission (dur- ing exocytosis and endocytosis, respectively) is thought to be important for lowering the energy barrier for these events. This has been proposed to be due to direct effects of PA on membrane curvature. The PA generated by PLD tends to contain one unsaturated fatty acid; the kink in the acyl chain generated by the double bond makes the shape of PA rather conical, with a small solvent-accessible head connected to a relatively broad hydrophobic anchor. As a result, high PA density on the inner leaflet of the bilayer pro- motes negative (concave) curvature in the membrane (Figure 7.12). The Figure 7.12 Effect of PA on membrane curvature. (a) space-filling model of Pa; note the kink induced by a double bond in one fatty acid chain. (b) The triangular cross- sectional shape of Pa induces negative curvature when concentrated on the inner leaflet of a membrane bilayer. (c) Transient areas of extreme negative curvature (blue arrows) at sites where a membrane vesicle fuses with or pinches off from a membrane. (a, adapted from B. alberts et al., Molecular Biology of the cell, 5th ed. garland science, 2008.) process of fusion of a vesicle with a membrane (or its reverse, the pinch- ing off of a vesicle from a membrane) involves passing through a transient intermediate stage with strongly negative membrane curvature at the “neck.” The localized activation of PLD at such sites can ease the transition through this energetically unfavorable state. Phospholipase D plays a role in mTOR signaling Phosphatidic acid generated by PLD is also required for activation of the mechanistic target of rapamycin (mTOR), a protein kinase that acts as a master regulator of cell growth, survival, and metabo- lism (Figure 7.13). The activity of mTOR is regulated by environmen- tal inputs such as the levels of mitogens, amino acids, oxygen, and ATP. Activated mTOR, in turn, phosphorylates and regulates the activity of proteins that control translation, such as ribosomal S6 kinase (S6K1) and eukaryotic initiation factor 4E binding protein (4E-BP1). In this way, the rate of protein synthesis is coupled to environmental conditions and the metabolic state of the cell. mTOR exists in two multiprotein complexes, termed mTORC1 and mTORC2, which are distinguished by the presence of the regulatory subunits Raptor or Rictor, respectively. The activity of mTORC1 is positively regulated by a small G protein, Rheb, which itself is negatively regulated by the TSC complex, consist- ing of TSC1 and TSC2. TSC was named for tuberous sclerosis complex, a genetic predisposition to non-malignant tumors, which is due to loss of function mutations in the genes encoding TSC1 or TSC2. It is not surprising that loss of a key negative regulator of the mTOR complex would lead to deregulation of cell growth and survival—a hallmark of cancer. The TSC complex receives and integrates inputs from a number of upstream signals. For example, TSC is phosphorylated and inhibited by the Akt kinase, providing a physiological link between the PI3K– Akt and mTOR pathways in regulating cell growth, proliferation, and survival. energy deprivation (AMPK) hypoxia mitogens (Akt) TSC1/2 small G proteins PIP2 PA rapamycin mTORC1 OUTPUT ( protein synthesis) Figure 7.13 The mTOR pathway and its regulation by PLD. activation of the mTor complex (mTorc1) is mediated by the g protein rheb, and requires phosphatidic acid (Pa). rheb is inhibited by the Tsc1/2 complex. a variety of environmental cues regulate mTor activity, including the protein kinases aMPK (regulated by cellular aTP levels) and akt (regulated by signals that promote proliferation, growth, and survival). in addition, phospholipase d (Pld) serves as a central hub for integrating cellular signals and regulating mTor activity. Pi5K, phosphatidylinositol 5-kinase; PiP₂, phosphatidylinositol 4,5-bisphosphate. Phosphatidic acid interacts directly with mTOR and is required for the activation of both mTORC1 and mTORC2. The immunosuppressive drug rapamycin binds to the same site on mTOR as PA, thus it is likely that rapamycin inhibits mTOR in mammals by blocking its interaction with PA (mTOR was first isolated as the target of rapamycin). Suppression of PLD activity, like rapamycin treatment, prevents the formation of mTOR complexes, thus it is likely that that PA facilitates the association of mTOR with the companion proteins Raptor and Rictor. The apparent requirement for PA indicates that PLD plays a critical role in regulating mTOR activity. Consistent with this, PLD activity is frequently elevated in cancer cells and PLD overexpression stimulates mTOR, while experi- mental inhibition or down-regulation of PLD suppresses mTOR activity and leads to apoptosis in many cancer cell lines. The details of how PLD activity responds to environmental signals are still under investigation (see Figure 7.13). PLD₁ is up-regulated in response to a variety of mitogenic signals, and the mechanism is likely to involve a network of small G proteins that are upstream activators of PLD₁, such as RalA and Rheb. PLD activity is elevated in response to amino acids and is dependent on Rheb, and RalA is similarly activated in response to amino acids, directly associates with PLD₁, and is required for elevated PLD activity. The dependence of elevated PLD activity on Rheb is an example of coherent feed-forward regulation: Rheb activates mTOR directly and also activates PLD, which generates a second mTOR activa- tor, PA (see Box 7.1). A significant conceptual and technical hurdle to clarifying these regula- tory relationships is that the PLD–PA–mTOR connection may be specific to mammalian cells, as PA and PLD have not yet been shown to regu- late the TOR pathway in more genetically tractable model systems such as yeast and Drosophila. Thus, integration of PLD into the regulatory networks controlling mTOR may have been a relatively recent evolution- ary innovation. The metabolism of sphingomyelin generates a host of signaling mediators The sphingolipid ceramide and its metabolites constitute a large and ever-expanding family of bioactive lipids that regulate a variety of cel- lular activities. Ceramide can be generated from the abundant membrane lipid sphingomyelin by the action of sphingomyelinase (SMase), or can be generated de novo from fatty acids. In turn, ceramide can be broken down by ceramidase (CDase) to sphingosine plus a fatty acid. Both ceramide and sphingosine can directly regulate effector proteins and, furthermore, can be phosphorylated by lipid kinases to generate bioactive derivatives: ceramide 1-phosphate (C1P) and sphingosine 1-phosphate (S1P), respec- tively (Figure 7.14). This network provides a specific example of the general principle that the levels of individual lipids are intimately interconnected and in dynamic equilibrium. Thus, any disturbance of the equilibrium (for example, acti- vation of a modifying enzyme) will ultimately affect the levels of many dif- ferent bioactive lipids. In this regard, it is useful to consider the relative abundance of the various players. Sphingomyelin is a rather abundant lipid (comprising up to 30% of plasma membrane phospholipid), and is thus thought to play a largely structural role. Its derivatives are much less abundant, however, so that fairly small changes in the rate of sphingomy- elin metabolism can have relatively large effects on the levels of these lip- ids. The least abundant products, such as S1P, have concomitantly higher CH ₂ CH ₂ O O O OH O high Figure 7.14 Major ceramide metabolic pathways involved in signaling. shown are structures of the ceramide lipids, along with the major enzymes that generate them. reverse reactions also occur in cells but are not shown here. sMase, sphingomyelinase; cdase, ceramidase; sK, sphingosine kinase; cK, ceramide kinase. CH CH NH CH ₂ O SMase OH CH CH NH C OH CH ₂ O OH CH CH NH2 OH CH ₂ sphingomyelin CDase ceramide CK O O P O OH O sphingosine SK CH CH NH C CH2 O O O P O OH O CH CH NH2 CH ₂ ceramide-1P sphingosine-1P low potencies; S1P binds to a GPCR with high affinity and thus can activate signaling even at very low (nanomolar) levels. The signaling activities of each of the products are further constrained by their biophysical properties. Sphingomyelin, with its bulky hydrophobic groups and highly polar head, is essentially confined to just one mem- brane leaflet. Once the polar head group is removed by SMase, however, ceramide is sufficiently nonpolar that it can readily flip between the inner and outer leaflets. Removal of the fatty acid chain from ceramide by CDase generates compounds with a single aliphatic chain, such as sphingosine and S1P, that are now sufficiently hydrophilic to be able to leave the mem- brane. Such compounds can exert their effects throughout the cell, or even to other cells throughout the organism. Sphingolipids play a role in the cellular response to stresses, largely through generation of ceramide and its further metabolites. There are a number of different SMases, which differ in their subcellular localiza- tion and specific activation properties. Environmental stresses such as ionizing and ultraviolet radiation, reactive oxygen and nitrogen species, and chemotherapeutic drugs can activate the so-called acid SMases, likely directly through reactive oxygen species and indirectly through phospho- rylation by PKCδ. Another class of SMases, the neutral SMases, is acti- vated in response to cytokines such as tumor necrosis factor α (TNFα) and interleukin-1 (IL-1). Downstream targets of the ceramide generated by SMase include the protein phosphatase PP2A and the protease cathepsin D, though many details remain unknown. S1P, which is generated by the action of sphingosine kinases (SKs) on sphingosine, has emerged as a very potent cell–cell signaling molecule with a variety of developmental and homeostatic roles in metazoans. For example, signaling by a number of cytokines leads to activation of SK and thus to an increase in S1P levels, and this is important for the proinflammatory effects of these cytokines. S1P signaling is also critical for the proper development and function of the endothelial and smooth muscle cells that form blood vessels. Most of the biological effects of S1P in cell–cell signaling are exerted through activation of a class of plasma membrane GPCRs termed the EDG or S1PR family. These receptors are activated by S1P at nanomolar levels, leading to the activation of a vari- ety of heterotrimeric G proteins. S1P may also function intracellularly as a direct signaling mediator, for example in Ca2+ homeostasis, but the specific intracellular effectors of S1P are not yet characterized. Phospholipase A ₂ generates the precursor for a family of potent inflammatory mediators Inflammation is a physiological response to infection, allergens, or trau- ma involving localized swelling, redness, and pain. These symptoms are caused in large part by the dilation and increased permeability of blood vessels, and by the recruitment of white blood cells (leukocytes) to the site. The eicosanoids are a class of lipid signaling molecules derived from ara- chidonic acid (AA) that play a critical role in the inflammatory response. They are generated locally by the action of cytosolic PLA₂ to generate free AA, which is then further modified by enzymes such as cyclooxygenases and lipoxygenase to generate prostaglandins and leukotrienes, respec- tively (Figure 7.15). These inflammatory mediators act locally (in the cell of origin or nearby cells) through binding to GPCRs, leading to physiologi- cal responses. Because of the impact of inflammation on human health, many commonly used drugs target this pathway. For example, aspirin and nonsteroidal anti-inflammatory drugs (NSAIDs) such as ibuprofen and naproxen target the cyclooxygenases (COX1 and COX2) that generate intermediates in prostaglandin synthesis. Although there is a wide variety of PLA₂ isoforms in animal cells, only the cytosolic PLA₂α (cPLA₂α) form is specific for phospholipid substrates with AA at the sn-2 position. The 20-carbon aliphatic chain of AA has four double bonds, and is the obligate precursor for eicosanoid biosyn- thesis. Mice lacking cPLA₂α are highly deficient in their inflammatory phosphatidylcholine lyso-PC O O O O platelet activating factor Figure 7.15 Generation of inflammatory mediators from arachidonic acid. regulatory enzymes are shown in b l u e. The action of cytosolic phospholipase a2α (cPla2α) on phosphatidylcholine generates arachidonic acid and lysophosphatidylcholine (lyso-Pc). O O cPLA2 O O OH arachidonic acid arachidonic acid is further processed by the cyclooxygenase (cox) and lipoxygenase (5-lo) pathways to generate potent bioactive lipids (green) that signal by binding to g-protein-coupled receptors. lyso-Pc can be further modified to generate platelet activating factor (PaF), which also signals through a g-protein-coupled receptor. Pgh2, prostaglandin h2; lTa4, leukotriene a4. OH PGH2 LTA4 prostaglandins, thromboxane, prostacyclin leukotrienes responses, firmly implicating cPLA₂α-derived AA in inflammation. In unstimulated cells, cPLA₂α is found in the cytosol but rapidly relocal- izes to perinuclear membranes upon activation by cytokines, mechanical trauma, or other proinflammatory stimuli. Activation requires binding of Ca2+ to the C2 domain of cPLA2α, which promotes its association with membranes. cPLA₂α is also activated by phosphorylation by MAP kinases such as Erk1/2, and by binding to C1P. Thus, eicosanoid generation is tied into a variety of signal transduction pathways. The other product of PLA₂ action on PC is lysophosphatidylcholine, which can then be acetylated to generate another extremely potent inflammatory mediator, platelet acti- vating factor (PAF), which exerts its biological effects at subnanomolar concentrations by binding to its own GPCR. In most cell types, cyclooxygenases co-localize with cPLA₂α on perinuclear membranes to convert AA into an intermediate, prostaglandin H₂ (PGH₂), which is then further processed in a cell-type-specific manner to gener- ate a wide variety of specific prostaglandins, prostacyclins, and thrombox- anes. In inflammatory cells such as macrophages and mast cells, cPLA₂α co-localizes with another enzyme, 5-lipoxygenase (5-LO), to generate the intermediate leukotriene A₄ (LTA₄), which is further processed to gen- erate a variety of leukotrienes. Specific plasma membrane transporters facilitate efflux of the finished bioactive lipids out of the cell. The potent biological effects of eicosanoids are due to their binding to a family of more than a dozen related GPCRs, which differ in the spectrum of specific mediators they bind, by their tissue distribution, and by their downstream effectors. Stimulation of these receptors, and the consequent activation of heterotrimeric G proteins, leads to cell-type-specific effects such as contraction of airway smooth muscle cells, vascular leakage, Nuclear receptors are discussed in Chapter 8 vasodilation, and pain. Interestingly, some eicosanoids have also been shown to bind to nuclear receptors of the PPAR (peroxisome proliferator- activated receptor) class. Although these receptors may be mostly involved in maintaining homeostasis of lipid biosynthetic pathways, it is possible that they may have signaling functions as well. sUMMary The unique chemical properties of lipid molecules cause them to self- organize into lipid bilayers, which form a barrier that is relatively imper- meable to most hydrophilic molecules. Lipid species in the membrane can serve as binding sites that are recognized by specific lipid-binding domains. The binding of protein molecules to the lipid bilayer can dra- matically enhance their interaction with one another, and many criti- cal signaling interactions take place at membranes. The individual lipid molecules can also serve to encode and transmit signaling information directly. The protein interactions in which lipids participate can be dra- matically regulated by modifying enzymes such as lipid kinases and phos- phatases, which can covalently modify lipid head groups, thus altering their recognition by distinct proteins. In addition, lipids can be degraded by specific lipases to release products which serve as diffusible signaling mediators that regulate various downstream targets. QUesTions What are the possible signaling consequences of a particular mem- brane lipid being confined to either the inner or outer leaflet of the membrane? Experimental evidence suggests that the plasma membrane is not homogeneous but contains regions with distinct lipid compositions. How might this heterogeneity affect the properties of proteins embed- ded in or associated with the membrane? What properties make phosphatidylinositol (PI) particularly useful in signaling? Eicosanoids have been shown to bind to G protein coupled receptors (GPCRs) and to nuclear receptors (NRs). What are the physical prop- erties of eicosanoids that allow this? How might you experimentally distinguish between GPCR-dependent and NR-dependent effects? reFerences Biological MeMBranes and Their ProPerTies Cho W (2006) Building signaling complexes at the mem- brane. Sci. STKE 2006(321), pe7. Groves JT & Kuriyan J (2010) Molecular mechanisms in signal transduction at the membrane. Nat. Struct. Mol. Biol. 17, 659–665. Hancock JF (2006) Lipid rafts: contentious only from simplistic standpoints. Nat. Rev. Mol. Cell Biol. 7, 456–462. Owen DM, Williamson D, Rentero C & Gaus K (2009) Quantitative microscopy: protein dynamics and mem- brane organisation. Traffic 10, 962–971. van Meer G, Voelker DR & Feigenson GW (2008) Mem- brane lipids: where they are and how they behave. Nat. Rev. Mol. Cell Biol. 2008; 9, 112–124. liPid-ModiFying enzyMes Used in signaling Aloulou A, Ali YB, Bezzine S et al. (2012) Phospholipases: an overview. Methods Mol. Biol. 861, 63–85. Bunney TD & Katan M (2011) PLC regulation: emerg- ing pictures for molecular mechanisms. Trends Biochem. Sci. 36, 88–96. references Burke JE & Dennis EA (2009) Phospholipase A2 struc- ture/function, mechanism, and signaling. J. Lipid Res. 50(Suppl), S237–S242. Michell RH (2008) Inositol derivatives: evolution and functions. Nat. Rev. Mol. Cell Biol. 9, 151–161. Suh PG, Park JI, Manzoli L et al. (2008) Multiple roles of phosphoinositide-specific phospholipase C isozymes. BMB Rep. 41, 415–434. exaMPles oF Major liPid signaling PaThWays Di Paolo G & De Camilli P (2006) Phosphoinositides in cell regulation and membrane dynamics. Nature 443, 651–657. Funk CD (2001) Prostaglandins and leukotrienes: advances in eicosanoid biology. Science 294, 1871–1875. Hannun YA & Obeid LM (2008) Principles of bioactive lipid signalling: lessons from sphingolipids. Nat. Rev. Mol. Cell Biol. 9, 139–150. Krauss M & Haucke V (2007) Phosphoinositide-metabo- lizing enzymes at the interface between membrane traf- fic and cell signalling. EMBO Rep. 8, 241–246. Roth MG (2008) Molecular mechanisms of PLD function in membrane traffic. Traffic 9, 1233–1239. Sun Y & Chen J (2008) mTOR signaling: PLD takes center stage. Cell Cycle 7, 3118–3123. Taylor & Francis Taylor & Francis Group https://taylorandfrancis.com information Transfer across the membrane Conveying information from the extracellular environment into the interior of the cell is perhaps the most fundamental hurdle that must be overcome in cell signaling. While the water-impermeable plasma membrane provides an essential function in shielding the cell’s con- tents from the environment, it necessarily imposes a barrier to free communication. The membrane physically isolates the cytosol from many environmental cues, particularly those provided by water-soluble signal- ing molecules such as peptides and hydrophilic small molecules but also other cues such as those from adjoining cells or the extracellular matrix. However, it is essential for cells to continuously adapt their behavior to their environmental conditions. This is especially true in the case of cells in multicellular organisms, where extensive communication between cells is essential for proper development and function. In this chapter, we will examine the various strategies that have evolved to allow selective signals to be transmitted through the plasma membrane, focusing primarily on the mechanisms used by transmembrane receptors. PrinciPles of Transmembrane signaling Transmembrane signaling depends on the ability of receptors on the cell to receive input from the environment. Receptors are proteins that, when bound to specific signaling molecules (ligands), undergo a change in activity that transmits a signal. Because their activity depends on whether or not they are bound to a ligand, they serve to convert infor- mation on extracellular ligand concentrations into an intracellular activity. Most receptors are found on the plasma membrane, where they are exposed to the extracellular environment. A few receptors, those that respond to signals that can freely cross the plasma membrane, are located within the cell. Figure 8.1 Cells receive a variety of environmental cues. adjacent cells, the extracellular matrix, and soluble factors from the extracellular fluid all provide signals that must be accurately received and interpreted by the cells of multicellular organisms. The cell must process and respond to a diversity of environmental cues Before discussing specific signal transduction mechanisms, it will be use- ful to consider the types of information the cell needs to gather from its environment (Figure 8.1). The simplest types of signals, which are important even for free-living unicellular organisms such as yeast, are the levels of nutrients, oxygen, and many other raw materials needed to build and maintain the cell. The levels of such compounds will determine the activity of cellular biosynthetic pathways, and will also affect behav- ior such as growth and motility, so the organism can adapt to take full advantage of abundant resources while avoiding suboptimal conditions when possible. The number of critical inputs is substantially larger for multicellular organisms, however. A wide variety of soluble signaling molecules secreted by other cells are present in the extracellular fluid bathing the cell, including factors such as hormones, cytokines, and mitogens. Hormones are secreted molecules that regulate cellular activities, often at distant sites in the body (in humans, familiar examples include growth hormone, insulin, and estrogen). In many cases, hormones are secreted by special- ized endocrine glands and tissues. Hormones can be either polypeptides, which are derived from the processing of larger proteins, or small mol- ecules synthesized from organic building blocks. More specialized types of peptide hormones include cytokines, which regulate various aspects of immune cell function, and mitogens, which induce the proliferation of cells. Mitogens are often also called growth factors, but this term can lead to confusion between mitogens and true growth factors, which specifi- cally promote cell growth (an increase in cell bulk). In addition to soluble signaling molecules, cells also must respond to more local, fixed cues, such as molecules present on the surface of adjoining cells. During development and in adult tissues, it is essential that cells be able to adapt their behavior in harmony with the properties of surround- ing cells. For example, adjoining epithelial cells must make tight contacts with their neighbors to form an epithelial sheet. And during develop- ment of the nervous system, the growth cones of elongating axons must interpret attractive or repulsive cues on adjacent cells, in order to make connections to the appropriate target cells. Cells need to be able to detect and react to both the presence of molecules on the surface of adjacent cells, as well as the local density of those molecules. In addition to cell-surface molecules, cells must also interpret signals from the extracellular matrix (ECM) in which they are embedded. The ECM is secreted by cells and consists of fibrillar proteins such as collagen that provide tensile strength, proteoglycans, the polysaccharide hyaluronic acid, and factors that modulate cell adhesion and motility such as fibronectin and its relatives. The ECM plays a structural role in tissues, but also plays a more active role in regulating activities such as cell migra- tion, differentiation, and tissue organization. Some ECM components directly signal to cells through their presence and local density, and the ECM can also serve as a reservoir for binding and presenting soluble hor- mones and signaling proteins to cells. In addition, the ECM can provide mechanical cues to the cell, through its resistance to being deformed as the cell makes attachments and applies forces to it. Cells must also respond to a wide variety of other types of physical sig- nals. In multicellular animals, among the most important of these are electrical signals propagated by neurons. These will not be explicitly dis- cussed in this chapter. In addition, specialized cells can respond to sensory PrinciPles of Transmembrane signaling membrane- permeable signals gated channels transmembrane receptors Figure 8.2 Three ways for a signal to cross the plasma membrane. (a) a few signaling molecules can passively diffuse across the membrane into the cytosol. (b) some signaling molecules bind to and regulate the opening of membrane channels. (c) many signaling molecules bind to transmembrane receptors, leading directly or indirectly to activation of intracellular enzymes. stimuli such as light, pressure, odors, and other cues important for the life of multicellular organisms. Three general strategies are used to transfer information across the membrane Although there is tremendous diversity in extracellular cues, the cell uses just three general mechanisms to transduce those cues through the plas- ma membrane to the cytosol (Figure 8.2). First, in a relatively few cases, the signal itself can passively cross the plasma membrane. This spe- cialized class of signaling molecules includes the membrane-permeable gas nitric oxide (NO) and lipid-soluble hormones including steroid hor- mones and their relatives. For these signals, no specialized mechanism is required to traverse the membrane, and the cytosolic effectors for these signals can bind them directly. The signaling molecules exert their influ- ence by binding to intracellular receptors. This is by far the simplest and most direct mechanism of transmembrane signaling. The second major strategy involves membrane channels. These are transmembrane pores, assembled from protein subunits, which selectively pass certain classes of small molecules such as ions. Gated channels can be opened and/or closed in response to specific signals, allowing them to function as receptors that transmit information across the membrane. Neurons conduct electrical signals by the rapid opening and closing of channels in response to changes in membrane potential, ion concentra- tions, and neurotransmitters. The final broad strategy involves transmembrane receptors. These proteins consist of a ligand-binding domain on the extracellular face of the membrane, one or more hydrophobic transmembrane segments, and an intracellular portion that couples to downstream signaling effectors. In many cases, the intracellular part of the receptor itself has intrinsic catalytic activity, for example a protein kinase domain, which is activated by ligand binding. In other cases, the receptor is associated noncovalently with other effector proteins, which relay the information that the ligand- binding domain is occupied. For most cells in multicellular organisms, this is by far the most prevalent and versatile mechanism for transmembrane signaling, coupling a wide variety of extracellular stimuli to changes in a number of intracellular activities. Many drugs target receptors The binding sites of receptors that bind to protein ligands (such as pep- tide hormones) are typical of other protein–protein interaction surfaces (see, for example, the complex between growth hormone and its receptor, Figure 2.4). However, receptors can bind and respond to a wide variety of other ligands, including simple ions and small organic molecules. Regard- less of the nature of the ligand, for binding to convey information it must lead to physical changes in the receptor itself, either in its conformation or its association with other molecules. Many drugs that are used to treat human medical conditions target the ligand-binding sites of cell-surface receptors. This is particularly true for the class of G-protein-coupled receptors (GPCRs), which control a host of medically relevant processes such as heart rate, blood pressure, inflam- mation, and neurotransmission. Compounds that mimic natural ligands that cause activation of the receptor and transmission of downstream sig- nals are termed agonists. Morphine and other opiates are agonists for the opioid receptor. On the other hand, some compounds may bind to a receptor and fail to evoke an activating response—they may, for exam- ple, bind to the ligand-binding site without inducing changes needed to transmit a signal. Furthermore, by binding the receptor at a site that is the same as (or overlaps with) the site for natural activating ligands or agonists, they can prevent normal receptor activation. Such compounds are termed antagonists. Antagonists are analogous to enzyme inhibitors, in that they inhibit receptor activity in a concentration-dependent fash- ion. Loratadine (Claritin®), an antihistamine, is a histamine H1 receptor antagonist; so-called beta blockers, used to lower blood pressure and con- trol heart rate, are β-adrenergic receptor antagonists. Figure 8.3 hydrophobic amino acid side chain hydrophobic fatty acid chains charged head groups TransducTion sTraTegies used by Transmembrane recePTors For receptors whose ligands cannot pass through the plasma membrane, information transfer depends on somehow conveying to the intracellular part of the receptor whether or not ligand is bound to the extracellular part. This is not a trivial problem, as the two parts of the receptor are separated by the hydrophobic lipid bilayer. The intracellular and extracel- lular portions of the receptor, which are hydrophilic, cannot move through the membrane to physically interact with each other, thus any communi- cation must be through the relatively featureless hydrophobic transmem- brane segments. These typically consist of rigid α helices ~20 amino acids in length, with their hydrophobic side chains projecting outward into the plane of the membrane (Figure 8.3). With a few exceptions, only two gen- eral solutions to this problem are used in biological receptors: concerted conformational changes for those receptors with multiple membrane- spanning segments (multiple-pass receptors), and the dimerization or oli- gomerization of receptors that span the membrane only once (single-pass A helical transmembrane segment. for proteins that traverse the lipid bilayer, the transmembrane portion almost always consists of a rodlike α-helical segment comprised of 20–25 amino acids. The side chains of the amino acids in the transmembrane segments are hydrophobic (blue circles), interacting favorably with the hydrophobic fatty acid chains in the interior of the membrane bilayer. receptors). Receptors with multiple membrane-spanning segments undergo conformational changes upon ligand binding Many receptors, including the large class of GPCRs and all membrane channels, contain multiple membrane-spanning segments. For GPCRs, binding of ligands to the extracellular portion of the receptor induces concerted conformational changes that involve alterations in the relative packing of the seven transmembrane helices, leading to conformational changes on the intracellular face of the receptor. As discussed in more detail below, these changes increase the affinity of the receptor for G proteins, thus activating downstream signaling. The fact that there are multiple transmembrane helices tightly packed against one another allows a change of conformation (relative orientation of the helices) to be transmitted through the membrane. By contrast, such a change in relative orientation is not possible for a single transmembrane segment, which has no adjacent helices to “push” against. Gated channels represent the other major class of receptors where lig- and binding is converted into conformational change. Membrane channels consist of a number of similar or identical subunits arranged in a ring structure in the plane of the membrane. Each subunit contains multiple membrane-spanning segments and, in the case of ligand-gated channels, a ligand-binding site is present on the side facing the extracellular envi- ronment. Ligand binding induces concerted conformational changes that alter the permeability of a pore at the center of the ring. Channel open- ing can thus lead to very rapid and dramatic changes in the intracellular concentration of the molecules that are allowed to pass through which, in turn, can have widespread effects on intracellular reactions. Receptors with a single membrane-spanning segment form higher-order assemblies upon ligand binding For those receptors that span the membrane only once, the extracellu- lar and intracellular domains are connected only by a single, hydropho- bic stalk bobbing in a fluid planar membrane. Concerted conformational changes cannot be transmitted across the membrane by single-pass recep- tors, and so they must adopt a completely different strategy for convey- ing to the cytosol the binding state of the extracellular domain. For these receptors, signaling depends on changes in the interaction of receptor mol- ecules with each other upon ligand binding (Figure 8.4). In many cases, this involves receptor homo- or heterodimerization: unliganded receptors are monomeric or only loosely associated, whereas ligand binding causes two receptor molecules to closely interact with each other. The simplest way that this can be done is if the ligand itself can simultaneously bind to two different receptor molecules. For example, platelet-derived growth factor (PDGF) is a dimer in its native state and so can bind to two recep- tors. Thus, low concentrations of PDGF will rapidly induce the formation of receptor dimers on the membrane. Receptor dimerization can also be induced by other specific mechanisms; for example, by a single ligand that has two distinct receptor-binding sites (as in the case of growth hormone), by ligands that are physically clustered on the surface of cells or the extra- cellular matrix, or by ligands that induce changes in the conformation of the receptor that increase the affinity of receptors for each other (as in the case of epidermal growth factor, EGF). Why does a receptor dimer, in which the intracellular domains of the two receptor molecules are closely apposed to each other, have different activi- ties from the individual receptors in isolation? In general, dimerization or oligomerization facilitates enzymatic reactions that would not be favored in the monomeric state. For example, when the intracellular portion of the receptor has protein kinase activity, dimerization makes it much more likely that the receptors will phosphorylate each other due to their proximity (Figure 8.5). An example is the PDGF receptor, which has tyro- sine kinase activity. Such transphosphorylation can have two impor- tant consequences, both of which play a role in downstream signaling. In most cases, transphosphorylation at a specific site in the activation loop of the kinase domain is sufficient to stably increase its catalytic activity. Figure 8.4 Multiple transmembrane segments allow receptors to transduce information about ligand binding through the membrane. (a) for a receptor that contains a single transmembrane segment, there is no straightforward mechanism by which the conformational changes in the extracellular domain induced by ligand binding can be transmitted to the intracellular portion. (b, c) When a receptor dimerizes or contains multiple transmembrane segments, ligand binding may induce conformational changes in the intracellular portion, for example, by rotation of the transmembrane portions relative to each other. Figure 8.5 Activation of receptor tyrosine kinases. ligand binding induces dimerization of the receptor. The tyrosine kinase catalytic domains then rapidly transphosphorylate each other on the “activation loop” site on the kinase domain itself, leading to increased kinase activity, and on other sites that serve as docking sites for cytosolic sH2 domain-containing proteins. Activation of kinases by activation- loop phosphorylation is discussed in Chapter 3 Once the receptor is activated, it can then phosphorylate other intracellu- lar targets, including its dimer partner. In addition, transphosphorylation can create binding sites for phosphodependent modular protein binding domains such as SH2 domains, thereby recruiting cytosolic effector pro- teins containing these domains to the receptor. Some receptors use a hybrid approach that combines aspects of concerted conformational changes and induced dimerization. For example, the insu- lin receptor (a receptor tyrosine kinase) consists of two identical sets of subunits covalently linked by disulfide bonds (see Figure 8.8, below). Thus, insulin binding does not affect the dimerization state per se, but instead induces conformational changes in the extracellular domain that are transmitted to the intracellular catalytic domains via changes in the relative orientation of the two transmembrane helices. These changes position the catalytic domains to facilitate transphosphorylation and cat- alytic activity only when insulin is bound. Another example is provided by bacterial chemotaxis receptors, which are constitutively dimerized and which transmit information on ligand binding through subtle changes in the relative orientation of their helical intracellular domains, which cou- ple to intracellular effectors. Receptor clustering confers advantages for signal propagation The general strategy of using receptor dimerization or oligomerization to transmit signals confers a number of interesting properties to the signal- ing pathways that they regulate. The primary effect of dimerization or oligomerization is to greatly increase the local concentration of the receptor for its partner(s), including any proteins that may be bound to it. Since the rate of a bimolecular reaction depends on the concentration of the two reactants, increasing the local concentration of receptors can greatly increase reaction rates when either the receptor itself or receptor-associated proteins can enzymatically mod- ify each other. Furthermore, this simple property provides a mechanism to make signal responses more switchlike, responding in an all-or-none fashion to changes in ligand concentration. This principle can be illustrated by the case of receptors that are cou- pled to nonreceptor tyrosine kinases, which include the receptors for a wide variety of cytokines and hormones, adhesion receptors, and the B cell and T cell receptors on the surface of lymphocytes. As will be discussed later in this chapter, all these different receptors share the property that the transmembrane receptor itself is noncovalently associated with an intracellular tyrosine kinase (the specific kinase varies depending on the particular receptor). In the absence of stimulation, these tyrosine kinases adopt an inactive conformation, but will occasionally “flip” into an active conformation for a moment before switching back to the more stable inac- tive conformation. Furthermore, because cytosolic phosphatase activity is relatively high, any substrate that might be phosphorylated during the brief period when the kinase is active is likely to be dephosphorylated rap- idly. Thus, in the unstimulated state, both kinase activity and substrate phosphorylation are low. The situation is quite different when receptors (and associated kinases) dimerize or associate into higher-order structures upon ligand binding (Figure 8.6). Now, whenever a kinase briefly adopts the active confor- mation, it is situated in very close proximity to another kinase molecule (associated with the dimer partner), as well as the second receptor mole- cule itself. It is therefore very likely that these substrates will be phospho- rylated before the catalytic domain reverts to the inactivate conformation. If the kinase manages to phosphorylate the “activation loop” site of the second kinase, that second kinase will now be stably activated, and will be highly likely to phosphorylate the first kinase and its associated recep- tor, thus stably activating the first kinase as well. Furthermore, if either kinase or receptor should be dephosphorylated by phosphatases, it is likely to be rapidly rephosphorylated by its neighbors. Thus, the net effect of dimerization or clustering is high kinase activity and high levels of receptor phosphorylation; both of these states can transmit downstream signals (by phosphorylating other proteins and by binding to downstream effectors, respectively). Another potential advantage of receptor clustering is that it allows infor- mation about the activity of one receptor or receptor dimer to be trans- mitted to multiple other receptors, allowing a higher order of information processing. In the example above, clustering of kinase-associated recep- tors ensures that activation of the entire cluster is likely to be relatively stable, even if an individual receptor becomes momentarily inactivated by ligand dissociation or dephosphorylation. More generally, if one receptor is bound to a particular downstream effector, that effector may be able to modify the activity of other receptors within the cluster. Figure 8.6 The effect of clustering on kinase activation. Typically, a kinase can exist in three different states: inactive (pale brown), partially active (pale pink), and phosphorylated and fully active (pink). The unphosphorylated kinase rapidly flips between the inactive and partially active states, with the majority of the kinase being in the inactive state at any given time. The partially active kinase can phosphorylate and activate any adjacent kinase molecule during the brief period when it is active (arrow). This phosphorylated kinase remains active until cellular phosphatases remove the phosphate, reverting the kinase back to the inactive state. (a) When the kinase molecules are sparsely and evenly t=0 t=1 t=2 distributed, partially active kinase rarely has the opportunity to activate other kinase molecules, and phosphatases rapidly inactivate the resulting activated kinase before it has a chance to activate other molecules. (b) by contrast, when kinases are clustered together, partially active kinase can phosphorylate and activate t=0 t=1 t=2 inactive kinase partially active kinase active kinase multiple adjacent kinases which, in turn, can activate additional adjacent kinases. soon, most of the kinases in the cluster are phosphorylated and activated. cellular phosphatases cannot easily inactivate the cluster, because any dephosphorylated kinase is rapidly rephosphorylated by its neighbors. Time steps are indicated by t=0, t=1, t=2. In a broader sense, clustering allows the formation of specific membrane domains, with high concentrations of receptors, receptor-binding pro- teins, and their substrates, that have very different properties than the surrounding areas of the membrane. Proximity makes intermolecular interactions and reactions much more efficient than if molecules needed to diffuse throughout the cell to encounter their substrates or binding partners. Some cells contain very specific subcellular compartments where particular signaling components are clustered together even in the absence of stimulation, which minimizes the time needed for components to diffuse and increases the efficiency of interactions. Such specializations are often seen in systems where speed of the response is important, for example neuromuscular junctions and the photosensitive organelles of photoreceptor cells in the eye. Clustering also can increase the apparent affinity of receptors for their ligands and for downstream effectors compared to isolated monomers, leading to more efficient binding. When receptors are sparsely distributed, whenever a binding partner dissociates from the receptor, it is likely to dif- fuse away before it has the opportunity to rebind. However, when there is a very high local concentration of receptors in the vicinity, as when receptors are clustered, the binding partner is much more likely to rebind to another receptor before it has a chance to escape by diffusion (see Figure 2.11). This effect is particularly strong when the overall density of receptors is relatively sparse, and when the fraction of receptors that are bound to ligand is low, both conditions that are common for physiological receptors. g-ProTein-couPled recePTors G-protein-coupled receptors (GPCRs) are by far the most abundant class of receptors in most eukaryotes, with ~900 distinct GPCRs found in the human genome. All of these receptors share the same overall topology, consisting of an extracellular N-terminus, an intracellular C-terminus, and seven membrane-spanning helices that are connected by hydrophilic loops that project into the cytosol or extracellular environment [thus they are often referred to as seven-transmembrane receptors (7-TMRs) or hep- tahelical receptors]. GPCRs respond to a wide variety of ligands, includ- ing small molecules, polypeptides, and lipids, and they signal by coupling ligand binding to the activation of heterotrimeric G proteins. G protein signaling is introduced in Chapter 3 G-protein-coupled receptors have intrinsic enzymatic activity The broad outlines of GPCR signaling are depicted in Figure 8.7. When an activating ligand binds, concerted conformational changes are trans- mitted to the intracellular aspect of the receptor. The activated receptor then associates with a GDP-bound heterotrimeric G protein and acts as a guanine nucleotide exchange factor (GEF) to promote release of GDP and binding of GTP to the Gα subunit. Conformational changes in the Gα subunit induced by GTP binding then lead to its dissociation from both the receptor and from the Gβγ subunits. Depending on the specific subu- nits and effectors, either the Gα or Gβγ subunits then bind to downstream effectors to modulate their activity. Direct effectors include ion channels, phospholipase C (PLC), adenylyl cyclases, phosphodiesterases, and GEFs for Rho family GTPases. It is important to note that while there are many hundreds of GPCRs, the number of heterotrimeric G proteins to which they couple is much more limited. For example, in humans there are fewer than 20 Gα subunits, (b) Figure 8.7 Signaling by G protein coupled receptors. (a) ribbon model of the gPcr rhodopsin docked with the transducin heterotrimer bound to gdP (orange space-filling representation). gα is depicted in green, gβ in blue, and gγ in pink, and rhodopsin in purple. The positions of the n- and c-terminal ends of each g protein subunit are indicated. (b) binding of ligand to the gPcr leads to nucleotide exchange on the gα subunit, and its dissociation from the gβγ subunits. both the activated gα and gβγ subunits can bind to and regulate downstream effector proteins. gα and gγ subunits associate with the membrane via covalently linked lipid groups (wavy lines). (a, adapted from W.m. oldham and H.e. Hamm, Nat. Rev. Mol. Cell Biol. 9:60–71, 2008. With permission from macmillan Publishers ltd.) divided into four classes based on their effectors. Since there are far more GPCRs than downstream G proteins and effectors, activation of many dif- ferent GPCRs is likely to lead to the same signal output. Furthermore, many GPCRs can interact with and activate multiple classes of hetero- trimeric G proteins. For these reasons, often one cell will contain only a few different GPCRs, and therefore be specialized to respond specifically to a certain class of signaling molecule. This concept is best exemplified in the olfactory system, which mediates our sense of smell. Many hundreds of different olfactory GPCRs are encoded in the vertebrate genome, each of which responds to a different class of odorants. All of these olfactory GPCRs are coupled to Gα_s and the activation of adenylyl cyclase. If a cell were to express many different odorant receptors, it would be unable to discriminate which one was activated because they all couple to the same effectors. For this reason, each sensory cell of the olfactory epithelium expresses just a single odorant receptor, so that it will respond only to a particular class of odorants. Light-activated signaling by pho- toreceptor cells is discussed in Chapter 12 Signaling by GPCRs can be very fast and lead to enormous signal amplification The speed, magnitude, and duration of downstream signals propagated by GPCRs are dependent on a number of different steps. First, ligand binding must induce conformational changes in the receptor that increase its affin- ity for heterotrimeric G proteins and render it able to act as a GEF. After G protein binding, nucleotide exchange must occur on the Gα subunit, where- upon the G protein subunits can dissociate and bind their effectors. Once the G protein dissociates from the receptor, another G protein can take its place. Repeated cycles of G protein activation allow signal amplification (multiple G proteins activated per activated receptor). Each activated G protein (either the Gα or Gβγ subunits) can stably bind to and activate a single effector. But because these effectors are either enzymes or ion channels, a single activated effector can further amplify a signal consid- erably. After activation, the system then resets back to baseline through several mechanisms. Activated Gα subunits eventually hydrolyze bound GTP through their intrinsic GTPase activity, and this can be stimulated considerably by regulator of G protein signaling (RGS) proteins, which act as GTPase-activator proteins (GAPs) for the Gα subunit. The receptor itself can be deactivated either by ligand dissociation or by desensitization pathways involving phosphorylation of the receptor by G-protein-coupled receptor kinases (GRKs), as discussed in more detail below. Signaling by GPCRs can be extremely rapid. For example, in the retinal rod and cone cells of the eye, activation of rhodopsin (a specialized GPCR that responds to the light-induced isomerization of bound retinal) can occur within a few milliseconds, and effectors are activated within a few hundred milliseconds. In this case, a single activated receptor can activate up to ~200 G proteins per second. Thus, not only is the signal very fast, but it is also highly amplified, as each activated G protein can go on to activate downstream effectors. This is not surprising, given the enormous evolutionary pressure on organisms to process visual stimuli as rapidly as possible. On the other hand, in more typical GPCR signaling pathways, such as stimulation of the β-adrenergic receptor leading to activation of adenylyl cyclase, activation of Gα_s takes ~0.5 s, so both the speed and potential signal amplification (the number of molecules of G protein acti- vated per second by activated receptor) are considerably lower. Transmembrane recePTors associaTed WiTH enzymaTic acTiviTy Except for the ion channels, all transmembrane receptors are directly or indirectly linked to intracellular enzyme activities. For many receptors, the enzymatic activity is encoded in the same polypeptide chain as the ligand-binding activity; such receptors are said to have intrinsic enzymatic activity. These include the GPCRs discussed above, in which ligand bind- ing activates GEF activity, and receptors with protein kinase, protein phosphatase, or guanylyl cyclase activity. Receptors can also be noncova- lently associated with enzymes such as protein kinases and proteases. Receptor tyrosine kinases control important cell fate decisions in multicellular eukaryotes Receptor tyrosine kinases (RTKs) are transmembrane receptors with intracellular protein tyrosine kinase domains. There are ~50 RTKs in the human genome, which can be divided into families based on their sequence similarity (Figure 8.8). These receptors bind polypeptide ligands that regulate cell proliferation, growth, differentiation, or migration; thus they play a critical role in development and tissue homeostasis in multicellular organisms (metazoans). Most RTKs are activated by dimerization induced by ligand binding, which allows the tyrosine kinase catalytic domains of the dimerized receptors to transphosphorylate each other on the “activation loop” and thus stably activate each other. In some cases, dimerization can also directly induce allosteric changes that increase kinase activity. The specific details of the catalytic activation mechanism vary among the different RTKs, but involve some combination of active-site rearrangement and unblocking of the substrate-binding cleft, leading to a catalytically active kinase domain capable of phosphorylating the dimer partner and other substrates. In some cases, for example the EGF and PDGF receptors, autophosphoryla- tion of the receptor on multiple sites is sufficient to recruit effector proteins containing SH2 or PTB domains, and thus transmit downstream signals. These effectors include the regulatory subunit of phosphatidylinositol 3-kinase (PI3K), PLC-γ, and adaptor proteins such as Grb2 and/or Shc that recruit activators of Ras (see Figure 4.11). In other cases, such as the insu- lin receptor and fibroblast growth factor (FGF) receptor, receptor activation leads to the phosphorylation of scaffold proteins on multiple sites, and it is these scaffold proteins that then serve as the primary sites for recruit- ment of downstream effectors. Such scaffolds include IRS1 for the insulin receptor and FRS2 for the FGF receptor. In some cases, the effectors that are recruited are themselves phosphorylated by the receptor, which further increases their activity or recruits additional effectors. But regardless of the specific details, activation of all RTKs leads to increased tyrosine phos- phorylation of the receptor and receptor-associated proteins, and to the recruitment of a variety of downstream effector proteins that couple to var- ious signaling pathways including the Ras/Raf/MAPK (mitogen-activated protein kinase), PI3K/Akt, and PLC/Ca2+/protein kinase C pathways. TGFβ receptors are serine/threonine kinases that activate transcription factors In metazoans, the transforming growth factor β (TGFβ) receptor family is the only class of receptors with intrinsic serine/threonine kinase activity. The TGFβ receptors bind to ligands such as TGFβ, activin, bone Figure 8.8 Receptor tyrosine kinases. The domain structures of the major families of receptor tyrosine kinases. egf, epidermal growth factor; igf, insulin-like growth factor; ngf, nerve growth factor; Pdgf, platelet- derived growth factor; m-csf, macrophage- colony-stimulating factor; fgf, fibroblast growth factor; vegf, vascular endothelial growth factor; eph, ephrin. (adapted from r. Weinberg, The biology of cancer. garland science, 2013.) Figure 8.9 Signaling by TGFβ receptors. Type i and type ii receptors exist as homodimers, which associate into a heterotetrameric complex upon ligand binding. The type ii receptor phosphorylates the type i receptor, activating it and increasing its affinity for r-smads ( li g h t p u r p l e ). Phosphorylation of r-smad by the type i receptor leads to its dissociation from the receptor and formation of a heterotrimeric complex with smad4 (light green). This complex is then free to translocate to the nucleus and regulate transcription of Tgfβ-responsive genes. TGFβ morphogenetic proteins (BMPs), and nodal that regulate developmental cell fate and proliferation. In humans, there are ~12 distinct receptors in the TGFβ family, which can be functionally divided into two classes (type I and type II). All have a similar overall structure, with a single membrane- spanning domain and an intracellular serine/threonine kinase domain. TGFβ receptors couple to transcriptional activation in a rather direct fashion, by phosphorylating and activating the SMAD family of transcrip- tion factors (Figure 8.9). As we have seen for other single-pass recep- tors, binding of the receptors to ligand induces dimerization, but in this case preexisting dimers of type I and type II receptors associate upon ligand binding, generating a heterotetrameric activated complex. In this complex, the type II receptor, which is constitutively active, can phos- phorylate the type I receptor through proximity. This has two important consequences: it greatly increases the affinity of the receptor for a class of SMAD proteins called the R-SMADs, and it increases the kinase activ- ity of the type I receptor itself. The type I receptor then phosphorylates the associated R-SMAD, leading to its dissociation from the receptor. The phosphorylated R-SMAD can then form heterotrimeric complexes with SMAD4, the so-called “co-SMAD,” and this complex is then competent to be transported into the nucleus and bind specific chromatin sites where it can activate transcription of TGFβ-responsive genes. It is interesting to note the very close parallels between the TGFβ receptor signaling mechanism and that used by receptors that couple to tyrosine kinases, despite the dissimilarity of the individual components. In both cases, ligand binding induces receptor dimerization/oligomerization, lead- ing to transphosphorylation. This, in turn, activates the intrinsic kinase activity of the receptor and recruits downstream effectors, which can then be phosphorylated and thus activated by the receptor. In plants, trans- membrane receptors with intracellular serine/threonine kinase domains are highly abundant and diverse. Clearly this is a robust and effective signal transduction mechanism that has been used multiple times in the course of evolution. Some receptors have intrinsic protein phosphatase or guanylyl cyclase activity A relatively large number of protein tyrosine phosphatases (PTPs) pos- sess transmembrane and extracellular domains, and thus are presumed to act as receptors, though in most cases their specific ligands are not known. Often the extracellular domains of the receptor tyrosine phos- phatases (RPTPs) resemble domains that mediate cell–cell or cell–sub- strate adhesion and, in at least a few instances, RPTPs have been shown to mediate cell–cell adhesion through homophilic interactions. This raises the possibility that these proteins may play a general role in modulat- ing local protein tyrosine dephosphorylation in response to cell–cell adhe- sion. This is plausible because many of the proteins that control cell–cell junctions, cell–cell and cell–substrate adhesion, and coupling to the actin cytoskeleton are regulated by tyrosine phosphorylation. The mechanism whereby ligand binding might regulate the phosphatase activity of the RPTPs is not yet fully understood. The cytoplasmic region for most of these receptors consists of two tandem PTP domains, though only the membrane-proximal domain is thought to be capable of catalytic activity (Figure 8.10). There is some evidence that dimerization or clus- tering of the receptor inhibits PTP activity, though the precise mechanism is not known. Thus the unliganded, monomeric receptor is constitutively active, and ligand binding leads to decreased activity. If this is the case, the net result of ligand binding to RTKs and RPTPs would be the same: a net increase in tyrosine phosphorylation in the vicinity of the receptors. Another class of receptors with intrinsic enzymatic activity consists of the membrane guanylyl cyclase receptors (mGCs). There are around six distinct mGCs in humans, including the receptors for atrial natriuretic peptide (ANP) and its relatives BNP and CNP (B-type and C-type natri- uretic peptide) which regulate kidney and smooth muscle function and guanylin and uroguanylin (which regulate intestinal water and electro- lyte transport). The ligands for other mGCs remain unknown, but the fact that several mGCs are found specifically in the olfactory epithelium or retina suggests roles in sensory transduction. Regulation of these enzymes by ligand binding is as yet poorly understood. The mGCs all contain a protein kinase-like domain (kinase homology domain, or KHD) located between the transmembrane and GC domains (Figure 8.11). The KHD does not appear to be capable of catalyzing phos- phate transfer itself, but is highly phosphorylated in the basal (unlig- anded) state. Ligand binding to the mGC receptors apparently shifts the equilibrium toward dimerization, and induces conformational changes in the dimer that promote ATP binding to the KHD and relieve inhibition of the GC domain, which becomes activated. Phosphorylation of the KHD seems to be required for receptor activation, and dephosphorylation by phosphatases leads to its inactivation (desensitization). The kinases and phosphatases that mediate these reactions are not yet known. Noncovalent coupling of receptors to protein kinases is a common signaling strategy Many cell-surface receptors do not have intrinsic catalytic domains, but instead their intracellular portions interact noncovalently with proteins with catalytic activity, such as protein kinases or proteases. A wide vari- ety of receptors couple to nonreceptor tyrosine kinases, exemplified by the cytokine receptors, T and B cell receptors, and integrins. Here, we will briefly outline some specific aspects of how these receptors activate their associated kinases, and the downstream effectors of that activation. Figure 8.10 Model for regulation of receptor tyrosine phosphatase activity. in the unliganded state (left), the phosphatase activity of the receptor tyrosine phosphatase (rPTP) is active. most rPTPs have two phosphatase homology domains, but only the membrane-proximal domain (purple) has enzymatic activity. upon ligand binding and clustering, for example by homotypic interactions with rPTPs from an adjacent cell (orange), conformational changes lead to inactivation of the phosphatase domain and thus localized increases in tyrosine phosphorylation. Figure 8.11 Regulation of receptor guanylyl cyclases. (a) in the basal (unactivated) state, receptor guanylyl cyclases such as the atrial natriuretic peptide (anP) receptor exist as homodimers. The kinase homology domain (KHd) is highly phosphorylated, and the guanylyl cyclase domain (gc) is inactive. upon ligand binding, conformational changes lead to binding of aTP to the KHd and to activation of the gc domain, leading to increased intracellular cyclic guanosine monophosphate (cgmP) levels. dephosphorylation of the KHd leads to release of aTP, inactivation of the gc domain, and release of bound ligand. The desensitized receptor cannot be activated until the KHd is phosphorylated again. T cell receptor signaling is described in more detail in Chapter 12 The cytokine receptors, which couple to STAT transcriptional activa- tors, have been introduced in Chapter 5. In general, these receptors are associated with nonreceptor tyrosine kinases of the JAK family (so-called “Janus kinases,” after the two-faced Roman god, because they contain tan- dem catalytic domains). Receptor dimerization induced by ligand binding leads to transphosphorylation of the cytoplasmic tails of the receptor; the phosphorylated receptor then serves to recruit the SH2 domain-containing STAT proteins, which are then phosphorylated by the associated JAK (Figure 8.12a). STAT phosphorylation induces conformational changes that promote its dimerization, nuclear transport, and DNA binding, and ultimately leads to changes in the transcription of STAT-responsive genes. Similar themes are seen in the transduction of signals from immune (B and T cell) receptors, and from adhesion receptors such as integrins. In the case of the T cell receptor—perhaps the best-studied example of a receptor that signals through nonreceptor tyrosine kinases—the CD4 co-receptor of the T cell receptor is associated with the Src family kinase Lck (Figure 8.12b). When receptor aggregation is induced by interac- tion of the T cell receptor with peptide–MHC (major histocompatibility complex) complexes displayed by antigen-presenting cells, Lck becomes activated through transphosphorylation. The proximal effect of Lck acti- vation is the phosphorylation of the receptor ζ chain on pairs of tyrosine residues separated by ~10 amino acid residues, which constitute a recog- nition site (termed the immunoreceptor tyrosine-based activating motif, or ITAM) for a second nonreceptor tyrosine kinase, ZAP-70. This kinase itself has tandem SH2 domains that engage the doubly phosphor- ylated ITAM; binding both localizes ZAP-70 to the liganded receptor and activates it so that it can then phosphorylate a number of other receptor- associated proteins. This promotes assembly of a large signaling complex (often termed the “immune synapse”), which leads to activation of a vari- ety of downstream signaling pathways. Integrins comprise a diverse family of cell-surface adhesion receptors that bind to cell-matrix- or cell-surface-associated peptides, such as fibronectin, laminin, and fibrinogen. Each integrin is a heterodimer con- sisting of α and β subunits, each with a large extracellular ligand-binding domain and a small intracellular portion. Integrins play a very impor- tant role in coupling cell adhesion to the actin cytoskeleton, so the cell (a) cytokine receptor (b) T cell receptor (c) integrins receptor aggregation leads to kinase activation kinase activated: JAKs substrate: receptor LCK TCR zeta chain FAK FAK Y397 phosphorylation leads to recruitment of second factor P P P P protein recruited: STATs subtrate: STATs TCR-associated SFK FAK, FAK-associated ZAP70 proteins proteins assembly of active complex effects: STAT dimerization and phosphorylation Figure 8.12 effects assembly of active TCR signaling complex assembly of focal adhesion Three examples of receptors coupled to activation of nonreceptor tyrosine kinases. examples depicted are (a) cytokine receptors, (b) T cell receptor (Tcr), and (c) integrins. for each receptor, the first step (upper panels) involves increased local concentration and activity of a nonreceptor tyrosine kinase (green) associated with the receptor, promoted by proximity in the dimerized/clustered receptor. This leads to increased phosphorylation of a substrate protein (arrows). in the second step (middle panels), this phosphorylated substrate serves to recruit an sH2 domain-containing effector protein (green). in the third step (lower panels), phosphorylation of additional substrates leads to downstream biological effects. see text for details of each pathway. in (b), for clarity, only one set of kinases in the middle and lower panels. faK, focal adhesion kinase; sfK, src family kinase. can apply traction force to its surroundings and thereby move. Integrins operate on a number of different levels, both as simple receptors and as sensors for forces, and as a mechanical linkage between the inside and outside of the cell. Here, we will consider two of these signaling modes, one of which involves clustering and activation of associated nonreceptor tyrosine kinases and another that involves allosteric changes. The first level of signaling from integrins is via nonreceptor tyrosine kinases, and is conceptually similar to other examples where receptor (b) outside-in extracellular matrix (ECM) (a) β α integrin [ECM] high affinity for talin (d) talin low affinity for talin [talin], talin activation (c) high affinity for ECM inside-out Figure 8.13 Bidirectional signaling by integrins. in the basal state, the unliganded integrin heterodimer has relatively low affinity for extracellular matrix (ecm) components via its extracellular domain, and for talin via its intracellular domain. conformational changes induced either (b) by binding to high concentrations of ecm components or (c) by binding high concentrations of talin or post-translationally modified talin, lead to (d) a receptor with higher affinity for both ecm and for talin. The effect is to generate clusters of activated integrins linking ecm to the actin cytoskeleton. dimerization/clustering is coupled to kinase activation. In this case, the integrin-associated kinase is focal adhesion kinase (FAK), a nonrecep- tor tyrosine kinase with a central catalytic domain flanked by rather large N-terminal and C-terminal regions with a number of protein interaction motifs (Figure 8.12c). These interaction domains mediate binding to the integrins and other proteins. Upon integrin clustering, FAK molecules are brought into close proximity, whereupon a key tyrosine residue is phos- phorylated, most likely via transphosphorylation. This phosphotyrosine then serves as a docking site for the SH2 domain of Src family kinases; binding has the dual consequence of localizing the kinase to the site of integrin engagement and of activating it by preventing it from adopting the closed, inactive conformation. The activated Src family kinase then phosphorylates a number of proteins, most notably FAK itself and other FAK-associated proteins such as p130Cas, altering their activity and cre- ating binding sites to recruit additional SH2 domain-containing proteins to the nascent focal adhesion. A second type of signaling by integrins serves to mechanically couple extracellular ligands and the actin cytoskeleton. Integrins undergo pro- found conformational changes when their extracellular domains bind to ligand, which alter the binding properties of the intracellular portion by changing the relative orientation of the α and β chains. The effect is to increase the affinity of the intracellular domain for proteins that cou- ple to the actin cytoskeleton, such as talin (Figure 8.13). Thus, ligand binding is intimately associated with linking the adhesion receptors to the cytoskeleton. Indeed, the same conformational changes can transmit information in the reverse direction—from inside the cell to the outside (so-called “inside-out signaling”). This is because binding of talin to the intracellular domain of integrin (promoted, for example, by increased local concentrations of talin due to clustering, or by post-translational modifications of talin that increase its affinity for the integrin) converts the integrin from the inactive to active conformation, which has higher affinity for extracellular ligands. This interplay between ligands and the actin cytoskeleton will tend to promote the localized clustering of acti- vated integrins into patches. These patches serve as the nucleators for the formation of focal adhesions, highly complex cellular structures that couple sites of cell adhesion to F-actin cables (stress fibers). Some receptors use complex activation pathways that involve both kinase activation and proteolytic processing A number of other receptors couple more indirectly to the activation of protein kinases; the activation mechanisms for such receptors are more complex and involve additional activities, such as regulated proteolysis and the remodeling of elaborate multiprotein complexes. In this section, we will consider the activation of the transcription factor NF-κB by Toll- like receptors; in the following section, we will discuss Wnt and Hedgehog, two other pathways in which transcriptional activators are regulated by a combination of phosphorylation and proteolysis. The Toll-like receptors (TLRs) play an important role in innate immu- nity by detecting pathogens and stimulating host defenses to fight off the invaders. There are at least 13 members of the TLR family in humans, each binding to a distinct pathogen-specific structural motif, such as lipopolysaccharide (LPS), flagellin, or single- or double-stranded RNA. Ligand-bound TLRs indirectly activate the NF- κ B family of transcription factors. NF-κB is a central mediator of responses to a wide variety of cel- lular stimuli. It exists in a latent, inactive form in unstimulated cells that can be rapidly mobilized without requiring new transcription or transla- tion. NF-κB is activated by the phosphorylation, ubiquitylation, and pro- teasome-mediated degradation of inhibitory subunits, termed IκBs, that are bound to the latent NF-κB. Phosphorylation of IκB, which triggers its degradation, is mediated by a heterotrimeric assembly termed the IKK (IκB kinase) complex. The mechanism whereby TLR ligand binding promotes IκB degradation is quite complex, involving a number of intermediates. The mechanism involves receptor dimerization and conformational changes leading to binding of a class of adaptor proteins containing a TIR domain, exem- plified by MyD88 (Figure 8.14). These adaptors also contain a so-called death domain (DD), which then binds to the death domains of the IRAK family of serine/threonine kinases, leading to activation of the IRAKs, pre- sumably through a combination of proximity and conformational changes. Activated IRAKs then promote downstream signaling by dissociating from the receptor and binding to and activating TRAF-6, a ubiquitin E3 ligase. Activated TRAF-6 causes Lys63-linked polyubiquitylation of itself and other substrates, which in turn leads to the recruitment and activa- tion of another kinase, TAK1, in complex with cofactors TAB2 and TAB3. NF-κB activation is described in more detail in Chapter 9 Figure 8.14 Signaling by Toll-like receptors. dimerization of Toll-like receptors (Tlrs) through ligand binding leads to recruitment of the myd88 adaptor by homotypic interactions of Tir domains. This leads to recruitment and activation of the kinase iraK4 via homotypic interactions of death domains ( b r o w n o b l o n g s ). iraK4 activates iraK1, which binds to Traf-6, activating its e3 ubiquitin ligase activity which leads to the polyubiquitylation of Traf-6. The Tab2 and Tab3 adaptors bind to ubiquitylated Traf-6 via ubiquitin-binding domains, leading to recruitment and transactivation of the kinase TaK1, which activates the iκb kinase complex (iKK), finally leading to activation of nf-κb. The result of TAK1 activation is, finally, activation of the IKK complex, which then is free to phosphorylate IκB and promote its polyubiquityla- tion and destruction. Despite the fearsome complexity of the overall mechanism (which is sim- plified here for clarity), it is important to note that it involves the same basic mechanisms that have been discussed over and over again in this chapter—receptor aggregation, conformational changes leading to chang- es in the binding of proteins to the receptor, and activation of associated kinases through proximity. Polyubiquitylation and proteasome- mediated degradation are dis- cussed in more detail in Chapter 9 Wnt and Hedgehog are two important signaling pathways in development The Wnt signaling pathway also uses a complex mixture of kinase activation and proteolysis (Figure 8.15). Members of the Wnt family con- trol many developmental cell fate decisions, and thus play an important role in normal development and in tissue homeostasis. The effector for canonical Wnt signaling (there are other modes of signaling from the Wnt receptors, such as the so-called planar cell polarity pathway, that will not be discussed here) is a latent transcription factor, β-catenin. In unstimu- lated cells, β-catenin is prevented from entering the nucleus and is rapidly degraded via sequestration in a multiprotein complex (the cytoplasmic destruction complex) that contains β-catenin, two scaffolding proteins (axin and APC), and two serine/threonine kinases (CK1 and glycogen synthase kinase 3, or GSK-3). Within this complex, β-catenin is phosphor- ylated, leading to recognition by β-TrCP (a specificity factor for a ubiquitin E3 ligase), polyubiquitylation, and ultimately targeting β-catenin to the proteasome for degradation. Upon Wnt stimulation, however, this com- plex is disrupted, allowing accumulation of intact, uncomplexed β-catenin, which is then free to translocate to the nucleus and stimulate transcrip- tion in association with nuclear coactivators. Figure 8.15 Wnt signaling. The receptor for Wnt consists of frizzled, which has the structure of a g protein coupled receptor, and lrP. in the absence of Wnt, β-catenin is rapidly degraded by a cytosolic destruction complex. The core of this consists of aPc and axin, which recruit β-catenin and the kinases glycogen synthase kinase 3 (gsK-3) and cK1, leading to phosphorylation of β-catenin. Phosphorylated β-catenin is recognized by β-TrcP, which leads to polyubiquitylation of β-catenin and targets it for degradation by the proteasome. in the presence of Wnt, however, the receptor heterodimer assembles; disheveled (dvl) is recruited, and lrP is phosphorylated, leading to axin recruitment. recruitment of axin to the receptor leads to disruption of the cytosolic destruction complex, through depletion of the pool of cytosolic axin. β-catenin is no longer phosphorylated, ubiquitylated, and degraded, and is free to translocate to the nucleus and regulate transcription of Wnt-dependent genes. Wnt Wnt receptors have two components: Frizzled, a member of the GPCR family, and a single-pass receptor called LRP. Wnt apparently binds to both, thereby promoting the close association of the two co-receptors. The receptor heterodimer has two new properties, both of which are impor- tant for downstream signaling. First, the tail of LRP becomes accessi- ble for phosphorylation by CK1 and GSK-3, which greatly increases its affinity for axin (note that in this case, CK1 and GSK-3 are not part of the destruction complex, but are part of a distinct, membrane-associated pool). Second, the liganded Frizzled is able to induce the phosphorylation and recruitment of yet another protein, Disheveled. This may involve the activation of a heterotrimeric G protein by liganded Frizzled. Phosphor- ylated Disheveled and axin can also associate with each other directly, so assembly of an activated LRP–Frizzled–Disheveled–axin complex on the membrane is likely to be highly cooperative. The assembly of this complex then leads indirectly to the dissociation of the destruction complex and release of free β-catenin. This is likely accomplished through simple mass action, by sequestering axin at the membrane and decreasing the pool that is available to form the destruction complex. The hedgehog (Hh) signaling pathway has a number of mechanistic similarities to the Wnt pathway. Hh signaling is important for the nor- mal development and patterning of virtually all tissues in multicellular organisms. Hh was first studied in Drosophila, where there is a single Hh ligand; in vertebrates there are three Hh ligands, the most widely dis- tributed of which is termed Sonic hedgehog (Shh). As in the case of Wnt, the ultimate effect of Hh signaling is transcriptional activation, mediated for vertebrate Hh by members of the Gli family (Figure 8.16). In cells not stimulated by Hh ligands, Gli is phosphorylated by protein kinase A, GSK-3β, and CK1. This leads to degradation of its C-terminus by the proteasome, generating a truncated form that acts as a transcriptional repressor. Hh ligands cause a change in the processing of Gli, such that a full-length or alternatively processed form that acts as a transcriptional activator predominates. It is thought that the balance in the cell between the repressor and activator forms of Gli determines signal output, allow- ing graded signals (for example, from spatial gradients of Hh ligands) to be read out by each cell in terms of transcriptional activity. One of the most interesting aspects of Hh signaling in vertebrate cells is that it is intimately connected to the primary cilium, a specialized filamentous organelle constructed from microtubules. The structure of the primary cil- ium is very similar to that of the flagella found on single-celled eukaryotes, and in vertebrates most cells contain a single primary cilium. More and more evidence is emerging that this organelle can serve as a signaling center, by co-localizing receptors, ligands, and cytosolic effectors, and also by actively transporting various components within the cilium via microtubule-based motor proteins. It is interesting to note that while the primary cilium is abso- lutely essential for Hh signaling in vertebrates, it appears to be unnecessary in Drosophila, where most cells are not ciliated. It is not known whether the dependency of the pathway on the primary cilium was lost in Drosophila, or whether it evolved de novo in the vertebrate lineage. How is Gli processing affected by Hh ligands? The receptor for Hh ligands is a protein with 12 membrane-spanning segments termed Patched (Ptc). Ptc normally functions to negatively regulate the activity of a seven- transmembrane protein of the GPCR family termed Smoothened (Smo). The precise nature of this negative regulation is still under investigation, but it appears not to be due to simple protein–protein interaction. One possibility is that Ptc (which is related to a family of bacterial membrane transporters) regulates the influx or efflux of a small-molecule ligand for Figure 8.16 Hedgehog signaling. signaling in vertebrates by sonic hedgehog (shh) occurs at the primary cilium. (a) When shh is absent, the shh receptor Patched (Ptc) is localized on the primary cilium and inhibits the activity of smoothened (smo). gli transcription factors undergo phosphorylation by protein kinase a (PKa), glycogen synthase kinase 3 (gsK-3), and cK1, leading to recognition by β-TrcP, polyubiquitylation, and partial proteolysis by the proteasome. The remaining fragment of gli (glir) has transcriptional repressor activity. (b) in the presence of shh, Ptc inhibition of smo is relieved, and Ptc relocalizes from the primary cilium. smo becomes activated and undergoes conformational changes that allow it to be phosphorylated by a g-protein-coupled receptor kinase (grK) and other kinases. Phosphorylated smo binds to arrestin and is transported up the primary cilium with the assistance of motor proteins such as Kif7. gli is also transported up the cilium and avoids phosphorylation. full-length, unphosphorylated gli translocates to the nucleus where it can activate gene transcription. Smo. Whether Smo actually functions as a GPCR is also still under inves- tigation, but what is clear is that its conformation dramatically changes upon binding of Hh ligands to Ptc. This causes Smo to become heavily phosphorylated on its C-terminus and to associate with other proteins such as arrestin (which normally functions to down-regulate and direct some aspects of downstream signaling from GPCRs, as discussed later in this chapter). In addition, in vertebrates, binding of Hh to Ptc leads to its relocalization from the primary cilium, where it is replaced by Smo. These changes lead to a rearrangement of the complexes that process Gli, such that Gli escapes phosphorylation and C-terminal processing and can thus enter the nucleus to activate transcription of its target genes. A variety of receptors couple to proteolytic activities After protein kinases, proteases are the enzymes most often used by recep- tors to convey information to the cell interior. Above, we discussed several examples where proteolysis plays a supporting role in receptor signaling. We will now consider two examples where proteolysis is the predominant mechanism. Proteolysis is unlike many other signaling transactions, in that it is essentially irreversible; resetting the system requires synthesis of new proteins and/or further degradation of signaling mediators gener- ated by proteolysis. Thus, this mechanism is most likely to be used for systems that must respond strongly and decisively to a signal, but for which temporal control on short time scales (that is, the ability to turn off or otherwise regulate the signal once initiated) is less important. The Notch signaling pathway was introduced in Chapter 5. Notch receptors are proteolytically processed upon binding to their ligands (such as Delta), which are presented on the surface of a neighboring cell. Ligand-induced cleavage liberates the Notch intracellular domain (NICD), which then translocates to the nucleus and promotes transcription of Notch-responsive genes. The specific mechanism whereby ligand engagement triggers release of the NICD is still imperfectly understood. It is known that ligand engage- ment first stimulates cleavage by an extracellular metalloprotease of the ADAM/TACE family at the “S2 site” of Notch, releasing the Notch extracel- lular domain (NECD) (Figure 8.17). The remaining part of the receptor is then cleaved within the hydrophobic transmembrane segment (at the “S3 site”) by the γ-secretase complex to generate the active NICD. Endocytosis of the Notch ligand by the signaling cell seems to be required for optimal signaling in the receiving cell, raising the interesting possibility that pull- ing forces due to endocytosis of the ligand expose the S2 cleavage site on the receptor. If true, this would be a rare example of signal transmission by a single-pass transmembrane receptor that does not involve dimerization or clustering. A number of transmembrane proteins, including the amyloid precursor protein (APP), whose cleavage products accumulate in neurons to cause Alzheimer’s disease, are cleaved in the membrane in a two-step process similar to Notch. Thus, the Notch pathway may be only one specific example of a more widespread but poorly understood signaling mechanism. Another example of receptors linked directly to proteolytic activities is pro- vided by the death receptors, which induce apoptosis, or programmed cell death, when activated by their ligands. These receptors include the tumor necrosis factor receptor (TNFR), TRAIL receptor, and Fas/CD95. Apoptosis is caused by the activation of a class of proteases termed caspases; the immediate consequence of death-receptor engagement is caspase activa- tion. The ligands for death receptors, such as tumor necrosis factor (TNF) and TRAIL, are all homotrimeric in structure, and their transmembrane receptors are also constitutively trimeric, even in the absence of ligand. When ligand binds to the extracellular domain of the receptors, however, The roles of proteases in signal- ing are discussed in more detail in Chapter 9 Apoptosis is discussed in more detail in Chapter 9 Figure 8.17 Notch signaling. dsl (delta-serrate- lag2) ligand on the signal-sending cell engages notch receptors on the signal- receiving cell. This exposes the s2 site of notch for proteolytic cleavage by the metalloprotease Tace, perhaps induced by pulling forces generated by endocytosis of dsl by the signal-sending cell. cleavage at the s2 site allows access to the s3 site by the γ-secretase complex, which then cleaves notch at the intramembrane s3 site. The notch intracellular domain (nicd) is then free to relocalize to the nucleus to regulate transcription of notch-responsive genes in the signal-receiving cell. FasL conformational changes are induced which expose protein interaction motifs on the intracellular portion of the receptor. All death receptors con- tain a modular domain termed the death domain (DD), which can dimer- ize with DDs in other proteins, as mentioned above for TLR signaling. In the case of Fas, an archetype for death receptors, the DDs that are exposed upon receptor activation bind to the DDs of an adaptor protein, FADD (Figure 8.18). In addition to its DD, FADD contains another modu- lar dimerization motif, the death effector domain (DED). Once the DD of FADD is engaged by the receptor, the DED is free to bind the DED of one of the so-called initiator caspases, caspase-8 or caspase-10. This, in turn, serves to activate the caspase, inducing its proteolytic processing from an inactive proenzyme form to an active holoenzyme. The supramolecular com- plex containing the receptor, adaptor proteins such as FADD, and initiating caspases such as caspase-8 is termed the death-inducing signaling complex (DISC). It is likely that caspase activation by the receptor can be explained by proximity—bringing multiple procaspases together in the DISC increases the likelihood that one will cleave the other into the activated form, initi- ating a cascade in which all bound procaspases will rapidly be activated. This is directly analogous to the mechanism of activation of tyrosine kinas- es by aggregation, discussed above. It is also likely that the association Figure 8.18 Signaling by death receptors. death receptors such as fas exist as constitutive trimers. binding of trimeric ligands such as fas ligand (fasl) induces conformational changes in the receptor that promote homotypic interactions of the death domains (light brown oblongs) of the receptor with those of adaptors such as fadd. This exposes the death effector domains (deds, green oblongs) of fadd, which interact with the deds of procaspase-8. The resulting large assembly, termed the death- inducing signaling complex (disc), promotes autocatalytic processing of procaspase-8, likely by a combination of proximity and conformational change. active caspase-8 is then free to diffuse throughout the cell to cleave substrates and effect the apoptotic cell death program. of procaspase molecules with each other induced by receptor-mediated aggregation induces conformational changes that facilitate activation and proteolytic processing. Consistent with the key role of proximity in the activation mechanism, it has been found that death receptor trimers aggregate into larger supramolecular structures upon ligand binding. gaTed cHannels A further major class of signaling receptors is the gated ion channels, which respond to ligands or other environmental cues by altering the permeability of the membrane to ions or other small molecules. Such gated channels play a fundamental role in neurotransmission, opening in response to neurotransmitters, membrane depolarization, or other stimuli. These channels allow specific ions to flood rapidly across the membrane, causing massive and very rapid changes in the electrical properties of the membrane, the basis for electrical signals. Because of their importance for neurotransmission, such channels are the targets of a number of familiar drugs. For example, local anesthetics such as novocaine block voltage- gated sodium channels, while calcium channel blockers (CCBs) are used to treat high blood pressure. Gated channels also regulate intracellular signaling, as in the case of calcium channels on the endoplasmic reticulum that open when bound to inositol trisphosphate, leading to the release of intracellular calcium stores. In this section, we will not deal with the electrophysiology of neuronal signaling, but will consider a few examples of gated channels that illustrate the mechanisms of their regulation by stimuli and their selectivity. Gated channels share a similar overall structure All gated channels, irrespective of the stimuli that regulate them or the solutes that they conduct, share similarities in overall topology. All are composed of multiple similar or identical subunits arranged in a ring structure in the plane of the membrane (Figure 8.19a). Depending on the specific channel, this ring may be composed of two to five subunits (with four or five subunits being the most common arrangement), though in some cases all subunits are combined in a single polypeptide chain. Each subunit can consist of anywhere from two to six hydrophobic α heli- ces that span the membrane, connected by loops of varying lengths. In Cys-loop family out in Kv, TRP families N (b) C C Figure 8.19 The structure of gated ion channels. diagrammatic depiction of the organization and subunit structure of gated ion channels. Pentameric subunit composition, in which each subunit has four transmembrane helices (top), is typical of a variety of ligand-gated ion channels of the cys-loop superfamily involved in neurotransmission. a tetrameric subunit composition, where each subunit has six transmembrane helices, is found in voltage-gated potassium (Kv) channels and TrP (transient receptor potential) channels. diagrammatic cross section of a channel embedded in the lipid bilayer. ions with particular properties (size and charge) can selectively pass through the pore of the open channel. the center of the ring is a narrow aqueous pore, which allows the selec- tive passage of certain molecules. This pore is lined with helices, arrayed perpendicular to the membrane like barrel staves. The specific properties of the pore determine which molecules can pass: the size of the opening, and in particular its electrostatic properties, determine which anions or cations will be attracted into the pore, and those that will be repelled or sterically excluded (Figure 8.19b). In addition to the physical structure of the pore itself, the gated channels must also be able to control the opening and closing of the pore in response to ligands or other signals (such as changes in voltage across the mem- brane, changes in temperature, or changes in extracellular pH). These chemical or physical changes must be converted into mechanical energy, to drive changes in the relative orientation of the helices that line the pore. In this way, a channel that in its resting state is sterically “closed” is converted to an open form that allows passage of small molecules across the membrane. We will briefly consider two families of gated channels for which structural data are available. Figure 8.20 The selectivity filter of the K+ channel. (a) open-channel model of the bacterial Kcsa K+ channel, showing the large aqueous pore and the selectivity filter near the extracellular (upper) face of the membrane. Potassium ions are shown as green spheres. (b) X-ray crystal structure of the Kcsa selectivity filter. Potassium ions are shown as green spheres, and water molecules as red spheres. note that K+ ions on either side of the selectivity filter are hydrated, but the ions passing through the filter have been dehydrated. carbonyl groups from the amino acids lining the filter coordinate to the K+, replacing water molecules. The size of the pore is a perfect fit for K+. (a, adapted from e. gouaux and r. macKinnon, Science 310:1461–1465, 2005. With permission from aaas; b, adapted from y. zhou et al., Nature 414:43–48, 2001. With permission from macmillan Publishers ltd.) The voltage-gated potassium channel provides clues to mechanisms of gating and ion specificity Voltage-gated potassium (Kv) channels play an important role in propa- gating action potentials in neurons. They respond to membrane depolari- zation by opening and allowing the passage of K+ out of the cell down its concentration gradient, thereby repolarizing the membrane to its resting state. Kv channels are one example of a large family of channels con- sisting of four identical subunits, each of which has six transmembrane helices. Pioneering work by the group of Rod MacKinnon has provided high-resolution x-ray crystal structures of a number of Kv channels in different states, providing perhaps the best view yet of how ion selectivity and gating can be achieved. The pore of the Kv channel contains a narrow constriction roughly in the middle that serves as a selectivity filter. This filter is lined by 20 oxygen atoms, arranged in four potential K+-binding sites in which their partial negative charges electrostatically shield the positively charged potassium ions (Figure 8.20). The way in which the oxygen atoms coordinate to the K+ is very similar to how the “hydration shell” of water molecules shields the ion when it is dissolved in water. Thus, there is little if any energetic cost to stripping the hydration shell from the ion as it passes through the filter. The actual size of the pore is a good fit for K+ (1.33 Å), while it would be too loose for Na+ (0.95 Å), the other abundant singly charged cation in the cell. There is evidence that the binding of multiple K+ ions to the selectivity filter induces a slight conformational change in the helices lining the pore, thus lowering the affinity of the ion for the filter (because some of the binding energy is used for the rearrangement). This, com- bined with electrostatic repulsion between K+ in the filter, ensures that the affinity is not so high that ions would have difficulty escaping from the filter, leading to low conductance (the number of ions that can pass per unit time). Similar principles are used in other channels to generate pores specific for various ions depending on their size and charge. Much has also been inferred about the gating mechanism of Kv chan- nels from crystal structures. While many of the details still are under investigation, it is clear that a number of positively charged residues in a so-called “voltage sensor” (consisting of helices 3 and 4 of the channel) are physically translocated through the membrane in response to depolariza- tion (Figure 8.21a). In the resting state, the external side of the mem- brane carries a slight positive charge relative to the inside, whereas when (a) (b) (i) open (ii) hypothetical closed resting potential (channel closed) depolarized (channel open) Figure 8.21 Mechanism of gating of the voltage-gated K+ channel. (a) schematic representation of the “voltage sensor” of the voltage- gated K+ (Kv) channel, based on crystal structures of bacterial and bacterial–vertebrate hybrid Kv channels. The voltage sensor is depicted in light pink; for clarity only one sensor is shown, while the tetrameric channel contains a total of four sensors arrayed around the perimeter of the channel. The voltage sensor has four positive gating charges (blue + signs), which are attracted to the negatively charged side of the membrane, thus driving movement of the sensor depending on the membrane potential. The charges move relative to two clusters of negative charges on the main body of the channel (pink – signs). The area between the two clusters of negative charges is hydrophobic, thus only two positions (“open” and “closed”) for the voltage sensor are stable. (b) Top left (i) is a representation of the voltage sensor and s4–s5 linker helix in the open conformation from the crystal structure. Helices are drawn as ribbons. The view is from the pore looking out, with the extracellular solution “above” and the intracellular solution “below.” The positive gating charges are shown as blue sticks. negatively charged residues in the external and internal clusters are pink; the hydrophobic phenylalanine in the middle is green. Top right (ii) is a depiction of a hypothetical closed conformation of the voltage sensor. The positive gating charges now reach toward the intracellular solution, and are stabilized through interactions with the internal negative cluster. The inward displacement of the s4 helix pushes down on the n-terminal end of the s4–s5 linker helix (orange), causing it to tilt toward the intracellular side and to close the pore. bottom left (iii) is a depiction of the open conformation of the s4–s5 linker helices and pore from the crystal structure. note that the voltage sensor (above) is rotated 180° relative to the pore as depicted here. bottom right (iv) is a hypothetical model of the s4–s5 linker helices and pore in a closed conformation based on the crystal structure of a closed K+ channel pore (Kcsa). (b, adapted from s.b. long et al., N a t u r e 450:376–382, 2007. With permission from macmillan Publishers ltd.) The modular design of signaling proteins is discussed in Chapter 10 the membrane depolarizes, the inside becomes positively charged relative to outside due to the rapid influx of sodium ions. Due to simple electro- static forces, the positively charged residues in the voltage sensor will always be attracted to the more negatively charged side of the membrane and repelled by the more positively charged side. In the crystal structures, which correspond to the open conformation, the positively charged resi- dues of the voltage sensor are accessible to solvent on the extracellular side of the membrane. Presumably in the closed conformation, at nor- mal membrane resting potential in which the inside is negatively charged relative to outside, those positively charged residues would be oriented differently so that they are accessible to the cytosol. Clusters of negatively charged residues are present on both sides of the channel to help stabilize the positively charged residues in either of the two orientations, and the space in between these two negatively charged clusters is entirely hydrophobic. Thus, only two conformations are likely to be stable, and the voltage sensor is unlikely to get “stuck” halfway open. This switchlike property, in which only the fully closed and fully open states are stable, is very important for channel function. It has been shown that the voltage sensor can be transferred to other channels to confer voltage gating, demonstrating the modularity of signaling proteins. Although a structure of a closed Kv channel is not yet available, a reasonable model can be generated for how the translocation of the volt- age sensor would affect the packing of the helices lining the pore. In this model, the pore is constricted when the sensor is oriented toward the cytosol, and opens when the sensor is oriented toward the extracel- lular environment, as expected from the known electrophysiology of the channel (Figure 8.21b). A number of other channels share the same overall topology of the Kv channel, but are gated by other stimuli. An important group of chan- nels for signaling is the TRP (transient receptor potential) family. These channels are permeable to Ca2+ (in addition to other cations, with vary- ing degrees of specificity), and are often activated by multiple signaling inputs, including small molecules, lipids and lipid metabolites, heat or cold, and voltage. Such channels can act in signaling as input integrators or coincidence detectors. Several interesting examples include those that respond to heat and cold (these channels are how sensory neurons in the body detect temperatures that are uncomfortably cold or hot). TRPV1, the so-called vanilloid receptor, opens in response to heat, and also to the binding of the compound capsaicin, which is what gives chili peppers their fiery taste. By contrast, the TRPM8 channel opens in response to cool tem- peratures, and also to the binding of compounds such as menthol that provide a sensation of cooling. Ligand-gated ion channels play a central role in neurotransmission Signal transmission at synapses involves the release of neurotransmit- ters (such as acetylcholine, glutamate, and γ-aminobutyrate, or GABA) from the signaling (presynaptic) cell, which bind to ligand-gated ion chan- nels on the receiving (postsynaptic) cell. The opening of these channels allows ions to flow across the membrane, either initiating an action poten- tial in the case of cation channels, or inhibiting it in the case of anion channels. Because of their central role in all aspects of the nervous sys- tem, the ligand-gated ion channels are the targets of a number of drugs. The Cys-loop superfamily of ligand-gated ion channels has been studied intensively, and includes the nicotinic acetylcholine receptor (nAChR) and the receptors for GABA, glycine, and serotonin. These channels share a common overall topology, consisting of five nonidentical subunits, each with four transmembrane helices (see Figure 8.19a). Ligand-binding sites have been mapped to the large N-terminal extracellular domain, in the cleft between adjacent subunits. High-resolution crystal structures of vertebrate ligand-gated ion channels are not yet available, but the structures of several prokaryotic channels with sequence and structural homology to vertebrate channels have been solved (Figure 8.22a). The structures of two closely related cation chan- nels show the receptor in the closed and open conformation. In the open conformation, a fairly wide funnel-shaped chamber lined with hydrophilic residues leads to a relatively narrow transmembrane pore (Figure 8.22b). At its narrowest point, at the intracellular membrane border, a ring of negatively charged glutamate residues presumably acts as a specificity filter, helping to attract and shield the cation after its hydration shell has been stripped away, and ensuring that only positively charged solutes of a certain size can pass. This ring is highly conserved in other ion channels that conduct cations, such as the nAChR. closed open Figure 8.22 Structure of a pentameric ligand- gated ion channel. (a) left, ribbon representation of glic (a bacterial cation channel with the same topology as mammalian ligand-gated ion channels) viewed from within the membrane with the extracellular solution above. This channel is in the open conformation. right, structure of the pentameric channel viewed from the extracellular side. each channel subunit is depicted in a different color. (b) left, view of the α2 helices of glic defining the pore region. The front subunit is removed for clarity. The molecular surface is shown as gray mesh. upper right, intracellular part of the pore region. shown are the inferred positions of cs+ (gray), rb+ (blue), and zn2+ (pink) from crystal structures. bottom right, schematic representation of the pore-opening mechanism. The α2/α3 helices of two subunits in the closed (left) and open (right) conformation are shown. The ion-coordinating glutamate residues are shown in pink, the permeating ions in blue. (adapted from r.J.c. Hilf and r. dutzler, Nature 457:115–118, 2009. With permission from macmillan Publishers ltd.) In the closed conformation, the top part of the pore, near the extracellular membrane border, is sterically blocked by a plug consisting of bulky hydro- phobic side chains. The difference between open and closed conformations appears to be due to a rotation of the pore-forming helices perpendicular to the plane of the membrane, which opens the pore on the extracellular side and narrows it on the intracellular side (Figure 8.22b). This rotation must be caused by ligand binding, and the structures reveal close physical linkages between the ligand-binding sites in the extracellular domain and the transmembrane helices that could translate conformational changes between the two. membrane-Permeable signaling The simplest and most straightforward way for information to cross the plasma membrane is for signaling molecules to diffuse passively through the membrane itself (see Figure 8.2a). Two types of molecules function this way: gases (NO, O₂, and, to a lesser extent, CO) and hydrophobic small molecules such as steroid hormones. It is the small size of these signaling molecules and their solubility in both aqueous and hydrophobic environments that gives them the relatively unusual ability to traverse freely and rapidly between the extracellular fluid and the cytosol. The receptors for these two classes of molecules are found in the cytosol and thus are fundamentally different from all other receptors, which must span the plasma membrane to function. Nitric oxide mediates short-range signaling in the vascular system As introduced in Chapter 6, nitric oxide (NO) is an important signaling molecule that regulates physiological processes such as blood flow through the vasculature. NO is a simple diatomic gas that can pass freely through cell membranes. The chemical reactivity of NO with heme or oxygen is very high—its half-life in tissues is only several seconds—and so its range of action is relatively short. This is an example of a paracrine signaling molecule, meaning that it primarily functions to signal between adjacent or nearby cells. In the best-understood physiological context, NO is syn- thesized by NO synthase in vascular endothelial cells; the newly gener- ated NO then diffuses to adjacent vascular smooth muscle cells, where it causes relaxation and thus dilation of the blood vessels. The receptor for NO in the target cell is soluble guanylyl cyclase (sGC), an enzyme that converts GTP into cyclic GMP (cGMP) and pyrophos- phate. sGC is cytosolic, and is distinct from the transmembrane receptor guanylyl cyclases (mGCs) discussed above. NO binding can activate sGC by several hundredfold, thus coupling increases in cytosolic NO levels to increases in cytosolic cGMP. cGMP then acts as a small, soluble signal- ing mediator to regulate the activity of a variety of downstream targets, notably cGMP-dependent protein kinase isoforms (Figure 8.23a). It is these downstream effectors of cGMP that ultimately mediate the cellular responses to NO, such as smooth muscle relaxation. Soluble guanylyl cyclase is a heterodimer that contains a regulatory domain with a heme functional group (a ring structure with a single cen- tral iron atom). In the absence of NO, this regulatory domain represses the activity of the catalytic cyclase domain of the enzyme. Binding of NO to the ferrous iron of the heme group displaces a histidine side chain from the heme, inducing conformational changes that relieve repression of the catalytic domain. Once bound to heme, dissociation of NO is relatively (a) NO (b) low affnity NO site high affnity NO site slowly dissociates rapidly dissociates inactive baseline [NO] partly active high [NO] fully active Figure 8.23 Signaling by nitric oxide. (a) nitric oxide (no) generated by one cell diffuses into a neighboring cell where it binds to and activates soluble guanylyl cyclase (sgc). activation of sgc leads to increased intracellular levels of cgmP which, in turn, regulates other signaling proteins. (b) sgc has two no-binding sites. The high- affinity site contains a heme group (feii), and is bound even at relatively low no concentrations. binding to this site partially activates sgc. at higher concentrations, no binds to the low-affinity site, leading to full activation of sgc. When no levels return to resting levels, no rapidly dissociates from the low-affinity site and sgc activity decreases. slow, with a half-life of approximately 2 or 3 minutes. For maximal activ- ity, however, a second molecule of NO must also bind to sGC; this can rapidly dissociate if NO concentrations go down, leaving a partially active “sensitized” form. It is thought that basal levels of NO in the vasculature are sufficient to maintain this sensitized form, setting the tonic level of vascular constriction (the Kd for binding of NO to the heme of sGC is in the picomolar range, while basal NO levels are in the nanomolar range). Acute signaling (the release of a bolus of NO from adjoining endothelial cells) leads to binding of NO to the second, lower-affinity site, leading to maximal sGC activity which returns to the basal level rapidly as tissue NO levels diminish (Figure 8.23b). Thus, this receptor system is set up to maintain relatively stable basal levels of activity, and to rapidly respond to local changes in ligand concentration. Carbon monoxide (CO), which has rather similar chemical properties to NO, can also bind to and activate sGC, though the affinity is much lower and the maximal activation of sGC is much less (two- to fourfold, as opposed to many hundredfold for NO). The physiological significance of CO signaling is not yet fully established, but CO levels can rise in tis- sues in response to stress. This has the potential to modulate sGC activity, either by moderately activating it or by acting as a partial antagonist of NO signaling. O ₂ binding regulates the response to hypoxia Molecular oxygen (O₂) can act as a signaling molecule in addition to its fundamentally important role in cellular respiration. Decreases in tis- sue O₂ levels (hypoxia) induce short-term physiological responses, such as the shutdown of inessential activities that consume ATP and an increased rate of anaerobic glycolysis. In the longer term, hypoxia induces transcrip- tional changes leading to tissue-level responses such as increased angio- genesis (the generation of new blood vessels). Transcriptional responses to hypoxia are mediated by a transcription factor, HIF-1α (hypoxia induc- ible factor), which under normal O₂ levels is rapidly degraded. Binding of O₂ activates heme-containing proline hydroxylase domain (PHD) proteins, which hydroxylate two prolines in HIF-1α. This modification allows bind- ing of VHL, a ubiquitin E3 ligase, leading to HIF-1α ubiquitylation and tar- geting to the proteasome. By contrast, under hypoxic conditions, the PHD proteins are inactivated, and HIF-1α begins to accumulate in the nucleus to promote transcription of hypoxia-dependent genes (Figure 8.24). Figure 8.24 Signaling by oxygen. (a) under normal tissue oxygen levels, molecular oxygen (o2) binds to the heme group (feii) of proline hydroxylase (PHd) and activates it. PHd hydroxylates two prolines in the transcription factor Hif-1α (hypoxia inducible factor), leading to its recognition by the vHl ubiquitin ligase. vHl adds long chains of ubiquitin (u) to Hif-1α, targeting it for degradation by the proteasome. in hypoxic conditions, tissue oxygen concentrations are too low to activate PHd. Hif-1α is not recognized by vHl, and is free to bind dna and regulate transcription of target genes. (a) O2 (b) high O2 PHD active low O2 HIF-1α hydroxylated, binds VHL HIF-1α degraded by proteasome PHD inactive HIF-1α stable, binds DNA transcription of target genes VHL takes its name from a genetic cancer predisposition, Von Hippel- Lindau syndrome. Afflicted individuals carry only one functional copy of VHL; the second has an inactivating mutation. Whenever a cell loses the functional allele through somatic mutation, HIF-1α is stable and consti- tutively active in that cell and its descendants. This leads to increased angiogenesis and altered metabolic activity, both of which promote tumor cell growth. The receptors for steroid hormones are transcription factors The other class of signaling molecules that can easily pass through the plasma membrane is comprised of hydrophobic hormones and related com- pounds, which include the steroid hormones (such as estrogen, progester- one, testosterone, and hydrocortisone), vitamins A and D, thyroid hormone, and retinoic acid. The receptors for these compounds are transcription fac- tors of the nuclear receptor (NR) superfamily. In humans, there are ~48 members of the NR family. The endocrine NRs bind to their specific ligands with high affinity, generally in the nanomolar range, consistent with the low circulating levels of their hormone and vitamin ligands. A number of other NRs bind to specific classes of lipids and lipid metabolites with rela- tively lower affinity, and are thought to function primarily in cellular lipid homeostasis. Finally, a number of NRs are considered orphan receptors, as their physiological ligands have not yet been identified. The NRs all bind to specific DNA sites as homo- or heterodimers via an N-terminal DNA-binding domain (Figure 8.25a). The ligand-binding domain is situated in the C-terminus of the molecule. The glucocorticoid receptor (GR) serves as a paradigm for NR action. When not bound by its ligand, GR is sequestered in the cytosol in an inactive complex with chaperone proteins such as hsp90, which presumably function to pre- vent the aggregation and/or degradation of the relatively unstable (and thus partially unfolded) unliganded receptor. Upon ligand binding, the receptor undergoes a conformational change, chaperone proteins and other co-repressors are released, and the receptor is now free to dimerize, relocate to the nucleus, and bind palindromic NR-binding sites on chro- matin (Figure 8.25b). The DNA-bound form can activate transcription by binding to a host of coactivator proteins, such as chromatin remodeling complexes including histone acetyl transferase (HAT). The net effect is to increase transcription from a battery of hormone-responsive genes. Thus, these receptors represent the most simple and straightforward mecha- nism of transcriptional regulation by an outside signal: direct activation of a transcription factor by its ligand. The simple model described above does not capture all of the variety and subtlety of regulation by different members of the NR family. For exam- ple, some of these receptors are constitutively bound to specific sites on the chromatin even in the absence of ligand. In these cases, the unlig- anded receptor acts as a transcriptional repressor by interacting with co- repressor proteins such as histone deacetylase complex (HDAC). When ligand binds, the conformational changes in the ligand-binding domain lead to release of the co-repressors and binding of a new set of coactivator proteins, such as HAT, which remodel the chromatin and convert the erst- while repressor into a transcriptional activator (Figure 8.25c). Note that The role of protein acetylation in transcriptional regulation is described in Chapter 4 nuclear receptor dimerization activation coactivators DNA binding ligand binding Figure 8.25 Signaling by nuclear receptors. (a) all nuclear receptor (nr) family members consist of an n-terminal dna-binding region ( o r a n g e ), and a c-terminal ligand-binding region ( p u r p l e ). (b) When their ligand is not present, nrs such as glucocorticoid hormone receptors reside in the cytosol in complex with hsp90 and other proteins. ligand binding leads to dissociation of hsp90, conformational changes, and dimerization. The receptor dimer is imported into the nucleus, binds dna, and associates with transcriptional coactivators. some nrs are constitutively bound to dna. in the absence of ligand, they are associated with transcriptional co-repressors such as histone deacetylase (Hdac). ligand binding induces conformational changes that lead to dissociation of co-repressors and binding of transcriptional coactivators such as histone acetyl transferase (HaT). while the specific effects of ligand binding are quite different from those described above for GR, the immediate consequence of ligand binding is the same; that is, to induce conformational changes in the ligand-binding domain that alter the binding partners and thus the biological activity of the NR. doWn-regulaTion of recePTor signaling While receptor activation is the basis for signal transmission through the membrane, the ability to turn receptor signaling off, or to adjust its sensitivity, is also critical for cells to respond to their environment. As we discussed in Chapter 2, very high-affinity complexes necessarily have long half-lives (on the order of several minutes or longer), and transmem- brane receptors tend to have very high affinity for their ligands, which are present at relatively low levels in the environment. Thus, the inactivation of a ligand-bound receptor by dissociation of the ligand is a relatively slow process. To circumvent this constraint, cells have adopted a number of other mechanisms to terminate the signal, or to allow the signal output to adapt to different levels of ligand. Here, we provide a few examples of specific mechanisms used to down-regulate receptor signaling. It should be emphasized that this is not meant to be a comprehensive list; indeed, it is likely that there are nearly as many ways to down-regulate receptor- mediated signals as there are ways to initiate the signal itself. In general, there are two ways in which the signals emanating from recep- tors can be modulated (Figure 8.26). The first and most straightforward is removing the receptor itself from the cell surface, often termed recep- tor down-regulation. Not only does this reduce the number of receptors that can respond to the signal over time, but activated receptors already bound to ligand can be targeted for degradation once they are internal- ized. Receptor internalization is a relatively slow process, operating on time scales of a few minutes. The second general mechanism is desen- sitization, or attenuation, in which the ligand-bound receptor becomes less active in transmitting downstream signals. Desensitization can be active ligand- bound receptor Figure 8.26 The fate of activated receptors. active, liganded receptor (center) is subject to down-regulation and desensitization. down-regulation involves internalization of the receptor–ligand complex, which can then be recycled to the cell surface or degraded in the lysosome. desensitization involves blocking downstream signaling from the liganded receptor. quite rapid, on the order of milliseconds. This is the case for ligand-gated ion channels, where relatively brief stimulation can cause the channel to reversibly adopt a closed (desensitized) conformation refractory to further stimulation. Many other receptors undergo some form of desensitiza- tion, such as GPCRs as discussed below. Desensitization can allow cells to adjust the baseline level of signal output to the current level of input signal, so they can respond to a much wider range of signal strength than would be possible otherwise. Ubiquitylation regulates the endocytosis, recycling, and degradation of cell-surface receptors In the absence of ligand, most cell-surface receptors undergo relatively slow cycles of endocytosis, sorting, and recycling back to the plasma membrane. Once receptors are activated by ligand binding, however, the dynamics of this process usually change dramatically: the rate of endocy- tosis increases, and many of the internalized receptor–ligand complexes are targeted for degradation in the lysosome. The net effect is to rapidly (within a few minutes) clear the activated receptor from the cell surface and extinguish its ability to transmit downstream signals. While this process is common to all types of cell-surface receptors, we will consider the specific example of the epidermal growth factor receptor (EGFR), a receptor tyrosine kinase that has been particularly well studied. Internalization of the activated EGFR can occur by a variety of specific mechanisms (see Figure 5.15). Of these, clathrin-mediated endocytosis is the most rapid, and is likely to be dominant under physiological conditions of relatively low receptor number and ligand concentration. A number of studies have shown that ligand-stimulated endocytosis through the clath- rin-mediated pathway requires receptor tyrosine kinase activity and, in particular, receptor autophosphorylation. For down-regulation, the key consequence of receptor autophosphorylation is the recruitment of Cbl, a ubiquitin E3 ligase. Cbl contains an SH2 domain that can bind direct- ly to phosphorylated sites on the receptor. In addition, Cbl also binds to the adaptor Grb2, which itself has an SH2 domain that binds to differ- ent phosphorylated sites on the receptor (Figure 8.27). In this way, Cbl recruitment is tightly coupled to receptor activation. The association of Cbl with the activated receptor leads to receptor ubiquitylation (either monoubiquitylation or Lys63-linked polyubiq- uitylation) and to rapid internalization via clathrin-coated vesicles. In other cases, ubiquitin is attached to receptor-associated adaptor proteins rather than the receptor itself, but in all cases the ubiquitylated recep- tor complex is rapidly recognized by the clathrin-dependent endocytic machinery, incorporated into clathrin-coated vesicles, and internal- ized. Once removed from the cell surface, the receptor–ligand complex (which still retains signaling activity) is transported to an organelle called the multivesicular body, where the decision is made whether to send it to the lysosome for degradation or recycle it back to the sur- face (Figure 8.28). This sorting process is accomplished by the so-called ESCRT complexes, which consist of a number of proteins containing ubiquitin-recognition motifs as well as lipid- and membrane-binding domains. Those proteins targeted for degradation (for example, those that are Lys63-polyubiquitylated) then pinch off from the external limit- ing membrane of the multivesicular body to form internalized vesicles. Because these vesicles are now topologically isolated from the cytosol, any receptor–ligand complexes found there can no longer signal. Ultimately, the multivesicular body fuses with a lysosome and the internalized vesicles, lipid and protein alike, are degraded. Ubiquitin and the enzymes that add and remove it from proteins are discussed in detail in Chapter 4 Figure 8.27 Recruitment of Cbl to the activated EGF receptor. upon ligand binding, the epidermal growth factor (egf) receptor is activated and autophosphorylates on a number of sites. some sites bind the sH2 domain of cbl directly; others bind the sH2 domain of the grb2 adaptor which, in turn, binds to the proline-rich tail of cbl (blue) via its sH3 domains. cbl has ubiquitin e3 ligase activity and, when recruited to the receptor, ubiquitylates the receptor and receptor- associated proteins, leading to engagement of the endocytic machinery. Figure 8.28 Sorting of receptor–ligand complexes in the multivesicular body. upon endocytosis, activated receptor–ligand complexes are transported to the multivesicular body (mvb). on the outer membrane of the mvb, escrT complexes, which contain ubiquitin- and membrane-binding domains, sort receptors for degradation or recycling. Those to be degraded are transported to vesicles inside the mvb, which are physically isolated from the cytosol and receptors can no longer signal. receptors to be recycled are transported by vesicles back to the cell surface. ultimately, the mvb fuses with a lysosome and its contents are degraded. While the specific details might vary for different receptors, this overall scheme of down-regulation is likely to be common for the vast majority of ligand-activated cell-surface receptors. Novel properties induced by activation, such as conformational changes or phosphorylation, lead to the recruitment of ubiquitin ligases, which mark the receptor for inter- nalization and target it for recycling or degradation. The wide variety of ubiquitin ligases and adaptor proteins that can be involved, as well as variations in the number of ubiquitin units conjugated and the linkage of polyubiquitin chains, provide ample scope for regulation of this proc- ess to adapt to the specific requirements of the situation. The key role of this pathway in modulating signal output is highlighted by the fact that a mutant version of Cbl was originally isolated as a viral oncogene. The oncogenic form of Cbl lacks ubiquitin ligase activity but still binds to acti- vated receptors, thereby preventing the binding of normal endogenous Cbl. Thus, activated receptors such as the EGFR cannot be efficiently down-regulated, leading to unrestrained signaling and inappropriate cell proliferation. G protein coupled receptors are desensitized by phosphorylation and adaptor binding G protein coupled receptors can be activated very rapidly and their acti- vation can result in tremendous signal amplification, so for this class of receptors it is particularly important to be able to control signal out- put by desensitizing activated receptors. Desensitization of GPCRs is accomplished by two families of proteins that specifically associate with the activated form of the receptor: the GPCR kinases (GRKs) and the arrestins. GRKs associate with activated GPCRs and phosphorylate them, generally on the C-terminus, to generate high-affinity binding sites for arrestins. Once arrestin is bound to the phosphorylated receptor, the binding of G proteins to the receptor is blocked, preventing activation of additional G proteins. Furthermore, the arrestin can mediate association with the endocytic machinery and ubiquitin-conjugating enzymes, thus promoting endocytosis and, in some cases, degradation of the liganded receptor (Figure 8.29). There are seven GRKs in humans, differing in their tissue distribution and specific domain structures. All have an N-terminal RGS homology (RH) domain that can interact with specific G protein α subunits, and a central serine/threonine kinase catalytic domain. The C-terminal region mediates association with membranes, either via a lipid-binding PH domain or sites for covalent lipid attachment, and, in some cases, can also interact specifically with G protein βγ subunits. The structure of GRK2, which was discovered by virtue of its ability to phosphorylate the acti- vated β-adrenergic receptor and thus was originally named β-adrenergic Figure 8.29 GPCR desensitization by GRK and arrestin. conformational changes induced by ligand binding to gPcrs recruit gPcr kinases (grKs), which phosphorylate the c-terminal tail of the receptor. The phosphorylated receptor then recruits and binds to arrestin, which prevents the receptor from activating g proteins, targets it for endocytosis, and also serves as a scaffold for assembly of g-protein- independent signaling complexes. Figure 8.30 GRKs bind to multiple products of GPCR activation. The kinase catalytic domain of the gPcr kinase (grK) binds to the activated g protein coupled receptor (gPcr; g r ee n) and phosphorylates the c-terminal tail. in addition, the n-terminal rH domain binds to activated (gTP-bound) g protein α subunits (purple), while the c-terminal region (c-ter) contains a PH domain that interacts with membrane lipids and a region that interacts with g protein βγ subunits (brown). RH kinase C-ter GRK receptor kinase (β-ARK), has been solved. The structure suggests that these three domains simultaneously bind to the activated receptor and its two products: the kinase domain binds the receptor itself, while the RH domain binds the GTP-bound Gα subunit, and the C-terminus binds the free Gβγ subunits (Figure 8.30). Each of these binding partners is only present upon receptor activation, making recruitment of the kinase to the membrane in the vicinity of the receptor cooperative and highly depend- ent on activation. The result of GRK recruitment is phosphorylation of the receptor (which leads to binding of arrestin), and also the sequestration of the receptor products. The net effect is to strongly block signal output from the receptor. For receptor down-regulation, the key property of the arrestins is their ability to bind specifically to the phosphorylated form of activated GPCRs. The immediate consequence of arrestin binding is to block the receptor from interacting with and thus activating G proteins. Binding to the receptor is also thought to induce major concerted conformational changes in arrestin which, in turn, unmask binding sites for a variety of other proteins that promote endocytosis and ubiquitylation of the arrestin–receptor complex, as well as promoting downstream signaling (Figure 8.31). Receptor-bound arrestin associates with clathrin and the AP-2 clathrin adaptor, and thus very efficiently promotes internalization of the complex via clathrin-coated vesicles. The fate of the internalized receptor then depends on specific details of the receptor and arrestin involved, but generally falls into two categories. In one category, arres- tin rapidly dissociates from the receptor after internalization, allowing receptor dephosphorylation and recycling back to the cell surface (resen- sitization). In the second instance, arrestin remains stably associated with the receptor, the arrestin and receptor are ubiquitylated by associ- ated ubiquitin ligases, and the complex is targeted for degradation in the lysosome. An interesting property of arrestins is their ability to initiate signal- ing on their own once they have bound to activated GPCRs. Activated arrestins can associate with and activate nonreceptor tyrosine kinases of the Src family, and can serve as scaffolds for activation of mitogen- activated protein (MAP) kinase cascades and other signaling mediators. For example, binding of arrestin to the Smo GPCR family member is thought to play an important role in Hh signaling, as described earlier in this chapter. These arrestin-dependent signals can, in some cases, cooper- ate with signals that are directly dependent on the G protein activated by the receptor, and can oppose the G-protein-mediated signaling in other cases. Thus, the GRK–arrestin system is not only a very efficient means to regulate the activity and dynamics of GPCR signaling, but also plays a larger role in sculpting the precise downstream signaling output from the activated receptor. summary summary Transmitting extracellular signals through the plasma membrane to the cell interior is one of the most fundamental challenges of cell signaling. Cells use a very limited number of mechanisms to accomplish this task. By far the most common mechanism involves transmembrane receptors, where ligand binding to the extracellular portion leads to altered enzy- matic activity of the receptor or its associated proteins in the cytosol. The other mechanisms involve ligand-gated ion channels, which open or close in response to stimuli, and in a few cases the use of signaling molecules that can passively cross the membrane and exert their effects directly in the cytosol. For transmembrane receptors, information about ligand bind- ing is conveyed to the cytosolic portion of the receptor either by concerted conformational changes in the case of receptors with multiple membrane- spanning segments, or dimerization or oligomerization in the case of receptors with a single membrane-spanning segment. Protein phospho- rylation and proteolysis are the most common activities that are coupled to receptor binding. Cells have also evolved a number of mechanisms to down-regulate the activity of receptors, providing an additional level of control over the extent and dynamics of signaling output. QuesTions What general classes of extracellular signals do cells need to detect? What are the three main ways for signals to cross the cell membrane? You are studying a transmembrane receptor with an intracellu- lar kinase domain, and observe by microscopy that it clusters into Figure 8.31 Model of arrestin–receptor interaction. Putative complex assembled from crystal structures of the active β₂- adrenergic receptor (from the complex with the gs heterotrimer) and arrestin-2. The receptor is depicted in blue and orange, and arrestin in yellow. Phosphate-binding residues of arrestin are indicated on the strands in pink, and other elements that experimental evidence suggests make direct contact with the receptor are indicated in green or purple. Phosphates from the receptor c-terminus (pink spheres) were added manually to position them near known phosphate-binding positive charges in arrestin. note that additional conformational changes in the receptor are likely in the actual receptor–arrestin complex. (courtesy of v.v. gurevich.) membrane microdomains upon stimulation by its ligand. Discuss the potential mechanisms by which clustering of such a receptor might sig- nificantly enhance phosphorylation and activation of its kinase domain. The human genome encodes hundreds of G protein coupled receptors (GPCRs), but individual cell usually express only a handful of GPCRs. What aspects of GPCR signaling may limit the number of GPCRs that can be simultaneously used in a cell? There are dozens of receptor tyrosine kinase (RTK) receptors and lig- ands in humans. By contrast, there is only a handful of ligands and receptors for Wnt and Hedgehog signaling. Why might this be the case? Many drugs act as antagonists of transmembrane receptors. If a small-molecule G protein coupled receptor (GPCR) antagonist inhib- its its target by 50% at a concentration of 2 × 10–8 M, what concentra- tion would be needed in the bloodstream to inhibit receptor activity by 99%? If the compound has a molecular mass of 500 daltons, how much of the drug would be needed to achieve this concentration in the blood- stream (assume that all of the drug enters the bloodstream, which has a volume of 5 liters)? Receptors that have intrinsic or associated kinase activity depend on dimerization and/or clustering in order to transmit downstream sig- nals. A key step in activating the associated kinase activity is usually input-dependent phosphorylation on the kinase activation loop. How- ever, there are a few receptor-associated kinases that do not require activation-loop phosphorylation for signaling. How might it be pos- sible to transmit signals without activation-loop phosphorylation? Similarly, consider the role of cellular phosphatase activity. In what ways would a much higher or lower phosphatase activity affect the signaling properties of the receptor? In most cells, the amounts of Gα and Gβγ subunits are closely matched. What effect might you expect if you were able to highly overexpress G protein β and γ subunits in the cell? Propose a mechanism whereby the cell normally keeps the amounts of different subunits in balance. What properties of nuclear receptors make them particularly well suited for regulating cellular lipid metabolism? In what way is this regulatory role similar and different from their role as receptors for extracellular signals? Signaling by G protein coupled receptors (GPCRs) is generally very rapid (half-time for changes in downstream signaling mediators can be milliseconds) while signaling by tyrosine kinases is generally quite a bit slower (half-time for recruitment of SH2 effector proteins of sev- eral minutes). What molecular mechanisms might explain this differ- ence in response dynamics? The erythropoietin (Epo) receptor of a famous long-distance runner is sequenced and it is found to have a mutation at a Ser residue known to be phosphorylated upon stimulation with Epo. The Ser residue is converted to Ala in the mutant allele. How might this mutation be linked to this person’s athletic endurance? references references TransducTion sTraTegies used by Transmembrane recePTors Cooper JA & Qian H (2008) A mechanism for SRC kinase-dependent signaling by noncatalytic receptors. Biochemistry 47, 5681–5688. Oh D, Ogiue-Ikeda M, Jadwin JA et al. (2012) Fast rebinding increases dwell time of Src homology 2 (SH2)- containing proteins near the plasma membrane. Proc. Natl Acad. Sci. U.S.A. 109:14024–14029. g ProTein couPled recePTors Johnston CA & Siderovski DP (2007) Receptor-mediated activation of heterotrimeric G-proteins: current struc- tural insights. Mol. Pharmacol. 72, 219–230. Lohse MJ, Hein P, Hoffmann C et al. (2008) Kinetics of G-protein-coupled receptor signals in intact cells. Br. J. Pharmacol. 153(Suppl 1), S125–S132. Oldham WM & Hamm HE (2008) Heterotrimeric G pro- tein activation by G-protein-coupled receptors. Nat. Rev. Mol. Cell Biol. 9, 60–71. Rasmussen SG, DeVree BT, Zou Y et al. (2011) Crystal structure of the β2 adrenergic receptor-Gs protein com- plex. Nature 477, 549–555. Wettschureck N & Offermanns S (2005) Mammalian G proteins and their cell type specific functions. Physiol. Rev. 85, 1159–1204. Transmembrane recePTors associaTed WiTH enzymaTic acTiviTy Clevers H (2006) Wnt/beta-catenin signaling in develop- ment and disease. Cell 127, 469–480. Le Borgne R (2006) Regulation of Notch signalling by endocytosis and endosomal sorting. Curr. Opin. Cell Biol. 18, 213–222. Schlessinger J (2000) Cell signaling by receptor tyrosine kinases. Cell 103, 211–225. Schmierer B & Hill CS (2007) TGFβ-SMAD signal trans- duction: molecular specificity and functional flexibility. Nat. Rev. Mol. Cell Biol. 8, 970–982. Shi Y & Massagú J (2003) Mechanisms of TGF-β signal- ing from cell membrane to the nucleus. Cell 113, 685–700. Thorburn A (2004) Death receptor-induced cell killing. Cell. Signal. 16, 139–144. Tonks NK (2006) Protein tyrosine phosphatases: from genes, to function, to disease. Nat. Rev. Mol. Cell Biol. 7, 833–846. Wilson CW & Chuang PT (2010) Mechanism and evolu- tion of cytosolic Hedgehog signal transduction. Develop- ment 137, 2079–2094. gaTed cHannels Gouaux E & MacKinnon R (2005) Principles of selec- tive ion transport in channels and pumps. Science 310, 1461–1465. Hilf RJC & Dutzler R (2009) Structure of a potentially open state of a proton-activated pentameric ligand-gated ion channel. Nature 457, 115–118. Long SB, Tao X, Campbell EB & MacKinnon R (2007) Atomic structure of a voltage-dependent K+ channel in a lipid membrane-like environment. Nature 450, 376–382. membrane-Permeable signaling Cary SP, Winger JA, Derbyshire ER & Marletta MA (2006) Nitric oxide signaling: no longer simply on or off. Trends Biochem. Sci. 31, 231–239. Jin L & Li Y (2010) Structural and functional insights into nuclear receptor signaling. Adv. Drug Deliv. Rev. 62, 1218–1226. Liu L & Simon MC (2004) Regulation of transcription and translation by hypoxia. Cancer Biol. Ther. 3, 492–497. Sonoda J, Pei L & Evans RM (2008) Nuclear receptors: decoding metabolic disease. FEBS Lett. 582, 2–9. doWn-regulaTion of recePTor signaling Premont RT & Gainetdinov RR (2007) Physiological roles of G protein-coupled receptor kinases and arrestins. Annu. Rev. Physiol. 69, 511–534. Shenoy SK & Lefkowitz RJ (2011) β-Arrestin-mediated receptor trafficking and signal transduction. Trends Pharmacol. Sci. 32, 521–533. Sorkin A & Goh LK (2008) Endocytosis and intracellular trafficking of ErbBs. Exp. Cell Res. 314, 3093–3106. Taylor & Francis Taylor & Francis Group https://taylorandfrancis.com regulated Protein degradation As we have seen in Chapter 4, the post-translational modification of proteins is one of the most common ways of transmitting signals in the cell. The most drastic form of post-translational modification is prote- olysis, the breaking of the peptide bonds that form the backbone of pro- teins. Proteolysis can be used to break down or degrade proteins for the purpose of eliminating their activity, or it can serve as an essential step in generating active enzymes or signaling proteins from longer, inactive precursors. Regulated proteolysis is central to two of the most fundamen- tal activities of the cell: control of the cell cycle and the decision to undergo programmed cell death (apoptosis). This chapter will explore some of the special properties of signaling pathways that involve proteolysis, and dis- cuss in greater detail the ubiquitin–proteasome pathway of protein deg- radation and the role of proteases in apoptosis. General ProPerties and examPles of siGnal-reGulated Proteolysis The most fundamental way in which proteolysis differs from other post- translational modifications is that it is essentially irreversible—the only way to regenerate the intact, uncleaved substrate is through the relatively slow and energetically costly process of translating a new polypeptide chain. This contrasts with other post-translational modi- fications, for example phosphorylation or histone acetylation, where the action of the writers (kinases and histone acetyl transferases) is opposed by erasers (phosphatases and histone deacetylases). Most other types of signaling inputs, such as the opening of membrane chan- nels to allow ions to flood in or out, are also rapidly reversible (in the case of channels, by channel closing and the action of ion pumps). The unique irreversibility of proteolysis is useful for signaling situations Figure 9.1 Signaling by proteases is irreversible. (a) the consequences of the activity of proteases (either the activation of enzymes from inactive zymogens, or the proteolytic destruction of proteins) are irreversible, and the system can be reset only by synthesizing new proteins. (b) Proteases can act like a ratchet- and-pawl mechanism, preventing a signaling pathway from easily reverting to its previous state. (a) (activation) (destruction) (b) where it is not desirable for the system to revert easily to the starting state (Figure 9.1). Two important examples of processes characterized by an irreversible commitment step are the cell cycle and apoptosis, both of which are dis- cussed in more detail later in this chapter. In the case of the cell cycle, it is essential that the process proceed in an orderly fashion from one step to another—DNA synthesis must occur once before mitosis, for example, and mitosis must occur once before a new round of DNA synthesis. If the cycle were to run “backward” (for example, if mitosis occurred before the genomic DNA could be replicated), the consequences for the cell and organism could be disastrous. Similarly, once the process of programmed cell death begins, it must go to completion to avoid damage to surrounding cells and tissues. Proteases are a diverse group of enzymes Proteases, the enzymes that hydrolyze the peptide bonds of proteins, are an evolutionarily ancient and highly diverse group. The peptide bond is inher- ently quite stable, so a variety of proteases have evolved to accomplish house- keeping functions such as breaking down the proteins in food (to generate energy and provide raw materials for synthesis of new biomolecules) or elim- inating cellular proteins that are damaged or no longer needed. Proteases can be classified on the basis of their reaction mechanism into the metallo- proteases, serine proteases, cysteine proteases, and aspartyl proteases. Most proteases exhibit pronounced preferences for the amino acid residues that flank the bond to be cleaved. Substrate specificity can be determined not only by the catalytic site itself but also, in many cases, by additional interactions with substrates outside of the catalytic site. This specificity is particularly important for those proteases that have been harnessed for signal transduction. Most cellular proteases are synthesized in an inactive precursor form, termed a zymogen. This is quite sensible, as it would be potentially dan- gerous to have fully active proteases rampaging about the cell as soon as they emerged from translation. Proteolytic cleavage of the zymogen at specific sites is needed to release the catalytically active enzyme. This pre- vents damage to the synthesizing cell, and permits precise activation at the time and place needed. Such an arrangement also allows the system to greatly amplify an initial signal, particularly in cases where a single activated protease can go on to cleave and thereby activate many more zymogen molecules of the same type. This sort of positive feedback loop is common where a rapid and explosive burst of activity is needed. Thus proteolysis confers distinct dynamic properties to signaling—a proteolyti- cally regulated process can switch from one state to another very rapidly, but it can only be reversed slowly through new protein synthesis. thrombus formation thrombolysis Blood coagulation is regulated by a cascade of proteases The principles of zymogen activation and signal amplification are illus- trated in the blood coagulation or clotting cascade of vertebrates. Clotting is, of course, important to prevent catastrophic loss of blood from dam- aged blood vessels after injury. The amount of clotting must be strictly controlled, however; too much clotting can cause blockage of blood vessels leading to tissue death, as in stroke, and is potentially life-threatening. This is a tricky problem for the organism, and so the clotting cascade has evolved under intense pressure to plug leaks rapidly and efficiently, to self-limit to only the areas where it is needed, and to remodel and ultimately break down over time to allow the repair and replacement of damaged tissue. The molecular machinery that drives this elaborate process consists of a cascade of serine proteases that are specifically activated by damaged tis- sue and by blood platelets that adhere to the site of damage. The ultimate effector of the clotting pathway is fibrin, a protein that self-assembles into a tangled fibrillar network (a thrombus) that physically fills the breach in the damaged tissue (Figure 9.2). Fibrin is generated from a soluble precursor, fibrinogen, which is a major component of serum (present at a concentration of ~3 mg/mL). It is converted by the proteolytic action of thrombin from its native soluble conformation to the conformation that self-assembles into fibrils. Thrombin also cleaves and activates a trans- glutaminase, Factor XIII, that cross-links the fibrin network in the clot. Thrombin is present at relatively high concentrations in the blood in a zymogen form, termed prothrombin, and must itself be proteolytically processed to become active. Thrombin activation must be very carefully controlled, and thus is depend- ent on a number of factors (Figure 9.3). One of these, an accessory factor (Factor V), is itself cleaved by thrombin to generate the active factor (Fac- tor Va), providing one positive feedback loop. The fact that thrombin itself can cleave and activate more molecules of prothrombin provides a sec- ond positive feedback loop. Other essential factors for prothrombin cleav- age are Ca2+ ions and negatively charged membrane lipids. Negatively charged lipids such as phosphatidylserine are almost always present only on the inward (cytosolic) face of cell membranes, so their exposure to the blood signals that ruptured cells and tissue injury are present. The very first molecules of thrombin that are activated to initiate clot formation are generated by another protease, Factor X, which must, in turn, be activated by zymogen cleavage, a process again facilitated by Ca2+ ions, negatively charged phospholipids, and additional accessory fac- tors. The entire multistep cascade, initiated either by negatively charged surfaces (the so-called “intrinsic pathway”) or by exposed tissue proteins (the “extrinsic pathway”), will not be further discussed, but is described in great detail in standard biochemistry texts. Once the process is initiated, Figure 9.2 The physiological role of the clotting cascade. tissue injury causing damage to blood vessels leads to contact between the serum and damaged cells, and to the binding and aggregation of platelets to damaged tissue. this activates the clotting cascade, leading to the deposition of a tangled network of fibrin (a thrombus, or clot; blue) that prevents the escape of blood cells from the damaged vessel. once the injury has been repaired, the clot is disassembled (thrombolysis). Figure 9.3 The clotting cascade. the pivotal event of clotting is the activation of thrombin by cleavage of its zymogen, prothrombin. activation of thrombin can be accomplished either by another molecule of activated thrombin or by a different protease, factor xa. factor x is similarly activated by other upstream proteases. once activated, thrombin cleaves and activates fibrinogen, leading to formation of fibrin filaments. thrombin generates more thrombin directly by cleaving prothrombin, and indirectly by cleaving and thus activating the cofactor factor V. thrombin also activates factor xiii, which cross-links the fibrin network to generate a more stable, hard clot. many steps require Ca2+ and negatively charged lipids such as phosphatidylserine (Ps). upstream factors Factor X Factor Xa Ca2+ PS prothrombin Factor Va Factor V thrombin fbrinogen fbrin cross-linked mesh however, a number of factors contribute to limiting the extent of clot for- mation to the immediate site of injury. One is blood flow, which constantly dilutes the activated proteases and accessory factors at the site of injury; another is the presence of circulating protease inhibitors such as anti- thrombin, which dampen protease activity away from the immediate site of injury. Finally, once tissues have had a chance to repair themselves, the clot must be removed, a process termed thrombolysis. Perhaps not surprisingly, this involves yet another serine protease, termed plasmin, that specifically cleaves fibrin and thereby dissolves the clot. The zymogen form of plas- min is plasminogen, which, in turn, is activated by other proteases such as tissue plasminogen activator (t-PA) and is inhibited by additional fac- tors. Because blood clots are a significant health risk (causing strokes and heart attacks, two of the main causes of death for humans), drugs that inhibit coagulation such as coumarin, which prevents the interaction of prothrombin with membranes, or those that enhance clot lysis (such as t-PA) have saved many lives. Regulated proteolysis by metalloproteases can generate signaling molecules and alter the extracellular environment Proteases play a number of roles on the extracellular surface of the cell, which include degrading extracellular matrix proteins to regulate cell motility, generating bioactive signaling molecules from larger precur- sors, and cleaving receptors as an essential step in their activation or down-regulation. Many of these activities are performed by metallo- proteases such as the matrix metalloproteases (MMPs) and disintegrin metalloproteases (ADAMs, A Disintegrin And Metalloprotease). These proteases all require a metal ion, generally zinc, for catalytic activity. ADAM ADAMTS MT-MMP MMP They are synthesized as inactive precursors and transported through the secretory pathway to the cell surface, where most remain associated via a C-terminal transmembrane segment or phospholipid anchor; some lack a membrane anchor and are released from the cell. The domain structures of the major families of metalloproteases are depicted in Figure 9.4. ADAMs regulate signaling pathways by cleaving membrane-associated proteins ADAMs are also known as “sheddases” for their ability to cleave membrane- associated proteins from the cell surface, a process termed ectodomain shedding. Such an activity can either promote signaling (for example, where a bioactive ligand is released) or inhibit it (such as when the extra- cellular ligand-binding region of a cell-surface receptor is cleaved off). Two particularly well-understood cases involve the processing by ADAM-17 of tumor necrosis factor α (TNFα) and the epidermal growth factor (EGF) family of soluble mitogens. In fact, ADAM-17 was first characterized as TNFα converting enzyme (TACE) for its role in processing the membrane- bound precursor form of TNFα into its soluble, active form, though later studies demonstrated that ADAM-17 is a sheddase with a particularly broad range of substrates. Gene knockout studies have shown that the biological phenotype of mice lacking ADAM-17 closely parallels that of mice with mutations in the EGF receptor (EGFR) family or their ligands. Since ADAM-17 cleaves a variety of EGFR ligands from the membrane, this provides convincing evidence that receptor signaling activity can be regulated at the level of ligand processing. Such a mechanism is prob- ably useful in generating localized signals, where a ligand exerts its influ- ence in the immediate vicinity of its shedding (this is termed juxtacrine signaling). Eph–ephrin signaling illustrates another way in which proteolysis can help sculpt signaling responses at the cell surface. Eph receptors are receptor tyrosine kinases; binding of the membrane-anchored ephrin to its corresponding Eph receptor on another cell generates bidirectional sig- nals impacting both cells. In the developing nervous system, Eph–ephrin signaling generally leads to repulsion of the two cells, an important cue in guiding axons to their proper targets. ADAM-10 specifically recognizes the Eph–ephrin complex, and binding induces a conformational change in the protease that increases its activity toward the bound ephrin, cleav- ing it from the cell surface (Figure 9.5). This allows the two interacting cell membranes to separate, an obvious requirement if two cells are to be repelled from each other. ADAM-10 also plays a major role in signaling by Notch receptors (discussed below). The potent biological activity of ADAMs raises the question of how they might be regulated by other signals. Control can be exerted by the cell at a variety of levels, including transcription, proteolytic activation, trafficking Figure 9.4 Structure of extracellular metalloproteases. a typical adam protease consists of a pro-domain (pro), a metalloprotease catalytic domain (protease), a disintegrin domain that mediates binding to cell-surface integrins, a cysteine-rich element (Cys), and an epidermal growth factor-like motif (eGf), followed by a transmembrane segment and cytoplasmic tail. adamts proteins are distinguished by multiple thrombospondin repeats (t) and lack the eGf motif and transmembrane and intracellular segments. a typical matrix metalloprotease (mmP) contains a pro- domain and catalytic domain followed by a hemopexin-like domain. While most mmPs are secreted, the membrane-type mmPs (mt-mmPs) are anchored to the membrane either by a transmembrane segment (depicted here) or by a glycosylphosphatidylinositol lipid group. ADAM-10 repulsion EphA3 ephrin A5 Figure 9.5 Role of ADAM proteases in Eph/ ephrin-mediated repulsive interactions. the epha3 receptor ( b l u e ) interacts with adam-10 (brown) on the surface of one cell. upon binding of epha3 to ephrin a5 (orange) on the surface of a second cell, adam-10 is activated and cleaves the associated ephrin. this allows the surfaces of the two cells to separate from each other. Binding of ephrin to eph also activates the intracellular tyrosine kinase domain of the eph receptor. to the cell surface, and internalization. Most intriguing, however, are cases where activation of other signaling pathways (by protein kinase C ago- nists or Ca2+ ionophores, for example) modulates ADAM activity. In many cases, regulation is likely to involve the intracellular domains of ADAMs, which typically contain potential phosphorylation sites and proline-rich regions that can bind proteins containing SH3 domains. In one exam- ple, G-protein-coupled receptors (GPCRs) have been shown to stimulate EGFR signaling in an ADAM-dependent fashion. This cross-talk involves activation of ADAM proteases by the GPCR, most likely via a pathway involving phosphorylation of their cytoplasmic domains. ADAM-mediated cleavage of EGFR ligands then leads to activation of their receptors on the same or nearby cells. MMPs participate in remodeling the extracellular environment Another important class of metalloproteases is the matrix metalloproteases (MMPs). As their name suggests, MMPs target extracellular matrix pro- teins such as collagen and fibronectin, in addition to a number of other substrates. They participate in a variety of normal events that involve remodeling of the extracellular environment, such as wound healing, ovulation, and angiogenesis. For example, the first MMP to be isolated degrades connective tissue in the tail of the tadpole during its meta- morphosis into a frog. MMPs also play a role in a number of diseases such as cancer, where they are important for the ability of tumors to invade adjoining tissues and to spread to distant sites (metastasis), and also in arthritis, where they contribute to the degeneration of connective tissue in the joints. MMPs are also a key component of podosomes, actin- rich cell-surface protrusions that mediate adhesion and invasion. These structures are found both in normal cells that invade tissues, such as macrophages and osteoclasts, and in cancer cells, where they are termed invadopodia (Figure 9.6). MMPs secreted by podosomes and invadopodia are thought to allow cells to eat their way through dense protein networks that would otherwise be impassible. Regulation of MMPs can occur at a number of levels, though in most cells their production is controlled transcriptionally and occurs only under spe- cific conditions. Like other metalloproteases, they are produced as inactive zymogens and must be proteolytically processed, either by active MMPs or by other proteases. Again, as in the thrombin cascade, complex regula- tory interrelationships allow both fine control of local activation and the potential for amplification of the signal. MMPs (and other metallopro- teases) are also regulated by specific factors, including the tissue inhibi- tors of metalloproteases (TIMPs). These proteins act as potent inhibitors cofilin cortactin gelatin merge Figure 9.6 Invadopodia are sites of matrix degradation. mda-mB-231 breast cancer cells expressing fluorescently tagged wild-type cortactin (red) were plated on alexa-405 gelatin matrix (cyan) for 4 hours and fixed. the cells were immunostained for cofilin (green). Cortactin and cofilin are two proteins that regulate the organization of the actin cytoskeleton. mature invadopodia are characterized by cortactin puncta that co-localize with cofilin and degrade the underlying gelatin matrix (shown by a lack of gelatin fluorescence). invadopodia seen on tumor cells are structurally and functionally similar to podosomes seen on normal invasive cells. (images courtesy of m. magalhaes and J. Condeelis.) by forming very high-affinity 1:1 complexes with their targets. Each TIMP differs in the spectrum of MMPs, ADAMs, and other metalloproteases that it can inhibit, contributing additional specificity that can be used to control the localized activation of metalloproteases. Proteolysis activates the thrombin receptor Extracellular proteases, in some cases, directly participate in receptor activation. The GPCRs activated by thrombin and other proteases pro- vide one example. These receptors, termed protease-activated receptors (PARs), play an important role in the cellular responses to vascular injury. They provide a direct mechanism for the proteases activated in the clot- ting cascade, such as thrombin, to signal directly to surrounding cells and to platelets. During normal clotting, for example, platelets respond to thrombin by becoming activated, a process that involves shape change, increased adhesiveness, and release of a variety of inflammatory media- tors, mitogens, and procoagulant factors. The PARs have the same seven-transmembrane topology typical of all GPCRs, but differ in that their N-termini contain a peptide sequence that can bind to and potently activate the same receptor in cis. In the unstim- ulated state, however, this autoactivating ligand is held in a conforma- tion that prevents interaction with its binding site (Figure 9.7). Protease cleavage releases the conformational constraints on the autoactivating ligand, allowing it to bind to the ligand-binding site and activate the receptor. As in all GPCRs, this then leads to activation of heterotrimeric G proteins and downstream signaling. Because the PARs are irreversibly activated by proteolysis (unlike typical GPCRs), their down-regulation and recycling take on particular importance. Before activation, PARs undergo continuous cycles of internalization and re-insertion into the GPCR signaling is discussed in Chapter 8 plasma membrane but, once activated, they are rapidly targeted for lyso- somal degradation upon internalization. This mechanism is apparently distinct from the typical GPCR down-regulation mechanisms that involve arrestin binding and monoubiquitylation. (b) platelets thrombin release of growth factors, clotting factors, inflammatory mediators Regulated intramembrane proteolysis (RIP) is an essential step in signaling by some receptors Another way in which proteases directly participate in receptor activation is illustrated by the Notch signaling pathway. As outlined in Figure 8.17, activation of Notch by its ligands involves multiple proteolytic steps: an initial “S2” cleavage liberates the extracellular domain of the receptor, allowing a second protease, the presenilin/γ-secretase complex, to cleave the receptor in the plane of the membrane, liberating the Notch intra- cellular domain (NICD) which thereupon translocates to the nucleus to regulate transcription. The S2 cleavage is performed by ADAM proteases, often ADAM-10. Indeed, in mice, ADAM-10 knockouts are phenocopies of Notch pathway mutations, strongly suggesting that ADAM-10 is prima- rily responsible for Notch activation in vivo. The process of ADAM-mediated ectodomain cleavage followed by fur- ther processing within the membrane by the γ-secretase complex is also used for a number of other proteins, and has been termed regulated Figure 9.7 Activation of protease-activated receptor (PAR) by thrombin. (a) Par has the seven-transmembrane topology of a typical G-protein-coupled receptor (GPCr). the activating ligand peptide (purple) cannot interact with its binding site until the receptor is cleaved by thrombin. after cleavage, ligand peptide binds the receptor intramolecularly and induces the active conformation. (b) signaling by the thrombin-activated Par on platelet plasma membranes leads to platelet activation. activated platelets adhere to cells and to each other, and secrete a wide variety of factors that regulate clotting, inflammation, and tissue repair. intramembrane proteolysis (RIP) (Figure 9.8). Other proteins sub- ject to RIP include Notch ligands such as Delta, the amyloid precursor protein (APP), and the EGFR family member ErbB4. Cleavage of APP can generate amyloid β, a hydrophobic extracellular peptide that aggre- gates into amyloid fibrils, which are the major component of the neurofi- brillary tangles and senile plaques found in the brains of Alzheimer’s disease patients. Normally, however, APP is likely to serve as an adhesion receptor and signal to the cytosol, both in its intact form and after RIP- mediated cleavage. Another interesting example of RIP-mediated signaling is ErbB4/HER4, one of the four members of the EGFR family of receptor tyrosine kinases. Binding of ErbB4 to its ligand, neuregulin-1, induces RIP via ADAM-17 and the γ-secretase complex. The products include the shed ectodomain, which may act as a “decoy” to bind ligand and sequester it from activating additional ErbB4 molecules, and the intracellular domain (ICD), which retains tyrosine kinase activity and has been reported to phosphorylate substrates in the nucleus such as Mdm2, a regulator of the tumor sup- pressor p53. The ErbB4 ICD has been shown to be essential for some biological activities, such as the formation of astrocytes in the developing mouse brain. It apparently performs this function by binding to other pro- teins, such as transcriptional coactivators or co-repressors, and helping escort them to the nucleus where they exert their effects. Neuregulin-1, the ligand that activates ErbB4, is also processed by the RIP pathway and generates an ICD that may have its own signaling outputs. Figure 9.8 Regulated intracellular proteolysis (RIP). in the first step of riP, activated adam protease cleaves the extracellular portion of a generic transmembrane protein, shedding the ectodomain. the membrane- associated product is then a substrate for the γ-secretase complex, which cleaves within the transmembrane segment. this results in a soluble intracellular domain (iCd), which can signal in the cytosol and nucleus (as in the case of notch and erbB4), and a small extracellular fragment. in the case of amyloid precursor protein, this extracellular fragment, amyloid β, can aggregate into fibrils and is associated with alzheimer’s disease. uBiquitin and the Proteasome deGradation PathWay The cell must be able to degrade cytosolic proteins rapidly and effi- ciently in some situations, for example, when a protein is unable to fold correctly or has become damaged. The accumulation and aggregation of such damaged proteins can be highly toxic to the cell. To avoid this prob- lem, eukaryotes have evolved an elaborate system that specifically tags damaged proteins for destruction and delivers them to the proteasome, a molecular machine that digests them into short peptides. This system for targeting specific proteins for removal from the cytosol has also been exploited extensively for signal transduction, providing a rapid and irre- versible way to change the protein constituents of the cell in response to various regulatory inputs. In this section, we will consider the machinery involved in this process, and discuss how it is used to control the cell cycle and other signaling events. The proteasome is a specialized molecular machine that degrades intracellular proteins While the cell has a need to hydrolyze damaged or misfolded proteins, it must simultaneously avoid the potential dangers of unrestrained pro- teolytic activity. For this reason, sites of proteolysis must be sequestered from other components of the cytosol. Lysosomes, which mostly mediate the degradation of proteins derived by endocytosis from the cell surface or extracellular environment, are shielded from the cytosol by a lipid membrane. The proteolyic destruction of cytosolic proteins, however, is mediated by a very large multisubunit protein structure termed the proteasome. This consists of a hollow cylinder lined on the inside with proteases, capped at each end by a “lid” structure that controls access to the proteolytic machinery of the inner chamber (Figure 9.9). Proteins (a) (b) central cylinder (protease) cap target protein with polyubiquitin chain active sites Figure 9.9 The proteasome. (a) three-dimensional reconstruction of the 26s proteasome from electron microscope images. side views and the view from the capped end of the proteasome complex (right) are shown, with the approximate location of proteasome components indicated by colors. (b) Processive protein digestion by the proteasome. the proteasome cap recognizes a substrate protein marked by a polyubiquitin chain, and subsequently translocates it into the proteasome core, where it is digested. (a, adapted from P.C. da fonseca and e.P. morris, J . B i o l . C h e m. 283:23305–23314, 2008. With permission from the american society for Biochemistry and molecular Biology; b, adapted from B. alberts et al., molecular Biology of the Cell, 5th ed. Garland science, 2008.) targeted for destruction are specifically bound by the lid of the protea- some, unfolded, and threaded inside, where they are reduced to small pep- tides that are released into the cytosol. In this way, the cytosol is shielded from the potent proteolytic activities of the proteasome, and only specifi- cally targeted proteins are ushered into the chamber for destruction. In eukaryotes, the intact proteasome (called the 26S proteasome, for its sedimentation behavior upon centrifugation) consists of a cylindrical core (the 20S core particle) and two 19S regulatory particles capping the ends of the cylinder. The core consists of 28 subunits stacked into four rings, and contains three distinct protease activities, each with different peptide- substrate specificity. Thus, any polypeptide that enters the core is likely to be efficiently cleaved into small oligopeptides by the combined action of multiple proteases. For the isolated core particle, entry to the inner cham- ber of the proteasome is sterically blocked. Entry is mediated by the 19S regulatory particle that, upon interaction with the core particle, forms a narrow channel just large enough to accommodate a polypeptide chain. Energy from ATP hydrolysis is required to unfold targeted proteins and to thread the unfolded chains into the inner chamber. The regulatory particle also performs the important task of select- ing which proteins will be passed through the proteasome and thereby destroyed. In eukaryotes, most proteasome targets bear a distinctive post- translational mark, the covalent addition of long chains of Lys48-linked polyubiquitin. These chains are added by cellular ubiquitin ligases in a process described in more detail in Chapter 4. The regulatory subunit con- tains several proteins with ubiquitin-binding domains; some of these are integral components of the regulatory particle, while some are accessory adaptor proteins that help recruit specific targets to the proteasome. The preference for substrates with long polyubiquitin chains is probably due to two factors. First, the presence of multiple ubiquitin-binding domains on the proteasome means that polyubiquitylated targets will bind much more tightly than singly or lightly ubiquitylated targets due to avidity. Second, there are deubiquitinase (DUB) activities associated with the reg- ulatory subunit that can remove ubiquitins from the end of the polyubiq- uitin chain; thus, lightly ubiquitylated substrates (with relatively short chains) may be deubiquitylated and released from the proteasome before the protein can be unfolded and stuffed into the inner chamber. In any case, once a protein is committed to destruction, it begins to unfold and other DUB activities remove the polyubiquitin chain at its base, allowing the ubiquitin to be recycled. The cell cycle and its regulation are described in Chapter 12 The cell cycle is controlled by two large ubiquitin- conjugating complexes The cell-cycle machinery ensures that each step in the cell cycle—mito- sis, G₁, S phase (where the genome is replicated), G₂, mitosis again—occurs only at the appropriate time and after the previous steps have been com- pleted. Two of the most fateful decisions that a cell must make are whether to initiate DNA replication (and thus commit the cell to divide), and when to segregate the chromosomes and other cell contents into two daughter cells through the process of mitosis followed by cytokinesis. For multicel- lular organisms in particular, mistakes here can be disastrous, leading to abnormal development or to diseases such as cancer. Not surprisingly, an elaborate mechanism has evolved to regulate these processes tightly. This involves the sequential activation of a series of cyclin-dependent kinases (CDKs), serine/threonine kinases consisting of a CDK catalytic subunit and a regulatory cyclin subunit that is needed for activity. Two large, multisubunit ubiquitin ligase complexes work along with the CDKs (a) substrate binding Skp1 Skp2 Skp1 Cul1 ubiquitin Cys E2 E3 Figure 9.10 Structure of the SCF complex and the APC. (a) the S a cc h a r o m y c e s cerevisiae sCf ubiquitin ligase complex: cullin scaffold subunit (Cul1, green), rinG-type e3 ubiquitin ligase (rbx1, p i n k), adaptor subunit (skp1, blue), specificity component (skp2, an f-box protein, p u r p l e), and an e2 ubiquitin ligase ( o r a n g e ). the active-site cysteine (Cys) of the e2, where ubiquitin would be covalently attached, is shown in space-filling representation (blue). (b) structure of the human aPC derived from cryoelectron microscopy, in the absence (left) or presence (right) of the Cdh1 specificity subunit ( p u r p l e ). the cullin subunit apc2 is colored green, and the doc1 subunit, which is involved in substrate recognition, is colored yellow. note that parts (a) and (b) are rendered at different scales. (a, adapted from n. Zheng to control progression through the cell cycle. Through their action, compo- nents necessary for the previous stage of the cycle are polyubiquitylated and targeted for destruction by the proteasome, ensuring that the cycle progresses to the next step and cannot run backward. The two ubiquitylating complexes are the anaphase-promoting com- plex (APC), which regulates the timing of mitosis, and the SCF complex, which acts at a number of steps in the cell cycle. Each of these complexes has a core architecture consisting of a RING-type E3 ubiquitin ligase sub- unit, a cullin scaffold subunit, and one or more adaptor subunits that, in turn, interact with specificity-determining subunits that bind and recruit substrates (Figure 9.10). Thus, for these complexes, at least four interact- ing proteins together perform the tasks normally performed by a single E3 ligase—recruiting the E2–ubiquitin and the substrate, and facilitating the transfer of ubiquitin to the substrate. For both the APC and the SCF complexes, multiple specificity-determining subunits can interact with the same core machinery, allowing each complex to be targeted to specific classes of protein substrates depending on the subunit bound. SCF recognizes specific phosphorylated proteins, targeting them for destruction The core components of the SCF are the E3 (Rbx1), a cullin (Cul1), and the Skp1 adaptor. The specificity components of the SCF complex are called F-box proteins, of which at least 50 are known in the human genome. Many F-box proteins have a generic WD40 repeat domain structure, but a key and defining characteristic of most is that they specifically bind to phosphorylated target proteins, many of which are substrates of specific cyclin–CDKs or other kinases regulated in the cell cycle. Recognition of a phosphorylated site by the SCF complex becomes the kiss of death— the protein is rapidly polyubiquitylated and consigned to destruction by the proteasome. In this way, a normally transient and reversible post- translational modification—phosphorylation—results in the permanent and irreversible elimination of the modified protein (Figure 9.11). This is but one of the many ways that CDKs and ubiquitin ligases work together in regulating cell-cycle progression. The large number of F-box proteins that can bind to the SCF complex, and the variety of substrates recognized by each F-box protein, make et al., Nature 416:703–709, 2002. With permission from macmillan Publishers ltd.; b, adapted from f. herzog et al., Science 323:1477–1481, 2009. Courtesy of holger stark and Jan-michael Peters.) (b) CDK substrate + SCF time Figure 9.11 SCF couples CDK activity to protein degradation. (a) substrate is phosphorylated by an activated cyclin–cyclin-dependent kinase (CdK) complex. upon phosphorylation, it is targeted by the f-box protein (fB) of the sCf complex and polyubiquitylated. the ubiquitylated substrate is then destroyed by the proteasome. (b) in such a system, the concentration of substrate is inversely related to the level of CdK activity. it difficult to generalize about the role of the SCF complexes in the cell cycle—virtually every step in the process is regulated in some way by SCF-mediated proteolysis. For example, the F-box protein β-TrCP is important in targeting both inhibitors of the cell cycle (for example, Wee1, a kinase that phosphorylates and thereby inhibits CDKs) and stimulators (for example, Cdc25, the phosphatase that removes the Wee1-mediated inhibitory phosphorylation). Which substrates are targeted by β-TrCP in a given situation depends on their phosphorylation state, which is medi- ated by yet other kinases and by the CDKs themselves. Another F-box protein, Skp2, specifically targets CDK inhibitors, which include p21 and p27. The proteolytic destruction of these inhibitors is a key step in pro- gressing from the G₁ to S phase of the cell cycle. Two APC species act at distinct points in the cell cycle The APC is an even more elaborate molecular assembly than the SCF complex: in addition to RING and cullin subunits, it contains roughly ten more proteins. In contrast to the SCF, however, it associates mainly with just two specificity subunits, Cdc20 and Cdh1. Thus, there are only two major species of APC (APCCdc20 and APCCdh1), each of which has a distinct role in the cell cycle. APCCdc20 is critical in driving the events of mito- sis, and in particular in triggering the metaphase-to-anaphase transition (when sister chromatids physically separate so they can migrate to oppo- site poles of the mitotic spindle, ultimately to be incorporated into the two daughter cells). Once anaphase is triggered, APCCdh1 takes over, and its activity is critical to maintaining a stable G₁ phase and repressing the initiation of a new round of DNA synthesis until appropriate signals are received (Figure 9.12). In late G₂ phase of the cell cycle, mitotic cyclin–CDK complexes accu- mulate and become activated, leading to phosphorylation of APC core subunits. This phosphorylation greatly increases the affinity of Cdc20 for the APC complex, leading to formation of active APCCdc20. Among the most important substrates for the active complex are the mitotic cyc- lins and a protein called securin. Polyubiquitylation and degradation of the mitotic cyclins allows progression from mitosis, but also serves to inactivate the kinase that activates APCCdc20—a negative feedback that leads ultimately to loss of APCCdc20. Degradation of securin leads to cleavage of the bonds that hold sister chromatids together, allowing them to move to opposite poles of the mitotic spindle. This is the most critical step in mitosis, because premature separation (before all chro- mosomes have had a chance to align on the mitotic spindle, with sister Cdc20 phosphorylation by cyclin B–CDK targets mitotic substrates, cyclin B APC loss of cyclin B–CDK activity Cdh1 phosphorylation by cyclin E–CDK Cdh1 targets cyclins A and B METAPHASE-ANAPHASE TRANSITION Figure 9.12 START Regulation of APC activity by phosphorylation. (a) during m phase, the cyclin B–cyclin-dependent kinase (CdK) complex phosphorylates the anaphase-promoting complex (aPC), increasing its affinity for the specificity subunit, Cdc20. the now-active aPCCdc20 targets substrates such as securin and cyclin B. decreasing cyclin B–CdK activity leads to dephosphorylation of aPC and dissociation of Cdc20. Cdh1 now binds and targets s-phase and mitotic cyclins (cyclins a and B, respectively) to maintain a stable G₁ phase. if G₁/s cyclin (cyclin e) begins to accumulate, Cdh1 is phosphorylated and no longer binds aPC, inactivating it until the next m phase. (b) aPCCdc20 activity (purple), aPCCdh1 activity (pink), and cyclin e accumulation (green) are plotted during m and G1 phases. metaphase-anaphase transition (where chromosomes separate in mitosis) and start (the point at which a cell is committed to undergoing dna replication and cell division) are indicated. chromatids attached to opposite spindle poles) can lead to chromosome loss or breakage. An elaborate control system, termed the spindle assem- bly checkpoint, ensures that APCCdc20 is not activated and securin is not polyubiquitylated and degraded until all chromosomes have assumed their proper positions. Once APCCdc20 has been activated and has caused mitotic cyclins to be degraded, its activity wanes due to loss of APC phosphorylation. At this Activation of IKK by receptors is described in Chapter 8 Figure 9.13 The NF-κB family. the rel homology domain, transcriptional transactivation domain (ta), ankyrin repeats (a), and death domain (dd) are depicted. p52 and p50 are derived by proteolysis of p100 and p105, respectively. point, Cdc20 can no longer bind and the second specificity subunit, Cdh1, can associate with the APC. APCCdh1 targets not only the cyclins necessary for M phase (cyclin B), but also those needed to trigger S phase (cyclin A), thus preventing premature entry into a new cell cycle. It does not, how- ever, ubiquitylate the cyclins needed to initiate exit from G₁. When tran- scription of these cyclins (such as cyclin E) is stimulated by mitogens and other proliferative signals, the activity of the G₁/S cyclin–CDK complex increases, leading to the phosphorylation of Cdh1. Once phosphorylated, it can no longer bind to APC, allowing S-phase cyclins to begin to accumu- late and initiate a new round of DNA synthesis. NF-κB is controlled by regulated degradation of its inhibitor Members of the nuclear factor κB (NF-κB) family of transcription fac- tors are activated in response to a wide variety of signals, particularly in the innate and adaptive immune responses. NF-κB is present in a latent, inactive form in the cytosol of unstimulated cells, from where it can be mobilized rapidly and in the absence of new protein synthesis. The acti- vation of NF-κB depends on the regulated degradation of inhibitory sub- units or domains which, in turn, is mediated by polyubiquitylation and proteasome-mediated proteolysis. The NF-κB family consists of five related proteins, which have in com- mon a Rel homology domain that mediates dimerization and DNA bind- ing (Figure 9.13). Three of these proteins [p65 (RelA), c-Rel, and RelB] have transcriptional transactivation domains, whereas the other two (p105/p50 and p100/p52) lack the transactivation domain but instead contain a C-terminal inhibitory domain consisting of a number of ankyrin repeats. For these subunits, proteolytic removal of the inhibitory domain is required for activity. A variety of heterodimers are possible, with dis- tinct but overlapping transcriptional activation profiles. The “canonical” NF-κB species consists of a p65/p50 heterodimer. The activity of NF-κB is held in check in unstimulated cells by association with inhibitory subunits, termed IκBs. There are several distinct species of IκB, all of which consist of multiple ankyrin repeat domains; for the canonical NF-κB pathway, the predominant inhibitory subunit is IκBα. These inhibitors function by retaining the ternary NF-κB/IκB complex in the cytosol, through the combined action of a nuclear export signal on IκB and steric blocking of nuclear localization signals present on the other subunits. The key to unleashing the transcriptional activity of NF-κB, therefore, is the physical removal of the inhibitory subunit. This is accomplished through signal-induced phosphorylation of the IκB, followed by recognition of the phosphorylated site by the SCF fam- ily E3 ligase, SCFβ-TrCP (already introduced above). Of course, this raises the question of how phosphorylation of IκB is regulated. This is the cul- mination of a rather complicated series of signaling events ultimately leading to activation of the IKK (IκB kinase) complex, which consists of two catalytic subunits, IKKα and IKKβ, and a regulatory subunit, p65, RelB, c-Rel p100, p105 p52, p50 IκBs, Bcl3 NEMO (Figure 9.14). It is interesting to note that the activation of IKK involves a number of proteins of the TRAF family that are themselves RING-family ubiquitin E3 ligases. However, the TRAFs mediate Lys63- linked polyubiquitylation, and thus do not target proteins for proteaso- mal degradation. Instead, the Lys63-linked modification of substrates such as TRAF itself seems to function primarily in helping to recruit and assemble signaling complexes that ultimately lead to the activation of IKK. Another interesting aspect of NF-κB regulation is that NF-κB activity strongly promotes the transcription of IκBα, which serves as a negative feedback loop to shut down NF-κB signaling after an initial burst of activity. In addition to the canonical NF-κB pathway, a noncanonical pathway involves activation of heterodimers containing p100 and RelB. Both p100 and p105 contain C-terminal domains that function as IκBs, so proteolytic processing of the precursor form is required for activation of complexes containing these subunits. In the case of p105, processing can be either constitutive or inducible, and the role of polyubiquitylation in processing is still not settled. In the case of p100, however, it is clear that p100–RelB heterodimers can be activated by a subset of NF-κB activators, in a proc- ess involving phosphorylation of p100 by IKKα (in the absence of IKKβ and NEMO). Once phosphorylated, p100 is recognized by SCFβ-TrCP, poly- ubiquitylated, and targeted to the proteasome. In this case, however, the entire protein is not degraded, just the C-terminal IκB-like region. Such partial degradation by the proteasome is quite unusual, and for p100 and p105 (and a few other known examples, such as the Gli transcription factors that are effectors of Hedgehog signaling) specialized sequence elements are responsible. These consist of a relatively unstructured glycine-rich region, which may form a hairpin loop and insert into the Figure 9.14 Canonical and noncanonical pathways of NF-κB activation. in the canonical pathway, a latent heterotrimeric complex consisting of two rel subunits and an iκB subunit (here, p65, p50, and iκBα) is phosphorylated by the heterotrimeric iκB kinase (iKK) complex. Phosphorylated iκB is targeted by sCfβ-trCP and polyubiquitylated, leading to proteasomal degradation. the released dimeric nf-κB translocates to the nucleus and induces transcription through its transactivation domain (ta). in the noncanonical pathway, a latent p100– relB heterodimer is phosphorylated by iKKα. the phosphorylated tail of p100 is polyubiquitylated and subjected to partial proteolysis in the proteasome, generating an active p52–relB heterodimer. Colors correspond to those in figure 9.13. proteasome, juxtaposed to a very stably folded region, which presumably thwarts the proteasome’s normal unfolding machinery and prevents fur- ther degradation. In the case of the noncanonical NF-κB pathway, partial degradation leads to the release of a p52–RelB heterodimer, which can then translocate to the nucleus and promote transcription of a unique set of NF-κB targets. CasPase-mediated Cell death PathWays In the development and day-to-day life of metazoans (multicellular organisms), occasions arise that necessitate the death of individual cells for the benefit of the whole organism. The contents of lysed cells can be quite toxic to surrounding tissues, and so a specialized physiological proc- ess termed apoptosis has evolved to ease cells through a relatively neat and tidy process of programmed cell death, minimizing the effects on neighboring cells and thus the organism as a whole. The decision to initi- ate this self-destruct program is obviously important both for the indi- vidual cell and for the organism—literally a matter of life and death. At the heart of the apoptotic machinery are the caspases, a specialized class of proteases, which execute the cell death program by cleaving key cell proteins. More recently, a broader role for caspases in other physiological responses, such as immunity, has emerged. In this section, we will discuss the regulation and action of caspases, focusing primarily on their role in apoptosis. Apoptosis is an orderly and highly regulated form of cell death For free-living unicellular organisms, cell death may be a tragedy for the individual cell but has little impact on the larger population. In meta- zoans, however, the fate of each cell is tied to the fate of the organism as a whole. Normal development requires that some cells die. For exam- ple, in Caenorhabditis elegans, a simple roundworm consisting of roughly 1000 cells, more than 100 cells undergo programmed cell death in the process of generating the finished worm. Furthermore, cells that have been damaged, particularly those that have sustained extensive or irrepa- rable damage to the genomic DNA, pose a considerable danger to multi- cellular organisms—cells that are allowed to pass on a damaged genome to their progeny may develop into cancers that can threaten the survival of the whole organism. Apoptosis provides an efficient way to remove such cells without generating cell debris that could initiate inflammatory or immune responses which could pose additional dangers to the surround- ing cells. Apoptosis is a highly orchestrated form of cell death initiated by specific sig- nals and leading to characteristic biochemical and morphological changes (Figure 9.15). The chromatin of a cell undergoing apoptosis condenses, and nucleases are activated that degrade the DNA into oligonucleosome- sized fragments (this “laddering” of degraded DNA, visible upon electro- phoresis, is a hallmark of apoptosis). The nuclear membrane breaks down, as do other membrane-bound organelles, and cytoskeletal changes lead to cell rounding and membrane out-pocketing or “blebbing.” The result- ing blebs ultimately pinch off, creating membrane-enclosed packets of cell contents, termed apoptotic bodies, which can be ingested by surrounding cells or by professional phagocytes such as macrophages. These cell frag- ments are then completely digested by lysosomes within the engulfing cell. In this way, the contents of the apoptotic cell, including its nucleic acids, can be recycled without ever being exposed to the extracellular necrotic cell death release of cell contents trauma apoptotic cell death apoptotic signals apoptotic bodies ingestion by other cells environment. This contrasts with cells that die by physical trauma or acute stress (a process called necrosis), which merely spill their contents into their surroundings. Apoptosis is initiated either by signals from the extracellular environ- ment, acting through plasma membrane receptors, or by signals from within the cell itself, such as physiological stresses or DNA damage. As discussed in more detail below, these extrinsic and intrinsic pathways dif- fer in important respects, but both lead to the activation of a specialized group of cysteine proteases, termed the caspases, that specifically cleave peptide bonds C-terminal to aspartic acid in their protein targets. Most caspase-mediated signaling pathways involve a hierarchy of two classes of caspases, initiator caspases and effector or executioner caspases. The initiator caspases are directly activated by apoptotic signals, as dis- cussed below. Once activated, they amplify the resulting signal by pro- teolytically activating the effector caspases, which go on to cleave cell proteins that mediate the physiological manifestations of the cell death program. As we have seen many times already, cascades of sequentially activated proteases are commonly used to generate widespread, irrevers- ible physiological changes. The activity of caspases is tightly regulated There are 13 distinct caspase genes in humans, which can be roughly divided into two classes (initiator and effector) based on their function, activation mechanism, and structural features (Figure 9.16). All caspases have a similar overall structure, consisting of an N-terminal pro-domain followed by large and small catalytic subunits. These three domains are connected by linker sequences that contain sites for caspase cleavage. All caspases are initially expressed as inactive zymogens, though the mecha- nism of their activation is different for the two classes. Initiator caspases (caspases 2, 8, 9, and 10) have an extended N-terminal pro-domain and are activated predominantly by dimerization or assembly into higher- order structures, whereas effector caspases (caspases 3, 6, and 7) have Figure 9.15 Apoptotic cell death. When a cell dies by trauma (physical insult, heat, toxins, and so on), the cell ruptures and its contents are released (necrosis). By contrast, a cell undergoing apoptosis dies by a programmed sequence of events, leading to its fragmentation into membrane-enclosed apoptotic bodies that are engulfed and digested by surrounding cells. (a) pro-domain catalytic domain caspase-2,9 caspase-8,10 caspase-3,6,7 caspase-1,4,5 initiator (intrinsic) initiator (extrinsic) effector inflammatory (b) initiator scaffold binding, conformational change processing active protease effector cleavage by initiator caspase active protease Figure 9.16 Caspase domain structure and activation mechanism. (a) all caspases have an n-terminal pro-domain followed by large and small protease catalytic domain subunits (blue). sites of cleavage by caspases are indicated by red a rr o w s. the caspase recruitment domain (Card) and death effector domain (ded) mediate homotypic protein interaction with scaffold and adaptor proteins during activation. (b) activation of initiator and effector caspases. interaction with scaffold complexes induces conformational changes that activate initiator caspases, while effector caspases must be activated by cleavage by upstream initiator caspases (red arrows). a shorter pro-domain and are activated by cleavage by upstream initia- tor caspases. Inflammatory caspases (caspases 1, 4, and 5) are similar in structure and activation mechanism to initiator caspases. In all cases, the mature, active caspase is a tetrameric complex consisting of two large and two small catalytic subunits. Initiator caspases must respond to and become activated by diverse upstream signaling inputs. This is accomplished by stimulus-induced aggregation, the same mechanism underlying many receptor-mediated signaling strategies. Recruitment of inactive, monomeric initiator caspas- es into a multimeric structure or scaffold leads to dimerization of the pre- cursors, along with conformational changes that rearrange the catalytic domain and allow cleavage of the dimer partner into its mature form. As discussed below, the multimeric complex that mediates the activation of the extrinsic pathway is the death-inducing signaling complex (DISC), whereas the intrinsic pathway is triggered by assembly of the apopto- some. More specialized signal-induced scaffolds also exist. For example, caspase-2 is activated in the nucleus in response to genotoxic stimuli by a structure termed the PIDDosome (PIDD is a major transcriptional target of p53, which is activated by DNA damage, cell-cycle checkpoints, and other stress responses). Similarly, pathogens activate caspases-1, 4, and 5 through the assembly of a scaffolding complex termed the inflammasome, which plays an important role in the innate inflammatory response (see below). By contrast, the effector caspases exist normally as inactive dimers. Their activation requires cleavage of the linker between the two catalytic seg- ments, allowing rearrangement of the catalytic site into its fully active conformation. This strategy makes the activation of effector caspases absolutely dependent on the levels of activity of the upstream initiator caspases. In lymphocytes, another unrelated protease, the serine protease granzyme B, can also cleave and activate effector caspases, but this seems to be an exception. In vitro studies showed that caspases are highly specific for peptide substrates with an aspartic acid (Asp) residue preceding the bond to be cleaved, followed by an amino acid with a small, uncharged side chain. Proteomic studies have shown that hundreds of different proteins are cleaved in cells where effector caspases are activated. Some of these are clearly important for the physical execution of apoptosis. For exam- ple, the caspase-activated DNase (CAD) and the Rho-associated kinase (ROCK) are activated by caspase-mediated cleavage, causing DNA deg- radation and reorganization of the actin cytoskeleton, respectively, while the lamins that provide structural integrity to the nuclear envelope are inactivated by cleavage by caspases. However, many of the substrates might be “innocent bystanders” that play no role in the physical events of apoptosis. It is interesting to note that, in many cases, cleavage by cas- pases modifies or stimulates the activity of a substrate, rather than sim- ply inactivating it by degradation. In this way, proteolysis can promote the irreversible activation of pathways that lead to the orderly death of the cell. Since inadvertent activation of caspases would have dire consequences for the cell, mechanisms are needed to keep basal activity levels low in the absence of strong apoptotic stimuli. A family of caspase inhibitors termed IAPs (inhibitor of apoptosis proteins) serves to repress activated caspases under most conditions. These proteins use two mechanisms to inhibit their targets. First, they can bind directly to and inhibit activated caspases though protein interaction modules termed BIR domains. Sec- ond, most IAPs also have a RING-type ubiquitin E3 ligase activity, which can mono- and polyubiquitylate the caspase and either inactivate it or target it for proteasome-mediated degradation. IAPs can themselves be down-regulated in order to sensitize cells to apoptotic stimuli. One way this is accomplished is by the translocation or increased expression of IAP antagonists such as Smac/DIABLO, which bind to and sequester IAPs and thereby prevent their binding to caspases. The extrinsic pathway links cell death receptors to caspase activation The extrinsic apoptotic pathway is induced by a number of lig- ands that interact with specific cell-surface receptors. These include Fas ligand (FasL) and its receptor Fas, tumor necrosis factor (TNF) and its receptor TNFR, and TRAIL and its receptors DR4 and DR5. These receptors are collectively referred to as death receptors. Binding of the corresponding trimeric ligand to the death receptor induces con- formational changes that expose death domains (DDs) on the intracel- lular portion of the receptor. Using the prototypic death receptor Fas as an example, the activated receptor then recruits an adaptor, FADD, through homotypic DD-mediated interactions. Receptor binding expos- es the death effector domain (DED) of FADD, which then recruits ini- tiator caspases (caspase-8 and to a lesser extent caspase-10) through homotypic interaction with the DED domain located in the caspase pro- domain (see Figure 8.18). Collectively, these interactions generate the active death-inducing signaling complex (DISC), which serves as a platform for the dimerization and allosteric activation of the bound caspase. Upon activation, bound caspases cleave each other to generate Signaling by death receptors is discussed in more detail in Chapter 8 soluble, heterotetrameric caspase that is free to activate caspase-3, the downstream effector caspase. In some but not all cell types, simultane- ous activation of the intrinsic pathway is also necessary for efficient induction of apoptosis. This is accomplished by caspase-8-mediated cleavage and activation of BID, a key promoter of the intrinsic pathway (discussed further below). Other death receptors use variations on this overall strategy for apoptotic activation (Figure 9.17). There are five receptors for the pro-apoptotic ligand TRAIL, but only two of these (DR4 and DR5) can activate caspases. The others can bind ligand, but because they lack intracellular DDs they are incapable of downstream signaling. Therefore they act as “decoys” to soak up and sequester TRAIL, thus blocking apoptosis in cells where they are expressed at high levels. In the case of the TNFR, the major signal- ing output under most conditions is activation of NF-κB, not apoptosis. NF-κB is activated by the recruitment of a different DD-containing adap- tor, TRADD, to the activated receptor. TRADD in turn recruits TRAF2 and the IKK complex, ultimately leading to the phosphorylation, ubiq- uitylation, and destruction of IκB. The active NF-κB that is released strongly suppresses apoptosis by inducing transcription of various apop- totic inhibitors, such as IAPs and a DED-containing decoy protein FLIP, which competes with initiator caspases to prevent their recruitment to the DISC. Only when NF-κB signaling is for some reason blocked is a second, pro-apoptotic complex formed. This complex contains TRADD, FADD, and caspase-8, and functions analogously to the DISC formed in FasL TRAIL TNFα Figure 9.17 Signaling by various death receptors. fas ligand (fasl) binding to fas leads to aggregation of the adaptor fadd, and recruitment/activation of initiator caspase. some trail receptors (dr4, dr5) signal through fadd in the same fashion, whereas other “decoy” receptors for trail do not contain intracellular death domain (dd) motifs and thus cannot activate caspases. the primary result of binding of tumor necrosis factor α (tnfα) to its receptor (tnfr) is the aggregation of the tradd adaptor, leading to recruitment of traf2, iκB kinase (iKK) activation, and activation of nf-κB. a secondary pathway leads to formation of a cytosolic complex containing tradd and fadd, which can recruit and activate initiator caspases. this pathway is normally strongly inhibited by nf-κB; thus it is only activated under conditions where nf-κB activity is suppressed. the domain structures of fadd and tradd are indicated in the inset. response to TRAIL or FasL. In this case, apoptosis is a “fail-safe” mech- anism that kicks in under circumstances where the primary response, NF-κB induction, is ineffective. TNFR signaling is just one example of many instances where the pathways leading to activation of NF-κB and to apoptosis intersect. Infection of cells with bacterial or viral pathogens can activate an alter- native DISC-like complex termed the inflammasome (Figure 9.18). The result of activation of this pathway is not apoptosis, but instead the processing and secretion of a host of proinflammatory cytokines. The (a) NALP1 ASC (b) conformation Δ, oligomerization bind ASC bind bacterial products, NTP activated caspase released Figure 9.18 INFLAMMASOME Assembly of the NALP1 inflammasome. (a) domain structure of nalP1 and asC. nalP1 contains an n-terminal Pyrin domain (Pyd), a domain that mediates nucleotide binding and oligomerization (nod), leucine-rich repeats (lrr) that bind microbial products such as cell-wall proteoglycans, and a Card domain (caspase recruitment domain). the asC adaptor contains Pyrin and Card domains. (b) Binding of bacterial products and nucleotide (ntP) to nalP1 induces conformational changes leading to oligomerization. nalP1 now binds asC and recruits inflammatory caspases such as caspase-1, which are activated by conformational change and dimerization. Processed caspase is then released from the complex. the organization of subunits shown is diagrammatic and not based on physical structural data. caspases that are activated include caspase-1 and caspase-5, which con- tain in their pro-domains another protein interaction module, termed the caspase recruitment domain or CARD domain, in place of the DED. Activation is achieved by assembly of a multiprotein complex consisting of a sensor protein (such as NALP1, NALP3, or IPAF) that is allosteri- cally activated by pathogen components. Activation induces oligomeriza- tion of the sensor, and exposes other protein interaction domains that can directly or indirectly (through adaptor proteins) interact with the CARD domain in the caspase pro-domain. In the case of the NALP1 inflamma- some, activation exposes its PYD domains, which recruit and bind to PYD domains of an adaptor such as ASC. This exposes the CARD domain of the adaptor, thus recruiting caspases through CARD–CARD interactions. This scheme is conceptually very similar to death receptor DISC forma- tion, with the PYD substituting for the DD and the CARD for the DED. Activated caspase-1 can directly cleave and activate proinflammatory cytokines such as interleukin-1β, and indirectly induce the processing and secretion of others. Figure 9.19 Bcl2 family of apoptotic regulators. representative structures are shown for the three major groups. Positions of Bcl2 homology domains (Bh1, Bh2, Bh3, and Bh4) and transmembrane helix (tm) are indicated. names of family members in each group are provided at left. note that other, less-well-characterized family members exist for each group. Mitochondria orchestrate the intrinsic cell death pathway The intrinsic apoptotic pathway responds to signals generated from within the cell, most of which fall into the general category of “stress responses.” For these pathways, the mitochondrion plays a sur- prising role in integrating the welter of potentially conflicting informa- tion on the status of the cell and, where necessary, executing the early phases of the apoptotic program. Central to this formidable task is a group of proteins termed the Bcl2 family, of which at least 12 are found in humans. The balance between pro-apoptotic and anti-apoptotic Bcl2 family members determines the fate of the cell. When pro-apoptotic Bcl2 family members predominate, they induce changes in the mitochondrial outer membrane that increase its permeability (termed mitochondrial outer membrane permeabilization, or MOMP), leading to the release of cytochrome c and other components from the mitochondria. Ultimately, it is these released mitochondrial proteins that nucleate the assembly in the cytosol of the apoptosome, yet another large cytosolic complex similar to the DISC that serves as a scaffold for the recruitment and activation of caspases. There are three major classes of Bcl2 proteins, defined by their biologi- cal activities and their structural elements. All contain at least one of four conserved sequence motifs, termed Bcl homology (BH) domains, and many also have a C-terminal transmembrane helix that anchors them to cellular membranes, particularly the mitochondrial outer membrane (Figure 9.19). These three categories are the anti-apoptotic Bcl2 proteins, which include Bcl2 and Bcl-XL; the pro-apoptotic or effector Bcl2 proteins, BAX and BAK; and the pro-apoptotic BH3-only proteins, which include BAD, BID, BIM, Noxa, and PUMA. Many of these Bcl2 family members can interact with each other, usually via insertion of the helical BH3 segment of one into a hydrophobic pocket on the surface of its partner (described below). It is the stoichiometry of the interactions among these classes of Bcl2, Bcl-XL anti-apoptotic BAX, BAK pro-apoptotic BAD, BID, BIM, Noxa, PUMA pro-apoptotic, BH3-only induce BH3-only proteins cytochrome c proteins that determines whether apoptosis will be activated. Under nor- mal conditions, there is an excess of anti-apoptotic proteins, and the activ- ity of the pro-apoptotic proteins BAX and BAK is held in check. Cellular stress signals, however, induce either the expression or post-translational modification of the BH3-only class of proteins, and this shifts the equilib- rium such that the pro-apoptotic Bcl2 proteins can aggregate on the outer membrane of the mitochondrion and induce MOMP (Figure 9.20). There is still some uncertainty whether it is the direct binding of BH3-only proteins to the effectors BAX and BAK that triggers MOMP, or whether the key role of BH3-only proteins is to titrate away anti-apoptotic Bcl2 proteins that normally inhibit BAX and BAK. In some cases, membrane localization of Bcl2 proteins also seems to be regulated by their dimer partners. In the case of BAX, for example, structural studies showed that the C-terminal membrane-insertion helix occupies the same hydrophobic pocket on the surface as exogenous BH3 peptides; thus, interaction with BH3-only proteins is likely to displace the membrane-insertion segment and expose it, leading to recruitment of BAX to the mitochondrial mem- brane (Figure 9.21). In any case, increased activity of BH3-only proteins directly or indirectly induces conformational changes in BAX and BAK that lead to membrane insertion and pore formation and, ultimately, to mitochondrial fragmentation and loss of mitochondrial function. The primary level at which the intrinsic pathway is induced is via increased activity or abundance of the BH3-only proteins. Different BH3-only proteins act as sensors for a wide variety of cellular stresses. For example, the transcription of PUMA and Noxa is induced by p53, and BIM transcription is stimulated by mitogen and growth factor deprivation and endoplasmic reticulum (ER) stress (excessive levels of unfolded proteins in the ER). Changes in post-translational modifica- tion can also lead to activation of BH3-only proteins, as in the case of Figure 9.20 Induction of apoptosis depends on the balance of Bcl2 proteins. normally, anti-apoptotic Bcl2 proteins are in excess over pro-apoptotic Bcl2 proteins. When Bh3-only protein activity or abundance increases, the balance shifts and pro-apoptotic proteins are in excess; momP is induced and cytochrome c and other pro-apoptotic factors are released from the mitochondrial intermembrane space. When levels of anti-apoptotic proteins greatly exceed those of pro-apoptotic proteins, as when Bcl2 is overexpressed, then levels of Bh3-only proteins that would normally induce apoptosis have no effect. (c) Figure 9.21 BH3 binding can regulate membrane insertion of pro-apoptotic Bcl2 proteins. (a and b) x-ray crystal structure of full-length Bax, a pro-apoptotic Bcl2 protein. in its soluble state, the C-terminal transmembrane helix of Bax (brown) is sequestered in a hydrophobic pocket. the structure in (b) is rotated 90° from that in (a). (c) diagrammatic representation of binding of Bh3-only protein (green) to Bax. the Bh3 helix binds to the same hydrophobic pocket as the C-terminal helix (brown). Bh3 binding would thus displace the C-terminal helix, promoting insertion of Bax into the mitochondrial membrane. (a and b, adapted from r.J. youle and a. strasser, Nat. Rev. Mol. Cell Biol. 9:47–59, 2008. With permission from macmillan Publishers ltd.) loss of phosphorylation of BAD by Akt in the absence of survival signals. As already mentioned, cleavage of BID into its active form by caspase-8 allows extrinsic pathway activity to be amplified by engaging the intrin- sic pathway. Expression levels of the anti-apoptotic Bcl2 proteins can also be regulated by survival signals and stress, thereby resetting the threshold for induction of apoptosis (see Figure 9.20). Indeed, Bcl2 itself was first discovered as an oncogene whose expression was elevated in B cell lymphomas; it promotes the uncontrolled overgrowth of B cells by inhibiting their apoptosis. Once MOMP is induced, cell death is induced through the assembly of the apoptosome, which serves as a platform for activation of the ini- tiator caspase-9. Apoptosome assembly depends on cytochrome c that is released from the intermembrane space of the mitochondria. In the cytosol, cytochrome c binds to the adaptor protein Apaf1 and, in concert with nucleotide binding, induces conformational changes that lead to the oligomerization of Apaf1 into a structure reminiscent of a seven-spoked wheel (the “wheel of death”) (Figure 9.22). Each spoke of the wheel con- sists of an Apaf1 molecule, oriented with the bound cytochrome c on the outside and with its CARD domain located near the hub. As in the case of the inflammasome, the exposed CARD domains serve to recruit initiator caspase (caspase-9 in this case) through homotypic CARD–CARD interac- tions. Bound caspase then is activated through conformational changes, undergoes autocleavage, and then is free to activate downstream caspases (caspase-3 and caspase-7). Cytochrome c is not the only pro-apoptotic protein released from mito- chondria during MOMP. Smac/DIABLO and related IAP antagonists promote apoptosis upon their release by sequestering and inhibiting the IAP proteins, which normally function to inhibit caspases. Some of the released mitochondrial proteins can also function to promote cell death in a caspase-independent fashion. Apoptosis inducing factor (AIF), a flavoprotein, translocates to the nucleus and induces chromatin con- densation and DNA fragmentation. EndoG is a nuclease that cleaves chromatin between nucleosomes upon its release from mitochondria. Finally, loss of the electrogenic potential of the mitochondria upon summary (a) Apaf1 MOMP contributes to cell death by depriving the cell of the means to generate energy. summary Proteolysis is unlike other post-translational modifications in that its effects cannot be reversed, other than by the relatively slow process of new protein synthesis. Most proteases are initially made as catalytically inactive precursors (zymogens), which are activated at the appropriate time and place by proteolytic processing. The regulated activation of proteases plays a key role in a number of signaling pathways, including extracellular processes such as blood clotting and in transmembrane sig- naling. The targeted degradation of cytosolic proteins is achieved by the proteasome, a molecular machine that recognizes and destroys proteins that have been tagged by polyubiquitin chains. The coupling of protein ubiquitylation to proteasome-mediated destruction is central to many signaling mechanisms, including those that regulate the orderly progres- sion through the cell cycle. Regulated activation of a specialized class of proteases, the caspases, serves both to initiate and execute the events of programmed cell death (apoptosis). questions Discuss the differences between phosphorylation and degradation as modes of regulation. What types of processes might be better regulated by one or the other? How does the functional coupling of phosphoryla- tion and ubiquitylation (by ubiquitin ligases that recognize phospho- rylated substrates, such as the SCF complex) affect signaling output from the phosphorylated substrates? Almost all proteases are activated by cleavage of an inactive precursor. In some cases, one activated protease molecule can cleave and activate other molecules of the same type, while in other cases, activation must be performed by a different (upstream) protease. Compare the effects on activation dynamics and localization in these two different cases. What mechanisms prevent the entire blood supply from clotting after a minor injury? Figure 9.22 The apoptosome. (a) domain structure of apaf1. Cytochrome c released from mitochondria binds to the Wd repeats, inducing conformational changes, nucleotide binding, and oligomerization of the nod (nucleotide binding and oligomerization domain). (b) three- dimensional reconstruction of the human apoptosome [apaf1 in complex with cytochrome c and the Card domain from procaspase-9] from cryoelectron microscopy images. seven apaf1 subunits assemble into a wheel-like structure. the Card domains of apaf1 and caspase-9 form a disc (purple) without visible connections with the rest of the structure. the regulatory regions, consisting of the apaf1 Wd repeats bound to cytochrome c, are located at the end of each spoke. (b, adapted from s. yuan et al., Structure 18:571–583, 2010. With permission from elsevier.) Many viruses alter regulation of the infected host cell. For example, several viruses induce the degradation of antiviral factors. Propose strategies that could be used by the virus to accomplish this. In addi- tion, many viruses have evolved mechanisms to inhibit cellular apop- totic pathways. Why might this be the case? Propose several distinct strategies that could be used by viruses to accomplish this. Proteasome inhibitors are increasingly used for cancer therapy. Why might such compounds be effective in preferentially targeting tumor cells? referenCes General ProPerties and examPles of siGnal-reGulated Proteolysis Drag M & Salvesen GS (2010) Emerging principles in protease-based drug discovery. Nat. Rev. Drug Discov. 9, 690–701. Edwards DR, Handsley MM & Pennington CJ (2008) The ADAM metalloproteinases. Mol. Aspects Med. 29, 258–289. uBiquitin and the Proteasome deGradation PathWay Finley D (2009) Recognition and processing of ubiquitin- protein conjugates by the proteasome. Annu. Rev. Bio- chem. 78, 477–513. Gao M & Karin M (2005) Regulating the regulators: control of protein ubiquitination and ubiquitin-like modifications by extracellular stimuli. Mol. Cell 19, 581–593. Hayden MS & Ghosh S (2008) Shared principles in NF-κB signaling. Cell 132, 344–362. Komander D (2009) The emerging complexity of protein ubiquitination. Biochem. Soc. Trans. 37, 937–953. CasPase-mediated Cell death PathWays Jin Z & El-Deiry WS (2005) Overview of cell death sign- aling pathways. Cancer Biol. Ther. 4, 139–163. Pop C & Salvesen GS (2009) Human caspases: activa- tion, specificity, and regulation. J. Biol. Chem. 284, 21777–21781. Youle RJ & Strasser A (2008) The BCL-2 protein fam- ily: opposing activities that mediate cell death. Nat. Rev. Mol. Cell Biol. 9, 47–59. Yuan S & Akey CW (2013) Apoptosome structure, assem- bly, and procaspase activation. Structure 21, 501–515. The Modular Architecture and Evolution of Signaling Proteins Proteins involved in signaling pathways are frequently composed of mul- tiple domains, each of which has a distinct biochemical activity. A domain represents a polypeptide sequence that has the ability to fold independ- ently into a functional unit, and is typically between 35 and 250 amino acids in length. Thirty-five residues represents a minimal size required to specify a stable three-dimensional fold, whereas 250 amino acids likely approaches an upper limit for what can be folded without error. Thus, one reason for the multidomain structure of eukaryotic proteins may be the inability of cells to support the folding of domains beyond a relatively modest size. Typically, signaling proteins consist of several different domains, joined by less ordered linker sequences. Because such domains can be assembled together into larger proteins while retaining their dis- tinct structure and activity, they are sometimes referred to as modules or modular domains. Protein domains generally have one of two primary functions: to medi- ate interactions with other molecules within the cell, or to catalyze enzymatic reactions. We will refer to these as interaction domains and catalytic domains, respectively. In Chapter 3, we considered cata- lytic domains such as those in kinases, phosphatases, GEFs, and GAPs. In this chapter, we will focus primarily on the many classes of interac- tion domains and their role in assembling signaling pathways. We will also examine the more complex signaling properties that can emerge from the linking of multiple domains, both within a single polypeptide chain and in multiprotein complexes. Finally, we will explore how mod- ules can be recombined in evolution to create new functions, or in other cases lead to disease. Figure 10.1 Interaction domains form compact, globular modules. (a) The Csk SH3 domain (green) bound to a proline-rich peptide (pink), illustrating how the n- and C-termini of the domains are located close together in space and on the face opposite the ligand-binding site. (b) Examples of different SH3 domain-containing proteins. SH3 domains can occur in varying numbers and combinations with other domains. (a, Adapted from T. Pawson and P. nash, Science 300:445–452, 2003. With permission from AAAS.) Nck Src ModulAr ProTEin doMAinS Protein domains, and the shorter peptide sequences that bind to them, are key building blocks of signaling proteins. By understanding the individual properties of these modules, we can often infer a great deal about a protein containing them: its enzymatic activities, its localization, its binding part- ners, and its mode of regulation. Here we look more closely at the general properties of these building blocks and how they are identified. Protein domains usually have a globular structure Interaction and catalytic domains typically fold into a globular structure, which is stabilized by a tightly packed core composed largely of hydropho- bic amino acids. Residues involved in binding or catalysis are exposed on the surface of the domain, or positioned in surface-accessible pockets. As we shall see, signaling proteins are constructed from a relatively limited number of domain types or families, each of which can be found in many different proteins and in various combinations with other domains. The amino acids that form the hydrophobic core of a domain are usually high- ly conserved between related members of a domain family, as are surface residues that play an essential function in binding or catalytic activity. More variable residues, involved in the detailed specificity of binding or substrate recognition, are often found in loops on the domain’s surface. In most interaction domains, the N- and C-termini of the domain are close together in space and are located on the face of the domain opposite from the ligand-binding site (Figure 10.1). In principle, this arrangement allows an interaction domain to be easily inserted into an existing protein, while leaving its ligand-binding surface exposed to the solvent. This likely facilitated the evolution of multidomain proteins, as we shall discuss fur- ther in this chapter. Bioinformatic approaches can identify protein domains The genome typically encodes many related copies of a protein domain, which likely arose from a single ancestral gene that was duplicated and modified in the course of evolution. As we have discussed, the members of a domain family possess a number of residues that are particularly important because they are required for the domain to fold correctly, or are essential for a critical function such as ligand recognition or catalysis. These residues are therefore conserved as a domain family gains new members during evolution. Practically, this means that pro- tein domains within the same family will usually show at least 15% identity in their primary amino acid sequences. This feature can there- fore be used to discover new domains by computationally aligning the sequences of multiple proteins and looking for regions of 35–250 amino acids that show significant similarities. Indeed, many domain families were originally discovered through this bioinformatic approach, and are frequently termed “homology domains” to denote their homologous Figure 10.2 Identification of domains by database analysis. The sequencing of the human genome has revealed around 20,000 protein-coding genes. databases of the amino acid sequences encoded by all these genes can be searched to identify groups of conserved residues that constitute a domain. in this example, pleckstrin contains two copies of the pleckstrin homology (PH) domain, which has also been identified in many other proteins, three of which are illustrated here. Light brown boxes denote other conserved domains. genomic sequence encoded proteome search for regions with sequence similarity (homology) sequences. Pleckstrin homology (PH) domains, for example, were orig- inally discovered because there are two PH domain sequences in the protein pleckstrin, and related sequences were subsequently discovered in many other proteins. (Figure 10.2). If the presence of a domain is suggested by a sequence, the predicted domain can be expressed using recombinant DNA techniques. The putative domain can then be ana- lyzed to see if it has a folded structure, as well as a biochemical function such as catalytic or binding activity. Once the characteristic conserved “consensus sequence” signature is defined for a domain family, this allows the identification of new fam- ily members. The amino acid sequence of a protein of interest can thus be compared with the conserved sequences of the library of previously established domains to identify its domain composition and organiza- PH PH PH pleckstrin two repeats of pleckstrin homology domain Akt kinase β-ARK VAV tion. Since different members of a domain family usually have similar functions, this approach often suggests the possible biochemical proper- ties of an individual protein. Global analysis of particular domains in an organism’s genome also allows a preliminary estimate of the number of gene products that an organism devotes to a particular signaling activ- ity. For example, sequence analysis indicates that the human genome encodes ≈518 protein kinase domains and ≈120 SH2 domains. Approxi- mately 70% of human proteins have one or more recognizable domains, and this percentage is likely to increase as more proteins are subjected to intensive investigation. Domains can be composed of several smaller repeats While most signaling domains fold as a single unit, there are a few classes of domains that are composed of several repeats of a smaller conserved unit, which by itself forms an element of secondary structure. This is a versatile arrangement, as the properties of such a domain can potentially be altered by varying both the amino acid sequences of the nonconserved loops and the number of the repeats. For example, WD40 repeats form β strands, which then are organized into a circular structure resembling a propeller with each repeat forming one blade (Figure 10.3a). The various specificity subunits that recruit substrate proteins to the SCF ubiquitin ligase complex are examples of WD40 repeat proteins. In several other cases, such as armadillo repeats, each repeated unit has an α-helical structure. Multiple linked repeats form a twisted, superhe- lical structure with an extended binding surface, which potentially can bind several different proteins. This device is used by the β-catenin pro- tein (Figure 10.3b). β-Catenin has a structural role at cell–cell junctions and is also a central component of the Wnt signaling pathway, in which it binds distinct protein ligands in the cytoplasm and nucleus. Thus, β-catenin must bind a variety of different binding partners, depending on its localization within the cell; this may be facilitated by the versatile properties of its armadillo repeat domain. identifcation of other proteins with pleckstrin homology domains The role of SCF complexes in reg- ulating the cell cycle is discussed in Chapters 9 and 12 Figure 10.3 Domains can be formed from smaller repeats. (a) The β-propeller from Cdc4, the substrate-binding component for a form of the SCF ubiquitin ligase complex in yeast. The propeller is composed of eight Wd40 repeats (labeled PB1–8), which are each composed of four antiparallel β strands. This protein recognizes a specific phosphorylated peptide motif (green, with phosphate group highlighted in pink). β-Catenin consists of 12 α-helical armadillo (Arm) repeats, which together form an extended superhelix. A positively charged groove that spans the entire superhelical repeat region forms the binding surface for the majority of β-catenin’s interaction partners. The structure of the armadillo repeat domain of β-catenin is illustrated complexed to the cytoplasmic domain of the cell adhesion protein E-cadherin (top) and to the catenin-binding domain (CBd) of the transcription factor Tcf3 (bottom). The β-catenin domain is shown in blue and the bound peptides in pink. (a, Adapted from S. orlicky et al., Cell 112:243–256, 2003. With permission from Elsevier; b, adapted from H.J. dyson and P.E. Wright, Curr. Opin. Struct. Biol. 12, 54–60, 2002. With permission from Elsevier.) Protein domains often act as recognition modules Many protein domains act as interaction modules that recognize and bind to short peptide sequences, sometimes in a fashion that requires a post-translational modification such as phosphorylation. These pep- tide sequences or motifs are typically located in unstructured regions of their host proteins. When removed from their normal location, the isolated peptides usually retain their ability to act as ligands for inter- action domains, as enzyme substrates, or both; for this reason they can be viewed as modular units of protein function. Signaling proteins often contain a combination of folded domains with enzymatic or binding properties as well as unstructured regions containing peptide motifs. Table 10.1 summarizes a number of interaction domains involved in signal transduction. Domains that are related by sequence similarity often have a similar rec- ognition function. For example, SH2 domains primarily bind phosphotyro- sine-containing sites, and so the identification of an SH2 domain sequence is strongly suggestive of its primary function. Despite this, however, there are instances where identifying a domain by computational or structur- al methods may not necessarily reveal its actual function. WD40 repeat domains, for example, have a wide range of binding properties, including binding to phosphorylated or methylated peptide motifs, and are there- fore not easy to categorize. There are also examples of domains with very different sequences and biochemical activities that fold into similar three-dimensional struc- tures. As we will discuss in more detail below, the fold (the overall arrangement of secondary structural elements and how they are con- nected) originally identified in PH domains, which often bind phospholi- pids, is also used by other domains, including phosphotyrosine-binding (PTB) domains and EVH1 domains that bind proline-rich sequences Table 10.1 Interaction domains involved in signal transduction 14-3-3 Phosphoserine/phosphothreonine signal transduction, subcellular 14-3-3 Vesicle trafficking, membrane dynamics, and receptor signaling neurobeachin BH1–BH4 dimerization apoptosis Bcl2 Bir repeat domain, caspases apoptosis XiaP BrCt Phosphoserine/phosphothreonine dna damage response and cell cycle BrCa1 Bromo acetyl-lysine Chromatin regulation Gcn5p BtB/PoZ Homodimerization and heterodimerization Chromatin regulation and protein Mel26 C1 diacylglycerol or phorbol ester Plasma membrane recruitment c-raf and vesicle trafficking deP Membranes, G-protein-coupled receptors signal transduction, protein targeting, and dsh eH Peptides with core nPF motifs endocytosis and vesicle trafficking eps15 eF-hand Calcium Calcium signaling Calmodulin Clathrin-dependent endocytosis and cytoskeletal regulation eVH1 Proline-rich sequences Cytoskeletal regulation, postsynaptic epsin Mena ubiquitin ligase substrates (phosphoserine/phosphothreonine) ubiquitylation Cdc4 FCH actin, microtubules Cytoskeletal regulation Fes FerM Phospholipids Cytoskeletal regulation and membrane PtlP1 FF Phosphoserine/phosphothreonine transcription, splicing Ca150 FH2 actin, homotypic interactions Cytoskeletal regulation mdia FHa Phosphoserine/phosphothreonine dna repair, signal transduction, vesicular trafficking, protein degradation MdC1 FYVe Phospholipids signal transduction, vesicular trafficking Hrs Gat ubiquitin (and other binding partners) Vesicle trafficking and protein sorting GGa1 Gel actin Cytoskeletal regulation Gelsolin GK Phosphoserine/phosphothreonine scaffolding Psd-95 Glue Phospholipids Vesicular trafficking Vps36 GraM Phospholipids Vesicular trafficking MtM1 GriP arf/arl family of small GtPases Golgi targeting Golgin-97 GYF Proline-rich sequences signal transduction, splicing CdBP2 Heat repeat domain, diverse binding partners Vesicular trafficking, protein translation importin β1 ubiquitin, e2 ubiquitin-conjugating enzymes ubiquitylation e6aP iQ Calmodulin Calcium signaling ras-GrF diverse binding partners, other liM domains diverse including gene expression and cytoskeleton organization hCrP (continued) Table 10.1 Continued Domain Binding target Cellular processes Example protein lrr repeat domain, diverse binding partners diverse rna1p MBt repeat domain, methyl-lysine Chromatin regulation CGi-72 MH1 dna and transcription factors transcription sMad2 MH2 Phosphoserine, homo-oligomerization signal transduction sMad2 Miu ubiquitin Vesicular trafficking rnF168 nZF ubiquitin ubiquitin-dependent processes ranBP2 Pas diverse binding partners signal sensor domain, detecting oxygen tension, redox potential, or light intensity PasK PB1 Heterodimerization signal transduction p67phox PdZ C-terminal peptide motifs diverse, scaffolding Psd-95 PH Phospholipids Membrane recruitment, vesicular traf- ficking, signal transduction, cytoskeletal akt regulation Polo-box Phosphoserine/phosphothreonine Cell cycle Plk1 PtB Phosphotyrosine tyrosine kinase signaling shc Pumilio repeat domain, rna Gene expression Pumilio PWWP Methyl-lysine, dna dna methylation, dna repair, transcription WHsC1 Protein sorting, vesicular trafficking, phospholipid metabolism p40phox rGs GtP-binding pocket of Gα proteins signal transduction rGs-4 ubiquitin, e2 ubiquitin-conjugating enzymes, transcription factors diverse, ubiquitylation, transcription Cbl saM Homotypic and heterotypic oligomeriza- tion, rna diverse ste11 sH2 Phosphotyrosine tyrosine kinase signaling src sH3 Proline-rich sequences diverse, cytoskeletal regulation src snare Components of snare complex Vesicle-membrane fusion syntaxin soCs box ubiquitin ligase substrates ubiquitylation socs-1 sPrY diverse binding partners diverse, including cytokine signaling, retroviral defense ranBPM start lipids lipid transport, transcription star sWirM acetyl-lysine Chromatin regulation and gene expression sMarC2 tir Homotypic and heterotypic interactions Cytokine and immune signaling tlr4 tPr repeat domain, diverse binding partners scaffolding function in diverse processes p67phox traF Components of tnF signaling pathways Cell survival, protein processing, and ubiquitylation traF-1 tuB dna and phospholipids Metabolism, transcription tulp-1 tudor Methyl-lysine and methyl-arginine Chromatin regulation and gene expression sMn uBa ubiquitin ubiquitylation HHr23a ueV ubiquitin and Pro-thr/ser-ala-Pro peptide Protein sorting tsG101 uiM ubiquitin ubiquitylation Hrs VHl Hydroxy-proline ubiquitylation VHl VHs ubiquitin endocytosis and protein sorting GGa repeat domain, phosphoserine/ phosphothreonine, dimethyl-lysine, others Proline-rich sequences, phosphoserine/ phosphothreonine diverse including cell cycle, ubiquitylation βtrCP diverse signaling processes YaP (Figure 10.4). The PH domain fold therefore represents a common structural framework with versatile binding properties. Conversely, a variety of domains with entirely distinct sequences and structures can converge on very similar biochemical activities. For example, a number of different domain types bind phospholipids such as phosphoinositides, and several distinct classes of domains, using different catalytic mecha- nisms, have protein phosphatase activity. Below, we will discuss three major classes of modular interaction domains: those that recognize peptides or proteins with post-translational Figure 10.4 Domains with the same structure can recognize different ligands. EVH1, PH, and PTB domains (green) have the same overall fold (as indicated by the very similar arrangement of α helices and β sheets in the three domains), but have entirely different binding properties and use different surfaces to engage their ligands (pink). (a) The EVH1 domain from the actin regulatory protein Mena is shown bound to a proline-rich peptide; the PH domain from the ArfgEF grP1 interacts with a phosphoinositol lipid head group; and (c) the PTB domain from the scaffold protein irS1 binds to a tyrosine-phosphorylated peptide. modifications, those that recognize specific unmodified peptide or protein motifs, and those that recognize specific phospholipid species. Signaling proteins also employ modular domains that recognize small-molecule sig- naling mediators, such as calcium and cAMP, but these will not be dis- cussed further here. inTErACTion doMAinS THAT rECognizE PoST-TrAnSlATionAl ModiFiCATionS Post-translational modifications, such as phosphorylation of a serine, threonine, or tyrosine residue, can produce binding sites for interaction domains. As first discussed in Chapter 4, the recognition of modified resi- dues provides a relatively simple molecular device through which the cell can respond to the activities of signaling enzymes such as protein kinases. In the case of protein phosphorylation, the post-translational modification is “written” by a kinase and “erased” by a protein phosphatase. A modu- lar interaction domain that recognizes the modified site essentially func- tions as a “reader” module that interprets the modification and causes downstream changes in function. In addition to phosphorylation, modular interaction domains recognize other types of post-translational modifica- tions, including acetyl- or methyl-lysine, methyl-arginine, hydroxy-pro- line, and ubiquitylated or sumoylated lysine (Figure 10.5). If the interaction domain and the modified peptide site are on two sepa- rate polypeptide chains, then the binding event will induce formation of a complex between the two proteins. Alternatively, the modified site and interaction domain may be in the same protein, in which case the modifi- cation will result in an intramolecular interaction. Such an intramolecu- lar association will suppress the ability of the interaction domain to bind sites on other proteins. It may also induce a conformational change that alters the activity of other domains within the same protein, as we have seen for regulation of the Src tyrosine kinase, in which the interaction of the SH2 domain with a phosphotyrosine site in the C-terminal tail leads to inhibition of the intervening kinase domain. SH2 domains bind phosphotyrosine-containing sites The Src Homology 2 (SH2) domain was the first modular interaction domain whose binding was shown to be dependent on post-translational modification. This domain was first recognized in protein tyrosine kinases including Src, the product of a viral oncogene that causes sarcomas in chickens. SH2 domains are found in over a hundred different human pro- teins, and in almost all cases are thought to bind specifically to tyrosine- phosphorylated peptides. SH2 domains are approximately 100 amino acids in length and fold into a compact structure with a central β sheet that separates the (a) (b) (c) EVH1 domain - proline-rich peptide PH domain - phospholipid PTB - pTyr peptide The regulation of Src family kinas- es by intramolecular interactions is discussed in Chapters 1 and 3 pTyr (Shc) ε-N-Me-Lys (Histone H3) modified peptide motif ε-N-Ac-Lys (Histone H4) ubiquitin OH-Pro (HIF-1α) SH2 (Grb2) Figure 10.5 Chromodomain (HP1) Bromodomain (Gcn5) interaction domain UIM (Vps27) VHL-β Examples of domains that recognize modified peptide motifs. The SH2 domain of grb2 recognizes a phosphorylated tyrosine residue on Shc. The chromodomain of HP1 recognizes a methylated lysine residue on histone H3, whereas the bromodomain of gcn5 recognizes an acetylated lysine residue on histone H4. ubiquitin is bound by the ubiquitin interacting moiety (uiM) of Vps27. The Von Hippel-lindau β protein (VHl-β) recognizes a hydroxylated proline residue on HiF-1α. in each case, the post-translational modification is highlighted in pink. note that different examples are not shown at the same scale. (Adapted from B.T. Seet et al., Nat. R e v . M o l . C e ll B i o l . 7:473–483, 2006. With permission from Macmillan Publishers ltd.) (a) (b) specificity pocket conserved pTyr pocket SH2 domain domain into two binding pockets (Figure 10.6). One of these is highly conserved among all SH2 domains and serves primarily to bind the phosphotyrosine, while the other—the “specificity pocket”—is more variable and binds the side chains of the adjacent amino acids. Thus, the affinity with which a particular peptide binds a given SH2 domain depends, in part, on its phosphorylation (about half of the binding ener- gy comes from recognition of the phosphotyrosine) and also on the fit of other peptide residues with the specificity pocket of the SH2 domain. Individual SH2 domains typically bind their preferred phosphorylated peptide motifs with a Kd of ~1 μM, although in some cases the inter- actions can be tighter (~100 nM). However, they also show relatively rapid on- and off-rates, indicating that SH2 domain–phosphopeptide interactions are highly dynamic, allowing signaling to be rapidly initi- ated and terminated. Binding to the phosphotyrosine is through a bidentate ionic interac- tion (such an interaction is illustrated in Figure 3.11b) with an essen- tial arginine residue (the fifth residue in β strand B, and hence termed ArgβB5). ArgβB5 sits at the base of the relatively deep pocket, and is pre- cisely positioned to interact with the phosphate group of a phosphotyro- sine. The shorter side chains of phosphoserine or phosphothreonine would be unable to project far enough into the pocket to encounter this arginine, explaining the selectivity of SH2 domains for phosphotyrosine. In the absence of tyrosine phosphorylation, the interaction between a peptide and an SH2 domain is typically very weak (Kd in the mM range), and so Figure 10.6 Structure of the SH2 domain of Src. (a) The central β sheet divides the domain into two binding pockets: a highly conserved pocket that is responsible for binding to the phosphotyrosine, and a more variable “specificity” pocket that interacts with the side chains of adjacent amino acids. Tyrosine-phosphorylated peptide ligand is depicted in yellow; phosphate group in pink. (b) Schematic of an SH2 domain; the presence of distinct phosphotyrosine and specificity pockets is reminiscent of a two- holed socket. is not relevant in cells. However, phosphorylation of the peptide leads to a 1000-fold increase in its affinity for the interaction domain. Phosphoryla- tion therefore acts as a switch to elicit complex formation. The sequence of amino acids flanking the phosphotyrosine sites strongly influences which SH2-containing proteins are recruited. Phosphopep- tides typically bind an SH2 domain in an extended conformation, so that they run across the central β sheet (see Figure 10.6), thereby position- ing the C-terminal amino acids to interact with the specificity pocket. Many SH2 domains engage only three C-terminal residues (in the +1 to +3 positions relative to the phosphotyrosine), while others bind more extended phosphorylated motifs up to eight amino acids in length, and can engage residues both N- and C-terminal to the phosphotyrosine. Dif- ferent SH2 domains prefer distinct amino acids in the +1 to +3 positions of the phosphopeptide. For example, the Src SH2 domain can accommo- date an isoleucine at position +3 into a hydrophobic specificity pocket. The Grb2 SH2 domain specifically binds phosphopeptides with aspar- agine in position +2, due to favorable hydrogen-bonding and the β-turn conformation favored by asparagine. The SH2 domains of the p85 adap- tor subunit of phosphatidylinositol 3-kinase (PI3K) prefer a methionine at the +3 position, and those of phospholipase C-γ (PLC-γ) prefer a run of hydrophobic amino acids following the phosphotyrosine (Figure 10.7). Figure 10.7 SH2 domain selectivity. (a) The SH2 domain of the Src tyrosine kinase binds to a phosphopeptide with the sequence pYEEi, which is accommodated because the isoleucine in the +3 position fits into a hydrophobic pocket in the variable surface of the SH2 domain. (b) The SH2 domain of the adaptor protein grb2 selects phosphotyrosine sites with an asparagine at the +2 position; the bulky tryptophan (pink) in the SH2 domains forces the binding peptide to form a β-turn structure rather than adopting an extended conformation. The phospholipase C-γ (PlC-γ) SH2 domain prefers peptides with hydrophobic residues following the phosphotyrosine, particularly an isoleucine at +1. The positively charged phosphotyrosine pocket is indicated in blue; peptide ligands are in yellow. Figure 10.8 Interaction domains can be joined to form new structural folds with novel properties. in Cbl, the SH2 domain (blue), (a) (b) Four-helix bundle R(p-6) four-helix bundle (green), and EF-hand (yellow) domains form a combined structural unit termed the TKB domain, which interacts with phosphorylated tyrosine residues on receptors. A tyrosine-phosphorylated peptide from the endocytic adaptor APS is depicted in pink. Shown are (a) a backbone structure and (b) a surface representation of the TKB domain. (Adapted from J. Hu and S.r. Hubbard, J . B i o l . C h e m. 280:18943– 18949, 2005.) N C SH2 domain C N EF-hand domain V(P-4) N(P-2) Q(P-1) pY S(P+1) A(P-5) E(P-3) F(P+2) The preference of a given domain for particular flanking residues has been investigated by probing degenerate peptide libraries in which phosphotyrosine is flanked by random combinations of amino acids. The consensus binding motif suggested by such an analysis can then be com- pared with experimentally determined phosphorylation sites and used to predict potential binding sites. T cell signaling is discussed in more detail in Chapter 12 Some SH2 domains are elements of larger binding structures There are some cases in which SH2 domains work with other adjacent domains to display somewhat atypical recognition properties. For exam- ple, in the ubiquitin E3 ligase Cbl, the SH2 domain is followed by a four- helix bundle and an EF-hand domain. These three domains actually fold together into an integrated structural unit, sometimes called the TKB domain (Figure 10.8). This larger domain binds phosphotyrosine through its SH2 subunit, but can also engage residues unusually distant from the phosphorylated sites. For example, peptide residues in the –5 and –6 posi- tions, relative to the phosphotyrosine, are recognized by the four-helix bundle subunit. Thus, the phosphopeptide-binding properties of the Cbl SH2 domain are altered and rendered more complex. Another example of an interaction domain with novel recognition prop- erties is the tandem SH2 module from the ZAP-70 tyrosine kinase, which functions in immune cell activation. ZAP-70 has two N-terminal SH2 domains that function cooperatively to engage an unusual dou- bly tyrosine-phosphorylated sequence (the immunoreceptor tyrosine- based activation motif, or ITAM) found in the signaling subunits of the T cell antigen receptor and other immune receptors. This interaction requires that the two SH2 domains be tightly coupled, because residues from the C-terminal SH2 domain contribute to the phosphotyrosine- binding pocket of the N-terminal SH2 domain, which is therefore non- functional unless the two SH2 domains are correctly oriented in space (Figure 10.9). Similar types of modified recognition properties have been observed in tandem domains that belong to other families besides SH2 domains. Several different types of interaction domains recognize phosphotyrosine Other types of domains with completely different structural folds can also bind tyrosine-phosphorylated peptide motifs. The interaction of PTB domains with phosphotyrosine-containing sites differs in several features (b) ITAM binding N-SH2 Figure 10.9 Domains can act cooperatively to regulate signaling. (a) Binding of MHC-bound peptide antigen (orange dot) to the T cell receptor (TCr, li g h t b r o w n) results in the phosphorylation of iTAMs (immunoreceptor tyrosine-based activation motifs) on the TCr. The tandem SH2 domains of zAP-70 can then interact with a doubly phosphorylated iTAM on the TCr. This activates the kinase domain of zAP-70, allowing it to phosphorylate substrates, such as the adaptor protein lAT (not shown). (b) Further details on zAP-70 activation. in the inactive state (left), interactions between the inter-SH2 region, linker sequence, and kinase domain serve to inhibit kinase activity and also keep the SH2 domains separated, preventing them from engaging phosphorylated tyrosines. upon iTAM phosphorylation, the SH2 domains orient themselves to dock with the phosphorylated motif. This releases the intramolecular interactions, allowing the kinase to be activated, and also positions the kinase in the correct orientation for phosphorylation of substrates. This extended conformation of zAP- 70 is further stabilized by phosphorylation of tyrosine residues (Tyr315 and Tyr319) within the linker region. from the interaction involving SH2 domains. First, PTB domains have a completely different fold from SH2 domains, with two orthogonal β sheets forming a β sandwich that is capped by a C-terminal α helix (see Figure 10.4c). As a consequence, the mode of peptide recognition is dif- ferent: the peptide ligand contacts the β-5 strand and C-terminal α helix of the PTB domain, and, in effect, adds an antiparallel strand to one of the β sheets. N-terminal to phosphotyrosine, the peptide forms a type 1 β turn, which is favored by the peptide motif NPxY (where “x” can be any amino acid); this element anchors the peptide and is the hallmark of most ligands for PTB domains. Stable peptide binding to the PTB domain of the scaffolding protein Shc depends on phosphorylation of the tyrosine in the NPxY motif. The bound phosphotyrosine is coordinated by three basic residues (two arginines and a lysine) of the PTB domain, which form a network of hydrogen bonds with the phosphate group. Unlike SH2 domains, many PTB domains bind with high affinity to unphosphorylated peptides (often with the NPxY consensus). Therefore it is likely that phos- photyrosine recognition evolved relatively late from an existing peptide- binding module. The PTB domains of Shc and other scaffolding proteins such as IRS1, Dok1, and FRS2 recognize phosphorylated NPxY motifs, and are fol- lowed by an unstructured region with several sites for tyrosine phos- phorylation. Autophosphorylated receptor tyrosine kinases recruit these proteins via their PTB domains, and then phosphorylate them on multiple tyrosine motifs, which consequently bind the SH2 domains Figure 10.10 Binding of one interaction domain can provide further targets for a catalytic domain. The PTB domain of irS1 docks to a phosphorylated tyrosine on the insulin receptor. The insulin receptor kinase then phosphorylates multiple tyrosine residues on irS1, which in turn form docking sites for specific SH2 domain-containing proteins, such as phosphatidylinositol 3-kinase (Pi3K), the adaptor grb2, and the tyrosine phosphatase Shp2. The binding of these SH2 domain- containing proteins leads to the activation of corresponding downstream signaling pathways. Figure 10.11 Structure of 14-3-3 proteins. The structure of the 14-3-3ζ dimer bound to the phosphopeptide rSHpSYPA. The individual 14-3-3 monomers are shown in green and blue and the phosphopeptide in yellow, with phosphates highlighted in pink. of other regulatory signaling proteins (Figure 10.10). PTB domain- containing proteins therefore extend and amplify a receptor’s ability to recruit cytoplasmic targets. The C2 domain is another interaction domain that can, in at least one case, recognize phosphotyrosine sites. Most C2 domains bind to phos- pholipids. However, the C2 domain of PKCδ, a protein serine/threonine kinase, binds a specific phosphotyrosine-containing motif in the cytoplas- mic tail of a transmembrane protein (CDCP1), again through a distinct structural mechanism. This directly links a phosphotyrosine signal, gen- erated by Src-mediated phosphorylation of CDCP1, to serine/threonine phosphorylation. Thus at least three separate families of interaction domains—SH2, PTB, and C2 domains—have independently converged on the selective recognition of phosphotyrosine sites. Multiple domains recognize motifs phosphorylated on serine/threonine The recognition of phosphorylated sites by specific interaction domains is a common feature of signaling pathways, and phosphoamino acids other than phosphotyrosine are also recognized. Like tyrosine, serine and threonine are hydroxyl amino acids, but they lack tyrosine’s bulky phenolic ring. Consequently, domains that recognize phosphotyrosine generally do not have the correct shape to accommodate the much small- er phosphoserine or phosphothreonine side chains. Serine and threonine are chemically and structurally very similar, however, so there are many domains that specifically bind to both phosphoserine and phosphothreo- nine motifs. At least ten different domain types, including FHA, WW, BRCT, Polo- box, MH2, and WD40 domains, can bind phosphothreonine- and/ or phosphoserine-containing motifs, and are found in the context of multidomain proteins. There are many more types of phosphoserine/ phosphothreonine-binding domains than phosphotyrosine-binding mod- ules, which may reflect the fact that phosphorylated serine/threonine residues are much more abundant in mammalian cells than phosphoty- rosine (the ratio is around 100:1). These modules bind selectively to pep- tide motifs in which a threonine or serine is phosphorylated (often more strongly to phosphothreonine), and in which the phosphorylated residue is flanked by a particular set of amino acids that are preferentially accom- modated by the domain’s ligand-binding site. 14-3-3 proteins recognize specific phosphoserine/ phosphothreonine motifs The binding of the 14-3-3 family of proteins to phosphorylated sites rep- resents a major mechanism through which cells respond to the effects of basophilic kinases (kinases that phosphorylate serine/threonine sites with nearby basic residues). The 14-3-3 proteins bind serine/threonine phosphorylated motifs with the general consensus RSxpSxP, RxY/FxpSxP, or SWpTx (the latter being a C-terminal sequence). Mammalian 14-3-3 proteins are very abundant and can bind at least 200 phosphorylated pro- teins involved in many different aspects of cellular regulation, including signaling pathways activated by cell-surface receptors, cell-cycle progres- sion, apoptosis, transcriptional control, cytoskeletal organization, metabo- lism, and protein trafficking. Each ~30 kD 14-3-3 polypeptide has nine α helices that form an amphipathic peptide-binding channel (Figure 10.11) that has a con- served, positively charged basic pocket composed of a lysine and two arginines, which directly bind the phosphate group. Unlike other phos- phopeptide-binding modules, 14-3-3 proteins are never found as compo- nents of larger proteins with additional domains, but they do interact noncovalently with one another to form homodimers or heterodimers. Each 14-3-3 dimer therefore has two phosphoserine/phosphothreonine- binding pockets and can interact simultaneously with two different phosphorylated sites; typically, these are located on the same polypep- tide chain, although a 14-3-3 dimer might potentially link two distinct phosphorylated proteins. Binding to a 14-3-3 dimer can modify the function of a phosphorylated protein in one of several ways. One consequence of 14-3-3 binding can be to interfere with the ability of a phosphorylated protein to bind another protein. For example, in the absence of survival signals, the pro-apoptotic protein BAD associates with the pro-survival protein Bcl-XL, inhibiting the survival effects of Bcl-XL and thus inducing cell death. Extracellu- lar stimuli that promote cell survival induce the phosphorylation of BAD on serine residues that consequently bind 14-3-3 dimers, dislodging BAD from Bcl-XL. Thus, when Bcl-XL is in a complex with BAD, its anti-apoptotic activity is suppressed; 14-3-3 binding to BAD relieves this inhibition and promotes cell survival. Association with 14-3-3 proteins can also alter the subcellular location of a phosphorylated protein, typically by anchoring it in the cytoplasm by blocking its nuclear localization signal. For example, association with 14-3-3 proteins can restrict the FOXO transcription factor to the cyto- plasm. FOXO normally induces the expression of genes that antago- nize cell-cycle progression and induce apoptosis. Thus, preventing its entry into the nucleus leads to increased cell proliferation and survival (Figure 10.12). Interaction domains recognize acetylated and methylated sites Phosphorylation is but one of several post-translational modifications that are selectively recognized by specific interaction domains. Lysine residues can be methylated or acetylated on the flexible N- or C-terminal tails of histones, for example, leading to changes in chromatin organiza- tion and the epigenetic control of gene expression. An ever-increasing number of non-chromatin proteins have also been shown to be subject to lysine methylation and/or acetylation. As with phosphopeptide rec- ognition, peptide motifs with an acetylated lysine flanked by a defined peptide sequence can be selectively recognized by particular domains, of which bromodomains are the cardinal example (see Figure 10.5). These are frequently components of proteins involved in chromatin remodeling, such as the histone acetyl transferases themselves, and are therefore intimately involved in controlling gene expression. Chro- modomains recognize specific peptide motifs in which the lysine is methylated, rather than acetylated. Chromodomains occur in proteins such as the heterochromatin protein 1 (HP1), which binds to histone H3 The role of Bcl2 family proteins in triggering programmed cell death is discussed in Chapter 9 The role of methylation and acetylation in regulating chromatin structure is discussed in Chapter 4 Figure 10.12 14-3-3 proteins can regulate subcellular localization. The FoXo transcription factor regulates the expression of genes leading to cell-cycle arrest and apoptosis. The serine/threonine kinase Akt phosphorylates multiple sites on FoXo, which form docking sites for 14-3-3 proteins. once bound to 14-3-3, FoXo is sequestered in the cytoplasm. This promotes cell survival and proliferation. (a) chromo tudor WD40 bromo histone H3 H ₂ N-A R T K Q T A R K S T G G K A P R K Q L A T K A A R K S N-terminal tail histone fold COOH Figure 10.13 Domains that bind modified histone residues. (a) A number of different interaction domains, including chromo, tudor, Wd40, and bromo domains, bind to histone lysine or arginine residues that have been methylated or acetylated. The binding of these domains regulates chromatin structure and gene expression. (b) Structure of the chromo domain from Drosophila HP1, showing the aromatic residues (green) that form a “cage” around the dimethylated lysine of the histone H3 tail peptide (pink). (a, Adapted from B.T. Seet et al., Nat. Rev. Mol. Cell Biol. 7:473–483, 2006. With permission from Macmillan Publishers ltd; b, adapted from A. Brehm et al., Bioessays 26:133–140, 2004. With permission from John Wiley & Sons, inc.) Ubiquitylation and its consequenc- es are discussed in more detail in Chapters 4 and 9 methylated at Lys9 and modifies chromatin structure to repress gene expression (Figure 10.13). Just as different types of interaction domains recognize phosphorylated sites, there are several different types of interaction domains that bind methylated lysines, including subsets of WD40, tudor, malignant brain tumor (MBT), and PHD finger domains. These different domains can use distinct structural mechanisms for binding to methylated lysines. Lysine methylation is more complex than protein phosphorylation, as a lysine residue can be mono-, di- or trimethylated. Interaction domains can bind preferentially to lysine residues carrying different degrees of methylation, increasing the potential sophistication of this type of bind- ing interaction. For example, chromodomains and tudor domains employ conserved aromatic residues to surround the ζ-methyl groups of a meth- ylated lysine residue with a hydrophobic cage (Figure 10.13b). In con- trast, the WD40 repeat domain of WDR5, a subunit of histone H3 Lys4 methyltransferase, preferentially forms hydrogen bonds to dimethylated Lys4 in histone H3; this interaction also depends strongly on recognition of an arginine residue at the –2 position relative to the modified lysine. Ubiquitylation regulates protein–protein interactions Lysine residues can also be modified by addition of the 76-amino-acid pro- tein ubiquitin, via an isopeptide linkage between the ε-amino group of lysine and the C-terminus of ubiquitin. Furthermore, ubiquitin can itself be ubiquitylated on one of its own lysines (for example, Lys48 or Lys63) or its N-terminus to form a polyubiquitin chain. Ubiquitin is in essence a transferable interaction domain that, once linked to its target protein, is recognized by binding modules (collectively termed “ubiquitin-binding domains” or UBDs). There are at least 11 structurally distinct families of UBDs, although these mostly bind the same hydrophobic patch on ubiq- uitin, centered on Ile44 (Figure 10.14). Although UBDs are referred to as “domains,” many of these are more akin to linear peptide motifs that engage the ligand-binding surface of ubiquitin. Ubiquitylation regulates a range of cellular processes, including pro- teasome-mediated proteolysis, protein trafficking to endosomes, post- replicative DNA repair, and signaling downstream of receptors leading to the activation of kinases that control the NF-κB transcription factor. In each of these pathways, the receptors for ubiquitylated proteins have one or more UBD. Since the affinity of ubiquitin for a UBD is typically rather weak, it is likely that multivalent interactions are important for the association of ubiquitylated proteins and their binding partners. Furthermore, there is often a close interplay between protein–protein interactions mediated by phosphorylation and ubiquitylation, in the sense that the substrate-binding domains of E3 protein-ubiquitin ligases often bind their target proteins in a phosphorylation-dependent manner. As a result, the target is only ubiquitylated after it has been phosphor- ylated (see Figure 4.8). inTErACTion doMAinS THAT rECognizE unModiFiEd PEPTidE MoTiFS or ProTEinS Thus far, we have discussed interaction domains that bind short pep- tide sequences that have been post-translationally modified. There are also a number of domain families that recognize unmodified peptide lig- ands. Here we focus on two domains of this sort—SH3 domains and PDZ domains—and their respective peptide ligands. Both of these domains are found in ~300 copies in the human proteome, making them among the most commonly used modules in signaling proteins. We will also consider domains that interact with each other to form homo- or heterodimers or larger oligomeric structures. Proline-rich sequences are favorable recognition motifs A number of interaction domains, including SH3, WW, EVH1, and GYF, bind short, proline-rich peptide motifs. Proline is unique among naturally occurring amino acids in that its side chain is fused to the nitrogen of the peptide backbone, forming a five-member ring. Due to the resulting con- formational constraints, proline-rich sequences tend to form a left-hand- ed helix with three residues per turn (a polyproline type II or PPII helix). In a PPII helix, both the side chains and the backbone carbonyls project outward from the axis of the helix, and so are available to contact an interaction domain (Figure 10.15a). A number of properties of the PPII helix make it ideal for mediating protein interactions. As opposed to more flexible peptides that are only locked into a single conformation upon binding, the PPII helix forms spontaneously; this reduces the entropic penalty involved in binding. Other non-proline resi- dues can be incorporated without disrupting the helix, and these can con- tribute to selective binding. In addition, a PPII helix has twofold rotational pseudosymmetry, meaning it has the potential to bind in either orientation Figure 10.14 Ubiquitin-binding domains. (a) Most ubiquitin-binding domains (blue), including the uiM, Miu, and uBA domains shown here, recognize the hydrophobic patch centered on ile44 of ubiquitin (colored green on the crystal structures). (b) The specificity of ubiquitin-binding domains depends on the ubiquitin chain length and linkage. For example, the K63-linked di-ubiquitin (top) has a more extended structure than the K48-linked di-ubiquitin (below). lysines are depicted in space-filled mode. (a, adapted from J.H. Hurley et al., Biochem. J. 399:361–372, 2006. With permission of Portland Press; b, adapted from K. newton et al., Cell 134:668–678, 2008. With permission from Elsevier.) Figure 10.15 SH3 domains bind polyproline helices. (a) A Sem-5 SH3 domain (green) is shown bound to a proline-rich peptide (orange). The cartoon below shows the mechanism of polyproline recognition. The core recognition surface of the SH3 domain has two grooves formed by aromatic amino acids (shown in blue) that each accommodate an xP peptide motif. Adjacent to this are two variable loops (rT and n-Src) that make contacts with residues flanking the PxxP motif, forming a “specificity” pocket (green). The gAdS SH3 domain ( g r ee n ) bound to a peptide motif from SlP-76 ( o r a n g e ) that is centered on rxxK, rather than PxxP. (a, Adapted from A. zarrinpar et al., Science STKE 179:re8, 2003. With permission of AAAS; b, adapted from B.T. Seet et al., EMBO J. 26:678–689, 2007. With permission from Macmillan Publishers ltd.) PPII helix SH3 (N- to C-terminal or C- to N-terminal) to a domain such as an SH3 domain. Proline-rich sequences were also likely selected as ligands for interaction domains because they are usually exposed on the surface of a protein, or located in unstructured regions, and so are readily accessible to a binding partner. Furthermore, the proline ring is relatively hydrophobic, which is unusual for side chains exposed on the surface of proteins, so shielding it from the solvent by binding to another protein is energetically favored. SH3 domains bind proline-rich motifs PPII helices that contain the consensus motif PxxP are bound by SH3 domains, which contain two antiparallel β sheets positioned at right angles to one another. The peptide-binding surface of the SH3 domain has two grooves, each of which accommodates one proline residue and an adja- cent amino acid that is usually hydrophobic. Two variable SH3 domain loops (called “RT” and “n-Src” for historic reasons) make numerous con- tacts with residues that flank the PxxP core motif and can be viewed as forming a specificity pocket. Many SH3 domains bind ligands containing the sequences R/KxxPxxP or PxxPxR/K (where R/K denotes either arginine or lysine), which bind in opposing orientations and therefore make simi- lar contacts with SH3 domains (Figure 10.15). In addition, these peptides frequently contain more variable sequences that extend beyond the basic residue and contribute to the specificity of SH3 domain interactions. As we have seen with other interaction modules, SH3 domains are relatively ver- satile in their binding properties, and in some cases bind peptides that lack a PxxP motif; one example is the C-terminal SH3 domain of the Grb2-like adaptor protein GADS, which serves a key role in signaling downstream of the T cell receptor (TCR) by linking two docking proteins, LAT and SLP- 76. The GADS SH3 domain binds with high affinity and specificity to a peptide motif on SLP-76 that is centered on an RxxK motif (Figure 10.15b). PDZ domains recognize C-terminal peptide motifs As with proline-rich sequences, the C-terminus of proteins is usually exposed and contains unique chemical features. These properties have made it a preferred site for recognition by a specific class of interaction modules termed PDZ domains. These domains have a β-sheet structure with a carboxylate-binding loop, which forms a pocket for the side chains of hydrophobic C-terminal residues such as valine (Figure 10.16). Indeed, the great majority of interactions mediated by PDZ domains involves the recognition of such C-terminal sequences, and only a few PDZ domains recognize motifs located within a protein. The key distinguishing feature of a PDZ domain-binding site is therefore the C-terminal hydrophobic res- idue, which conceptually has the same defining role as phosphotyrosine in binding to an SH2 domain, or a PxxP motif in recognition by an SH3 domain. As with these other domain–peptide interactions, the residues adjacent to the core feature, in this case residues located N-terminal to the C-terminal amino acid, provide a degree of specificity in determining which PDZ domains are recruited to a particular motif. The amino acid at the –2 position (that is, two amino acids from the C-terminus, which is denoted as the “0” position) is especially important, but significant contri- butions can be made by residues more distant from the C-terminus. Many transmembrane receptors, such as receptor tyrosine kinases, G-protein-coupled receptors, ion channels, and adhesion proteins, have C-terminal PDZ-binding motifs, as do many cytoplasmic proteins involved in functions such as cell polarity and the regulation of Rho fam- ily GTPases. Furthermore, PDZ proteins frequently have multiple tan- dem PDZ domains (up to 13 in the case of the MUPP1 protein), and can therefore simultaneously bind several different proteins with appropri- ate C-terminal motifs. PDZ domain proteins therefore commonly serve as scaffolds to co-localize transmembrane receptors and cytoplasmic signal- ing proteins at a specific site within the cell. Protein interaction domains can form dimers or oligomers Most protein interactions discussed above involve a folded protein domain interacting with a peptide ligand. By contrast, some protein domains medi- ate dimerization or oligomerization as part of their role in the assembly of functional signaling complexes. Dimerization can occur between two identical proteins (homodimerization) or two distinct domains from the same family (heterodimerization), which can bring two different proteins into a common complex to regulate a signaling pathway. Higher-order oligomeric complexes are also possible. We have already introduced how dimerization of 14-3-3 proteins plays an important role in their ability to bind to multiply phosphorylated partners (see Figure 10.11). The SAM (sterile alpha motif) domain is an example of a module that can form a head-to-tail dimer as well as self-associate into an extended oligomer. SAM domains are found in a variety of proteins ranging from receptor tyrosine kinases and cytoplasmic signaling proteins to transcrip- tion factors and polypeptides that regulate chromatin. SAM domains usu- ally interact through two distinct surfaces, termed EH (for end-helix) and ML (for mid-loop); the EH surface of one domain interacts with the ML surface of a second domain (Figure 10.17). In addition to forming dimers, a SAM domain with mutually compatible EH and ML sites can assemble into long polymers, as has been observed for the human transcription fac- tor TEL, a member of the Ets family that mediates transcriptional repres- sion. A similar open-ended polymer is formed by the SAM domain of the Drosophila protein polyhomeotic, a member of the polycomb group of pro- teins that maintain chromatin in a transcriptionally repressed state. In both cases, the polymers formed by SAM domains may transmit a signal for transcriptional repression over an extended region of chromatin. This type of domain-mediated oligomerization can also be dynamical- ly regulated, as in the case of the Drosophila SAM domain-containing Figure 10.16 PDZ domains bind C-terminal motifs. The Pdz domain (green) of Erbin bound to its optimal ligand (yellow), WETWV-CooH. The C-terminal hydrophobic residue (valine in this case, with the terminal carboxylate group shown in pink, numbered 0) is the key distinguishing feature of Pdz ligands. Adjacent residues (tryptophan and threonine in this example) provide a degree of specificity. The amino acid at the –2 position (threonine) is of particular importance. (Courtesy of Megan Mclaughlin and Sachdev Sidhu, university of Toronto.) (a) (b) Figure 10.17 SAM domains can form polymers. SAM domains have distinct (a) EH and (b) Ml surfaces, which can (c) dimerize in a “head-to-tail” fashion and also form longer oligomers. residues implicated in dimer–dimer interactions are colored pink. (Adapted from J.J. Kwan et al., J. Mol. Biol. 342:681–693, 2004. With permission from Elsevier.) Figure 10.18 Multimerization of SAM domains regulates the transcription factor Yan. in the absence of receptor tyrosine kinase activation, the majority of Yan molecules (green) form a SAM domain- mediated homopolymer, which has a strong affinity for dnA. dnA binding of Yan prevents the binding of transcriptional activators, resulting in the repression of target genes. A small amount of Yan forms a SAM–SAM heterodimer with the regulatory protein Mae (orange). receptor tyrosine kinase stimulation leads to the activation of MAP kinase (MAPK), which can dock with Mae and phosphorylate Yan. Phosphorylated Yan is exported to the cytoplasm. To maintain the equilibrium, Yan polymers dissociate from the dnA, allowing transcription to occur. Since Mae expression is regulated by Yan, a positive feedback cycle is initiated, ensuring full de-repression of target genes upon MAPK activation. (Adapted from F. Qiao et al., Cell 118:163– 173, 2004. With permission from Elsevier.) Membrane lipids and their role in signaling are discussed in detail in Chapter 7 transcriptional repressor Yan. Receptor tyrosine kinase activation pro- motes the heterodimerization of Yan with a second SAM domain protein, Mae. Binding to Mae leads to the depolymerization of Yan associated with chromatin, its phosphorylation, and its export from the nucleus. In this way, relief of transcriptional repression can be coupled to extracellular signals during development (Figure 10.18). inTErACTion doMAinS THAT rECognizE PHoSPHoliPidS We have discussed how extracellular signals can induce the dynamic post- translational modification of intracellular signaling proteins, and that these modifications frequently exert their effects by creating binding sites for the interaction domains of target proteins. However, signaling information can also be carried by non-protein molecules, such as lipids or small molecules, which are modified in response to a stimulus. For example, external signals can induce the phosphorylation of phospholipids that are embedded in mem- branes through their fatty acid side chains and present their phosphorylated head groups to the cytoplasmic face of the membrane. In the same way that interaction domains can recognize phosphorylated sites on proteins, there are a number of interaction domains that selectively bind the head groups of phospholipids, particularly members of the phosphoinositide family. In this section, we will discuss several examples of domains that recognize specific membrane phospholipids, thereby playing a critical role in regulating the interaction of signaling proteins with membranes. PH domains form a major class of phosphoinositide-binding domains Phosphoinositide-binding domains, such as PH domains, can selectively recognize one or more forms of the phosphorylated inositol head group of phosphoinositides. A protein can therefore be localized to a region of the membrane enriched in a phospholipid preferentially recognized by its PH domains. These domains can target a protein to a particular subcel- lular organelle or, by recognizing one of the signaling phosphoinositides, can induce translocation of a protein to a membrane (usually the plasma membrane) following activation of a cell-surface receptor, such as a interaCtion doMains tHat reCoGniZe PHosPHoliPids Figure 10.19 The PLC-δ PH domain is recruited to regions of the membrane rich in PI(4,5)P ₂ . After recruitment to the membrane, the enzymatic activity of PlC-δ then converts Pi(4,5)P₂ to inositol 1,4,5-trisphosphate (iP₃) and diacylglycerol (dAg), which activate protein kinase C (PKC) and calcium signaling pathways. PI(4,5)P₂- containing membrane PI(4,5)P₂ P DAG + P PH EF P P P IP₃ receptor tyrosine kinase or G-protein-coupled receptor. Such phosphoi- nositide-dependent localization therefore positions enzymes or adaptors with appropriate phospholipid-binding domains in the proximity of their PLC-δ PLC C2 targets. The enzyme phospholipase C-δ (PLC-δ), for example, has a PH domain that selectively binds phosphatidylinositol 4,5-bisphosphate [PI(4,5)P₂], which is also the substrate for its catalytic domain. This inter- action therefore concentrates PLC-δ at regions of the membrane that are enriched for its substrate (Figure 10.19). In contrast, the PH domains of the serine/threonine kinase Akt and its activator PDK1 bind selectively to PI(3,4,5)P₃ and PI(3,4)P₂ . PDK1 and Akt are cytosolic until PI(3,4,5)P₃ levels in the plasma membrane rise—in response to an external signal that activates PI 3-kinase—providing docking sites for their PH domains. Once at the membrane, PDK1 phosphorylates and activates Akt, which has numerous substrates involved in the control of cell growth and survival. As mentioned earlier in this chapter, PH domains have the same overall fold as PTB domains, consisting of two orthogonal β sheets capped by a C-terminal α helix (see Figure 10.4). The phosphoinositide-binding pocket of PH domains is on the opposite surface from the α helix, and is typi- cally formed by basic residues located on the loop between the first two β strands, which can make up to 19 hydrogen bonds with the phosphates of the inositol ring. As a consequence, the affinity with which phospho- inositides bind PH domains can be relatively high (Kd as low as 30 nM). Subtle differences in the binding pocket can alter the specificity of phospho- inositide recognition, and some PH domains use a second, distinct bind- ing pocket adjacent to the canonical site to engage phosphoinositides or phosphorylated sphingolipids. It should be noted that the majority of PH domains identified on the basis of sequence similarity cannot be shown to bind with high affinity to phosphoinositides, leaving open the possibility that they bind to other, currently unknown targets. FYVE domains are phospholipid-binding domains found in endocytic proteins Just as multiple different classes of domains bind phosphopeptide sites in proteins, there are several distinct types of interaction domains that bind phosphoinositides in membranes. One of these is the FYVE domain, which is stabilized by two zinc-binding clusters and has a shallow, positively charged pocket that interacts with phosphatidylinositol 3-phosphate [PI(3)P] on endosomes (Figure 10.20a). Because they make fewer hydro- gen bonds with bound phosphoinositides than do PH domains, the affinity of FYVE domains for soluble forms of PI(3)P is rather weak. However, additional features of FYVE domains greatly strengthen their interac- tions with membranes. First, they insert nonpolar residues into the inte- rior of the membrane, adding binding energy to that provided by selective phosphoinositide recognition. Second, FYVE domain proteins can form dimers and oligomers, increasing their avidity for PI(3)P-rich membranes. Examples of FYVE domain proteins include Hrs, which is localized to endosomes through binding of its FYVE domain to PI(3)P. Hrs also has a protein kinase C and calcium signaling pathways Akt activation is described in Chapter 5 (a) Figure 10.20 Interaction domains that bind phospholipids. (a) A model of the interaction of a FYVE domain homodimer with the membrane. FYVE domains typically bind endosomal membranes enriched in Pi(3)P; the domain is stabilized by the binding of zinc (orange spheres). Basic residues that interact with the phosphorylated lipid head group are shown in blue. note the potential insertion of hydrophobic side chains from the FYVE domain (yellow) into the membrane (approximate position indicated by horizontal line). For crystallization, the water-soluble inositol 1,3-bisphosphate [ins(1,3)P2] head group was used as an analog for the insoluble membrane lipid Pi(3)P. (b) BAr domains interact strongly with curved membranes due to their shape and positively charged binding surface. Binding to a curved membrane increases electrostatic interactions between the domain and the negatively charged membrane, and is energetically favored. BAr domains can thus stabilize and/or promote membrane curvature, for example in the formation of endocytic vesicles. (a, Adapted from M. lemmon, N a t . R e v . M o l . Cell Biol. 9:99–111, 2008. With permission from Macmillan Publishers ltd; b, adapted from H.T. McMahon and J.l. gallop, N a t u r e 438:590–596, 2005. With permission from Macmillan Publishers ltd.) ubiquitin interaction motif and thus it acts as a receptor on endosomes for endocytosed, ubiquitylated receptors. SARA, in contrast, is a scaffold that associates with the TGFβ receptor serine/threonine kinase and R-SMAD, and thereby localizes the activated receptor complex to endosomes through its FYVE domain. This prevents degradation of internalized TGFβ recep- tors and stimulates signaling through the SMAD2/4 pathway. SARA binding also distinguishes signaling-competent TGFβ receptors internal- ized through clathrin-coated pits from those that have been internalized through a caveolin-mediated pathway, which are targeted to lysosomes for degradation (see also Figure 5.15). BAR domains bind and stabilize curved membranes In addition to PH and FYVE domains, many other protein domains bind negatively charged lipids, and these are particularly prevalent in proteins involved in endocytosis and in protein/vesicle trafficking. These domains can have more complex properties. BAR domains, for example, form dimers of coiled-coil sequences that adopt a banana-shaped structure (Figure 10.20b). The concave surface of this dimer is typically positive- ly charged, and thus has some affinity for biological membranes (which tend to have a net negative surface charge due to the phosphate groups of phospholipids). Their curved shape means that they bind with highest affinity (lowest energy) to membranes with a specific degree of curvature, matching the curvature of the domain. BAR domains thus can stabilize or even induce curved membranes, and thereby promote the formation of vesicular or tubular structures of the sort required for endocytosis. CrEATing CoMPlEX FunCTionS BY CoMBining inTErACTion doMAinS Thus far, we have primarily discussed individual binding domains—the minimal sequence necessary to fold independently and recruit the appro- priate ligand. However, it is quite common, through the process of evolution, for distinct domains to recombine and become linked with other domains. In this section, we discuss how such domain combinations can lead to pro- teins and complexes with more sophisticated signaling behaviors. Recombination of domains occurs through evolution Although members of a family of related domains generally share simi- lar biochemical activities, they can be found in proteins containing a wide Src RasGAP STAT SH3 SH2 Tyr kinase SH2 SH3 SH2 PH C2 RasGAP DNA BD SH2 tyrosine phosphorylation Ras inactivation transcriptional activation Figure 10.21 SH2 domains are found in numerous proteins with varied roles in signal transduction. The domain arrangements of a subset of the ~110 human SH2 domain-containing proteins are depicted. The primary signaling activity of each protein is indicated on the right. note that in PlC-γ, both the phospholipase C catalytic domain Vav CH RhoGEF PH C1 SH3 SH2 SH3 cytoskeletal regulation (Rho G protein activation) and a PH domain are split into two halves, indicated by n and c. PLC γ PH PLCn PHn SH2 SH2 SH3 PHc PLCc C2 phospholipid signaling Grb2 Cbl SH3 SH2 SH3 SH2 E3 ubiquitin ligase linking receptors to targets ubiquitination variety of other kinds of functional domains with either binding or catalytic properties. The range of combinations in which a single class of domains is found suggests that evolutionary diversification has occurred by shuffling domains in different ways. Clearly this provides a ready mechanism for gen- erating new signaling behaviors through the repeated re-use and recombi- nation of a limited set of modular functional units of protein structure. We can see this combinatorial diversification by examining the modular domain architectures of SH2 domain-containing proteins. Figure 10.21 illustrates a selection of proteins that contain SH2 domains, and their varied functions. These include regulating the GTP-binding state of small G proteins, phospholipid metabolism, the dynamic reorganization of the actin cytoskeleton, ubiquitylation, protein tyrosine phosphorylation, and transcription. While this figure shows only a small sample of the com- plement of SH2-containing proteins found in humans, the combinatorial diversity of the family is apparent. Similar combinatorial diversity is observed for many other modular domain families. In some proteins, SH2 domains are linked exclusively to other interaction domains, such as SH3 domains that bind proline-rich sequences. These types of proteins act as adaptors that link phosphotyrosine-containing receptors to downstream targets that bind the adaptor’s SH3 domains (see below). SH2 domains can also be directly linked to enzymatic domains. Proteins such as phospholipase C-γ1 have both catalytic domains and sev- eral different interaction domains that confer binding to both peptide and phospholipid ligands; the activities of such proteins are therefore poten- tially regulated by multiple inputs. The cellular response to changes in tyrosine phosphorylation is therefore determined largely by the ability of the resulting phosphotyrosine motifs to associate with specific SH2 domain proteins, and also the downstream pathways that these proteins activate or repress. Combinations of interaction domains or motifs can be used as a scaffold for the assembly of signaling complexes One of the major classes of multidomain signaling proteins is the scaffold proteins, which contain multiple interaction domains or peptide motifs within the same polypeptide. In general, such proteins can be used to Figure 10.22 Flexibility of adaptor protein function. (a) The grb2 adaptor couples activated receptor tyrosine kinases such as the PdgF receptor (PdgFr) to downstream effectors such as the ras activator Sos. (b) grb2 can bind multiple effector proteins with distinct activities through its SH3 (b) (c) pY Grb2 PH PTB YYY PIP Dok domains. recruitment of Sos promotes ras signaling, while recruitment of the ubiquitin ligase Cbl promotes receptor ubiquitylation and down-regulation. (c) Members of the dok family of adaptors contain a PH domain that binds to phosphoinositides, a PTB domain that binds to tyrosine-phosphorylated peptides, and potential tyrosine phosphorylation sites (Y). Ras Sos Ras signaling Cbl pY ubiquitylation Thus dok family adaptors can integrate signals from lipid kinases that generate phosphoinositides such as PiP, and tyrosine kinases that generate phosphotyrosine (pY). coordinate the assembly of multiprotein complexes, either in a static way, or in a way that is dynamically modulated by signaling. One of the simplest examples of creating new functions is illustrated by proteins such as Grb2, which are constructed from two different recog- nition domain types, linked together to form an adaptor protein. Grb2 contains a single SH2 domain flanked by two SH3 domains. The Grb2 SH2 domain specifically recognizes phosphorylated motifs, such as the pYSNA peptide motif that is generated in the activated PDGF receptor, while the SH3 domains from Grb2 recognize specific proline-rich peptide motifs, such as those in Sos, a GEF for the small G protein Ras. Thus, Grb2 functions as an adaptor because it takes a phosphotyrosine-based input and converts it into a specific output: activation of Ras on the plas- ma membrane (Figure 10.22a). Additional outputs are also possible, as the Grb2 SH3 domains can bind to effectors other than Sos—for example, Cbl (a ubiquitin E3 ligase) or dynamin (a protein involved in endocytosis). Further flexibility is provided because, in principle, different adaptor pro- teins can be used to convert the same input signal into different potential outputs, depending on the domain combinations found in the adaptor pro- tein (Figure 10.22b and c). Below, we discuss how other types of domain combinations can lead to even more complex signaling behavior. Scaffold proteins containing PDZ domains organize cell–cell signaling complexes such as the postsynaptic density Proteins containing multiple interaction domains can function as scaf- folds to organize complex signaling pathways. Examples include the PDZ domain-containing scaffolds that organize cell–cell signaling junctions. In multicellular organisms, specific signaling must often occur between adjacent cells, and highly specialized complexes may be assembled at the junctions between cells. For example, neurons connect with each other at junctions called synapses. The postsynaptic side contains a special- ized cellular substructure known as the postsynaptic density (PSD), named because it is so densely packed with proteins that it can be easily imaged by electron microscopy. The PSD is typically found on protruding actin-rich structures known as dendritic spines, and it is where neuro- transmitter receptors and other signaling proteins are concentrated in the cell. Synaptic scaffolding proteins are critical for organizing the PSD. Some of these proteins, exemplified by the synaptic scaffold protein PSD95, contain multiple PDZ domains, as well as SH3 and phosphoserine/ phosphothreonine-binding guanylyl kinase-like (GK) domains, which all Figure 10.23 PDZ-containing proteins at the postsynaptic density. Many of the proteins located at the postsynaptic density (PSd) contain Pdz domains. These form supermolecular complexes required for proper synapse structure and function. Some of these interactions are illustrated in this figure. The major Pdz-containing protein, PSd95 (also known as dlg4), can link to many receptors and ion channels such as the nMdA receptor (nMdAr), AMPA receptor (AMPAr), and potassium channel (KCh). PSd93 (also known as dlg2) also interacts with receptors, such as the kainate receptor, and can form heteromultimers with PSd95. PSd95 links to the Pdz domain- containing Shank protein via the adaptor gKAP. Shank also interacts with the metabotropic glutamate receptor (mglur) via the protein Homer and connects with the actin cytoskeleton via cortactin. PSd95 connects with microtubules via binding to MAP1A. PiCK1, a Pdz and BAr domain-containing protein, can also connect with receptors and ion channels. grip1, with seven Pdz domains, can bind to AMPArs and regulates the localization of these and other receptors to PSd membranes. grip1 can bind grASP to regulate ras signaling, which can also be affected by SYngAP binding to PSd95. Kalirin also binds PSd95 and stimulates rho signaling. CaMKii phosphorylates a number of PSd proteins including PSd95. AMPAr, AMPA (α-amino-3-hydroxyl-5-methyl-4-isoxazole propionic acid) receptor; CaMKii, calcium/calmodulin-dependent kinase ii; nMdAr, nMdA (N-methyl-d-aspartate) receptor. (Adapted from W. Feng and M. zhang, N a t . R e v . N e u r o s c i . 10:87–99, 2009. With permission from Macmillan Publishers ltd.) form different interactions (Figure 10.23). Most of the neurotransmitter receptors in the PSD have a C-terminus that is recognized by these PDZ domains. In addition, the other domains form interactions that anchor the complex to the cytoskeletal structure of the dendritic spines. Thus these scaffolds, in effect, organize a large supermolecular complex in which the receptors and their downstream signaling effectors form a preassem- bled structure properly localized in the cell to allow for an efficient and rapid response to neurotransmitters released from the presynaptic cells. Notably, other cell–cell junctions, such as the tight junctions between adjacent epithelial cells, also use related PDZ scaffolds for organization. Proteins with multiple phosphotyrosine motifs function as dynamically regulated scaffolds scaffold protein inducible scaffold protein phosphorylation-dependent recruitment interactions As has been discussed above, there are many examples of proteins that contain multiple phosphotyrosine motifs (see Figure 10.10). Such pro- teins can act as scaffolds that transiently organize larger signaling com- plexes in response to phosphorylation. The scaffolding activity of such proteins is dynamically regulated by the activities of tyrosine kinases and phosphatases, in contrast to the constitutive activity of scaffolds such as the PDZ-containing proteins discussed above (Figure 10.24). Examples include the receptor tyrosine kinases—such as the PDGF and EGF recep- tors, which contain multiple tyrosine motifs that are phosphorylated Figure 10.24 Properties of scaffold proteins. Scaffolds have binding sites for multiple partners and function to nucleate assembly of large multiprotein complexes. An inducible scaffold requires signaling input to promote binding activity. For example, binding may be dependent on phosphorylation by an upstream kinase. The role of scaffold proteins in TCR activation is illustrated in Chapter 12 Regulation of PKA is discussed in Chapter 6 Figure 10.25 Allosteric switch proteins. Many signaling enzymes are regulated by modular interaction domains (orange). in the example illustrated here, activity of the catalytic domain (green) is repressed by intramolecular interactions in the basal (inactive) state. disruption of these intramolecular interactions, for example by ligands for the interaction domains (purple), leads to activation of the catalytic domain. upon receptor activation—and receptor-associated scaffold proteins such as IRS1 (see Figure 4.11). Further examples are provided by immune sig- naling molecules like LAT and SLP-76, which contain multiple tyrosine motifs that are phosphorylated by the tyrosine kinase ZAP-70 upon its recruitment to the activated T cell receptor (TCR). In this case, phospho- rylation of LAT and SLP-76 transiently creates SH2 binding sites, nucle- ating the formation of a large scaffold complex involving LAT, SLP-76, the SH2/SH3 adaptor GADS, the SH2-containing kinase ITK, and the SH2-containing enzyme PLC-γ. Formation of this complex is required for proper activation of PLC-γ, which in turn is critical for inducing the over- all TCR response. rECoMBining inTErACTion And CATAlYTiC doMAinS To Build CoMPlEX AlloSTEriC SWiTCH ProTEinS The catalytic domains involved in signaling must often be regulated by specific upstream signals. The modular architecture of signaling proteins provides a flexible solution to this problem: coupling catalytic domains with specific interaction domains can generate allosteric molecules in which upstream regulatory signals, localization, and catalytic activity are coordinately regulated (Figure 10.25). Such molecules can be termed allosteric switch proteins. Below, we describe several examples of com- plex multidomain switches to illustrate the diversity of mechanisms and regulatory relationships that can emerge from using these components in new combinations. Many signaling enzymes are allosteric switches We have already seen several examples of this type of regulation for pro- tein kinases. For example, we saw in Chapters 1 and 3 that intramolecu- lar interactions mediated by SH2 and SH3 domains lock the catalytic domain of Src family kinases in an inactive conformation. Src is there- fore an allosteric switch protein, in which kinase activity is specifically induced by disruption of the autoinhibitory interactions, either through dephosphorylation of the SH2-binding motif, or by binding to competing SH2 and SH3 ligands (see Figure 1.9). We have also seen how in PKA the kinase subunit is noncovalently associated with a regulatory (R) subunit made up of two cAMP-binding domains and a pseudosubstrate peptide inhibitor. This multidomain complex results in an inhibited, inactive kinase that can be activated by binding of cAMP to the R subunits. Thus we can see how this kind of modular allosteric regulation can be achieved, not only through intramolecular autoinhibitory interactions, but also through noncovalent interactions with other molecules. These types of modular, autoinhibitory schemes are observed for many classes of catalytic functions, including kinases, phosphatases, GEFs, GAPs, and many others. There is even growing evidence that multidomain scaffold proteins can exist in inactive states involving intramolecular interactions. autoinhibition intermolecular ligand input catalytic domain OFF ON OUTPUT reCoMBininG interaCtion and CatalYtiC doMains to Build CoMPleX allosteriC sWitCH Proteins 14-3-3 Protein regulates the Raf kinase by coordinately binding two phosphorylation sites Above, we described how association with 14-3-3 proteins can alter a phosphorylated target protein’s localization or can occlude its interactions with other partners. However, 14-3-3 domains can also be used to regu- late a protein’s catalytic activity in a phosphorylation-dependent man- ner. In the Erk-MAP kinase pathway, the c-Raf serine/threonine protein kinase (a MAPKKK) is activated by the small G protein Ras and, in turn, stimulates signaling through the MAP kinase pathway. In the basal state, 14-3-3 proteins bind to phosphorylated c-Raf, inhibiting its activity. In this complex, each protomer of a 14-3-3 dimer binds a distinct phospho- rylation site in an inactive c-Raf molecule: one N-terminal to the kinase domain (Ser259 in human c-Raf) and a second at the C-terminus (Ser621) (Figure 10.26). This two-point interaction acts as a clamp to stabilize an autoinhibited conformation in which kinase activity is suppressed and the Ras-binding domain is blocked. Dephosphorylation of the N-terminal site is required to release the 14-3-3 clamp and allow c-Raf activation. Mutation of Pro261 of c-Raf to Ser has been observed in Noonan syn- drome, a human genetic disorder associated with heart defects, facial abnormalities, short stature, and other features. This mutation prevents Ser259 phosphorylation and thus 14-3-3-mediated inhibition of c-Raf activity. The aberrantly active c-Raf then signals through the Erk-MAP kinase pathway, ultimately leading to disease. This example underscores the modular nature of signaling proteins, the regulation of a key signaling kinase by its association with a 14-3-3 dimer, and the severe phenotypic consequences when this inhibitory interaction is lost. Certain plant protein kinases are regulated by modular light-gated domains An example of the range and general utility of these modular systems is provided by plants, where some serine/threonine protein kinases can be activated by light. Such a mechanism is particularly useful in plants, which derive their energy from light and must therefore adjust their phys- iology and orientation depending on its availability. In one well-studied case, the ability to respond to blue light is provided by the LOV domain, a light-gated interaction module. LOV domains tightly bind flavin mono- nucleotide, and are found in a diverse set of light-regulated signaling pro- teins in plants. In the case of kinases, the LOV domain inhibits the kinase catalytic domain through an intramolecular interaction when in the rest- ing (dark) state. Blue light creates a covalent adduct between the flavin and a conserved cysteine in the LOV domain, leading to a conformational change that releases the LOV domain from the catalytic domain, and thereby stimulates kinase activity (Figure 10.27). The modularity of this regulatory mechanism has been demonstrated in the laboratory, as inves- MAP kinase cascades are described in Chapter 3 Ras-GDP Ras-GTP Figure 10.26 Regulation of Raf kinase by 14-3-3 proteins. 14-3-3 proteins hold raf in an inactive state by binding simultaneously to phosphorylated Ser259 and Ser621. dephosphorylation of Ser259 releases 14-3- 3, allowing the g protein ras to activate raf, leading in turn to activation of the MAPK signaling pathway. LOV domain flavin mononucleotide Jα-helix kinase tigators have been able to engineer recombinant proteins containing LOV domains that can be regulated in cells by light (further discussed below). dark light Figure 10.27 Activation of kinases with LOV domains by light. light interacts with flavin mononucleotides of the loV domain. This triggers a conformational change, including disruption of a critical α helix (Jα), that leads to dissociation and activation of the kinase domain and phosphorylation of substrates, including the loV domain-containing regulatory region. (Adapted from J.M. Christie, Annu. Rev. Plant Biol. 58:21–45, 2007. With permission from Annual reviews.) Figure 10.28 Regulation of the neutrophil NADPH oxidase by modular interactions One of the most fascinating molecular machines that is allosterically regulated by modular interaction domains is the neutrophil NADPH oxi- dase. This enzyme is activated during the phagocytosis of bacteria to pro- duce the toxic superoxide anion (O – ) within specialized internal vesicles termed phagosomes, which are used to kill the bacteria. Given the toxicity of superoxide, it is particularly critical that the oxidase catalytic activity only be induced at the right place and time. Fittingly, a complex assembly Multiple interaction domains regulate NADPH oxidase activation. (a) The nAdPH oxidase complex consists of multiple subunits: p22phox and p91phox are transmembrane proteins, whereas p67phox, p47phox, and p40phox are cytosolic regulatory components. The p47phox subunit is held in an inactive state by its two SH3 domains (blue and orange) being locked together via interaction with a C-terminal polybasic motif. other domains indicated are the TPr repeat domain, a proline-rich region (Prr), PX domain, and PB1 domain. (b) Phosphorylation of the polybasic motif of p47phox releases the SH3 domains, allowing them to interact with the polyproline motif (PxxP) of p22phox. This allows interaction of additional domains, thus activating the complex. (c) Crystal structure of the SH3A and SH3B domains of p47phox, showing interaction with the polybasic motif (depicted in yellow). Phosphorylation of residues in this region, including Ser303 (arrow), disrupts this intramolecular interaction. (a and b, adapted from S.S-C. li, B i o c h e m . J . 390:641–653, 2005. With permission of Portland Press; c, from Y. groemping et al., Cell, 113:343– 355, 2003. With permission from Elsevier.) reaction is required to form the active enzyme. The NADPH oxidase has multiple components, including two transmembrane proteins—p91phox and p22phox—that form a heterodimeric flavocytochrome that medi- ates electron transfer to molecular oxygen, and three cytosolic regulatory proteins—p47phox, p67phox, and p40phox. Upon stimulation by phago- cyte receptors, these three cytosolic proteins translocate to the membrane and associate with the integral membrane proteins to yield a functional oxidase complex (Figure 10.28a and b). Each of the cytosolic proteins has multiple interaction domains, prima- rily SH3 domains, through which they interact with one another, and with the integral membrane phox proteins, following a phagocytic signal. However, in unstimulated cells, these domains are sequestered through intramolecular interactions that prevent adventitious oxidase activa- tion. For example, two adjacent SH3 domains of p47phox bind to the tail of p47phox, which blocks the ability of these SH3 domains to associate in trans with another protein. Activating signals, which include binding of the small G protein Rac and phosphorylation of multiple serine resi- dues in p47phox, result in disruption of the intramolecular SH3-binding interaction. The released p47phox SH3 domains are now free to bind to an alternative proline-rich sequence in the membrane-associated sub- unit p22phox, yielding a functional oxidase that is essential for limiting microbial infections. In this system, the two tandem SH3 domains from p47phox pack together to form a single binding surface that recognizes a binding motif that is longer and more specific than typically seen for a single SH3 domain. This likely confers particularly tight control over the enzyme activity (Figure 10.28c). inhibited complex (b) active complex CrEATing nEW FunCTionS THrougH doMAin rECoMBinATion We have seen how evolution has harnessed the amazing flexibility of mod- ular catalytic and interaction domains to build complex signaling devices and to create new behaviors and phenotypes. It is easy to envision how the molecular processes that drive evolution, such as gene duplication, exon shuffling, and point mutation, could lead to the generation of new signaling functions. Of course, in most cases, we can only infer such events indirectly. There are, however, instances in which the generation of new functions through domain rearrangements can be directly demonstrated. We discuss below several examples from human disease, and from direct- ed efforts to engineer new functions in the laboratory. Some modular domain rearrangements can lead to cancer Although domain rearrangement provides a powerful mechanism for generating new functions, there is a potential trade-off in this functional flexibility, in that genetic rearrangements that lead to novel domain combi- nations can also lead to misfunction and disease in individuals. A random domain rearrangement may in some cases offer a fitness advantage, while in most other instances it is likely to confer a fitness disadvantage, either at the level of the individual mutated cell, or at the level of the organism. Here, we discuss two examples of how oncogenes—genes that can cause cancer when mutated or overexpressed—can be generated by chromosome translocations. Oncogenes generally lead to the uncontrolled proliferation or increased survival of cells in which they are expressed, so these muta- tions provide a fitness advantage for the cell, but clearly are a disadvantage for the organism, as the resulting tumor might ultimately lead to its death. Above, we introduced the propensity of SAM domains to oligomerize into extended complexes. This behavior can also induce aberrant signal- ing. In some human leukemias, for example, a chromosomal transloca- tion leads to the fusion of the self-oligomerizing SAM domain from the gene TEL to the catalytic domain of the nonreceptor tyrosine kinase Abl (Figure 10.29a and b). The resulting chimeric protein is constitutively clustered, resulting in persistent activation of the tyrosine kinase domain and oncogenic transformation (remember that dimerization or clustering is the most common mechanism for activating tyrosine kinases, as it pro- motes transphosphorylation and stabilization of the active conformation of the catalytic domain). Several other oncogenic fusion proteins (involv- ing tyrosine kinases other than Abl) are similarly activated by the oli- gomerizing function provided by the SAM domain of TEL. Activation of the Abl tyrosine kinase by chromosome translocation is also seen in the vast majority of patients with chronic myelogenous leuke- mia (CML). In this case, a coiled-coil oligomerization domain derived from the BCR gene on chromosome 22 is fused to the tyrosine kinase domain and C-terminus from the Abl gene on chromosome 9 (Figure 10.29c). The characteristic 9;22 chromosomal translocation that generates this fusion protein is termed the “Philadelphia chromosome,” and was one of the first specific genetic abnormalities to be directly correlated with human cancer. A functional coiled-coil domain in the BCR portion has been shown to be required for oncogenic transformation by the BCR–Abl fusion. Thus, as in the case of the TEL–Abl oncogene, constitutive dimerization or clustering of BCR–Abl is thought to lead to persistent kinase activity and oncogenic transformation. BCR–Abl is well known as the target for the drug imatinib (Gleevec®), which specifically inhibits the Abl tyrosine kinase domain. The effectiveness of imatinib in treating CML is the most dramatic success to Figure 10.29 Activation of Abl by fusion with oligomerization domains. (a) Abl is normally held in an inactive state by intramolecular interactions, mediated primarily by its SH2 and SH3 domains. When fused to TEl, SAM domains of the fusion protein oligomerize, leading to autophosphorylation and activation of the Abl catalytic domain. (c) The Philadelphia chromosome is generated by translocation of the tip of chromosome 9 (encoding Abl) to chromosome 22 (encoding BCr). The translocation generates a hybrid protein containing a portion of BCr at the n-terminus fused to Abl. dimerization or oligomerization of the coiled-coil domains from BCr leads to constitutive activation of the Abl kinase domains. (a) SH2 date in the effort to develop rational cancer therapies that target specific signaling proteins. The role of scaffold proteins in controlling MAP kinase output is discussed in Chapter 3 Modules can be recombined experimentally to engineer new signaling behaviors Work in the emerging field of synthetic biology, which is aimed at using natural biological components to build novel functional systems, has dem- onstrated that signaling modules can be used to generate non-natural, custom-designed signaling proteins, pathways, and cellular behaviors. Many eukaryotic cells use scaffold proteins to properly wire their MAP kinase pathways. Most cells contain several of these closely related path- ways that mediate distinct responses; scaffolds are thought to assemble the correct partners into a single complex, thus promoting their efficient interaction, and preventing their interaction with incorrect but related partners elsewhere in the cell. Recombination of individual domains from MAP kinase scaffold proteins has been used to construct chimeric scaf- fold proteins that assemble novel combinations of MAPK components (Figure 10.30). These chimeric scaffolds can reroute signaling in a living cell so that a specific input now produces a normally unrelated response. For example, a yeast cell can be reprogrammed such that stimulation with a mating pheromone induces the response normally associated with high osmotic stress. Engineered domain recombination involving catalytic and interaction domains can also be used to build new allosteric signaling proteins and new cellular behaviors. For example, the small G protein Rac is a master regulator of actin polymerization, which drives membrane protrusion and cell movement. Recombination of a Rac-specific GEF domain (a Dbl homol- ogy or DH catalytic domain) with different interaction domains can result native scaffold pathways A chimeric scaffold protein new pathway linkage Figure 10.30 Chimeric scaffolds can rewire signaling pathways. in this example (modeled on MAP kinase cascades), two scaffolds assemble distinct protein complexes, such that stimulus A leads to output X and stimulus B leads to output Y. A chimeric, engineered scaffold protein can redirect signal output such that stimulus A A A B INPUT now leads to output Y. X B Y X Y OUTPUT Y in novel allosteric switches in which Rac activation and actin polymeri- zation are controlled by novel, non-natural inputs. For example, recom- bination with an autoinhibitory intramolecular PDZ domain interaction that can be relieved by PKA phosphorylation leads to a PKA-induced GEF (Figure 10.31a). Alternatively, recombination of Rac itself with a light- gated LOV domain module from plants generates a novel light-induced Rac (Figure 10.31b). (a) (b) LOV OFF DH PDZ + PKA hν ON PDZ DH Figure 10.31 Engineering allosteric switch proteins that respond to novel signaling inputs. (a) Coupling rac activation to protein kinase A (PKA) activity. dH domains act as gEFs for rho family g proteins. in this example, activity of the dH domain is repressed by an engineered intramolecular interaction between a Pdz domain and the C-terminus. Phosphorylation of the C-terminus by PKA disrupts this intramolecular interaction, leading to gEF activity and rac activation. Activation of rac by light. in this example, a constitutively gTP-bound form of rac is fused to a loV domain, thus blocking its interaction with downstream Jα-helix effectors OFF (blocked) ON effectors. light induces conformational changes that release the loV domain, leaving rac-gTP free to bind effectors. These kinds of engineered signaling proteins clearly illustrate the func- tional flexibility of modular domains. Such engineered proteins also have several potential uses. First, examples like light-controlled signaling proteins can be used as powerful research tools to activate pathways in a spatially and temporally controlled manner using specific patterns of activating light. Second, these engineered signaling proteins might, in the longer term, allow reprogramming of cells so that they carry out custom- designed therapeutic sensing/response behaviors, such as those involved in detecting and eliminating diseased cells or pathogens. SuMMArY The modular architecture of signaling proteins, coupled with the bind- ing properties of interaction domains, provides a flexible solution to the problem of how to control signaling protein activities, and to coordinately couple them to upstream regulatory signals and downstream targets. As we have seen, this is achieved not only by the association of proteins with one another, but also through interaction domains that recognize a range of biomolecules such as phospholipids. Modular interaction domains can either recognize unmodified peptide motifs, or can specifically bind only to peptide motifs after they have been post-translationally modified. In the latter case, such domains serve to couple post-translational modification to the localization or activity of downstream effector proteins. Proteins with multiple modular interaction domains or binding motifs can serve as scaffolds to assemble signaling proteins into larger molecular complexes. When combined with catalytic domains, modular interaction domains can be used to generate allosteric switches that can be regulated by a variety of inputs. The modular archi- tecture of signaling proteins also provides a ready mechanism to explain the evolution of more complicated signaling behaviors from a relatively limited toolkit of modular functional units. QuESTionS Most eukaryotic signaling proteins have modular structures composed of several domains that each carry out a distinct subfunction. What evolutionary constraints or advantages may have led to this type of organization? A particular modular protein interaction domain is often found in many different signaling proteins in a given organism. Hypothesize how a modular protein interaction domain family may have expanded over the course of evolution. What issues may arise as such a domain family expands in size? You are analyzing an SH2–phosphotyrosine peptide interaction and find that the Kd for the unphosphorylated peptide is 1 mM, while the Kd for the phosphorylated form is 100 nM. This peptide is found on the C-terminal tail of a receptor tyrosine kinase that is expressed at low levels. What is the approximate concentration range for the SH2- containing protein in the cell? Protein interaction modules, such as SH2 domains or PDZ domains, often recognize a key physical feature in their cognate peptides, such as a phosphotyrosine side chain (SH2) or a free C-terminus (PDZ). How do individual members of the domain family establish distinct specificities? reFerenCes Describe the general ways in which protein interaction domains can be combined with catalytic domains to regulate their activity and function. How can fusions between interaction and catalytic domains lead to dis- ease? How can they be harnessed to engineer new cellular behavior? If modular recombination of signaling domains can lead to diseases such as cancer, why would these features not have been selected against by natural selection? The human genome encodes over 100 different SH2 domains, most of which bind specifically to phosphotyrosine peptides. By contrast, the budding yeast has a single SH2 domain, which binds to RNA polymer- ase II in a serine/threonine phosphorylation-dependent fashion. Yeast has no dedicated tyrosine-specific kinases and few tyrosine-specific phosphatases. Based on this information, provide a model for the evo- lution of the SH2 domain family and tyrosine kinase signaling. Most signaling proteins contain multiple functional domains (either bind- ing or catalytic domains). In many cases, expression in the cell of a mutant form of the protein, in which one of these domains has been inactivated, can act as a dominant negative mutation; that is, the mutant protein can inhibit the activity of the normal, endogenous protein. Explain how this can occur. In other cases, mutation of one domain can sometimes lead to the opposite effect—constitutive activity. When might this be the case? In the laboratory, how might you create a modular domain that has a novel binding activity? rEFErEnCES ModulAr ProTEin doMAinS Dyson HJ & Wright PE (2002) Coupling of folding and binding for unstructured proteins. Curr. Opin. Struct. Biol. 12, 54–60. Orlicky S, Tang X, Willems A et al. (2003) Structural basis for phosphodependent substrate selection and orienta- tion by the SCFCdc4 ubiquitin ligase. Cell 112, 243–256. Pawson T & Nash P (2003) Assembly of cell regulatory systems through protein interaction domains. Science 300, 445–452. inTErACTion doMAinS THAT rECognizE PoST-TrAnSlATionAl ModiFiCATionS Au-Yeung BB, Deindl S, Hsu LY et al. (2009) The struc- ture, regulation, and function of ZAP-70. Immunol. Rev. 228, 41–57. Brehm A, Tufteland KR, Aasland R & Becker PB (2004) The many colours of chromodomains. Bioessays 26, 133–140. Hu J & Hubbard SR (2005) Structural characteriza- tion of a novel Cbl phosphotyrosine recognition motif in the APS family of adapter proteins. J. Biol. Chem. 280, 18943–18949. Hurley JH, Lee S & Prag G (2006) Ubiquitin-binding domains. Biochem. J. 399, 361–372. Newton K, Matsumoto ML, Wertz IE et al. (2008) Ubiq- uitin chain editing revealed by polyubiquitin linkage- specific antibodies. Cell 134, 668–678. Pascal SM, Singer AU, Gish G et al (1994) Nuclear mag- netic resonance structure of an SH2 domain of phospho- lipase C-gamma 1 complexed with a high affinity binding peptide. Cell 77, 461–472. Seet BT, Dikic I, Zhou MM & Pawson T (2006) Read- ing protein modifications with interaction domains. Nat. Rev. Mol. Cell Biol. 7, 473–483. inTErACTion doMAinS THAT rECognizE unModiFiEd PEPTidE MoTiFS or ProTEinS Kwan JJ, Warner N, Pawson T & Donaldson LW (2004) The solution structure of the S. cerevisiae Ste11 MAPKKK SAM domain and its partnership with Ste50. J. Mol. Biol. 342, 681–693. Qiao F, Song H, Kim CA et al. (2004) Derepression by depolymerization; structural insights into the regula- tion of Yan by Mae. Cell 118, 163–173. Seet BT, Berry DM, Maltzman JS et al. (2007) Efficient T-cell receptor signaling requires a high-affinity interac- tion between the Gads C-SH3 domain and the SLP-76 RxxK motif. EMBO J. 26, 678–689. Zarrinpar A, Bhattacharyya RP & Lim WA (2003) The structure and function of proline recognition domains. Sci. STKE 2003(179):re8. inTErACTion doMAinS THAT rECognizE PHoSPHoliPidS Kutateladze TG (2006) Phosphatidylinositol 3-phos- phate recognition and membrane docking by the FYVE domain. Biochim. Biophys. Acta 1761, 868–877. Lemmon MA (2008) Membrane recognition by phos- pholipid-binding domains. Nat. Rev. Mol. Cell Biol. 9, 99–111. McMahon HT & Gallop JL (2005) Membrane curvature and mechanisms of dynamic cell membrane remodelling. Nature 438, 590–596. CrEATing CoMPlEX FunCTionS BY CoMBining inTErACTion doMAinS Chothia C, Gough J, Vogel C & Teichmann SA (2003) Evo- lution of the protein repertoire. Science 300, 1701–1703. Christie JM (2007) Phototropin blue-light receptors. Annu. Rev. Plant Biol. 58, 21–45. Feng W & Zhang M (2009) Organization and dynamics of PDZ-domain-related supramodules in the postsynap- tic density. Nat. Rev. Neurosci. 10, 87–99. Kim E & Sheng M (2004) PDZ domain proteins of syn- apses. Nat. Rev. Neurosci. 5, 771–781. Li SS (2005) Specificity and versatility of SH3 and other proline-recognition domains: structural basis and impli- cations for cellular signal transduction. Biochem. J. 390, 641–653. Peisajovich SG, Garbarino JE, Wei P & Lim WA (2010) Rapid diversification of cell signaling phenotypes by modular domain recombination. Science 328, 368–372. Vogel C, Bashton M, Kerrison ND et al. (2004) Structure, function and evolution of multidomain proteins. Curr. Opin. Struct. Biol. 14, 208–216. rECoMBining inTErACTion And CATAlYTiC doMAinS To Build CoMPlEX AlloSTEriC SWiTCH ProTEinS Freed E, Symons M, Macdonald SG et al. (1994) Binding of 14-3-3 proteins to the protein kinase Raf and effects on its activation. Science 265, 1713–1716. Fu H, Subramanian RR & Masters SC (2000) 14-3-3 proteins: structure, function, and regulation. Annu. Rev. Pharmacol. Toxicol. 40, 617–647. Groemping Y, Lapouge K, Smerdon SJ & Rittinger K (2003) Molecular basis of phosphorylation-induced acti- vation of the NADPH oxidase. Cell 113, 343–355. Light Y, Paterson H & Marais R (2002) 14-3-3 antago- nizes Ras-mediated Raf-1 recruitment to the plasma membrane to maintain signaling fidelity. Mol. Cell. Biol. 22, 4984–4996. Lim WA (2002) The modular logic of signaling pro- teins: building allosteric switches from simple binding domains. Curr. Opin. Struct. Biol. 12, 61–68. CrEATing nEW FunCTionS THrougH doMAin rECoMBinATion Bashor CJ, Helman NC, Yan S & Lim WA (2008) Using engineered scaffold interactions to reshape MAP kinase pathway signaling dynamics. Science 319, 1539–1543. Bashor CJ, Horwitz AA, Peisajovich SG & Lim WA (2010) Rewiring cells: synthetic biology as a tool to interrogate the organizational principles of living systems. Annu. Rev. Biophys. 39, 515–537. Chau AH, Walter JM, Gerardin J et al. (2012) Designing synthetic regulatory networks capable of self-organizing cell polarization. Cell 151, 320–332. Dueber JE, Yeh BJ, Chak K & Lim WA (2003) Reprogram- ming control of an allosteric signaling switch through modular recombination. Science 301, 1904–1908. Golub TR, Goga A, Barker GF et al. (1996) Oligomeriza- tion of the ABL tyrosine kinase by the Ets protein TEL in human leukemia. Mol. Cell. Biol. 16, 4107–4116. Karginov AV, Ding F, Kota P et al. (2010) Engineered allosteric activation of kinases in living cells. Nat. Bio- technol. 28, 743–747. Kim CA, Phillips ML, Kim W et al. (2001) Polymeriza- tion of the SAM domain of TEL in leukemogenesis and transcriptional repression. EMBO J. 20, 4173–4182. Lim WA (2010) Designing customized cell signalling cir- cuits. Nat. Rev. Mol. Cell Biol. 11, 393–403. McWhirter JR, Galasso DL & Wang JY (1993) A coiled- coil oligomerization domain of Bcr is essential for the transforming function of Bcr-Abl oncoproteins, Mol. Cell. Biol. 13, 7587–7595. Park SH, Zarrinpar A & Lim WA (2003) Rewiring MAP kinase pathways using alternative scaffold assembly mechanisms. Science 299, 1061–1064. Wu Y, Frey D, Lungu OI et al. (2009) A genetically en- coded photoactivatable Rac controls the motility of living cells. Nature 461, 104–108. Information Processing by Signaling Devices and Networks A living cell must take in a vast amount of information about its envi- ronment and internal state, and use this information to respond in vari- ous ways. What genes should be turned on or off? Should it grow, divide, differentiate, or die? Should it move or change its shape? Thus, from a signaling perspective, cells are complex information-processing systems, akin to computers and other human-made devices. How then do cellular components process information and respond appropriately? The previous chapters have introduced the fundamental molecular parts and mechanisms that make up cell signaling systems. In this chapter, we try to understand how these molecular parts can be used to build devices that are capable of executing particular information-processing tasks. Our understanding of signaling design principles is still relative- ly primitive—we are just starting to reach beyond a descriptive under- standing of signaling systems to grasp the fundamental principles of how molecular systems execute complex information-processing tasks. In part, our knowledge is limited because relatively few signaling systems have been quantitatively analyzed at high resolution. However, in recent years, computational systems biology and synthetic biology—approaches that combine analysis, simulation, and construction of molecular systems— have become powerful tools in understanding how signaling systems achieve complex behaviors. In this chapter, we will examine a series of specific information-process- ing tasks that are common to many cellular processes, and provide examples of how molecular systems are able to accomplish these tasks. This will not be an encyclopedic examination of all possible cellular behaviors, but instead we will focus on a few commonly used design principles. We consider how cell signaling devices process various input stimuli, including input amplitude, duration, and combinations (multi-input control), and how they control output responses, including output amplitude and duration. We will examine how this diverse array of processing behaviors can be generated using signaling molecules as the basic components. SIgNalINg SyStemS aS INformatIoN-ProceSSINg DevIceS In order to discuss the general properties of information processing in cells, it is useful to step back from the seemingly overwhelming complex- ity of the biological system to a more abstract perspective. The fields of computer science and communications, with their focus on the design of highly complex information-processing devices and systems, provide a helpful theoretical framework for such a discussion. Signaling devices can be considered as state machines All information-processing systems are built from state machines— devices that can exist in multiple discrete states, and which can change their state in response to specific instructional inputs (Figure 11.1a). If these states have different functional properties (what we will refer to as outputs), then the state machine will serve as an input/output device. It is easy to understand how human-made machines serve as input/out- put devices (Figure 11.1b). Take for example an automatic door, which can exist in the open or closed state; the open state has a distinct func- tional output—it is the only state in which a person can pass through the door. This automatic door is an input/output device because a particular input—in this case, motion detected by the door’s sensor—serves as an instruction to convert from the closed state to the open state. Thus, the input of motion results in the functional output of allowing a person to walk through the door. Another example of a human-made state machine is the transistor—the component that enables most of our digital electronic systems. A transis- tor can switch between a high-resistance state and low-resistance state. The application of a high voltage to the transistor (the input) switches it to the low-resistance state, which in turn allows current to flow through the transistor (the output). When linked in complex ways, these simple input/ output devices can yield immensely complex decision-making machines, such as computers. Living cells can also be considered as state machines (Figure 11.1c). For example, a cell can exist in distinct states, such as a quiescent (survival) state, a growing/dividing (proliferation) state, and a dying (apoptotic) state. These different fates are controlled by environmental inputs (such as growth factors, metabolic state, and mechanical inputs) that serve as the instructions that induce changes in state. Often it is a complex combi- nation of signals that serves as the instructions to change state in cellular systems. At a more microscopic scale, the individual proteins and pathways within a cell can also act as state machines. For example, a simple allosteric kinase can exist in an active or inactive conformation, and this conformation might be controlled by the input of phosphorylation/dephosphorylation. In another example, a transmembrane receptor involved in detecting a mitogen is also a state machine—its input is the concentration of mitogen, while its output is the activated state of the receptor. generic state machine INPUTS OUTPUTS 0 1 1 0 human-made devices AUTOMATIC DOOR biological devices CELL INPUTS STATES OUTPUTS EXIT cannot walk through door INPUTS STATES OUTPUTS division motion no motion can walk through door EXIT growth factors death signals survival death TRANSISTOR INPUTS voltage no voltage STATES OUTPUTS no current flow current flow KINASE INPUTS phosphorylation dephosphorylation STATES OUTPUTS substrates unphosphorylated substrates phosphorylated Figure 11.1 Signaling devices as state machines. (a) a generic state machine switches between two different states (0 and 1) when provided with specific instructions (inputs). output depends on the state. (b) examples of human-made devices that function as state machines. an automatic door will open only when it detects motion. a transistor will allow current to flow only when an external voltage is applied. examples of biological state machines involved in cell signaling. a cell can be viewed as a device that can exist in many distinct states, such as dividing (proliferating), quiescent, and apoptotic, with specific signals providing instructions to switch from one state to another. a signaling molecule, such as a kinase, can be viewed as a state machine, with specific inputs (e.g., phosphorylation) serving as instructions to switch from one state to another. Signaling devices are organized in a hierarchical fashion Information-processing systems, both biological and human-made, are often organized in a hierarchical fashion. State machines can exist at a variety of scales, and often a larger-scale state machine is constructed from many smaller-scale state machines. Take the example of an automatic Figure 11.2 Signaling input/output devices are organized in hierarchical fashion. a single signaling molecule is an input/ output device when its activity is controlled by specific inputs. this molecule may be part of a larger multiprotein pathway, which itself functions as a modular input/ output device. finally, this pathway device may be a component of a larger, more complex signaling network at the cellular level. Important signaling behaviors can be mediated by events that span this whole range of scales. In this example, input from binding of ligands to cell-surface receptors leads to the output of transcriptional regulation of specific genes. INPUTS OUTPUTS cellular network device pathway device molecular device door—while this system can be viewed as a single state machine, if one looks deeper into the inner workings of the door’s sensor and processor, one will find many smaller state machines (including transistors) that are linked together to yield the whole system. Cellular information-processing machines are also built in a hierarchi- cal manner. For example, although the whole cell can be considered as one state machine, it is clear that the individual proteins (such as recep- tors and kinases) that make up a growth-control pathway are themselves smaller-scale state machines (Figure 11.2). Similarly, multiple proteins and pathways interact with each other to yield a larger network capable of more complex signal processing. In summary, cells use a range of scales and different types of solutions to solve information-processing tasks. In some cases, a signal-processing device can comprise just a single signaling protein, whereas in other cases, a larger network of interacting molecules is involved. We will explore how particular tasks can be accomplished by devices that operate both at the molecular scale and at the network scale. In the case of networks, many of the systems we will examine also incorporate transcriptional ele- ments, which lead to changes in the abundance of different proteins. This is because at a systems biology level, it is often this whole ensemble of integrated molecular elements that yields a particular functional cellular behavior. Signaling devices face a variety of challenges in input detection If one considers the multitude of different stimuli that bombard a cell, it is natural to wonder how the cell is able to use this information to respond appropriately. To accomplish this, cellular devices, no matter what specific sustained input increase transient input increase combination of input changes time distinct output decisions Figure 11.3 Signaling systems respond to different types of input changes. Signaling systems respond to the changes in the amplitude of inputs, which may be sustained or transient. most signaling systems must monitor and respond to changes in the amplitude of many different inputs simultaneously. physiological role they play, must perform a range of common tasks. At the most basic level, input stimuli must be detected and transformed into a specific response output. More specifically, often the amplitude (strength) of the input must be accurately measured, as well as how the input changes over time (Figure 11.3). In many cases, a cellular device must measure multiple inputs and make a coordinated response that integrates these multiple signals. While the exact molecules that receive and transmit information will be different for a particular response in a particular cell, one often uncovers similar solutions to accomplish these common tasks. Despite these commonalities, all signaling systems must be highly adapted to their specific roles in the cell. Depending on their function, for example, many signaling systems have evolved to respond only to a spe- cific type of input stimulation, effectively ignoring the many other pos- sible inputs vying for attention. Some signaling systems show responses that are proportional to the input level, whereas others only become acti- vated above a given threshold. Some signaling systems respond imme- diately upon stimulation, whereas others require sustained input before switching on—if, for example, the response is very costly and the cell needs to ensure that it does not trigger the output in response to ran- dom fluctuations in stimuli. Throughout this chapter, we consider some of the ways that signaling systems have evolved to address each of these challenges. Proteins can function as simple signaling devices How are signaling molecules built to function as input/output devices? Throughout the previous chapters in this book, we have reviewed vari- ous mechanisms by which a molecule can change its functional output based on input stimuli (Figure 11.4). Many signaling proteins act as a simple allosteric switch, whereby the active conformation of the protein is stabilized in the presence of a stimulatory covalent modification (for example, phosphorylation) or ligand binding (Figure 11.4a). For example, the kinase domain of the insulin receptor is activated by phosphorylation of its activation loop, which in turn leads to repositioning of the activation loop to allow binding of peptide substrate. Allosteric control mechanisms like this are common to many protein kinases. Some proteins achieve input/output control by acting as a modular allos- teric switch. In these types of proteins, the catalytic domain of the protein itself may not function as an allosteric switch. However, other regula- tory domains (usually protein–protein interaction domains) can provide switching function by participating in autoinhibitory interactions with Regulation of protein kinases by activation-loop phosphorylation is discussed in Chapter 3 allosteric switches EXAMPLES INPUT phosphorylation insulin receptor kinase - Chapter 3 G proteins - Chapter 3 modular allosteric switches INPUT ligand binding tyrosine phosphatases - Chapter 3 nonreceptor Tyr kinases - Chapters 1, 3 RhoGEFs - Chapter 3 protein-complex switches INPUT ligand binding protein kinase A (PKA) - Chapters 3, 6 neutrophil NADPH oxidase - Chapter 10 localization switches Pho4 (nuclear localization) - Chapter 5 GEFs (membrane localization) - Chapter 3 Figure 11.4 Examples of molecular input/output devices. one example is illustrated for each type; other examples, and chapter numbers where examples are discussed in more detail, are listed on the right. (a) allosteric switch proteins are activated by conformational changes in response to inputs such as post- translational modification or ligand binding. (b) modular allosteric switches are regulated by intramolecular interactions involving modular binding domains. (c) Protein-complex switches are regulated by interactions between different protein subunits. (d) localization switches are driven by changes in subcellular localization. the catalytic domain (Figure 11.4b). For example, in the tyrosine phos- phatase Shp2, the protein tyrosine phosphatase (PTP) catalytic domain is by itself constitutively active. However, the intact protein contains two SH2 domains. In the folded tertiary structure, the N-terminal SH2 domain binds to the PTP domain, blocking its active site—the SH2 domain autoin- hibits the phosphatase activity. This multidomain protein then acts as an input/output device because binding of tyrosine-phosphorylated ligands to the Shp2 SH2 domains leads to the release of the SH2–PTP domain autoinhibitory interaction, resulting in activation of the phosphatase activity. Multiple polypeptides can also interact to form a protein-complex allos- teric switch. For example, protein kinase A (PKA) has a catalytic kinase domain and a regulatory domain (Figure 11.4c; see also Figure 6.7). The regulatory domain binds to the catalytic domain, blocking its substrate- binding site with a pseudosubstrate sequence, thus inhibiting its activity. This complex now acts as a switch that can be activated by the input of the small signaling mediator, cAMP. cAMP specifically binds to the regu- latory subunit, which causes the subunit to release the catalytic domain, which is now active. This mechanism is not that different from the autoin- hibitory mechanism described above, except that regulation is achieved by domains on a separate polypeptide chain. In addition to changes in conformation or interactions, another simple and common way to change protein activity in response to inputs is to change its subcellular localization (Figure 11.4d). For example, the tran- scription factor Pho4 is normally localized in the cytoplasm, where it is inactive because it cannot interact with the promoters that it can regu- late. However, signal input (dephosphorylation of Pho4) leads to its import into the nucleus, where it can now exert its output activity on transcrip- tion. Similarly, phosphorylation of FOXO transcription factors by kinases such as Akt leads to their export from the nucleus and sequestration in the cytosol, leading to transcriptional changes. INtegratINg multIPle SIgNalINg INPutS In both cellular and electronic signaling, many devices are more com- plex than those described above. For example, cellular responses are highly dependent on the integration of multiple inputs. In responding to inputs, the cell’s ultimate output must take account of a wide range of conditions—not just a single signal, but often multiple external signals or stresses, as well as internal states (such as the energy status of the cell or what stage of the cell cycle it is in). It would be hard for a cell to function if it did not have complex signaling systems capable of monitoring and integrating multiple signals. Thus, it is not surprising that the majority of signaling proteins and networks function as multi-input state machines. They are capable of responding to many different inputs, some of which act in a positive manner (activate output), and some of which act in a negative manner (repress output) (Figure 11.5). Cells use many different strategies to build signal-integrating devices. In this section, we explore several examples of signaling devices that can integrate multiple signals. Logic gates process information from multiple inputs To understand how information from multiple inputs can be processed, it is helpful to draw parallels between cellular signaling systems and human-made digital control systems, which depend on logic gates. Logic gates specify outputs depending on the combination of two inputs; they come in a variety of types, such as AND, OR, NOR, or XOR gates, depend- ing on their input/output behavior (Figure 11.6). For example, in a simple two-input AND gate, the output is switched on only when both input 1 and input 2 are present, and not in the presence of either input 1 or input 2 alone. In a NOR gate, output is switched on only in the absence of both inputs 1 and 2. Either input alone (or together) is sufficient to switch out- put off. When these types of digital control gates are linked together, the resulting circuits can be capable of highly complex responses—the micro- processor chips in computers and other electronic devices, for example, can contain millions of logic gates. Regulated changes in subcel- lular localization are discussed in Chapter 5 ACTIVATING REPRESSING A B C D E signaling node Figure 11.5 Most signaling proteins integrate many different inputs. these include combinations of activating and repressing signals. Figure 11.6 INPUT INPUT INPUT INPUT Signal integration by multi-input A B logic gates. examples are shown of various types of digital two-input gates. for each type of gate, the standard symbol used to represent the gate in schematic drawings is indicated above, and how output depends on each input (a and B) is shown below. A B A B A B OUTPUT OUTPUT OUTPUT OUTPUT Many signaling proteins or networks show behavior analogous to logic gates. For example, some proteins will be strongly activated only in the presence of two different inputs and weakly activated in the presence of either individual input—such proteins are analogous to an AND gate. Of course, most signaling proteins do not show the absolute on–off properties of digital systems (Figure 11.7), because their physiological inputs (the concentrations of proteins, lipids, and other biomolecules, and the activity of enzymes) usually vary continuously over a range of values, as opposed to being either present or absent. Despite this, logic gates provide a useful analogy for information processing by signaling proteins and networks, which is critical for precise cellular decision-making. output The consequences of multiple pro- tein modifications are discussed in more detail in Chapter 4 biological ‘‘AND gate’’ Simple peptide motifs can integrate multiple post-translational modification inputs One simple but effective way for signaling proteins to integrate multi- ple inputs is to have multiple post-translational modifications within the same peptide. The functional output of having a single modification ver- sus two modifications could be very different, especially if this peptide serves as a modification-dependent binding site for downstream effectors. The combination of multiple possible modifications greatly expands the possible states of a protein, providing the means for more sophisticated regulation. An example of antagonistic modifications occurs at the N-terminal tail of histone H3 (Figure 11.8). When Lys9 is methylated by histone methyl transferases, the modified site acts as a signal to recruit the chromodo- main of the protein HP1, which acts to silence nearby transcription by altering chromatin structure. However, during mitosis, the Aurora-B kinase can phosphorylate the nearby Ser10 residue of histone H3. This Figure 11.7 A biological AND gate. a biological signaling device, although not truly digital in response, approximates an aND gate if high output activity is only obtained when two inputs (a and B) are both present at high levels. Note that there is no output activity when either of the two inputs is absent. modification blocks HP1 binding, even if Lys9 is methylated. Thus, this simple histone-tail peptide can act as a sophisticated signal-integration point—its recruitment of the regulatory protein HP1 is activated by methylation but then can be overridden by phosphorylation. This is only one example of how the interplay between post-translation- al modifications can be used to integrate signals. Such relationships can be antagonistic, as described in the example above, or cooperative TWO INPUTS methyl transferase kinase histone tail Figure 11.8 methylation methylation and phosphorylation binds HP1 protein OUTPUT chromatin silencing blocks HP1 binding no chromatin silencing Peptide motifs can integrate multiple post-translational modifications. the tail of histone H3 can be trimethylated on lys9 by a histone methyl transferase, which leads to recognition by the HP1 protein. In turn, HP1 binding leads to chromatin remodeling and silencing. However, phosphorylation of Ser10 in the histone H3 peptide by the aurora-B kinase blocks HP1 recognition. this behavior is analogous to another type of two-input logic gate, an aND Not gate. or sequential (that is, modification A is a priming modification which is required for modification B to occur). For example, the protein kinase GSK-3 can only efficiently phosphorylate substrates that have been pre- viously phosphorylated by a second kinase, such as CK1 or CK2. Cyclin-dependent kinase is an allosteric signal-integrating device An allosteric kinase such as a cyclin-dependent kinase (CDK) responds to multiple inputs, including post-translational modifications and ligand binding (Figure 11.9). These input-integration properties are critical for the central role of CDKs in controlling progression in the cell cycle—many distinct inputs must be weighed before CDK activation can proceed. More specifically, to be activated, Cdk2 must both be phosphorylated on the activation loop (Thr160 in human Cdk2) by the CAK complex, and also bind to a cyclin partner. These two positive inputs are both required for Cell-cycle control is explored in Chapter 12 INPUTS REPRESSING ACTIVATING inhibitory ligands (p27) inhibitory phosphorylation (Wee1 kinase) removal of inhibitory phosphorylation (Cdc25 phosphatase) cyclin binding activating phosphorylation (CAK) OUTPUT phosphorylation of cell-cycle substrates Figure 11.9 Cyclin-dependent kinase (CDK) acts to integrate a wide variety of both activating and repressing inputs. activating inputs include binding of the cyclin subunit, phosphorylation by caK, and dephosphorylation by cdc25. repressing inputs include phosphorylation by Wee1 and binding of inhibitory ligands such as p27. full CDK activity, and thus mimic AND-gate control. At a mechanistic level, these two inputs work together to properly position the C-helix and activation loop of CDK, to allow for catalytic activity (see Figure 3.14). The activity of CDK can be overridden by several negative inputs, how- ever. These inputs act like safety mechanisms to prevent activation before the appropriate time in the cell cycle, or to halt cell-cycle transitions when checkpoints have indicated that something is amiss and must be correct- ed. Phosphorylation at two other sites within the CDK (residues Thr14 and Tyr15 in Cdk2) by the kinase Wee1 and its relative Myt1 shuts off the kinase activity. In addition, binding of CDK inhibitor proteins, such as p27, to the CDK–cyclin complex can also shut off kinase activity by causing large conformational changes in the kinase domain, thus blocking ATP binding, and also by occupying the substrate-docking-site groove on the cyclin subunit. Thus, the negative phosphorylation input and nega- tive regulator input have a NOR-gate relationship—either is sufficient to prevent activation of the CDK–cyclin complex. Figure 11.10 Modular signaling proteins act as switches that integrate multiple inputs. (a) two domains of the actin regulator N-WaSP are involved in autoinhibition of the output domain. relief of autoinhibition requires the cooperative binding of two inputs: gtP-bound cdc42 to the g protein binding domain (gBD), and the phospholipid phosphatidylinositol 4,5-bisphosphate (PIP₂) to the basic domain (B). each input alone is a poor activator. (b) the adaptor Nck can substitute for cdc42. N-WaSP thus behaves like a more complex input/output device containing both aND and or gates. Modular signaling proteins can integrate multiple inputs Modular proteins can also serve as signal integrators. An example is the signaling protein WASP and its homologs such as N-WASP, which are key regulators of the actin cytoskeleton (Figure 11.10). N-WASP has a catalytic domain that can interact with and activate the actin-related pro- tein 2/3 (Arp2/3) complex—a seven-protein machine that nucleates new actin filaments that grow as branches from existing filaments. Activation of N-WASP (and subsequently Arp2/3) must occur with high spatial and temporal precision in order to yield specific actin-driven morphological changes, such as cell movement, endocytosis, and phagocytosis. Not sur- prisingly, N-WASP is regulated by multiple inputs. These inputs include proteins, such as the active (GTP-bound) form of the small G protein Cdc42, and lipids, such as phosphatidylinositol 4,5-bisphosphate (PIP₂). N-WASP is normally only strongly activated when both Cdc42 and PIP₂ are present: thus, the two inputs have an AND relationship. This dual control is thought to target actin polymerization precisely to sites on the membrane that contain both Cdc42 and PIP₂. This AND-gate control is only observed in the intact N-WASP protein. As with many other examples, N-WASP uses an autoinhibitory mechanism to achieve this complex input control. N-WASP has multiple modular domains, including a small G protein binding domain that can bind to (b) INPUTS Cdc42 PIP₂ Arp2/3-mediated actin polymerization OUTPUT Arp2/3 activity activated Cdc42, and a basic domain that can bind to PIP₂. In the intact protein in its basal state, these two domains participate in autoinhibitory interactions that repress the ability of the N-WASP catalytic domain to stimulate actin nucleation. These autoinhibitory interactions are cooper- ative; the complex is tightly held in the OFF state and it is very difficult for either single input to release the autoinhibited state. However, if both Cdc42 and PIP₂ are present, they can cooperate to release autoinhibition, resulting in much higher activity when both ligands are present. Examined more closely, regulation of the activity of WASP family proteins is even more complex. For example, SH3 domains from proteins such as the Nck adaptor can bind to sites within a proline-rich region of N-WASP and stimulate its catalytic activity. Nck seems to be able to substitute for Cdc42; either of these two inputs, together with PIP₂, promotes activation. This three-input system contains both OR and AND gates: the output is dependent on (Cdc42 OR Nck) AND PIP₂ (Figure 11.10b). Furthermore, N-WASP can also be regulated by tyrosine phosphorylation and other inputs. We can see that, much as in the case of complex electronic cir- cuits, increasingly complex biological behaviors can be derived by linking together simple input/output relationships. This type of general structure—a switch composed of a modular output catalytic domain combined with autoinhibitory input domains—is highly flexible and is observed in many different signaling proteins. Such a mod- ular allosteric framework can, in principle, lead to integration of many different inputs and many different types of integrating relationships. The evolutionary flexibility of this system has been demonstrated in the laboratory by swapping input domains in N-WASP, resulting in novel syn- thetic proteins that respond to different input stimuli specific for the new autoinhibitory domains. Transcriptional promoters can integrate input from multiple signaling pathways Processing of multiple inputs can also be achieved at a network level by transcriptional promoters that integrate the activities of multi- ple transcription factors. For example, when T cells are activated by antigen-presenting cells, one of the important outputs is activation of the interleukin-2 (IL-2) promoter and the subsequent production and secretion of the cytokine IL-2. IL-2 promotes survival and proliferation of the activated T cell. The IL-2 promoter serves as a critical integration point, acting as an AND gate that detects two distinct branches of the T-cell-activation pathway (Figure 11.11). T cell activation is described in more detail in Chapter 12 INPUTS Figure 11.11 Ca2+/ calcineurin pathway Erk (MAPK) pathway OUTPUT Promoters can integrate the input of multiple signaling pathways. In t cell signaling, expression of the cytokine Il-2 requires input from two pathways. the ca2+/calcineurin pathway leads to dephosphorylation and thus activation of the transcription factor Nfat. the erk maP kinase (maPK) pathway leads to phosphorylation and thus activation of the transcription factor aP-1. expression from the Il-2 promoter only occurs when Nfat and aP-1 interact synergistically. thus, the IL-2 promoter IL-2 cytokine production Il-2 promoter functions like an aND gate. One of the pathways activated upon T-cell-receptor stimulation is the Erk MAP kinase (MAPK) pathway. Activated Erk phosphorylates and activates the transcription factor AP-1. However, this factor alone is not sufficient to activate transcription from the IL-2 promoter. T-cell-receptor activation also activates a second pathway branch involving Ca2+ signaling. Release of Ca2+ from intracellular stores activates the phosphatase calcineurin, which dephosphorylates the transcription factor NFAT, allowing it to be transported into the nucleus. The IL-2 promoter has binding sites for both AP-1 and NFAT, and only when these factors bind simultaneously is transcription activated. The binding of the two factors is cooperative: the complex binds DNA much more stably than either AP-1 or NFAT alone. Thus, the IL-2 promoter functions as an AND gate that confirms that both the MAPK and Ca2+ branches of the T-cell-signaling network have been activated, something that will only occur with robust and sustained T cell activation. reSPoNDINg to tHe StreNgtH or DuratIoN of aN INPut In addition to simply detecting the presence of an input, it can be very important for a cell to measure its amplitude (strength) or duration. In this section, we will look at different ways that signal output can be related to the strength or timing of input signals. We will also introduce ways of describing how different components of a signaling system (usu- ally different signaling proteins) interact with each other and modify each other’s behaviors to process information. The specific nature and wiring of these interrelationships—the network architecture —can provide the basis for surprisingly sophisticated and complex signaling devices (see Box 11.1 for a primer on network architecture). Throughout this chapter, we describe examples of network architec- tures that are capable of performing specific information-processing tasks. However, it is important to note that there is not a simple one-to- one relationship between network architecture and network function— a given architecture might actually perform a number of different processing functions, depending largely on the exact parameters that define the network links. One can only state that a particular archi- tecture type can perform a function or is often associated with that function. In Table 11.1, we list several commonly observed cellular network architectures and the range of behaviors with which they are often associated. Table 11.1 Network architectures and their associated behaviors cascade amplification Switchlike activation negative feedback limit output level increase precision of output level adaptation oscillation (nonrobust) Positive feedback amplification Switchlike activation Bistability, memory interlinked positive/negative feedback robust, tunable frequency oscillations interlinked, multiple positive feedback increased synchronization, precision of feed-forward, coherent Delay/filter for sustained input feed-forward, incoherent Pulse response, adaptation Box 11.1 Signaling network architectures At the simplest level, nodes in a signaling network represent individual signaling components, such as individ- ual proteins, while links are the regulatory relationships between these components. For example, a node could represent an enzyme like a kinase or phosphatase, while the link from this node to another node (a substrate) represents the reaction in which the enzyme modifies the substrate (Figure 11.12). Figure 11.12 Network representation of regulatory connections. (a) links between nodes can be either positive or negative. (b) a series of two negative links can be represented by a single (a) positive link node negative link node positive link. double negative positive = Links between nodes can be either positive or negative, depending on whether the action of the upstream node on the downstream node results in activation or repression. Often a link in a network diagram might actually represent a series of relationships that are mediated by a chain of one or more intermediate links. In these cases, a series of two negative links could also be represented by a single positive link (in other words, the sign of a series of links is the mathematical sum of the signs of the individual links). Aside from forming cascades (that is, a number of nodes linked in series), nodes can be linked in fan-in and fan- out configurations. A fan-in configuration is when one node is controlled by multiple upstream inputs. A fan-out configuration is when one node controls multiple output nodes (Figure 11.13). Figure 11.13 Fan-in and fan-out network architectures. FAN OUT FAN IN As described in the main text, feedback and feed-forward network architectures are responsible for many of the complex biological behaviors we see in signaling. Feedback is observed when the output from a given node fol- lows a path of links that returns to regulate the node of origin. Feedback loops can be either negative or positive (Figure 11.14). POSITIVE FEEDBACK NEGATIVE FEEDBACK Figure 11.14 Examples of positive and negative feedback loops. (continued) Signaling systems can respond to signal amplitude in a graded or a digital manner Cells can respond to changes in input levels in different ways. In some cases, the situation may require what is referred to as a linear or graded response: a response in which the system detects a range of input levels and produces an output that is proportional to the input level. For exam- ple, in a stress-response pathway or a hormone-detection pathway, it may be useful for the cell to tune the degree of its response to match the level of the stress or hormone. Alternatively, in other physiological situations, a switchlike or digital response may be needed: the cell must ignore inputs below a certain threshold and only respond to strong signals above this threshold. Such systems are often described as having nonlinear or all-or-none behavior. An all-or-none response system might be advanta- geous in development, for example, where a cell only adopts a specific developmental fate when a morphogen input is present above a specific threshold. What determines whether a cell signaling system will respond in a graded manner or a digital manner? First, it is useful to consider the input/out- put behavior of an enzyme that is, for example, regulated by the binding of a single activating ligand (Figure 11.16a). Here, we consider the concen- tration of ligand as the input and the fractional activity of the enzyme as the output. In this case, activation will follow a hyperbolic curve (identical to the binding curve or isotherm for the binding reaction—see Figure 2.7). At low to medium input concentrations, output activity will be approxi- mately linear with respect to input, although as the system approaches input ligand saturation, the output activity will level off to the maximal level. Thus, one can see that such simple signaling devices are intrinsi- cally linear in their response, at least when significantly below their satu- ration point. In the next two sections, we will consider different ways of (a) LINEAR DOSE–RESPONSE (low sensitivity) (b) SWITCHLIKE DOSE–RESPONSE (ultrasensitive) INPUT INPUT INPUT inactive active OUTPUT INPUT inactive active OUTPUT inactive INPUT INPUT inactive Figure 11.16 OUTPUT OUTPUT Linear versus switchlike activation curves. (a) a simple binding-driven activation process will yield a hyperbolic dose–response curve which, in the early part of the activation curve, is approximately linear. the sensitivity (change in output divided by change in input) of this type of response is relatively low. (b) more complex signaling systems can show a more switchlike response, in which the dose–response curve is sigmoidal. Within a narrow range of input values, switchlike devices have a high response sensitivity (ultrasensitivity). (c) Individual signaling molecules can show ultrasensitive responses if they are activated by input in a cooperative manner. this includes multisubunit allosteric enzymes (top) or enzymes regulated by multiple autoinhibitory domains (bottom). cooperative switches are illustrated on the right; the corresponding noncooperative enzymes are illustrated on the left for comparison. generating more switchlike responses; it is important to remember that there are often diverse molecular approaches for achieving a particular type of signaling behavior. An enzyme can behave as a switch through cooperativity It is possible for an enzyme to show switchlike behavior if it is activated in a highly cooperative manner (Figure 11.16b,c). For example, if the enzyme (or enzyme complex) has multiple binding sites for the activating ligand, and if binding of ligand to one site allosterically increases the favorability of binding at the other sites (that is, ligand binding is cooperative), then this enzyme will be activated in a highly switchlike manner. The enzyme Cooperative binding is discussed in more detail in Chapter 2 Figure 11.17 INPUT response curve (with increasing ligand) will be sigmoidal; that is, below a threshold point, adding input will lead to a sublinear increase in out- put. However, near and at the threshold point, adding input will lead to a very steep nonlinear increase in output level. Output will level off again, however, as ligand binding nears saturation and the enzyme approaches maximal activity. Such a response is said to be ultrasensitive: in some portion of the signal–response curve, a relatively small change in the level of input leads to a much larger than proportional change in the response (Figure 11.17). An example of a cooperative enzyme switch is protein kinase A (PKA), which was mentioned earlier in this chapter (see Figure 11.4c). The regu- latory subunit of PKA (which binds and inhibits the catalytic subunit) Graded (linear) versus ultrasensitive (nonlinear) responses. Increasing ultrasensitivity corresponds to increasingly sigmoidal activation curves. this sharper transition can approximate a threshold point—a level of input that is required for all-or-none activation. Processive and distributive phosphorylation are discussed in Chapter 4 has two cAMP-binding sites. Binding of cAMP to these sites is cooperative and leads to cooperative release and activation of the catalytic subunit. This mechanism results in a sigmoidal, switchlike activation of PKA in response to increasing cAMP concentration. This effect is amplified by the fact that inactive PKA is a heterotetramer (two catalytic and two regu- latory subunits, with a total of four cAMP-binding sites), increasing the cooperativity of cAMP binding. Networks can also yield switchlike activation There are also network solutions to generating a switchlike detection sys- tem (Figure 11.18). For example, specific types of signaling cascades can lead to ultrasensitive, switchlike responses. A simple cascade of enzymes that successively activate one another will generally produce a linear input/output response (assuming all enzymes are operating below sat- uration). In some cases, however, cascades that involve multisite phos- phorylation can show much more switchlike input/output responses (Figure 11.18a). Multisite phosphorylation may be distributive (one enzyme–substrate encounter leads only to one phosphorylation event, so two encounters are needed for multiple phosphorylation) or proces- sive (one enzyme–substrate encounter leads to multiple phosphorylation events). Increasing the number of distributive events, as well as increas- ing the number of cascade steps, can lead to a sharper, more switchlike input/output transition of the cascade as a whole. For each step, the input is the concentration of the upstream activating kinase. In the case of a distributive multisite phosphorylation event, the effect of an increased amount of activating kinase is multiplied, because it contributes to each of the individual stepwise phosphorylation reactions (this assumes, as is the case in actual cells, that phosphorylation can be reversed at a constant rate by the action of phosphatases). This leads to a more sigmoidal activa- tion curve. This is quite analogous to the way that increasing ligand con- centration will contribute to each step of binding in an allosteric enzyme with multiple ligand-binding sites leading to cooperative activation. This type of distributive kinase cascade is observed to contribute to switchlike behavior of the MAP kinase cascade in Xenopus oocytes. However, it is also important to point out that not all MAP kinase cas- cades show switchlike activation, despite the fact that all such path- ways involve three two-site phosphorylation steps. Other MAP kinase cascades, such as the yeast mating-response pathway, show a primarily linear response. In these cases, it is thought that factors such as scaffold proteins might increase the processivity of this process (making each step less distributive). Switchlike activation can also occur with what is termed zero-order ultrasensitivity (Figure 11.18b). Reversible modifications typically (a) INPUT (b) INPUT (modifying enzyme) constant demodifying enzyme OUTPUT zero-order ultrasensitivity ([substrates] >> K m) cascade of multistep phosphorylation (nonprocessive) OUTPUT (c) inhibitor INPUT INPUT INPUT inhibitor titration (high affinity) Figure 11.18 OUTPUT OUTPUT POSITIVE FEEDBACK Network mechanisms that can yield ultrasensitive (nonlinear) input/output responses. a multistep activation reaction, and the organization of several of these reactions into a cascade, can lead to ultrasensitive input/output behavior for the intact cascade. (b) Zero-order ultrasensitivity. If a protein is subject to a reversible enzymatic modification (such as phosphorylation), then a change in the amount or activity of modifying enzyme can lead to an ultrasensitive change in the level of modified target protein. this behavior requires that the target protein concentrations be higher than the Km of each of the two upstream enzymes (that is, both reactions are zero-order and thus insensitive to substrate concentration). (c) Inhibitor titration can also generate nonlinear activation. Increasing input will have no effect on output until the inhibitor is titrated out, whereupon there will be a sharp change in output. (d) a positive feedback loop can also increase ultrasensitivity. use different enzymes for the forward and reverse reactions—in the case of phosphorylation, kinases and phosphatases. Under some situ- ations, an increase in the amount or activity of the forward modifying enzyme will lead to an ultrasensitive increase in the amount of the modified target protein. For this to occur, the target protein concen- trations need to be higher than the Km of the two modifying enzymes (that is, they are operating at zero-order saturation kinetics—when the enzymes are fully saturated with substrate, so reaction velocity is at Vmax and independent of substrate concentration). Over time, the slight increase in modifying-enzyme activity exerts a disproportionate effect on the level of the modified target because the demodifying enzyme is already operating at full capacity and therefore cannot counteract the increase. Nonlinear activation can also be simply generated if there is a high-affin- ity inhibitor of the input. Increasing input will have virtually no effect on output until the inhibitor is titrated out, whereupon there will be a relatively sharp change in output (Figure 11.18c). Another common and powerful solution for achieving switchlike activa- tion is through a network that has strong positive feedback. Let us consider a cascade in which the output of the cascade acts to further activate the cascade itself (Figure 11.18d and Box 11.1). This is often the case when the system involves a protein kinase, where one acti- vated kinase molecule is able to phosphorylate and activate multiple additional kinase molecules (remember that most kinases are activat- ed by phosphorylation of activation-loop residues). Positive feedback networks of this type are often observed to display extremely sharp, switchlike input/output behavior. At low input levels, the pathway is poorly activated. However, near the threshold point, the pathway is sufficiently activated that its output begins to generate positive feed- back, which in turn acts to further provide input to the pathway. In this way, the pathway becomes explosively activated in a sharp, all-or-none manner. Positive feedback of this type can also lead to other interest- ing behaviors, which we will discuss below. Note that a constant, low level of an opposing activity (for example, a phosphatase) is required to prevent the spontaneous, uncontrolled activation of such a system at low stimulus levels. Signaling systems can distinguish between transient and sustained input In some cases, it is very important for a cell signaling system to meas- ure the duration of an input. For example, it may be advantageous to avoid turning on a costly response program in response to transient, noisy inputs, and instead only respond to sustained stimulation. This type of response is somewhat similar to the behavior of an automatic door that is programmed to stay open until a sustained period has passed during which no motion is detected, to avoid accidentally closing the door on a person. How might a signaling system be programmed to distinguish between a sustained and a transient input? One solution to this problem is a network with a coherent feed-forward architecture (Box 11.1 and Figure 11.19a). In such a network, the output from an upstream node fans out to two different pathway branches. In a coherent feed-forward architecture, these two branches reconverge to regulate a downstream node in the same direction or sign (thus the two branches have a “coherent” effect on the downstream node). A detector for a sustained input can be built from a coherent feed-for- ward network with two features: first, the speeds at which the signal travels down the two divergent branches of the feed-forward network must differ; and second, the downstream converging node must act as an AND gate that is only activated when both branches of the feed-forward architecture are transmitting positive signals. The resulting system will generate output only when stimulated by an input whose duration (a) INPUT (b) input fast branch slow branch transient input sustained input delay between fast and slow branch (c) active ERK MAPK INPUT early transcription factor fast branch AND gate output time FAST: phoshpho- stabilization P Fos1 gene of Fos1 protein SLOW: Fos 1 expression stable unstable OUTPUT Figure 11.19 OUTPUT Fos-mediated transcription rapid proteolytic degradation A network that responds only to sustained inputs: coherent feed-forward loop. (a) a network architecture that can distinguish sustained inputs from transient inputs is a coherent feed-forward loop. critical features are a time delay between a fast branch (blue) and a slow branch (purple), and a fan-in node that acts as an aND gate to integrate the signals from the two branches. temporal profile of input, fast branch, slow branch, and output of this network. a transient input pulse (shorter than the delay between the two branches) results in no output, since the signals from the two branches do not arrive at the terminal aND gate at the same time. a sustained input (longer than the delay between branches) results in output. In this case, both signals arrive at the aND gate simultaneously. (c) a signaling network with this architecture is used to detect sustained erk maP kinase (maPK) activity. erk has two effects on the transcription factor fos1. first, erk induces expression of fos1. this branch is slow, since it is dependent on transcription and translation. In the fast branch, erk directly phosphorylates fos1 and stabilizes the protein. Because of the time delay in the transcriptional branch, only sustained erk activity will lead to stable accumulation of fos1, and to fos-mediated transcriptional output. is longer than the difference between the time scales of the two pathway branches. If stimulated by a more transient input, the fast and slow branches will be activated at different times, and the convergent AND- gate node will not be activated. Only with stimulation that is more sus- tained than the time difference between the two branches will there be dual input activation of the convergent AND-gate node (Figure 11.19b). This type of coherent feed-forward architecture appears to be used to allow mammalian cells to detect the differences between sustained and transient activation of the Erk MAP kinase (Figure 11.19c). Some of the responses downstream of Erk are mediated by the transcription fac- tor Fos, which serves the role of an AND gate. This transcription factor requires two inputs: first, its protein expression must be induced; and second, because it is inherently proteolytically unstable, its stability must be enhanced. Both of these inputs are mediated by activated Erk, but with different time scales. Direct phosphorylation of the Fos transcrip- tion factor by Erk leads to protection against proteolysis; this represents a fast, direct link. Expression of Fos, however, is mediated by a cascade of events downstream of Erk activation (Erk phosphorylates intermediate transcription factors that increase transcription of the Fos gene, leading to increased Fos protein production). This expression path represents a slow, indirect link between Erk and Fos. Thus, only when there is a suf- ficiently sustained pulse of Erk activation will the Fos transcription factor be both highly expressed and stabilized, ultimately resulting in expres- sion of Fos-activated genes. INPUT Figure 11.20 Amplification in signaling cascades. Input can be amplified by a cascade of enzymes, such as kinases or proteases, that activate one another. amplification occurs because one activated enzyme molecule can activate many downstream substrate molecules. cascades only result in amplification under conditions in which the enzymes are freely diffusible and each downstream enzyme is present at higher concentration than its upstream activator. Protease cascades are discussed in Chapter 9 moDIfyINg tHe StreNgtH or DuratIoN of outPut It is critical for a signaling system to control the amplitude and duration of its output. Some responses in cells are transient, lasting only as long as the stimulus, or in some cases even less (such as in the case of adaptation, discussed below). Other signaling responses can last for very long time periods, even after the input stimulus is gone, thus representing a type of cellular memory (long-term changes in output after transient input). Here, we describe several common molecular and network mechanisms for tuning output and generating specific classes of behavior. Signaling pathways often amplify signals as they are transmitted A common challenge facing signaling systems is that an initiating stimu- lus might only be detected by a small number of receptor molecules, yet the signal needs to be propagated in a manner that leads to an output of altered cellular behavior involving huge numbers of downstream mol- ecules. Amplification of signals as they are transmitted along a pathway yields high final output levels from relatively low input levels. One fun- damental mechanism for signaling systems to amplify output is through the use of enzymes—one activated enzyme can act on many substrate molecules. For example, activation of one kinase molecule can lead to the phosphorylation of many-fold larger numbers of substrate molecules. Similarly, one activated molecule of adenylyl cyclase can generate many- fold higher equivalents of cAMP, which can then propagate signals to many more molecules of downstream effectors. Indeed, small signaling mediators such as cAMP often function to amplify the signal from a rel- atively small number of activated receptors to generate a massive and widespread response throughout the cell. In principle, even higher levels of signal amplification will occur if mul- tiple amplifying enzymes are linked into a cascade, such as a MAPK cas- cade, with each step potentially multiplying the degree of amplification (Figure 11.20). Proteases can also be linked in cascades, as in the acti- vation of thrombin in blood clotting, or caspase activation in apoptosis. However, this type of exponential amplification is not always observed in endogenous cascades. First, significant amplification at every step of a cascade requires that there be a higher concentration of each succes- sive enzyme in the cascade. This is not always the case, and the degree of amplification can be limited by saturating activation of the available downstream enzyme. Second, the organization of cascade components into multiprotein signaling complexes by scaffold proteins can cause some of the individual cascade steps to no longer serve as independent reactions, thereby dampening amplification. Thus, although most pathways amplify their signals at some point, the degree of amplification observed for each individual step can vary significantly. Negative feedback allows fine-tuning of output Signaling pathways are not always designed to maximally amplify their output; instead, it is often more important to control output levels to make them more precise. Relatively little is known about the ways in which the precision of pathway output levels is controlled, but evidence points to an important contribution of negative feedback loops. Negative feed- back loops can dampen signaling through a pathway, leading to a steady- state output that is lower than would be expected without the negative feedback. Such negative feedback loops are also thought to increase the precision of output levels by counteracting fluctuations in input level. For example, in pathways that utilize Ca2+ as a diffusible signaling media- tor, upstream receptor activation often leads to an influx of Ca2+ into the cytosol. The maximal increase in cytosolic Ca2+ concentration, however, is limited by a negative feedback loop: when cytosolic Ca2+ exceeds 0.6 μM, Ca2+ uptake into the mitochondria is induced. This negative feedback loop therefore acts to limit maximal output of the system (Figure 11.21a). Without such a mechanism, Ca2+ levels might easily rise to toxic levels. An elegant and clear example of how negative feedback loops can tune the precision of output was observed in a synthetic gene expression net- work (Figure 11.21b). When an externally inducible promoter was used to express green fluorescent protein (GFP), cell-to-cell variability in GFP levels was high, due to random events such as differences in the copy number of the plasmid containing the GFP gene. However, the addition (a) INPUT Ca2+ channel negative feedback limits output level high Ca2+ mitochondria OUTPUT >0.6 μM Ca2+ induces mito- chondrial uptake INPUT OUTPUT (b) no feedback GFP promoter GFP fluorescence/cell + negative feedback rep GFP promoter Figure 11.21 GFP fluorescence/cell Negative feedback control of output level and precision. (a) example of a negative feedback loop that limits output level. Stimuli that open ca2+ channels result in increased concentrations of cytoplasmic ca2+. However, ca2+ concentrations of >0.6 μm induce mitochondrial uptake of ca2+, thereby acting as a negative feedback loop to limit cytoplasmic ca2+ despite increasing input stimulation. (b) Negative feedback can also increase output precision. one example involves two synthetic gene expression circuits, one with a simple promoter, and another with a repressor (rep)-controlled negative feedback loop (negative autoregulation). on the right, the distribution of fluorescence per cell is depicted. expression of a green fluorescent protein (gfP) reporter is more precise in the presence of the negative feedback circuit—that is, there is less cell-to-cell variation in gfP fluorescence. Standard deviation from the average is represented by black bars. of a negative feedback loop, in which the inducible promoter not only expressed the GFP reporter but also a transcriptional repressor of that promoter, resulted in significant reduction of cell-to-cell variability. This is because the inhibitor sets up a kind of steady state or set point, where the activity of the inducer is balanced by the activity of the repressor. When- ever expression level rises beyond this set point, the negative feedback loop provides a self-correcting mechanism that represses transcription in proportion to the deviation; similarly, when transcription goes down, repression will be relieved and transcription will tend to revert to the set point. Presumably, similar negative feedback architectures are used in natural signaling pathways to increase the precision of output and damp- en the effects of random events, or “noise.” Adaptation allows cells to control output duration It can also be critical for signaling systems to control the duration of their output. Many sensory systems therefore display a property known as adaptation—upon stimulation with a sustained input, the system ini- tially responds with a burst of output, after which output levels return to their basal levels, even if the input stimulus remains at a high level (Figure 11.22). An adaptive response can be used to shut off output when it is harmful or toxic in excess, or costly in some other way for the organism. Perhaps more importantly, an adaptive system couples response to relative changes in the input, not to its absolute level, thus allowing the system (a) nonadaptive response adaptive response INPUT INPUT OUTPUT time OUTPUT time negative feedback with buffering adaptive node incoherent feed-forward with proportioner adaptive node FAST FAST FAST turns down upstream or output nodes by integrating system output Figure 11.22 turns down output nodes in proportion to system input Adaptation circuits limit duration of output and increase dynamic range of sensory systems. (a) In a normal, nonadaptive response, output level increases with input and remains high during sustained input. the dynamic range of sensing is limited as the system reaches saturation. By contrast, in an adaptive system, input-induced increase in output level is transient—the system automatically resets to basal steady state, even with sustained input. (b) two basic network architectures that can achieve adaptation. first, negative feedback loops (left) can allow a regulatory node (r) to monitor and integrate activity of the output node (o), and to down- regulate upstream or output nodes to return to basal steady state. the negative feedback loop must be slow enough that there is a transient delay before return to steady state. Second (right), a regulatory node (r) in an incoherent feed-forward loop can sense input levels and proportionally down-regulate output (o) to return to the basal steady state. again, the negative branch of the feed-forward loop must be relatively slow to allow for a transient output response. Note that the specific architectures depicted are examples, and that other specific architectures are possible for each class of adaptation circuit. to respond to a much wider range of input levels. By resetting output to its original basal level, adaptation allows the system to respond to further increases in input (as opposed to becoming quickly saturated, as occurs with nonadaptive sensing systems). The ability of adaptation to increase the dynamic range of input sensing is critical for many sensory systems, such as in our own vision, allowing our eyes to reset and operate under dramati- cally different ambient conditions, such as a dark room or bright daylight. Theoretical analyses suggest that there are two basic ways to achieve adaptation (Figure 11.22b). The first is with a negative feedback loop mediated by a regulatory node that acts to buffer output. In this case, the system sums over time (integrates) how much output is being trans- mitted through the pathway, then inhibits upstream signaling in a manner that returns it to a constant steady state. To achieve this, the two enzymes that act on the regulatory node (the output activity, and a basal enzyme that opposes it) must operate under saturating conditions (substrate concentration > Km) such that regulatory-node activity only reflects the change in output enzyme activity and is independent of sub- strate concentration. As with the coherent feed-forward examples above, the forward signaling loop needs to be faster than the negative feedback loop to generate a transient response (deviation from steady state) fol- lowed by adaptation. A second theoretical way to achieve adaptation is through an incoherent feed-forward loop (Figure 11.22b) in which a receptor node transmits a positive signal to both the output node and a regulatory node, while the regulatory node, in turn, acts to negatively regulate the output node in a feed-forward loop. Here, the regulatory node acts as a proportioner—it will negatively regulate the output node in proportion to the strength of the input signal that it receives, thus resulting in adjustment of the sys- tem back to a constant steady-state output. Once again, differences in the kinetics of the two branches of the feed-forward loop allow for transient output responses that deviate from steady state. One of the best biochemically characterized examples of adaptation in cell signaling is found in bacterial chemotaxis (Figure 11.23). Bacteria can sense a gradient of chemorepellent molecules and swim away from the source. To do this, the bacteria perform a biased random walk—they switch between a straight swimming phase and a tumbling phase in which they reorient direction. When moving to a region of higher chemorepellent concentration, the bacteria show an increased probability of tumbling, allowing them to sample new directions; when moving away from chemo- repellents, however, the probability of tumbling is lower, allowing them to swim with higher persistence. In essence, the cells sample in time wheth- er they are moving to higher or lower chemorepellent concentrations, and adjust their behavior accordingly. Mechanically, tumbling is caused by clockwise rotation of the cell’s flagella, which forces the flagella to act independently, while swimming is caused by counterclockwise rotation of the flagella, which then intertwine (because of the handedness of the heli- cal flagella) and act in concert to propel the cell forward. In a controlled setting, if a bacterium is exposed to a step increase in chemorepellent, a transient increase in tumbling frequency is observed, followed by a rapid (within seconds) return to a basal tumbling frequency. This adaptation is the fundamental basis of the biased random walk. How does the bacterial signaling system yield this critical adaptive behav- ior? The biochemical circuit for chemotaxis is shown in Figure 11.23d and e, and has the architecture of a negative feedback loop with adaptive node, as described above (see Figure 11.22b). Chemorepellent is detected by a (a) bacterial chemotaxis: a biased random walk (b) tumbling: clockwise rotation of flagella (cell reorients) swimming: counter-clockwise rotation of flagella (straight movement) (d) chemo- repellent tumbling probability cell behavior (e) INPUT: time OUTPUT: CW rotation of flagellar motor (tumbling) adaptation module: methylation- mediated negative feedback loop chemorepellent receptor methylation receptor FAST SLOW CheB CheR CheA OUTPUT: CheY Figure 11.23 Bacterial chemotaxis is controlled by a negative feedback adaptation system. (a) Bacteria avoid chemorepellents by a biased random walk. the cell alternates between a straight swimming phase and a tumbling phase ( p u r p l e c i r c l e s ) in which it reorients before swimming in a new direction. If the bacteria swim to a higher chemorepellent concentration, the probability of tumbling increases. (b) tumbling occurs when the bacterial flagella rotate clockwise. Swimming occurs when the flagella rotate counterclockwise; because of the natural helicity of the flagella, in this direction they intertwine and act in concert to propel the bacterium. (c) a stepwise increase in chemorepellent results in a transient increase in the probability of tumbling, followed by a return to the steady-state basal probability. thus, increasing chemorepellent always induces a transient increase in tumbling, regardless of starting chemorepellent levels. (d) a negative feedback circuit underlies the adaptative behavior. chemorepellent activates a chemotaxis receptor, which, in turn, activates the protein chea. chea activates two targets. first it rapidly phosphorylates chey, which then binds to the flagellar motor to induce clockwise (cW) rotation (tumbling). However, chea also activates cheB, a demethylase that inhibits the receptor. a counteracting enzyme, cher, promotes receptor methylation and activation. thus, the methylation system forms a negative feedback loop that can restore output to low levels after a delay. (e) the chemorepellent sensing network in (d) is presented schematically so the “negative feedback with buffering adaptive node” architecture can be more readily appreciated (compare with figure 11.22b). chemotaxis receptor, which in turn activates the histidine kinase CheA, which can then activate the response regulator CheY via phosphorylation. Active (phosphorylated) CheY binds to the flagellar motor to cause clock- wise rotation and tumbling. Thus, the increase in active CheY after stimu- lation explains the resulting increase in tumbling frequency. However, the bacteria’s receptors quickly undergo adaptation, returning to basal output levels even at a constant high level of input. Adaptation is mediated by a second set of reactions—methylation of the receptor on several aspar- tate residues, which increases the output activity of the receptors. Recep- tor methylation acts as the “regulatory” node that buffers system output. Methylation is controlled by two opposing enzymes: the methylase, CheR, has constant basal activity, while the demethylase, CheB, requires acti- vation by CheA. Thus CheA, which is activated by the receptor, has two activities: it will activate CheY, leading to tumbling; and it will activate CheB, which leads to receptor demethylation and down-regulation. This system will return the level of active CheY (and tumbling frequency) to a constant steady state but only after a transient deviation caused by the slower kinetics of the CheB-mediated negative feedback loop. Feedback can cause output levels to oscillate between two stable states Another important temporal response observed in biological regulatory systems is oscillation, when output levels fluctuate between states of high and low activity in a periodic manner. Examples of biological oscilla- tion include the cell cycle and circadian rhythms (daily cycles, for exam- ple in sleep or activity, that can persist even in the absence of external light/dark cues). Biological oscillators have been studied by numerous approaches, including the construction of synthetic oscillator systems. We will focus on several of these synthetic systems because of the key princi- ples that they demonstrate. Negative feedback is a core requirement for oscillations, but a simple negative feedback loop without sufficient steps or delays cannot achieve oscillation (Figure 11.24a). Such a system will move monotonically toward a reduced, single steady-state output (compared to the equiva- lent circuit without negative feedback). A circuit with at least three steps in the negative feedback loop, or with an explicit delay, can destabilize this steady state and yield minimal oscillators. However, such oscillators have been constructed in synthetic biology experiments and are found to depend on highly precise parameters, and often show damped oscillations. Biological systems, however, typically require oscillators that are highly robust—that is, they maintain their performance under a wide range of conditions and are relatively insensitive to perturbations of the system. More recent studies have shown that overlaying positive feedback on these simpler circuits can result in far more robust oscillators, with consistent amplitudes. Positive feedback, or any type of ultrasensitivity acting on the nodes that form the core negative feedback loop, acts to prevent an approach to a single steady state and makes the system more bistable— able to exist in two distinct stable output states, without a stable interme- diate steady state (bistability is further discussed below). In principle, any of the mechanisms of ultrasensitivity outlined in Figures 11.16 and 11.18 can be used to increase oscillator performance. A natural example of a robust oscillator is the cell cycle in cleavage- stage embryos of the frog Xenopus laevis (Figure 11.24b). Immediately after fertilization, the large amphibian egg undergoes repeated rounds of synchronous DNA replication and mitosis until ~1000 cells are gen- erated; remarkably, this cycle can be reproduced in vitro using cell-free Bacterial two-component signaling systems of this type are discussed in Chapter 4 (a) negative feedback negative feedback + delay time no oscillation poor, damped oscillations time delay destabilizes approach to steady state negative feedback + delay + positive feedback more robust oscillations with consistent amplitude time nonlinearity of positive feedback destabilizes approach to steady state (b) example of natural robust oscillator: Xenopus oocyte cell cycle negative Figure 11.24 positive feedback Wee1 cyclin feedback Robust oscillators require a core negative feedback loop with (nonlinear) Cdc25 APC/Cdc20 CDK bound to cyclin OUTPUT delay and nonlinear (ultrasensitive) nodes. (a) a negative feedback loop is required for oscillation, but on its own leads to monotonic approach to a reduced steady-state output, compared to a circuit with no negative feedback (dotted line). these minimal circuits cannot yield oscillation. a three- component negative feedback loop, or a two-component negative feedback loop with an intrinsic delay, can yield oscillations, but these are often damped and have inconsistent amplitudes. addition of positive feedback can yield robust oscillations with consistent amplitudes. Positive feedback causes nodes to respond in a nonlinear (ultrasensitive) manner, which systematically robust oscillations in CDK activity destabilizes approach to a single steady state. (b) the X e n o p u s embryo cell cycle is an example of a robust oscillator constructed from interlinked positive and negative feedback loops. the core negative feedback loop involves activation of the aPc/cdc20 complex by cDK–cyclin, which leads to destruction of cyclin. there are two positive feedback loops, which revolve around the inhibitory phosphorylation site in cDK. aPc, anaphase-promoting complex; cDK, cyclin-dependent kinase. Xenopus egg extracts. At the core of this cycle is a negative feedback loop in which the active cyclin–CDK complex phosphorylates the anaphase- promoting complex (APC), promoting the binding of its Cdc20 subunit (see Figure 9.12). The APC/Cdc20 complex then polyubiquitylates cyclin and targets it for destruction, thus turning the system off after a delay. However, overlaid on this system are two positive feedback loops that make the cyclin–CDK node act in a more ultrasensitive manner. Both of these act through a distinct inhibitory phosphorylation site on CDK (remember CDK is a complex multi-input node; see Figure 11.09). First, the Wee1 kinase that phosphorylates and inhibits CDK–cyclin is itself phosphorylated by CDK–cyclin and thereby inactivated and targeted for degradation (a double negative loop). Second, the Cdc25 phosphatase that dephosphorylates and activates CDK–cyclin is itself activated by phos- phorylation by CDK–cyclin. These two positive feedback loops are thought to enhance the performance of this critical cell-cycle oscillator, resulting in consistent amplitude oscillations. Cell-cycle oscillations are most appar- ent in rapidly dividing embryos; in most other cells, additional signaling inputs are required before a new cell cycle is initiated. Bistable responses also underlie more permanent outputs In simple signaling systems, changes in output are transient, persisting only as long as the input perturbation is maintained. When the input perturbation is removed, the system returns to its basal level—behavior akin to a buzzer that only stays on as long as the user pushes the input button. In some cases, however, signaling systems respond to inputs with a permanent change in output that persists beyond the duration of a tran- sient input. This behavior is analogous to a toggle switch, such as those commonly used to turn on a light—the light stays on after the switch is flipped, in the absence of any further input. This type of sustained change in the system after transient stimulation plays a critical role in complex behaviors such as development, learning, and immune response. How can this form of molecular memory—the conversion of a transient input into a permanent (or semipermanent) change in output—be achieved? A sufficiently strong positive feedback can lead to a system that shows bistability. In these cases, the state of the system depends on its history and starting conditions. These systems often show hysteresis; that is, the input level at which the system switches between the two output states will be different if one is moving up from low input to high input, or mov- ing down from high input to low input (Figure 11.25a). A hysteretic system could display memory by locking into a high-output state upon a transient increase in input (Figure 11.25b). If the basal input level is above the point required to reset the system, then the output will remain high even after a return to the basal input level. The memory can only be reset (reverting the system to the low-output state) by lowering input levels past the transition point for decreasing input. An extreme form of hysteresis is an irreversible system, in which no decrease in input level is sufficient to restore the system to the low-input state (Figure 11.25c). The power of this positive feedback architecture has been demonstrated in several synthetic biology experiments, in which transcriptional or sig- naling circuits have been engineered to show this type of “lock-on” mem- ory (Figure 11.25d). In these cases, activation of the system above an ON input threshold can stably toggle the system to a new activated state. A second distinct OFF input trigger must be induced to return the system to the inactive output state. Most permanent cellular changes, such as those that occur during development, are thought to involve this kind of strong positive feedback, lock-on circuitry. Cell-cycle regulation is discussed in more detail in Chapters 9 and 12 (a) INPUT positive feedback loop INPUT double negative feedback loop bistable switch increasing input decreasing input ON OFF INPUT (b) OUTPUT A B C OUTPUT hysteresis A start B C start ON OFF ON OFF INPUT OUTPUT INPUT OUTPUT “memory” C B A time C B A time (d) TOGGLE SWITCH: double negative feedback loop with two triggers INPUT 1: “write” IPTG “write” OUTPUT “write” “erase” IPTG aTc OUTPUT Figure 11.25 INPUT 2: “erase” aTc “erase” GFP time Circuits capable of memory: a sustained change in output level after a transient change in input. (a) Strong positive (or double negative) feedback systems can lead to bistability, in which the system can only exist in two distinct stable states (low and high output; no intermediate states). Bistable systems often show hysteresis—the transition point between low- and high-output states will differ with increasing versus decreasing input (dark brown arrow). (b) a hysteretic system could display memory by locking into a high-output state upon a transient increase in input (labeled c). If the basal input level (B) is above the point required to reset the system, then the output will remain high even after a return to the basal input level. the memory can be reset only by lowering input levels past the decreasing input transition point (a). (c) an extreme form of hysteresis is an irreversible system, in which no decrease in input level is sufficient to restore the system to the low-input state. (d) Synthetic toggle switches with memory can be constructed using this type of network design. one example involves a double negative feedback loop, in which the tet repressor (tetr) inhibits expression of the lac repressor (lacI), and the lac repressor inhibits expression of both the tet repressor and a green fluorescent protein (gfP) reporter gene. the system can exist in a high lac repressor (and gfP) state or a high tet repressor state. the system can be induced to switch between states by adding small molecules that block the activity of one of the repressors (IPtg blocks lacI; atc blocks tetr). Summary Summary Cell signaling systems must be able to take in information and adjust their output states accordingly. Individual signaling molecules can them- selves function as complex gates and switches that process information. These proteins can also be hierarchically organized into networks that can perform higher-order information processing. Signaling molecules and pathways can integrate information from multiple inputs, and can exhibit different linear versus nonlinear input/output responses. These systems can also be organized to build networks that can monitor input amplitude and duration. Signaling networks can also be used to precisely control output amplitude and duration, resulting in complex dynamic behaviors such as adaptation, oscillation, and memory. Current work sug- gests that common network architectures are used to achieve particular functional signaling behaviors, even if the precise molecular implementa- tion is different for each case. QueStIoNS Describe the difference between feedback and feed-forward loops. What different classes of feedback loops are possible? What different classes of feed-forward loops are possible? When is it useful for signaling systems to behave as an OR rather than an AND gate? Describe different strategies by which the modular arrangement of domains or motifs within signaling proteins can facilitate the integra- tion of two separate inputs to make a coordinated decision. When is it advantageous for a cell to respond to input in a graded fash- ion, as opposed to a switchlike (or ultrasensitive) fashion? What are the mechanisms by which an individual signaling molecule can yield an ultrasensitive response? What are the mechanisms by which a signal- ing pathway or network can yield an ultrasensitive response? In what contexts is it useful for a cell to distinguish between transient and sustained input? What are the signaling network mechanisms that do this? When is it useful for a cell to display adaptation after sensing an input? What are the general molecular strategies for achieving precise adaptation? What is a bistable system? What kinds of physiological cellular res- ponses require bistable behavior, and why? For what physiological functions might a bistable response be nonoptimal? refereNceS SIgNalINg SyStemS aS INformatIoN- ProceSSINg DevIceS Alon U (2006) An Introduction to Systems Biology: Design Principles of Biological Circuits. Boca Raton, FL: Chapman & Hall/CRC. Lim WA, Lee CM & Tang C (2013) Design principles of regulatory networks: searching for the molecular algo- rithms of the cell. Mol. Cell 49, 202–212. INtegratINg multIPle SIgNalINg INPutS Hirota T, Lipp JJ, Toh BH & Peters JM (2005) Histone H3 serine 10 phosphorylation by Aurora B causes HP1 dis- sociation from heterochromatin. Nature 438, 1176–1180. Macián F, López-Rodriguez C & Rao A (2001) Partners in transcription: NFAT and AP-1. Oncogene 20, 2476–2489. Prehoda KE & Lim WA (2002) How signaling proteins integrate multiple inputs: a comparison of N-WASP and Cdk2. Curr. Opin. Cell Biol. 14, 149–154. Prehoda KE, Scott JA, Mullins RD & Lim WA (2000) Integration of multiple signals through cooperative regulation of the N-WASP-Arp2/3 complex. Science 290, 801–806. reSPoNDINg to tHe StreNgtH or DuratIoN of aN INPut Alon U (2007) Network motifs: theory and experimental approaches. Nat. Rev. Genet. 8, 450–461. Brandman O & Meyer T (2008) Feedback loops shape cellular signals in space and time. Science 322, 390–395. Dueber JE, Mirsky EA & Lim WA (2007) Engineering synthetic signaling proteins with ultrasensitive input/ output control. Nat. Biotechnol. 25, 660–662. Ferrell JE Jr (1996) Tripping the switch fantastic: how a protein kinase cascade can convert graded inputs into switch-like outputs. Trends Biochem. Sci. 21, 460–466. Murphy LO, Smith S, Chen RH et al. (2002) Molecular interpretation of ERK signal duration by immediate early gene products. Nat. Cell Biol. 4, 556–564. Tyson JJ, Chen KC & Novak B (2003) Sniffers, buzzers, toggles and blinkers: dynamics of regulatory and signal- ing pathways in the cell. Curr. Opin. Cell Biol. 15, 221–231. Whitty A (2008) Cooperativity and biological complexity. Nat. Chem. Biol. 4, 435–439. moDIfyINg tHe StreNgtH or DuratIoN of outPut Barkai N & Leibler S (1997) Robustness in simple bio- chemical networks. Nature 387, 913–917. Becskei A & Serrano L (2000) Engineering stability in gene networks by autoregulation. Nature 405, 590–593. Elowitz MB & Leibler S (2000) A synthetic oscilla- tory network of transcriptional regulators. Nature 403, 335–338. Gardner TS, Cantor CR & Collins JJ (2000) Construc- tion of a genetic toggle switch in Escherichia coli. Nature 403, 339–342. Ma W, Trusina A, El-Samad H et al. (2009) Defining net- work topologies that can achieve biochemical adapta- tion. Cell 138, 760–773. Novák B & Tyson JJ (2008) Design principles of bio- chemical oscillators. Nat. Rev. Mol. Cell Biol. 9, 981–991. Stricker J, Cookson S, Bennett MR et al. (2008) A fast, robust and tunable synthetic gene oscillator. Nature 456, 516–519. Tigges M, Marquez-Lago TT, Stelling J & Fussenegger M (2009) A tunable synthetic mammalian oscillator. Nature 457, 309–312. Tsai TY, Choi YS, Ma W et al. (2008) Robust, tunable bio- logical oscillations from interlinked positive and nega- tive feedback loops. Science 321, 126–129. Yi TM, Huang Y, Simon MI & Doyle J (2000) Robust perfect adaptation in bacterial chemotaxis through inte- gral feedback control. Proc. Natl Acad. Sci. U.S.A. 97, 4649–4653. How Cells Make Decisions In the previous chapters of the book, we introduced the molecular components of cell signaling systems and examined how these components link together to build more complex devices and networks. In this chapter, we turn to a more classical physiological view, and examine several model signaling pathways in an effort to synthesize earlier concepts and to provide a broader context of how cells employ signaling machinery to make important physiological decisions. We focus on four physiological processes: Vision—how our eyes detect and process light signals; Growth factor signaling—how cells detect and process signals that instruct them to proliferate; The cell cycle—how cells control their replication and division; T lymphocyte activation—how key cells in our immune system are mobilized to fight infection. While these examples represent only a few of the myriad complex decisions that the cells in our body must make with extreme preci- sion, they epitomize common problems that cells face, such as how sig- nal integration is used to make very specific decisions; how signals are propagated and amplified through cascades; and how feedback networks control the amplitude and timing of signaling. The chapter is organized in a visual-interrogative style, as a series of panels with questions. The first panels of each section introduce the phys- iological signaling system at different scales—from the organismal, to the cellular, to the molecular levels. Subsequent panels in each section are introduced with questions and focus on key functional problems that cells need to address to perform specific tasks. Within these panels, we dis- cuss the molecular and network-level mechanisms that cells have evolved to solve these problems. It is important to keep in mind that these sys- tems are extremely complex and still the subject of considerable ongoing research. Thus, the models shown here represent simplified versions of our current understanding, and there is little doubt that future research will lead to modifications of these models. Nevertheless, the types of mechanisms and solutions shown in these pathways are also employed by many other cells in diverse physiological processes, and exemplify funda- mental questions and principles of cell signaling. Figure 12.1 The human eye. When light enters an eye and hits the retina, it initiates a signaling cascade that transmits a visual signal to the brain. The first section of this chapter explores this transformation. Figure 12.3 Cell division. The image to the right shows a pair of cells dividing synchronously during mitosis. Control of the cell cycle is discussed in the third section of this chapter. The final section of the chapter explores how our bodies fight infection. Courtesy of George von Dassow. Figure 12.1 A healing wound. Rapid cell proliferation is essential to wound healing. However, cell proliferation must also be controlled and cease. The second section of this chapter explores these processes. VERTEBRATE VISION: how photoreceptor cells sense and amplify light inputs As organisms evolved and became more complex, they also developed increasingly sophisticated sensory systems which improved their ability to interact with the surround- ing environment. One of the most sophisticated and best understood sensory systems is vision. Light that we sense enters the eye, which is an extremely complex organ, and this light hits specific photoreceptor cells in the retina. There are two basic types of photoreceptor cells, named for their overall cell shape: rods, which have the highest sensitivity to low levels of light, and cones, which are used primarily for color vision. Activation of these photoreceptor cells by light ultimately leads to propagation of a neuronal signal through the optic nerve to the visual cortex in the brain, where these signals are processed to yield the overall perceived image that we “see.” Here, we focus specifically on the question of how the molecular signaling machinery in the photoreceptor cells of the eye functions to detect light and convert it to a signal that can be efficiently transmitted to the optic nerve. Our everyday experience reveals many aspects of vision that make it a particularly useful sensory system. We can see even when it is quite dark, because rod photoreceptor cells are very sensitive to low levels of light. We can also see when it is much more bright, although it can take a few sec- onds to adjust to dramatic changes in brightness, because the visual system has a wide dynamic range and can adapt to prolonged stimulation by ambient light levels. At the same time, the visual system has high temporal resolu- tion—when we close our eyes, it immediately appears dark because signals are turned off rapidly. This fast resolution allows us to respond to rapidly changing cues, such as an approaching baseball, which can take less than a second to travel from the pitcher’s hand to the batter. Here, we describe how a rod photoreceptor cell functions, focusing on the cellular and molecular signaling events that occur in the cell upon light stimulation. In particular, we will focus on the following questions: how does the photoreceptor cell convert light into a biochemical signal that can be transmitted to the brain? how does the photoreceptor cell detect low levels of light, even a single photon? how can the response be so rapid? how does the photoreceptor cell reset itself quickly to allow for detection of further increases in light? PanEL 1 PanEL 2 organ | vertebrate eye CELL | photoreceptor cell lens cornea retina DARK STATE LIGHT STATE cation channels OPEN cell cation channels CLOSED pupil iris sclera cell depolarized hyper- polarized dis membrane optic nerve retina epithelial cells horizontal cells nucleus light bipolar cells amacrine cells ganglion cells synaptic region OUTPUT light activation of ganglion cells; signal to brain via optic nerve high glutamate release inhibition of ganglion cells low glutamate release activation of ganglion cells The verTebraTe reTina | light input detected by the retina leads to an action potential in ganglion cells that transmits information to the brain Light from the environment enters the eye and passes through the outer layers of the retina until it reaches the photorecep- tor cells. Light stimulation of photoreceptor cells initiates a cascade of cell–cell communication that leads to the activation of ganglion cells, which transmit the visual signal to the brain via the optic nerve. Note that light must first pass through the ganglion cells to reach the photoreceptor cells, and then the signal propagates back to the ganglion cells. The cell-signaling processes begin when light induces hyperpolarization of the photoreceptor cell. Rod cells are the class of photoreceptor cells most sensitive to low levels of light. phoTorecepTor cell (rod cell) | light causes hyperpolarization of photoreceptor cells, inhibiting glutamate release In the dark state, cation channels in the photoreceptor cell membrane are open, and the cell is depolarized. In this depolarized state, the cell releases high amounts of the neurotransmitter glutamate, which activates the bipolar cells, which in turn inhibit the ganglion cells. Light leads to hyperpolarization of the photoreceptor cells by closing cation channels in the plasma membrane. Hyperpolarization in turn inhibits glutamate release from the cell, ultimately activating the downstream ganglion cells. This light-induced response is initated through a molecular signaling network in the photoreceptor cells, as shown in the next panel. PanEL 3 MoLECULar nETWorK | visual transduction cascade DARK STATE LIGHT STATE PHOTORECEPTOR CELL light disc lumen rhodopsin HIGH cGMP cGMP cGMP cGMP cGMP G cGMP cGMP DE cGMP cGMP generates G α PDE6 transducin degrades cGMP LOW disc membrane guanylyl cyclase cGMP Na+ K+ cyclic- nucleotide- gated (CNG) channel guanylyl cyclase high cGMP OPEN channel: depolarization low cGMP CLOSED channel: hyperpolarization signaling neTwork in The phoTorecepTor cell disc membrane | light activates a signal transduction cascade in the photoreceptor cell that leads to closing of cation channels and hyperpolarization of the cell In the dark state, the enzyme guanylyl cyclase is consti- tutively active and produces a high level of the second messenger cyclic GMP (cGMP) from GTP. cGMP binds to and allosterically opens the cyclic-nucleotide-gated (CNG) channel, which allows cations such as sodium and potas- sium into the cell, leading to a constitutively depolarized state. The photoreceptor cell is thus constantly utilizing energy, in the form of GTP, to maintain itself in a highly responsive state. When light input strikes the photoreceptor cell, it acti- vates the G-protein-coupled receptor (GPCR) rhodopsin, which is highly concentrated in the disc membranes on the outer segment of the cell. Activated rhodopsin in turn activates the heterotrimeric G protein alpha subunit (Gα) named transducin. One of the major downstream targets of activated transducin is the enzyme phosphodiesterase 6 (PDE6). PDE6 hydrolyzes cGMP (to GMP) leading to a rapid decrease in cGMP concentrations. Under these low-cGMP conditions, the CNG channel closes, leading to the transient light-induced hyperpolarization of the cell. 1 How does the photoreceptor cell detect light and convert it to a biochemical signal? lighT-indUced conFormaTional change The photoreceptor cell contains signaling proteins that can convert light into a series of protein conformational changes. These conformational changes trigger changes in enzymatic function. THE RECEPTOR RHODOPSIN CONVERTS LIGHT INTO A CHANGE IN CONfORMATION, AND A CHANGE IN ENzy- MATIC ACTIVITy Incoming light is sensed by rhodopsin, a membrane- bound G-protein-coupled receptor (GPCR). Rhodopsin is composed of retinal, a light-sensitive cofactor, covalently linked to the protein opsin. When retinal absorbs a photon, it isomerizes from the 11-cis to the all-trans conformation. This results in a conformational change in the opsin protein which activates rhodopsin for downstream signaling by reorganizing the transducin-binding site. Active rhodopsin acts as a GEf enzyme that activates transducin. INACTIVE conformation rhodopsin light ACTIVE conformation transducin- disc lumen cytosol Rhodopsin binds the cofactor retinal, which isomerizes upon stimulation with light. The change in retinal structure forces a major change in the protein conformation. binding surface light 11-cis retinal all-trans retinal A G PROTEIN AND SECOND MESSENGER CASCADE LEADS TO CHANNEL CLOSING The activated state of rhodopsin PDE6 GDP GTP cGMP CNG channel Na+ a starts a biochemical cascade that leads to CNG channel opening and hyperpolarization of the photoreceptor cell. α γ α β b c decrease in K ⁺ cGMP GMP d Like other GPCRs, activated rhodopsin acts as a GEF to activate a heterotrimeric G protein, in this case the protein transducin, by causing it to exchange bound GDP for GTP. The activated alpha subunit of transducin (GTP bound) dissociates from its beta and gamma subunits, and binds to the effector protein phosphodiesterase 6 (PDE6), allosterically activating it. Active PDE6 hydrolyzes the second messenger cyclic GMP (cGMP) into GMP. Reduction in the concentration of cGMP leads to closing of the CNG channel, blocking entry of Na+ and Ca2+ ions, thus leading to hyperpolarization of the cell. G proteins are discused in Chapter 3 Second messengers like cGMP are discussed in Chapter 6 GUANyLyL CyCLASE AND PHOSPHODIESTERASE ACT AS OPPOSING “WRITER” AND “ERASER” ENzyMES CONTROLLING cGMP LEVELS. cGMP level in the photoreceptor cell ultimately determines if the cell propagates a signal to the brain. The second messenger molecule, cyclic guanine monophosphate “WRITER” guanylyl cyclase light “ERASER” PDE (cGMP), is one of the key regulatory nodes in photoreceptor sig- naling. The concentration of cGMP ultimately determines whether the photoreceptor cell is depolarized and sends a signal to the brain. The intracellular concentration of cGMP is controlled by the balance between two opposing enzymes, the “writer” enzyme gua- GTP HIGH LOW cGMP GMP no signal to brain signal to brain nylyl cyclase that synthesizes cGMP from GTP, and the “eraser” enzyme PDE that degrades cGMP to GMP. In the dark state, the writer is more active, leading to high cGMP. With light stimulation, the eraser also becomes highly active, thus resulting in a transient decrease in the level of cGMP. Note that, although not explicitly shown, the cell is constantly paying energy for this highly respon- sive signaling system, in the form a constant resynthesis of GTP. Second messengers like cGMP are discussed in Chapter 6 2 How is the photoreceptor cell able to detect low light, even a single photon? enZYmaTic ampliFicaTion Several steps in the signaling pathway help convert a single photon into hundreds of active molecules. The visual transduction cascade relies on enzymes and small signaling mediators, which can convert a small number of input molecules into many more active output molecules. This allows the system to respond in a robust manner even when stimulated by very few photons in dim light. In fact, even a single photon can lead to a measur- able response from the photoreceptor cell. Amplification occurs primarily at two steps. First, when a rhodopsin molecule is activated by a single photon, it can activate over a hundred molecules of transducin per second, each of which can go on to activate a PDE6 molecule. Second, each PDE6 molecule, in turn, hydrolyzes about 1000 molecules of cGMP per second. The resulting change in cytosolic cGMP concentration results in the closing of a few hundred cation channels. This is sufficient to hyper- polarize the membrane and suppress glutamate release by the photoreceptor cell. Amplification is discussed in Chapters 3 and 11 light one rhodopsin molecule absorbs one photon 100 transducin molecules are activated by rhodopsin 100 PDE molecules are activated by transducin 100,000 cGMP molecules are hydrolyzed by active PDE 250 CNG channels close How can the response be so rapid? spaTial organiZaTion The organization of the signaling proteins and their properties lead to efficient and rapid communication. ONE Of THE MOST REMARKABLE ASPECTS Of VISUAL SIGNALING IS ITS SPEED. SEVERAL fACTORS ACCOUNT fOR THIS RAPIDITy Signaling proteins are very densely packed on the surface of the rod disc membrane (rhodopsin, transducin, and PDE6 cover 25%, 10%, and 1% of the membrane surface, respectively). This ensures that activated rhodopsin almost immediately encounters trans- ducin, and that activated transducin almost immediately encounters PDE6. Two-dimensional diffusion on the membrane increases the likelihood of productive interactions compared to three-dimensional diffusion in solution. PDE6 is almost a “perfect enzyme”: once activated, it operates nearly at the diffusion-limited rate, so virtually every cGMP molecule it encounters is converted to GMP. Signal output depends on cGMP and cations, which diffuse very rapidly due to their small size. disc membrane rhodopsin transducin PDE Diffusion and its role in reaction rates is discussed in Chapters 6 and 7 How does the photoreceptor cell reset itself to enable detection of further changes in light? adapTaTion The photoreceptor cell signaling network contains several negative feedback loops that mediate adaptation. NEGATIVE fEEDBACK LOOPS The activation of the photoreceptor cell (hyperpolari- zation) is only transient, even in constant light. The cell is able to automatically reset itself to its original baseline output (depolarized state). This sensory adaptation is critical to allow the cell to respond to further increases in light stimulus, thus giving the cell a much higher dynamic range of light detection. These adaptive mechanisms are part of what allow the visual system to function in a wider range of ambient light conditions. The adaptation of the photoreceptor cell involves at least three characterized negative feedback loops, outlined below. NEGATIVE FEEDBACK LOOPS IN THE VISUAL CASCADE b light rhodopsin transducin PDE6 cGMP--> GMP CNG channel closing hyperpolarization arrestin recruitment a GPCR kinase (GRK) Ca2+ c guanylyl cyclase G protein down-regulation is discussed in Chapters 3 and 8 Second messengers like cGMP are discussed in Chapter 6 PhoSPhoRyLATIoN FEEDbACk Activated rhodopsin binds and allosterically activates a GPCR kinase (GRk), which then phospho- rylates rhodopsin on multiple sites. These phosphorylation sites recruit the protein arrestin, which prevents rhodopsin from activating trans- activated rhodopsin activated GPCR stimulates GRK P P P arrestin binds to phosphorylated desensitized rhodopsin P P P ducin. This usually occurs within 200 milliseconds of activation. Arrestin also acts as an adaptor to couple rhodopsin to the endocytosis to phosphorylate the GPCR on multiple sites 1 ATP GPCR 2 ADP 3 arrestin machinery, resulting in internaliza- tion into endosomes. From the endosomes, rhodopsin may be recycled to the rod disc membranes, or it can be degraded in lysosomes. Thus, arrestin serves to desensitize rhodopsin after activation. GPCR kinase (GRK) • blocks transducin binding promotes receptor endocytosis 2 PDE6 has GAP activity that inactivates transducin GAP FEEDbACk Activated PDE6, in addition to hydrolyzing cGMP, has some GTPase-activator protein (GAP) activity that reciprocally catalyzes inactivation of transducin. GTP active form of transducin 1 (GTP bound) stimulates PDE6 GDP CALCIuM–GuANyLyL CyCLASE FEEDbACk Closing of the CNG channels after photoreceptor-cell activation also leads to a decrease in Ca2+ concen- tration in the cell. This decrease in Ca2+ also acts to increase guanylyl cyclase activity. Guanylyl cyclase activity counteracts PDE6 activity to restore a high level of cGMP, thus opening the CNG channels and restoring the photoreceptor cell to a depolarized state. PDE6 G α further activates guanylyl cyclase 3 increases cGMP cGMP Ca2+ lowers Ca2+ 2 CNG channel closes 1 In summary, the negative feedback loops rapidly restore the cell to a high cGMP/depolarized state that is ready to respond to further input stimuli. Negative feedback is discussed in Chapter 11 SUMMarY The g-protein-coupled receptor rhodopsin senses light, which triggers an intracellular signaling cascade leading to hyperpolarization of photoreceptor cells. The visual system is very sensitive, and uses amplification to convert small numbers of photons into a robust response. Protein density, diffusion, and reaction rates are optimized for rapid response. Multiple negative feedback loops rapidly down-regulate the phototransduction cascade, allowing changes in light to be detected with high temporal resolution. rEFErEnCES Burns ME & Pugh EN Jr (2010) Lessons from photoreceptors: turning off G-protein signal- ing in living cells. Physiology (Bethesda) 25, 72–84. Calvert PD, Govardovskii VI, Krasnoperova N et al. (2001) Membrane protein diffusion sets the speed of rod phototransduction. Nature 411, 90–94. Fu Y & Yau KW (2007) Phototransduction in mouse rods and cones.Pflugers Arch.454,805–819. Leskov IB, Klenchin VA, Handy JW et al. (2000) The gain of rod phototransduction: recon- ciliation of biochemical and electrophysiological measurements. Neuron 27, 525–537. Palczewski K (2012) Chemistry and biology of vision. J. Biol. Chem. 287, 1612–1619. PDGF SIGNALING: triggering controlled cell proliferation during wound healing While a developing and growing organism has many prolif- erating cells, in an adult body, most cells are quiescent (not growing), and rapid growth often is a hallmark of diseases like cancer. However, there are specific situations in which cells in the adult must be able to rapidly grow. Here we examine how fibroblasts, cells which most of the time are quiescent, are able to initiate rapid proliferation when the body is wounded. We explore how these cells are able to ini- tiate this response only when they receive precise signal- ing instructions, and how this response is transmitted and controlled. After an injury, a wound in the skin and underlying connec- tive tissue must be rapidly repaired to restore integrity of the tissue and to reconstruct a barrier to microorganisms, which will otherwise infect and colonize the wounded site. The process of wound healing requires that numerous sig- nals be transmitted between different cell types to elicit a coherent response. Upon wounding, blood spills into the site and the platelets it contains are exposed to components of the extracellular matrix. Integrin-mediated signaling then triggers platelets to release clotting factors, leading to the formation of the initial clot. The platelets also release growth factors, including platelet-derived growth factor (PDGF) and transforming growth factor β (TGFβ), which have several effects. One outcome is that they precipitate an inflammatory phase by recruiting neutrophils and mac- rophages to the wound. These cells engulf and kill invading bacteria and, in the case of macrophages, produce more PDGF. But how is new tissue made to repair the wound? Wound repair is mediated largely by a class of cells called fibro- blasts, which normally lie dormant in a quiescent state. However, upon injury, the fibroblasts near the site of the wound detect the release of PDGF and awaken. They migrate to the area of the wound, begin to proliferate, and secrete extracellular matrix proteins such as collagen that are needed to repair the damaged tissue. Thus, fibroblasts rapidly sense a chemical signal and initi- ate a diverse program of behaviors, which includes directed migration toward the wound, proliferation to produce more fibroblasts, and repair and remodeling of extracellular matrix. In this section, we will focus on the proliferation response, which must be exquisitely regulated because aberrant proliferation could lead to cancer. This is one example of many analogous mitogen response pathways that trigger similar sets of potent but tightly controlled proliferative responses. Here, we will focus on the following questions: how do fibroblasts detect and respond to the extracellular PDGF signal? how is this signal propagated within the fibroblast to trigger cell proliferation? how is the proliferative response terminated? PANEL 1 PANEL 2 TISSUE | process of wound healing CELL | fibroblast response to wounding & platelet activation INPUT tissue injury PDGF + TGFβ INPUT PDGF fibroblast cell platelets inflammation recruit neutrophils and macrophages to prevent infectio activate fibroblasts OUTPUTS cell migration to wound proliferation collagen deposition new extracellular matrix EPIDERMIS wound healing keratinocytes DERMIS fibroblasts collagen and elastin fibers OUTPUT wound healing fibroblasts Courtesy of Jan Schmoranzer WouND HEALING | platelets mobilize fibroblasts fIbrobLAst rEspoNsEs | migration, proliferation, and collagen deposition At the site of injury, damaged tissue activates blood platelets, which initiate clotting and also secrete factors such as PDGF that act upon fibroblasts in the connective tissue (dermis). Collagen secreted by the fibroblasts is then cross-linked to repair and strengthen the extracellular matrix, as well as to support the regrowth of epithelial cells of the overlying epidermis. The gradient of PDGF released by platelets is sensed by fibroblasts, and these cells respond is several ways. First, they migrate to the site of the wound. Second, they begin to rapidly proliferate. Third, they begin to deposit new collagen and other extracellular matrix proteins. PANEL 3 MOLECULAR NETWORK | control of fibroblast proliferation INPUT PDGF kinase P P P P P P P P SH2 Grb2 SH3 SH3 GEF Sos GDP GTP Raf Ras P Raf PDGF receptor SCAFFOLD ADAPTOR RasGEF P P MEK P OUTPUT expression of transcription factors P P P Erk P P Erk P proliferation genes proliferative genes fIbrobLAst proLIfErAtIoN pAtHWAy | low PDGF stimulation leads to expression of proliferation control genes such as Myc Human platelets contain two closely related chains of PDGF, A and B, which are synthesized as precursors, pro- cessed by proteolytic cleavage, and are then linked through disulfide bonds to form homo- or heterodimers. The mature PDGF dimers exert their effects on target cells by binding to the extracellular regions of the PDGF receptor (PDGFR), which has intrinsic tyrosine kinase activity. PDGF binding induces dimerization and autophosphorylation, producing at least nine phosphotyrosine sites that serve as docking sites for signaling proteins with SH2 and PTB domains. One such adaptor protein is Grb2, which recognizes phosphoryl- ated PDGFR through an SH2 domain and contains two SH3 domains that recruit Sos, a guanine nucleotide exchange factor (GEF) for Ras. Localization to the plasma membrane allows Sos to activate Ras, which itself can then activate the Erk MAP kinase pathway. Ultimately, activated Erk translocates to the nucleus, where it induces the transcrip- tion of proliferative genes, including cyclins (which drive the cell cycle) and Myc. Myc is a transcription factor that orchestrates a complex gene expression program required for cell growth and proliferation. Among other effects, Myc induces transcription of genes involved in glycolysis and metabolism, ribosome biogenesis, mitochondrial biogen- esis, DNA replication, and the cell cycle. 1 How do fibroblasts detect the local occurrence of a wound? A trANsmEmbrANE rEcEptor IN fIbrobLAsts sENsEs pDGf AND trANsmIts tHE sIGNAL Across tHE mEmbrANE INto cELLs The platelet-derived growth factor receptor (PDGFR) is a receptor tyrosine kinase. PDGFR ACTIVATION REQUIRES LIGAND BINDING AND DIMERIZATION PDGF binds to PDGFR with very high affinity (Kd ~10–10 M) so binding is favored even at low PDGF con- centrations. This is important so that fibroblasts can respond robustly to relatively weak input signals. Binding to PDGF induces dimerization of PDGFR monomers, and juxtaposition of the two cytoplasmic kinase domains stimulates their catalytic activity. Prior to activation, the kinase domain is inhibited through intramolecular interactions, with the net result that the activation segment of the kinase domain adopts a nonproductive con- formation so that residues important for formation of the active site are not properly positioned, and the substrate-binding site is blocked. This multilayered inhibitory device is important because the inappropriate activation of the PDGFR could have pathologic effects, as indeed is seen in diseases such as fibrosis, sclero- derma, and cancer, in which these constraints are overridden. Catalytic activation is achieved because the two adjacent kinase domains of the ligand-bound dimer are able to cross-phosphorylate one another on tyrosine residues that are essential for maintaining the autoinhibited state, and whose inhibitory effects are negated once they are phosphorylated. Thus, the proximity effect, initially induced by binding of PDGF to the PDGF receptor, is thereby converted by a phosphorylation-driven allosteric switch into catalytic activation of the receptor’s kinase domain. model of PDGF receptor extracellular domain dimer bound to PDGF PDGF PDGF receptor monomer PDGF receptor monomer Kinase activation is discussed in greater detail in Chapter 3 Information transfer across a membrane is discussed in greater detail in Chapter 8 2 How are PDGF signals propagated within the cell to generate outputs such as cell proliferation? muLtIpLE pHospHoryLAtIoN sItEs oN pDGfr ALLoW mEmbrANE rEcruItmENt of ADAptors/ EffEctors, LEADING to co-LocALIzAtIoN of sIGNALING mAcHINEry Receptor phosphorylation creates docking sites for SH2-domain-containing proteins. SH2-domain-mediated recruitment assembles key signaling complexes that propagate the signal intracellularly. PDGFR AUTOPHOSPHORYLATION CREATES BINDING SITES FOR MULTIPLE EFFECTORS, ULTIMATELY LEADING TO MULTIPLE OUTPUTS PDGFR is usually the most abundantly tyrosine-phosphorylated protein in PDGF-stimulated cells. It has at least nine autophosphorylation sites that are not directly involved in regulat- ing kinase activity, but rather serve as docking sites for cytoplasmic signaling proteins with one or more phosphotyrosine-binding domains (mostly SH2 domains). The autophos- phorylated receptor is therefore converted into a scaffold that recruits a range of proximal targets based on selective phosphopeptide–SH2 domain interactions. Phosphoinositide 3-kinase (PI3K) catalyzes the conversion of PIP₂ to PIP₃, an early step in the migration response. PI3K binds PDGFR through SH2 domains on its p85 regulatory subunit which, in turn, binds to the p110 catalytic subunit. Grb2 is an adaptor protein that recruits Sos to the membrane, leading to the proliferation response. INPUT PDGF PDGF PI3K p85 SH3 PI3K p110 P P P P SH2 kinase P kinase P Grb2 P SH3 P SH3 SH2 OUTPUTS PI3K Sos/Ras migration proliferation MEMBRANE LOCALIZATION OF SOS DRIVES THE PROLIFERATION RESPONSE In addition to its SH2 domain, Grb2 also possesses N- and C-terminal SH3 domains that engage proline-rich sequences in the C-terminal tail of Sos, which acts as a guanine nucleo- Ensuring that Ras is activated only in the correct time and place is critical. Indeed, hyperactive Ras mutations are frequently observed in cancer. tide exchange factor (GEF) for the Ras GTPase. These Grb2-mediated interactions concentrate Sos at the membrane, where it has access to its substrate Ras, which is itself anchored in the plasma membrane by its C-terminal isoprenyl modifica- tions. Upon recruitment, Sos catalyzes the exchange of Ras-bound GDP to GTP, thereby converting Ras from the inactive form to the active form. P P kinase P P SH3 P P P P SH2 Grb2 GEF Sos GDP GTP Ras GTPases and their activation by GEFs are discussed in greater detail in Chapter 3 Phosphorylation-dependent interactions are discussed in greater detail in Chapter 4 The role of subcellular localization in signaling is discussed in greater detail in Chapter 5 3 How is misactivation of the proliferation response prevented? KEy sIGNALING moLEcuLEs LIKE sos AND rAf fuNctIoN As sWItcHEs tHAt rEquIrE muLtIpLE INputs to bEcomE ActIvAtED Combinatorial gating of signaling molecules is a common theme for tightly controlling cellular response. THE GEF PROTEIN SOS IS REGULATED BY COMBINATORIAL INPUTS AND FORMS A POSITIVE FEEDBACK LOOP WITH RAS Sos is a multidomain protein with an N-terminal histone-like domain, followed by a Dbl homology (DH) domain, a pleck- strin homology (PH) domain, a helical linker, the catalytic REM-Cdc25 domain, and the Grb2-binding region at the C-terminus. In addition to its interaction with Grb2, Sos is also recruited to the membrane and activated by binding to the phospholipid phosphatidylinositol 4,5-bisphosphate (PIP₂) through its PH domain. In addition to the catalytic PIP2 GEF activity Grb2 site, Sos also binds Ras-GTP at a second site: an allosteric regulatory site in the REM-Cdc25 domain. Binding at this allosteric site is enhanced by binding of PIP₂, and it increases Sos catalytic activity. This constitutes a positive feedback histone DH Sos PH REM Cdc25 proline-rich motifs loop, as the product of Sos activity, Ras-GTP, serves to further activate Sos, leading to greater Ras-GTP production. PIP₂ and Ras-GTP at the plasma membrane cooperatively stimulate Sos, and these interactions may be initiated or stabilized by Grb2. The dependence of Sos on multiple inputs may ensure that it is buffered against erroneous acti- Ras GTP vation by weak upstream signals, while the positive feedback afforded by allosteric Ras provides a mechanism to turn on Sos rapidly and maintain it in the active state when a signal such as PDGF exceeds a threshold level. MULTIPLE INPUTS CONTROL ACTIVATION OF RAF The primary downstream targets for Ras-GTP are the Raf serine/threonine kinases, which possess an N-terminal domain that selectively interacts with Ras in the GTP-bound form. In the allow for additional upstream inputs, and co-localize it with downstream substrates. For example, the inactive Raf conformation is maintained by a 14-3-3 dimer that simultaneously binds two phosphorylated threonine sites that flank the kinase domain and clamps the kinase in the inactive state. For Raf activation, the N-terminal site must be dephosphorylated by the phosphatases PP1 or PP2A, relieving the 14-3-3 clamp. absence of a signal, the N-terminal region of Raf interacts with the kinase domain to repress catalytic activity; binding to Ras-GTP coordinately releases this autoinhibition and relocal- izes Raf to the plasma membrane. dephosphorylation 1 2 P GTP Ras Although binding to Ras is the primary signal for activation, Raf is also subject to several other regulatory controls that P PP1 P PP2A 14-3-3 Raf prevent its inappropriate activation, OFF OFF ON GTPases and kinases are discussed in greater detail in Chapter 3 The modular domain architecture of signaling proteins is discussed in greater detail in Chapter 10 Positive feedback loops are discussed in greater detail in Chapter 11 4 How is the proliferative response terminated? NEGAtIvE fEEDbAcK Loops ALLoW sIGNALs to bE tErmINAtED, so tHAt proLIfErAtIoN oNLy occurs for A LImItED tImE Both the active receptor and the downstream MAPK Erk trigger several negative feedback loops. P P RasGAP GTP Ras Ras PDGFR ACTIVATION ALSO RECRUITS PATHWAY DOWN-REGULATORS In addition to recruiting positively acting effectors such as PI3K and Sos (via Grb2), PDGFR phospho- tyrosine sites also recruit SH2-domain-containing enzymes that down-regulate the pathway, including a Ras GTPase-activator protein (which inacti- vates Ras by promoting the hydrolysis of GTP to GDP) and tyrosine phosphatases Shp1 and Shp2 1 P P P P Shp1/2 phosphatase P P receptor inactivation (which remove phosphotyrosine sites on PDGFR). Sustained activation of PDGFR also ultimately leads to its internalization into endosomes and possible 2 dephosphorylation degradation. This is, in part, mediated by recruitment of the ubiquitin E3 ligase Cbl. Cbl 3 receptor internalization ERK PARTICIPATES IN MULTIPLE NEGATIVE FEEDBACK LOOPS The Erk MAP kinase is also involved in several negative feedback loops. First, Erk suppresses signaling near the beginning of the pathway by phosphorylating the C-terminal tail of P P Grb2 P P P P P P GEF Sos P 1 negative GDP GTP Ras P Raf P Sos, and interfering with its binding to the Grb2 SH3 domains. Second, in addition to inducing proliferative genes, Erk induces the transcription of genes encoding dual-specificity phosphatases (MAP kinase phosphatases, or MKPs), which dephosphorylate the activation loop of the Erk and thus inhibit its activity. These relatively slow negative feedback loops allow attenuation of the pathway after activation. 2 dephosphorylation MKP phosphorylation P MEK P P Erk P Negative feedback loops are discussed in greater detail in Chapter 11 SUMMARY Platelet-derived growth factor (PDGF) is released by platelets at the site of a wound . In fibroblasts, PDGF is detected by the PDGFR, a receptor tyrosine kinase. Upon activation by PDGF, the PDGFR autophosphorylates, translating an extracellular input into intracellular signals. Autophosphorylation of PDGFR creates binding sites for several effectors containing SH2 domains, allowing one signal to fan out into multiple responses. The proliferative response to PDGF requires membrane localization of the guanine nucleotide exchange factor Sos, which leads to activation of the G protein Ras and subsequently to activation of the Erk MAP kinase signaling pathway. MAPK phosphorylation of key transcription factors leads to expression of proliferation genes. Combinatorial regulation and positive feedback loops prevent misactivation by inappropriate or weak inputs, but produce strong activation when appropriate. Multiple negative feedback loops coordinate attenuation of signaling to ensure proliferation is not erroneously maintained. REFERENCES Andrae J, Gallini R & Betsholtz C (2008) Role of platelet-derived growth factors in physiol- ogy and medicine. Genes Dev. 22, 1276–1312. Boykevisch S, Zhao C, Sondermann H et al. (2006) Regulation of ras signaling dynamics by Sos-mediated positive feedback. Curr. Biol. 16, 2173–2179. Demoulin JB & Essaghir A (2014) PDGF receptor signaling networks in normal and cancer cells. Cytokine Growth Factor Rev. March 19 (Epub ahead of print). doi: 10.1016/j. cytogfr.2014.03.003. Lemmon MA & Schlessinger J (2010) Cell signaling by receptor tyrosine kinases. Cell 141, 1117–1134. THE CELL CYCLE: generating sharp, irreversible transitions during cell proliferation Cell division is a hallmark of living organisms. For a single- celled organism, cell division is a means of reproduction. For multicellular organisms, cell division is required for development from an embryo, but is also crucial for main- taining the health of the organism. Thus, it is critical that cell division occurs in a precise and regulated manner. The eukaryotic cell cycle is a series of committed steps. For example, once a cell initiates DNA replication, the entire process must be completed; having two copies of some genes and only one copy of others for a prolonged period could lead to misregulation of many important pathways. Thus many cell-cycle steps occur in a switchlike (ultrasensitive) and irreversible manner. In this section, we will discuss some signaling mechanisms that promote sharp and deci- sive transitions between key phases in the cell cycle. It is also critical that each step in the cell cycle only occurs after the previous one is successfully completed. Failing to follow this strict order could result in severe consequences for the progeny of the division. In particular, the cell must ensure that genomic DNA is accurately replicated and seg- regated so that each daughter cell receives the appropriate complement of genes. Therefore, the cell cycle incorporates several checkpoints; each process cannot proceed until some prerequisites have been completed. Here, we will review two such checkpoints: the spindle assembly check- point and the DNA damage checkpoint. The cell cycle is a beautiful example of tightly choreographed and precisely timed signaling events. In this section, we will review how cyclin-dependent kinases act as a central switch to control the cell cycle, and how their interactions with key regulatory factors drive progression through the distinct phases of the cell cycle. Specifically we will address the following questions: how do the cyclin proteins regulate the cyclin-dependent kinases and control distinct phases of the cell cycle? how does the circuitry of the cell cycle drive sharp and irreversible transitions between the phases of the cell cycle? how does the cell arrest progression through the cell cycle when it detects critical problems? PANEL 1 CELL | distinct phases of the cell cycle KEY CELLULAR EVENTS AND TRANSITIONS IN THE CELL CYCLE S PHASE DNA replication 1 x DNA 2 x DNA G ₁ PHASE point of arrest of nonproliferative cells G1/S transition mitosis G ₂ PHASE G ₂ /M transition commitment to mitosis The four phases of the cell cycle are: (1) the G₁ (gap 1) phase; (2) S phase, when the DNa is replicated; (3) G₂ (gap 2) phase; and (4) M phase, when the replicated chromosomes separate (mitosis) and cell division takes place. Nondividing cells usually are paused in the G₁ phase. Mitosis can be further cytokinesis division anaphase M PHASE metaphase sister chromatids begin to separate sister chromatids held together by cohesin divided into several subphases; the two most central subphases are metaphase, when the paired sister chromatids align, and anaphase, when the sister chromatids are pulled apart by the mitotic spindle. Mitosis is followed by cytokinesis—when the cell constricts to become two new daughter cells. Key commitment transition points in the cell cycle are highlighted and include: STarT (just prior to the G₁/S transi- tion), which is a commitment to DNa replication; the G₂/M transition, which is a commitment to start mitosis; and the metaphase-to-anaphase transition, which represents a commitment to chromosome segregation and division. We will be focusing on what drives these major transitions, how they occur in a sharp and irreversible manner, and how they are blocked if something goes wrong. KEY TRANSITIONS CORRESPOND TO THE PERIODIC RISE AND FALL OF CYCLIN PROTEINS Changes in the phases of the cell cycle are driven by the rise and fall in the abundance of a related family of key regulatory proteins known as the cyclins. The levels of cyclins change periodically, with each one associated with a distinct phase of the cell cycle. The plot to the left shows the periodic fluc- tuations in major classes of cyclins: the G₁/S cyclins, which spike during STarT; the S cyclins, which rise during S phase; G1 cyclins G1 G1/S S M time metaphase anaphase cytokinesis and the M-phase cyclins, which rise during the beginning of mitosis, but disappear when anaphase begins. Note that these cyclins have different names in different organisms, and some organisms have multiple cyclins of the same class. Functionally, these cyclins play a crucial role in activating the cyclin-dependent kinases (CDKs)—the central switch regulat- ing the cell cycle—which are described in more detail in the following panel. Cyclins bind to the CDK and have two effects: first, the cyclins act as allosteric activators to shift the CDK into an active conformation (although full activation of CDK also requires phosphorylation on the CDK activation loop); second, the cyclins can act as adaptors that bind to specific substrates, thus, in part, determining what specific substrates CDK phosphorylates for a given phase of the cell cycle. PANEL 2 MOLECULAR NETWORK | cyclin-dependent kinase (CDK) is a central switch whose activity is modulated by the different cyclins CYCLINS ARE EXCHANGEABLE SUBUNITS THAT ACTIVATE AND DIRECT CDK ACTIVITY IN A PHASE-SPECIFIC MANNER OTHER FACTORS THAT REGULATE CDK ACTIVITY cyclin- dependent kinase complex phosphorylation G1/S cyclin output program activating phosphorylation CDK binding of CDK inhibitors cyclin (CDK) OFF S cyclin M cyclin ON (necessary for CDK activity in addition to cyclin binding) P ON OFF P OFF inhibitory phosphorylation The central molecular switch controlling the cell cycle is the cyclin- dependent kinase (CDK). This family of closely related kinases is defined by low activity unless associated with a cyclin subunit. In addition to activating the bound CDK, each cyclin recognizes distinct docking motifs on preferred substrates. Thus each cyclin directs CDK to phosphoryl- ate a distinct set of targets. These characteristic sets of substrates form the physiological program of that particular phase of the cell cycle: the G₁/S program triggers the start of the cell cycle; the S-phase program phosphorylates targets that trigger DNa replication; the M-phase program triggers the initiation of mitoisis, while the destruction of the M-phase cyclin then allows progression to anaphase and cell division. Cyclins are not the only way to regulate CDK activity. There are several other modes of regulation: CDK inhibitors: CDK–cyclin com- plexes can be inhibited by the binding of specific CDK inhibitor subunits. Thus, destruction of inhibitors can be critical for activating CDK. activating phosphorylation site: activation of CDK also requires phos- phorylation of Thr160 in the activation loop of the kinase (both pThr160 and CDK allosteric regulation is discussed in Chapter 3 MAKING A CYCLE: FEEDBACK LOOPS DRIVE RISE AND FALL OF CYCLINS, LEADING TO PHASE TRANSITIONS cyclin binding are required for CDK activity). regulatory kinases and phos- phatases can modulate this site. inhibitory phosphorylation site: CDK activity can be inhibited by phosphoryl- ation on specific Tyr and Thr residues. Cyclins and other regulatory factors control the cyclin-dependent kinase CENTRAL KINASE SWITCH regulatory kinases and phosphatases can also modulate these sites. (CDK) and direct it to execute the current cell-cycle phase program. How then does the cell transition to the next phase of the cell cycle? This system can progress through multiple phases because in addition to execut- ing the current phase program, CDK CDK regulators cyclins reg. kinases & phosphatases phosphorylation execute current phase program phosphorylation also leads to feed- back regulation that can help to end the current cycle phase and to drive entry into the next cycle phase. For example, active CDK complexes can proteolytic degradation of previous phase factors synthesis of next phase factors ubiquitin ligase regulatory proteins transcription regulators phosphorylate and activate ubiquitin ligase regulatory proteins, which can in FEEDBACK drives transition to next phase program turn lead to targeted ubiquitylation and degradation of factors associated with maintaining the prior cell-cycle phase. Destruction of these factors leads to exit from the prior phase. active CDK complexes can also phosphorylate and activate transcriptional regulators, thus driving the synthesis of factors that are associated with the next phase of the cell cycle. Expression of new factors thus helps drive entry into the next phase. 1 How are sharp and committed transitions between cell-cycle phases achieved? POSITIVE FEEDBACK LOOPS AND PROTEOLYTIC DEGRADATION MAKE KEY CELL-CYCLE TRANSITIONS SHARP AND IRREVERSIBLE Progression through many of the key transitions in the cell cycle is driven by positive feedback loops, which help make the transitions sharper and more switchlike (all-or-nothing). although the exact proteins involved are different at each stage of the cell cycle or in different organisms, many of the network-level circuit architectures are well conserved. In addition, several key steps in the cell cycle are driven by decisive proteolytic events that are inherently irreversible. Together, these mechanisms make transitions in the cell cycle reliable and committed. EXTERNAL STIMULI starting state of CDK increased phosphorylation of Whi5 POSITIVE FEEDBACK new state of CDK sharp entry into START transcriptional inhibitor transcriptional activator Whi5 SBF Whi5 phosphorylation inactive P Whi5 SBF expression of G ₁ /S cyclins (Cln1/2) transcription of G ₁ /S genes G ₁ /S-CDK OFF ON POSITIVE FEEDBaCK rEGULaTION OF G₁/S TraNSCrIPTION PrODUCES SHarP COMMITMENT TO STarT inactivate Whi5. Since Whi5 is a transcriptional inhibitor, this results in the initiation of gene expression by the positive transcription factor SBF. Genes associated with the G₁/S Entering STarT (leaving G₁ and beginning the G₁/S transi- program are expressed, begining entry into STarT. tion) represents the commitment to replicate the cell’s genome. This needs to be a sharp and decisive transition. an important transcriptional positive feedback loop con- tributes to making this a sharp transition. When the cell is in G₁ phase, genes associated with the G₁/S program are not expressed because of the presence of a transcriptional inhibitor—in yeast, this inhibitor is the protein Whi5. Whi5 interacts with and represses the function of the transcrip- tional activator SBF. at the initiation of STarT, the G₁-CDK enzyme is stimulated to initiate phosphorylation of Whi5. Phosphorylation acts to among these newly activated genes are those express- ing the G₁/S cyclins (in yeast, Cln1 and Cln2). Expression of these new cyclins leads to formation of the new G₁/S cyclin–CDK complex. This new form of CDK efficiently phosphorylates Whi5 (which leads to further SBF-driven expression of G₁/S genes). This positive feedback loop further drives the transition, such that even a small initial threshold amount of Whi5 phosphorylation will lead to an all-or-nothing commitment. The ultrasensitive increase in the activity of the G₁/S cyclin–CDK complex leads to a sharp entry into the STarT phase of the cell cycle. Ultrasensitive responses and positive feedback circuits are discussed in Chapter 11 POSITIVE-FEEDBaCK-DrIVEN DEGraDaTION OF CDK INHIBITOr PrODUCES SHarP ENTrY INTO S PHaSE an analogous positive feedback loop occurs during the transition to S phase. In yeast, S-cyclin–CDK complex begins accumulating during the G₁/S transition, but is maintained in an inactive state by a specific CDK inhibitor protein called Sic1. Sufficient G₁/S cyclin–CDK, however, can initiate phos- phorylation of Sic1 on several key sites. Phosphorylated Sic1 is recognized by the ubiquitin ligase complex SCF–Cdc4 and targeted for proteolytic degradation. as more Sic1 is degraded, the active form of S cyclin–CK is released. This newly activated CDK complex provides positive feedback by efficiently phosphorylating more Sic1, leading to further Sic1 degradation. Thus an initiating threshold level of Sic1 phos- phorylation will trigger this positive feedback and a sharp, ultrasensitive entry into S phase. starting state of CDK new state of CDK sharp entry into S phase POSITIVE FEEDBACK active S-phase CDK Sic1 S-CDK CDK inhibitor: Sic1 inactive pool of CDK (next phase) phosphorylation of Sic1 P P P P P phospho-Sic1 targeted for destruction by SCF^Cdc4 G ₁ /S-CDK Ubiquitin-directed proteolysis is discussed in Chapter 9 POSITIVE FEEDBaCK Or DOUBLE NEGaTIVE FEEDBaCK LOOPS arE a COMMON THEME IN DrIVING SHarP CELL- CYCLE TraNSITIONS The two examples of positive feedback loops described above, involved in two successive cell-cycle transitions, are shown in circuit form here. These circuits represent common themes in sharp cell-cycle transitions. Both are positive feedback loops based on double negative feedback (where activation occurs through inhibition of an inhibitor). In the case of the G -to-G /S transition, the transcriptional repressor Whi5 is inactivated by CDK phosphorylation. In the case of the G₁/S-to-S transition, the CDK inihibitor Sic1 is degraded in response to CDK phosphorylation. In addition, a negative feedback loop in which S-cyclin–CDK phosphorylates and inactivates the SBF transcriptional activator provides another key link that drives exit from the G₁/S program. Here, we have focused on the yeast circuits for these transitions, although analogous feedback circuits are observed for the same transitions in the mammalian cell cycle. 1 1 drives committed entry into G ₁ /S positive feedback (inhibition of transcriptional inhibitor) drives committed entry into S positive feedback (degradation of CDK inhibitor) Whi5 SBF G ₁ /S-cyclin expression Sic1 negative feedback (inhibition of transcriptional activator) drives exit from G ₁ /S How are sharp and committed transitions between cell-cycle phases achieved? (continued) POSITIVE FEEDBACK P M inhibitory phosphorylation Cdc25 phosphatase M Wee1 kinase mitosis OFF ON DOUBLE NEGATIVE FEEDBACK (= POSITIVE) time ultrasensitive transition DIrECT POSITIVE FEEDBaCK: CDK CaN POSITIVELY rEGULaTE ITS OWN aCTIVaTION TO DrIVE SHarP ENTrY INTO M PHaSE a complex comprising the M-phase cyclin (cyclin B) and Cdk1 is critical for initiating spindle assembly and other key mitotic processes. Prior to the onset of mitosis, Cdk1 is maintained in an inactive state by phosphorylation catalyzed by the inhibitory kinase, Wee1. When mitosis is set to begin, the inhibitory phosphorylation is removed by the activating phos- phatase, Cdc25. The activated Cdk1 molecules then participate in two positive feedback loops: Cdk1 phos- phorylates and activates more Cdc25 molecules (activator of Cdk1 activ- ity), while it also phosphorylates and inhibits Wee1 (inhibitor of Cdk1 activ- ity). as a result of these two positive feedback loops, entry into mitosis is an ultrasensitive event (see graph). The exact nature of the trigger that initiates the positive feedback loop is not fully understood. Feedback loops are discussed in greater detail in Chapter 11 PrOTEOLYSIS PrOVIDES aN IrrEVErSIBLE SWITCH: M CYCLINS aND SECUrIN arE KEY TarGETS FOr DEGraDaTION IN THE METaPHaSE-TO-aNaPHaSE TraNSITION Unlike most regulatory post-trans- lational modifications, proteolytic degradation is essentially irrevers- ible because regenerating the intact protein requires further protein synthesis. Ubiquitin ligases, which mediate polyubiquitylation of target substrates and their subsequent M Cdk1 cyclin B phosphorylation Cdc20 M P proteolytic degradation decrease in M-cyclin–CDK activity degradation by the proteasome, are key cell-cycle regulators. Degradation- based regulation ensures that the cell cycle progresses to the next stage and cannot go backward. a key cell-cycle regulator is the anaphase-promoting complex (aPC), which ubiquitylates substrates and targets them for degradation during the metaphase-to-anaphase transi- tion. The aPC is a large, multi-subunit P APC securin P release of separase: separation of sister chromatids progression to anaphase time complex, whose specificity is con- trolled by two primary activators: metaphase anaphase Cdc20 and Cdh1. When chromosomes are properly aligned in mitosis, the aPC is activated by Cdc20, marking the metaphase-to-anaphase transition. Two primary substrates of aPCCdc20 are securin and cyclin B. Degradation of securin leads to activation of separase, a protease that degrades the linkage that holds sister chromatids together (allowing their separation in anaphase). Degradation of cyclin B allows exit from mitosis. Regulated protein degradation is discussed in Chapter 9 2 How does the cell cycle ensure that each transition proceeds only under appropriate conditions? THE CELL HAS KEY CHECKPOINTS THAT WHEN TRIGGERED LEAD TO CELL-CYCLE ARREST The cell has sensor proteins that detect if critical problems arise during the cell cycle. If triggered, these sensor proteins initiate checkpoint programs that lead to cell-cycle arrest. Below, we discuss checkpoints for DNa damage (which results in blocking entry into or progression through S phase) and spindle misassembly (which results in blocking progression to anaphase and chromosome segregation). These checkpoint blockades give the cell time to correct problems that would otherwise lead to serious consequences, such as genomic instability or chromosome mis-segregation. DNa DaMaGE CHECKPOINT BLOCKS INITIaTION OF S PHaSE BY aCTIVaTING KINaSES THaT SENSE aND aMPLIFY DNa LESIONS One of the most important functions of the cell cycle is to replicate the genome accurately and segregate the chromo- somes equally between the two daughter cells. Failure to detect damage to genomic DNa could result in the trans- mission of mutations to a cell’s progeny. Indeed, cells have several proteins that link DNa damage to regulation of the cell cycle. In G₁, DNa damage pauses the cell cycle before S phase, thereby preventing any errors in DNa replication. Detection of DNa damage in G₂ results in a similar pause before initiating chromosome segregation. rounds of phosphorylation P ATM/ATR Chk2 Chk2 P DNA damage * P P P P P FHA P P P active P Chk2 Cdc25 degradation CDK-activating phosphatase expression of p21 CDK inhibitor G ₁ /S-CDK complex inhibited G1/S OFF DNa DaMaGE INDUCES FOrMaTION OF KINaSE COMPLEXES THrOUGH PHOSPHOrYLaTION-DEPENDENT INTEraCTIONS The response to DNa damage requires the action of aTM/aTr kinases. aTr and aTM phosphorylate oligomeric adaptor proteins, including rad9 and 53BP1, which also localize to sites of DNa damage. These phosphorylated adaptors are recognized by the Chk2 kinase through an FHa domain. Chk2 is then activated by phosphorylation, both by aTr and through autophos- phorylation, and released, thereby allowing additional molecules of Chk2 to localize to the signaling complex. This allows amplification of the response: a single site of DNa damage can activate several molecules of Chk2. activated Chk2 phosphoryl- ates Cdc25, which targets Cdc25 for degradation. In the absence of Cdc25, phosphorylation by Wee1 inhibits CDK2, the G₁/S-CDK. Chk2 also indirectly induces the transcription of p21, a CDK inhibitor, further ensuring that S phase is not initiated. Kinase activation and phosphorylation-dependent interactions are discussed in greater detail in Chapters 3 and 4 How does the cell cycle ensure that each transition proceeds only under appropriate conditions ? THE SPINDLE aSSEMBLY CHECKPOINT BLOCKS THE METaPHaSE-TO-aNaPHaSE TraNSITION USING a SENSOr PrOTEIN THaT CONFOrMaTIONaLLY DETECTS UNaTTaCHED KINETOCHOrES Initiation of anaphase before proper spindle assembly can result in after all the kinetochores are properly missegregation of chromosomes spindle assembled kinetochores not fully attached Before sister chromatids are sepa- rated, they must be properly aligned and attached to both spindle poles. Each chromatid is attached to the microtubules of the mitotic spindle by a specialized structure called the kinetochore. If the kinetochores proper segregation mis-segregation are not properly secured to spindle microtubules, sister chromatids could be mis-segregated, resulting in one daughter cell receiving two copies while the other daughter receives none. The spindle check¬point acts to ensure chromosome segregation occurs only attached. Mad2, a critical component of the spindle assembly checkpoint, has two primary binding partners: Mad1, which binds to unattached kinetochores, and Cdc20, which is also a key activator of the aPC. Cdc20 is cannot activate the aPC when it is bound to Mad2. In the absence of a binding partner, Mad2 exists in an open conformation (O-Mad2), where a “safety belt” region (pink, in the figure below) is bound tightly to the Mad2 core. In order to bind Mad1 or Cdc20, Mad2 must undergo a conformational change where the safety belt loosens so that it can wrap around the binding partner. In the ligand-bound closed conformation (C-Mad2), the safety belt interacts with another region of Mad2. safety belt N closes Mad1 or Cdc20 Spindle assembly checkpoint protein Mad2 can exist in two conformations, OPEN and CLOSED C N C N O-Mad2 intermediate ligand-bound C-Mad2 Kinetochores that are not attached to the spindle bind to a stable complex of Mad1 and C-Mad2. This copy of C-Mad2 can dimerize with soluble O-Mad2, and this binding loosens the safety belt, which allows binding to Cdc20 and converts the O-Mad2 molecule into a C-Mad2 molecule. In a similar fashion, C-Mad2 bound to Cdc20 will also potentiate the binding of O-Mad2 to Cdc20. Thus, C-Mad2 acts as a catalyst: a single unattached kinetochore can help convert many unbound O-Mad2 molecules to com- plexes of C-Mad2 bound to Cdc20. The binding of Cdc20 to Mad2 acts to block the metaphase-to-anaphase transition by sequestering Cdc20 and preventing it from activating the aPC. Thus anaphase cannot be induced until all the kinetochores are attached. Unattached kinetochores potentiate Mad2 binding to Cdc20 resulting in inhibition of the APC 4 Mad2 (closed) binds Cdc20 adaptor protein 5 titrates Cdc20 adaptor micro- tubules unattached kinetochore binds Mad1 binds Mad2 3 converts from open to closed conformation P OFF E3 protein away from APC metaphase sister chromatids open closed free Mad2 bound Mad2 catalytically converts free Mad2 to closed conformation APCCdc20 ubiquitin ligase complex BLOCKS Protein interactions are discussed in greater detail in Chapter 2 Conformational changes are discussed in greater detail in Chapter 3 SUMMARY Stages of the cell cycle are controlled by the cyclin-dependent kinases (CDKs) which associate with a series of phase-specific factors known as the cyclins. Cyclin levels rise and fall through the cell cycle, with distinct cyclins associated with each phase of the cycle. Specific cyclin–CDK complexes phosphorylate phase-specific targets, leading to the execution of that phase’s program. These specific CDK complexes also control the timing of the eventual transition to the next phase in the cell cycle. CDK complexes initiate many positive feedback loops (or double negative feedback loops), which play a central role in generating sharp and committed transitions between distinct phases of the cell cycle. These committed transitions are required for ratchet-like forward progression of the cell cycle. Some key steps in the cell cycle are regulated by ubiquitin-mediated proteolysis, which makes these transitions irreversible and committed. Cell-cycle progression can be arrested by checkpoint programs that detect key problems such as DNA damage or unattached kinetochores. These checkpoints provide time for the cell to fix problems that would be exacerbated if the cell cycle had continued progressing. REFERENCES A number of the figures in this section have been adapted from David Morgan’s book The Cell Cycle: Principles of Control. London: New Science Press. Bertoli C, Skotheim JM & de Bruin RA (2013) Control of cell cycle transcription during G1 and S phases. Nat. Rev. Mol. Cell Biol. 14, 518–528. Craney A & Rape M (2013) Dynamic regulation of ubiquitin-dependent cell cycle control. Curr. Opin. Cell Biol. 25, 704–710. Ferrell JE Jr (2013) Feedback loops and reciprocal regulation: recurring motifs in the systems biology of the cell cycle. Curr. Opin. Cell Biol. 25, 676–686. Fisher D, Krasinska L, Coudreuse D & Nov\aacutek B (2012) Phosphorylation network dynamics in the control of cell cycle transitions. J. Cell Sci. 125, 4703–4711. Johnson A & Skotheim JM (2013) Start and the restriction point.Curr.Opin.Cell Biol.25,717–723. Morgan DO (2007) The Cell Cycle: Principles of Control, London: New Science Press Ltd. Musacchio A & Salmon ED (2007) The spindle-assembly checkpoint in space and time. Nat. Rev. Mol. Cell Biol. 8, 379–393. Reinhardt HC & Yaffe MB (2009) Kinases that control the cell cycle in response to DNA damage: Chk1, Chk2, and MK2. Curr. Opin. Cell Biol. 21, 245–255. Teixeira LK & Reed SI (2013) Ubiquitin ligases and cell cycle control. Annu. Rev. Biochem. 82, 387–414. Artistic rendering of a human dendritic cell illustrating the sheet-like processes on its surface. Courtesy of Donald Bliss and Sriram Subramaniam. T CELL SIGNALING: controlling the launch of an adaptive immune response Some of the most remarkable examples of complex cellu- lar signaling processing are displayed by the lymphocytes of the mammalian adaptive immune system. The adaptive immune system recognizes molecules unique to invading pathogens, and subsequently activates sophisticated down- stream response programs that eliminate specific pathogen types, kill infected cells, and provide long-term immunity against future infection with that pathogen. In the course of an infection, pathogens initially encounter chemical mechanisms and specialized cells, such as neutro- phils and macrophages, which try to kill and eliminate the invading microbes and infected cells. This first, more gener- alized response is known as the innate immune response. In vertebrates, pathogens also initiate a second response, known as the adaptive immune response, which is much more specifically aimed at the particular invading pathogen. An adaptive immune response typically begins when antigen- presenting cells, such as dendritic cells, engulf pathogens or pathogen-derived molecular products. Inside dendritic cells, the pathogens or their derivative proteins are broken down into small peptide fragments. As a part of normal protein turnover in the cell, these foreign peptides bind to Major His- tocompatibility Complex (MHC) receptors in the endoplas- mic reticulum, and are then transported to the cell surface, where they are displayed on the outside of the cell as MHC– peptide complexes. Thus MHC receptors, in effect, display the intracellular protein diversity of the cell on its extracellular surface. The majority of MHC complexes expressed on the cell contain self-derived peptides; pathogen-derived peptides are referred to as antigens because of their ability to be rec- ognized as foreign, and to eventually generate antibodies and trigger other aspects of the adaptive immune response. Dendritic cells displaying a pathogen-derived antigen will migrate from sites of infection to lymph nodes where they interact with T lymphocytes, also known as T cells. T cells have specialized receptors on their surfaces, called T cell receptors (TCRs), that can bind with great specificity to foreign peptides displayed in MHC–peptide complexes on dendritic cells. Each individual T cell expresses a unique TCR variant that can recognize and bind only a very specif- ic set of antigenic peptide sequences. Therefore, each T cell recognizes only one or a few antigens. However, since ver- tebrates typically produce over a million different variant TCRs within an organism, the repertoire of T cells is capa- ble of identifying nearly all foreign antigens. (T cells that recognize self-antigens are eliminated during development, as such self-reactive cells would give rise to autoimmunity.) A T cell that has not yet been activated through recogni- tion of a specific foreign peptide is said to be “naïve.” A T cell becomes activated when it recognizes the specific, cognate antigen for its TCR, in addition to other co-stimulatory sig- nals provided by the antigen-presenting cell. Once a naïve T cells is activated, it begins to proliferate, in order to create more T cells with receptors specific to that particular anti- gen. These cells also differentiate into more specialized types of T cell, called killer (or cytotoxic) T cells and helper T cells, which perform specific functions that enable the adaptive immune system to eliminate pathogens. Helper T cells stim- ulate antibody production through B cells and also stimulate neutrophils. As the name suggests, killer T cells attack and kill infected cells. This multipronged response stimulated by antigen recognition and T cell activation is extremely power- ful and can eliminate many infections. In this section, we focus on the signaling events that allow T cells to detect foreign antigens presented in MHC–peptide complexes. Specifically, we will address the following questions: how does the TCR detect and propagate signals in the T cell? how does the T cell signaling system, which can be triggered by a very small number of non-self MHC–peptide com- plexes (≤10), lead to a potent adaptive immune response? since the vast majority of MHC–peptide complexes display self-peptides and not foreign antigen, how does the T cell sign- aling system recognize only foreign antigens and filter against activation by weak or transient self-peptide signals? PANEL 1 ORGANISM | launching the adaptive immune response LYMPH SYSTEM INVADING PATHOGENS INNATE RESPONSE neutrophils macrophages viruses phagocytosis bacteria ADAPTIVE RESPONSE dendritic cell mature dendritic cell (antigen-presenting cell) LYMPH NODE T cell activation antigen presentation naïve T cell proliferation & differentiation of effector T cells antibodies neutralize pathogens kill infected cells killer T cells LYMPH NODE ADAPTIVE IMMUNE RESPONSE stimulate B cells to produce antibodies stimulate neutrophils helper T cells The adapTive immune response is launched during an infecTion The adaptive immune response begins when specialized antigen-presenting cells, such as dendritic cells, take up pathogen-derived material in the periphery through phagocytosis. They then traffic to the lymph nodes, where they interact with naïve T cells. The body has an incredible diversity of naïve T cells, each with a distinct T cell receptor that can recognize potential antigenic peptides. If the antigen- presenting cell encounters a cognate T cell that recognizes one of the mol- ecules that it presents on its surface as an antigen, then that T cell becomes activated. Through a cytokine-stimu- lated autocrine loop, stimulated by the cytokine IL-2, that T cell will clonally proliferate and differentiate into killer and helper T cells (effector T cells). Together, these cells will launch a series of complex responses, including the killing of infected cells presenting the cognate antigen, and stimulation of B cells to produce cognate antibodies. These responses help neutralize the infection. PANEL 2 CELL | engagement of T cell and antigen-presenting cell anTigen presenTaTion To The T cell multiple inputs from the dendritic cell are required for T cell activation In the lymph node, dendritic cells and T cells will directly contact each other, literally crawling over one another, scan- ning for proper molecular interaction partners (panel a). Once the cells rec- ognize each other, they form extensive contacts, and full activation of a cognate T cell can require hours of sustained interaction with the antigen-presenting cell (here a dendritic cell). (a) (b) T cell activation occurs when cognate peptide antigen is presented by a dendritic cell dendritic cell digests proteins & presents peptides on surface through MHC complex MHC–peptide complexes NO TCR RECOGNITION NO RESPONSE Human dendritic cell (blue) interacting with T cell (yellow). Olivier Schwartz/Science Photo Library at the heart of the recognition process is the T cell receptor (TCR), a multi-protein cell-surface complex on the T cell which directly binds and recognizes the peptide–MHC complex on the dendritic cell (panel b). Most peptides displayed on the dendritic cell will be self-peptide– MHC complexes, which TCRs do not recognize. If however the dendritic PEPTIDE–MHC COMPLEXES DISPLAYED ON DENDRITIC CELL SURFACE complex, and are necessary for TCR activation. The CD28 co-receptor recog- nizes the ligand B7, which is a partner cell-surface protein in the dendritic cell. The activation process then leads to major cytoskeletal rearrangements in the T cell (leading to more extensive interac- tions between the cells), changes in COGNATE T CELL PROLIFERATES AND DIFFERENTIATES transcription (including expression of key secreted cytokines), and cell proliferation and survival. all of the receptor systems cooperate, providing tight control over whether the T cell actually undergoes sustained activation. The T cell thus acts like an anD gate that requires multiple inputs for activation. cell displays an antigen peptide that is recognized by the TCR with sufficient (c) co-receptors work with MHC–TCR complex to trigger T cell activation affinity and duration, then that T cell will become activated, leading to clonal proliferation and differentiation. Proper interaction between the TCR and MHC complex, however, is not sufficient for activation of the T cell (panel c). Multiple other cell–cell interactions are required, including the interaction of cell adhesion receptors (LFa1 with ICaM), co-receptor molecules such as CD4 (in helper T cells) and CD8 (in cytotoxic T cells), and other co-stimulatory recep- tors, such as CD28. The cell adhesion molecules are necessary to form a tight and extensive area of contact between the dendritic cell and the T cell. The cell adhesion LFA1 ICAM co-receptor CD4/CD8 co-receptor T cell receptor peptide–MHC TCR co-stimulatory receptor B7 CD28 co-stimulatory receptor CD4 and CD8 co-receptors participate directly in the formation of the TCR–MHC T cell activation PANEL 3 MOLECULAR NETWORK | T cell receptor (TCR) signaling network T cell acTivaTion signal propagation involves four main molecular complex modules dendritic cell 1 PEPTIDE–MHC ENGAGES TCR & CO-RECEPTOR 2 CD4 co-receptor Lck phosphatase: CD45 P kinase: Csk SH3 SH2 active Lck phosphoryl- ates ITAMS on TCR Lck P P P P P P P P P P P P P P P P P P P P ITAMs (inactive) Lck kinase is held in inactive basal state Lck (active) (kinase) ZAP-70 (kinase) P Co-receptor Lck complex TCR ZAP70 binds to phosphorylated ITAMS Lck phosphorylates & activates ZAP-70 activated TCR module T cell OFF STATE. Lck is a src family tyrosine kinase that is asso- ciated with the co-receptor (in this case, for a helper T cell, the co-receptor is CD4). Lck is the key kinase that phosphorylates and activates the TCR upon activation. Lck’s activity is basally regulated by opposing enzymes—the tyrosine kinase Csk, which phosphorylates and inhibits Lck, and the membrane- associated phosphotyrosine phosphatase CD45, which “erases” this phosphorylation and activates Lck. In the basal state, Lck is inactive. Exactly how peptide–MHC engagement with the TCR triggers Lck activity is unclear, but it is thought to involve the engagement of the co-receptor CD4 with the TCR– peptide–MHC complex, leading to transphosphorylation by other Lck molecules in the complex. These changes ultimately shift the balance in the amount of activated Lck. In the next panel, we discuss how Lck then activates the TCR. Activated TCR complex. Within minutes of stimulation by engagement with the antigenic peptide–MHC complex, Lck phosphorylates key tyrosine motifs in the TCR known as immunoreceptor tyrosine-based activation motifs (ITaMs). Phosphorylation of the ITaMs converts them into binding sites that are recognized by the tandem sH2 domains from the cytoplasmic tyrosine kinase ZaP-70. Once recruited to the TCR, ZaP-70 is also phosphorylated and allosterically activated by the co-localized Lck kinase. Thus the Lck kinase is responsible for both recruitment and activation of ZaP-70. In the next panel, we describe how ZaP-70 leads to assem- bly of a major signaling complex. Assembly of LAT/SLP-76 scaffold complex. The acti- vated and localized ZaP-70 kinase is now in a position to phosphorylate two of the most important substrates, the scaffolding proteins LaT and sLP-76. LaT is a membrane- associated protein with five tyrosine phosphorylation sites, while sLP-76 is a soluble protein with four tyrosine phospho- rylation sites. Phosphorylation of the key tyrosines in these scaffold proteins creates a set of sH2-domain-binding sites, and thus allows the assembly of a large multiprotein signaling assembly that is essential for T cell activation. The sH2- and sH3-domain-containing protein gaDs acts as an adaptor to mediate this assembly. The enzyme phospholipase Cγ (PLCγ), which also has sH2 and sH3 domains, forms a 3 4 assembly of LAT & SLP-76 scaffold complexes P PLC γ PH PLCγ cleaves PIP ₂ to generate small-molecule mediators (DAG and IP ₃ ) PIP2 P SH2 P SH3 SH3 SH2 SH3 DAG RasGRP (GEF) PKCθ Ras (G protein) Ca2+ calcineurin LAT GADS P P P SLP-76 ITK Sos Raf (MAPKKK) Mek (MAPKK) ZAP-70 phosphorylates scaffold proteins LAT & SLP-76 Nck VAV Erk (MAPK) LAT/SLP-76 scaffold/signaling complex CYTOSKELETAL REARRANGEMENTS AP-1 NF-κB transcription factors NFAT part of this complex and is activated as a result (PLCγ is illustrated here using a simplified domain representation; see Figure 10–21 for the complete domain structure). This complex is also responsible for activating the enzyme sos (a gEF for the gTPase Ras). The proteins nck (an sH2/ sH3 domain adaptor) and VaV (a gEF for the Rac g protein) are also recruited to this complex. nck and VaV are part of the machinery that triggers cytoskeletal rearrangements in the T cell (not discussed in detail here). In the next panel, we describe how PLCγ and sos trigger the core signaling pathways that ultimately lead to activation. Core signaling pathways lead to activation of key T cell genes. The activated LaT/sLP-76 complex leads to activation of a number of core signaling pathways. One of the key effectors is the enzyme PLCγ, which when activated in this complex, generates the signaling media- tors diacylglycerol (Dag) and inositol trisphosphate (IP₃) by hydrolyzing the lipid signaling molecule phosphatidylinositol (4,5) bisphosphate (PIP₂). Dag leads to the activation of the Ras-MaP kinase pathway and the protein kinase AP-1 NF-κB NFAT T CELL ACTIVATION (E.G., CYTOKINE PRODUCTION) Cθ pathway (activated sos also helps activate Ras). IP₃ stimulates the release of calcium from the ER, leading to activation of the calcium-dependent phosphatase calcineu- rin. The net result is the activation of a set of transcription factors, aP-1, nF-κB, and nFaT, that cooperatively mediate a number of critical transcriptional responses that are part of T cell activation (co-stimulatory signals are also important for nFaT activation, but the details are not discussed here). Here we illustrate one key transcriptional output, which is the increased expression of the cytokine IL-2. 1 How does the T cell receptor transmit signals after peptide–MHC recognition? T cell recepTor phosphorylaTion iniTiaTes acTivaTion The initial stages of signal transmission upon T cell activation are controlled by tyrosine phosphorylation. Engagement of the TCR with the peptide–MHC complex leads to phosphorylation of the TCR by the Lck kinase (in a process that is still poorly understood). Phosphorylation of the T cell receptor occurs on peptide motifs known as ITaMs (immunoreceptor tyrosine-based activation motifs). Upon phosphorylation, the ITaMs act as binding sites for the sH2 domains from the kinase ZaP-70. This recruitment of ZaP-70 to the TCR complex then triggers the next set of phosphorylation events in the T-cell-activation cascade. The TCR is composed of eight sub- units: the α and β chains contain the extracellular domains that interact with the MHC–peptide complex, while the CD3 γ, δ, ε, and ζ chains contain intra- cellular regions that communicate with downstream signaling proteins. The intracellular segments of the TCR col- lectively contain 10 ITaMs—a specific dual tyrosine motif that is phosphoryl- ated by Lck kinase upon activation. The P P P P P P P P P P P P . . . Y P X X [L/I] X6–9 Y P SH2 domains of ZAP-70 kinase bind phosphorylated tyrosine sites on phosphorylated ITaMs become docking sites for the tandem sH2 domains from the ZaP-70 kinase. Once bound to the TCR, ZaP-70 is itself phosphorylated and activated by Lck. In turn, activated CD3γ CD3ε P P P P P P P P CD3ε CD3δ X X [L/I] . . . ITAMs and localized ZaP-70 phosphorylates a number of critical downstream targets that initate the T cell response. CD3ζ CD3ζ ITAM tandem phosphotyrosine motif (ITAM) P C-SH2 P ZAP-70 interdomain The ITaMs act as recruitment sites for the tandem sH2 units from the kinase ZaP-70. The tandem recognition unit has higher affinity and specificity than many other sH2–peptide interactions due to avidity effects. Because the interaction is dependent on phosphorylation of the tyrosine residues in the ITaMs, the interaction only occurs after the receptor has been activated. Binding of the ITaMs also relieves auto- inhibition of the ZaP-70 kinase. The ZaP-70 kinase is now in an activated conformation and localized to the activated receptor complex, where it is in position to phosphorylate a number of key downstream targets as shown in panel 3. N-SH2 Receptor activation is discussed in Chapter 8 SH2-mediated phosphopeptide recognition is discussed in Chapter 10 Multivalent protein interactions are discussed in Chapter 2 How does the T cell launch a robust response when stimulated by as few as ten antigenic peptide complexes? posiTive feedback loops in The T-cell-acTivaTion neTwork can amplify a small anTigenic inpuT signal To Trigger a robusT response T cells are remarkable in that as few as 10 antigenic peptide–MHC complexes presented by an antigen-presenting cell can lead to robust activation of a cognate T cell. Positive feedback loops in the T-cell-activation pathway play an important role in amplifying this small stimulus to result in full activation. Here we review four examples of positive feedback loops involved in T cell activation. These examples illustrate how feedback regulation of signaling processes can operate at many differ- ent levels, including intracellular molecular interactions, cellular reorganization and localization, and intercellular (paracrine) communication. POsITIVE FEEDBaCK THROUgH CELLULaR REORganIZaTIOn: FORMaTIOn OF THE IMMUnE sYnaPsE LEaDs TO RECEPTOR anD sIgnaLIng-PROTEIn CLUsTERIng Upon antigen presentation, the antigen-presenting cell and the T cell undergo major structural reor- ganization, forming an extensive cell–cell adhesion junction known as the immune synapse or the supramo- lecular activation complex (sMaC). The immune synapse structure forms within a few minutes of the encounter, but can remain stable for well over an hour. Within this new cellular junction struc- ture, cell adhesion molecules like ICaM segregate to the periphery (psMaC), while signaling receptors, including and antigen-presenting cell, is thought to enhance and amplify their com- munication with one another, allowing them to accurately assess the presence of even a few molecules of the cognate antigenic peptide–MHC complex. T cell immune synapse antigen- pre enting cell pSMAC ICAM cSMAC TCR CD4/CD8 (co-receptor) CD28 the TCR, CD4 and CD8 co-receptors, and CD28 co-stimulatory receptors, are corralled and clustered in the center (csMaC). This clustering of key signaling receptors, especially during a sustained interaction between T cell T cell (blue) interacting with antigen- presenting cell (red). The zone of contact between the cells is the immunological synapse (green). Courtesy of Tomasz Zal, M. Anna Zal , and Nicholas R.J. Gascoigne. supramolecular activation complex (SMAC) (co-stimulatory receptor) agonist strong weak self- peptide POsITIVE FEEDBaCK THROUgH RECEPTOR–RECEPTOR sPREaDIng although an antigen-presenting cell–T cell interface may only contain a few agonist (non-self) peptide complexes within a sea of non-agonist (self) peptide complexes, recent evidence suggests that agonist–receptor complexes can interact with non- Lck P P P P P P P P P P P P P P P P P P P amplification P P P P P P P P P P P P P P P P P P P P agonist–receptor complexes. Thus, activated Lck associated with an agonist–receptor complex may be able to phosphorylate other nearby recep- tors, thus leading to amplified signal transmission. T cell activation activation POsITIVE FEEDBaCK THROUgH IL-2 CYTOKInE aUTOCRInE & PaRaCRInE sIgnaLIng One of the key outputs of naïve T cell activation is the secretion of the cytokine IL-2. IL-2 is a potent mitogen and survival factor for T cells. Thus IL-2 secreted by an activated T cell can then form an autocrine loop (stimulates the same cell) and a paracrine loop (stimulates its neighboring T cells). self-activation by this paracrine loop is detected by the IL-2 receptor in T cells. dendritic cell T cell IL-2 proliferation IL-2 autocrine signaling paracrine signaling IL-2 activation itself leads to increased expression of both IL-2 and IL-2 recep- tor in the signal-receiving cell, thus forming an even more potent positive feedback mechanism for amplifying activation. Ultimately this process leads to expansion of a single cognate T cell into a clonal army of activated T cells. initial activation autocrine/paracrine amplification POsITIVE FEEDBaCK THROUgH a Ras-sOs aLLOsTERIC FEEDBaCK LOOP LEaDs TO sWITCHLIKE aCTIVaTIOn OF THE MaPK ERK activation of the Ras-MaP kinase pathway is a central part of T cell activation. Initial activation of Ras is thought to occur via activation of the gEF RasgRP in response to diacylglycerol (Dag) production (see molecular network, panel 3). RasgRP leads to a linear increase in Ras activa- tion. However, a second RasgEF, the molecule sos, is known to be allo- sterically activated by its own product (gTP-bound Ras). Thus, it is postulated that the initial amount of Ras that is activated by RasgRP is sufficient to allosterically activate sos gEF activity, thus triggering a positive feedback loop that can strongly amplify output once past a threshold of activation. Consistent with this model, increasing stimulation of T cells is observed to lead to a bistable (all-or-none) response, with respect to phosphorylation of the MaPK Erk (right). Disruption of the Ras–sos positive feedback loop disrupts this all-or-none response. flow cytometry GDP RasGRP GEF GTP GTP Sos GEF Sos DIGITAL OUTPUT Erk MAPK P module phospho-Erk Erk Erk-PP How does the T cell prevent misactivation? negaTive feedback loops help prevenT and limiT T cell acTivaTion It is critical that T cells are not activated by improper inputs. We now know that there are a number of negative feedback loops that may help to keep the system relatively quiet and prevent misactivation. These negative feedback loops can be used as a way to abort misactivation of a T cell by transient or partial input signals (e.g., absent co-stimulation). Full activation of a T cell requires several hours of sustained interaction with the antigen-presenting cell. Here we describe two such negative feedback systems. P P P P P P CD28 CTLA-4 competes for B7 binding CTLA-4 ITIMs CD28 co-receptor signaling appears to be an important target for nega- tive feedback regulation to prevent misactivation of the T cell. When the TCR is stimulated, in addition to the better-characterized positive effects, this also leads to increased cell-surface expression of molecules known as co-inhibitory receptors, such as CTLa-4. The CTLa-4 co-inhibitory P P P P P P P P P P P P P P co-stimulatory receptor co-inhibitory receptor inhibitory receptor binds to the same ligand (B7) in the antigen-presenting cell as that recognized by the CD28 co- receptor, but with much higher affinity. Lck TCR intracellular signaling pathways negative feedback signaling Thus CTLa-4 competitively blocks co-receptor stimulation of the T cell. In addition, the CTLa-4 receptor has intracellular motifs known as immune tyrosine inhibitory motifs (ITIMs), which, upon phosphorylation, recruit inhibitory factors such as inhibitory T CELL ACTIVATION T CELL INHIBITION nEgaTIVE FEEDBaCK: aCTIVaTIOn OF CO-InHIBITORY RECEPTORs THaT BLOCK CO-sTIMULaTORY sIgnaLs activation of T cells requires co- stimulatory signals from the dendritic cell, in addition to the MHC–peptide signal detected directly by the TCR. For example, a co-stimulatory receptor in the T cell, CD28, is activated by the co- stimulatory cell-surface molecule, B7, expressed on the antigen-presenting phosphatases, which turn down T cell activation (ITIMs have an opposing function to ITaMs). Monoclonal antibodies that block inhibitory co-receptors like CTLa-4 have provided a major breakthrough in potential cancer therapies, because they can effectively stimulate the host’s own immune response against tumor cells. These antibodies, however, are also associated with autoimmune-like side effects, as might be expected P P P P P P P P P P P P P P P P P P P P degradation negative cell. Dual activation of the TCR and the CD28 co-receptor is necessary for T cell activation. to occur if T cells lack this negative feedback mechanism to prevent misactivation. TCR feedback SH2 ubiquitinylation PPPP Cbl activation of the TCR leads to its tyrosine phosphorylation. While these phospho- rylation events lead primarily to recruitment of activating downstream factors, they can also lead to recruitment of sH2-domain-containing ubiquitin ligases of the Cbl family. Recruitment of Cbl leads to ubiquitylation and ultimately internalization and degradation of the TCR, creating a negative regulatory loop. Negative feedback loops are discussed in Chapter 11 How might the T cell discriminate between antigenic and non-antigenic peptides? coordinaTed fasT negaTive feedback and slow posiTive feedback could yield a discriminaTory filTer for T cell acTivaTion as a T cell encounters and contacts an antigen-presenting cell, it will engage millions of self-peptide–MHC complexes, scanning for a potential antigenic peptide that is a match for its specific TCR. The affinities of antigenic versus non-antigenic peptide complexes for the TCR are not significantly different (e.g., they may only differ by a few fold excess of self-peptides presented on MHC low affinity short complex lifetime (<2 sec) <10 agonist peptide–MHC complexes are sufficient for strong activation low to moderate affinity longer complex lifetime (2–10 sec) antigen peptide in Kd), so receptor occupancy alone cannot account for the remarkable specificity of T cell activation—over a million non-antigenic peptide complexes are unable to activate the T cell, while as few as ten antigenic peptide complexes are sufficient to fully activate the T cell. We do not have a complete answer to this complex problem, but some data suggest that a critical way in which non- self peptides differ from self-peptides NO ACTIVATION is the lifetime of the complex that they form with the cognate TCR—non-self peptides bind with lifetimes at least tenfold longer than self-peptides. Thus the TCR may use lifetime of receptor engagement to discriminate between self and non-self signals. Below we ACTIVATION discuss models for how dynamically coordinated positive and negative feedback loops might act in concert to achieve both effective filtering against misactivation by self-peptides, yet still allow robust activation by antigenic (longer-lifetime) peptides. FasT nEgaTIVE FEEDBaCK MEDIaTED BY sHP1 Can BLOCK aCTIVaTIOn BY sHORT sTIMULaTIOn TIME InPUTs We have described how negative feedback loops in T cell signaling may abort and filter against improper activa- tion, while positive feedback loops may amplify a sustained signal to yield full activation. Yet it remains unclear how such positive and negative feedback loops can yield discrimination between self and antigenic peptide inputs. It is postulated that one way in which positive and negative feedback loops can act in a discriminatory way is by having tightly coordinated fast nega- tive feedback loops and slow positive feedback loops. such a network could act as a sharp time filter in which only signals that are sustained enough (i.e., longer-lifetime antigenic peptide binding) will switch the system on. Here we describe a set of coordinated positive and negative feedback loops involving the Lck tyrosine kinase and shp1 phosphotyrosine phosphatase that have been postulated to contribute to T cell temporal discrimination. Lck SH3 SH2 P phosphorylated Shp1 binds to Lck SH2 domain P P The Lck tyrosine kinase is activated upon TCR engagement and generates the first positive signal of TCR activation, ITaM phosphorylation. nevertheless, in addition to this positive signal, Lck also generates a negative signal—it phosphorylates the shp1 phosphotyrosine phosphatase. Phosphorylated shp1 is recruited to the Lck sH2 domain, where it can exert negative effects by catalyzing the dephos- (inactive) FAST NEGATIVE FEEDBACK from Shp1 phosphatase suppresses activation by transient inputs Lck (active) phorylation of the TCR, ZaP-70, and Lck itself. This negative feedback loop may act to filter out transient stimulation. transient signaling TCR and ZAP-70 phosphorylation NO ACTIVATION Shp1 phosphatase sLOW POsITIVE FEEDBaCK In WHICH aCTIVaTED ERK OVERRIDEs sHP1 nEgaTIVE REgULaTIOn MaY aLLOW FOR FULL aCTIVaTIOn WITH sUsTaInED InPUT sTIMULaTIOn If activating signals received by the T cell are sustained enough (and there are repeated rounds of receptor engagement), then this should lead to the gradual accu- mulation of the activated form of the MaPK Erk. Erk has been found to phosphorylate Lck at residue ser59, which disrupts the interaction of shp1 with Lck. Thus activated Erk can disrupt the Lck→shp1 negative feedback loop, acting as a double-negative (or positive) feedback loop. Thus Erk provides a slower, delayed positive feedback loop that overrides shp1 fast negative feedback. P P P active TCR and ZAP-70 phosphorylation Shp1 phosphatase sustained Erk P SLOW POSITIVE FEEDBACK displaces Shp1 phosphatase stimulation ACTIVATION from Lck 100 0 0.1 1 10 10 ² 10 ³ 10 ⁴ 10 ⁵ number of MHC–peptide ligands on antigen-presenting cell Theoretical plot of how inter- locked fast negative and slow positive feedback loops could lead to a high level of T cell discrimination between long-lifetime and short-lifetime peptide–MHC ligands. long- lifetime peptide short- lifetime peptide COORDInaTED FasT nEgaTIVE FEEDBaCK anD sLOW POsITIVE FEEDBaCK Can LEaD TO an aLL- OR-nOnE TIME FILTER These negative and positive feedback loops differ in time scale, and are mutually exclusive (they both determine whether shp1 is or is not recruited to Lck). Quantitative modeling indicates that such a circuit could yield activation that is highly dependent on the lifetime of the stimulating input (peptide–MHC engage- ment with the TCR). Because Erk activation in T cells is such a switchlike response, one expects to see a sharp threshold for this Erk-mediated time filter. such a dual-timescale feedback circuit could explain how a system can be activated fully by a very small number of long-lifetime peptide–MHC ligands, but can ignore stimulation by large numbers of short-lifetime peptide– MHC ligands. Positive and negative feedback loops are discussed in Chapter 11 SUMMARY The T cell scans for antigenic peptide–MHC complexes and co-stimulatory signals on the surface of an antigen-presenting cell. Activation of the naïve T cell only occurs with the proper combination of signals. The T cell receptor (TCR) recognizes the peptide–MHC complex and propagates signals via tyrosine phosphorylation on ITAM peptides. The phosphorylated ITAMs then serve as recruitment sites for the ZAP- 70 kinase (via its SH2 domains), thereby triggering further downstream signaling. Positive feedback loops in T cell signaling amplify weak but correct inputs to yield robust activation, even by as few as 10 non-self peptide–MHC complexes on an antigen-presenting cell. Negative feedback loops in T cell signaling prevent misactivation by transient or partial signals, such as those presented by the large excess of self-antigen peptide–MHC complexes (millions of which are presented on any antigen-presenting cell). Coordination between fast negative feedback and slow positive feedback loops may allow the T cell to discriminate between self and non-self (antigenic) peptides, only yielding a full response for non-self peptides that form a longer-lived complex with the TCR. REFERENCES Altan-Bonnet G & Germain RN (2005) Modeling T cell antigen discrimination based on feedback control of digital ERK responses. PLoS Biol. 3, e356. Das J, Ho M, Zikherman J et al. (2009) Digital signaling and hysteresis characterize ras activation in lymphoid cells. Cell 136, 337–351. Dustin ML & Groves JT (2012) Receptor signaling clusters in the immune synapse. Annu. Rev. Biophys. 41, 543–556. Kortum RL, Rouquette-Jazdanian AK & Samelson LE (2013) Ras and extracellular signal- regulated kinase signaling in thymocytes and T cells. Trends Immunol. 34, 259–268. Morris GP & Allen PM (2012) How the TCR balances sensitivity and specificity for the recognition of self and pathogens. Nat. Immunol. 13, 121–128. Sherman E, Barr V & Samelson LE (2013) Super-resolution characterization of TCR-depen- dent signaling clusters. Immunol. Rev. 251, 21–35. Sykulev Y (2010) T cell receptor signaling kinetics takes the stage. Sci. Signal. 3, pe50. Tkach K & Altan-Bonnet G (2013) T cell responses to antigen: hasty proposals resolved through long engagements. Curr. Opin. Immunol. 25, 120–125. Zarnitsyna V& Zhu C (2012) T cell triggering: insights from 2D kinetics analysis of molecu- lar interactions. Phys. Biol. 9, 045005. Zehn D, King C, Bevan MJ & Palmer E (2012) TCR signaling requirements for activating T cells and for generating memory. Cell. Mol. Life Sci. 69, 1565–1575. methods for studying signaling proteins and networks Our current understanding of cell signaling is the result of decades of exper- imental research in a variety of fields. In this chapter, we describe some of the most commonly used methods and approaches for studying cellular signaling. These range from tools for the analysis of individual signaling proteins to those that can characterize entire networks within living cells. Biochemical and Biophysical analysis of proteins Changes in the physical state of proteins are central to cell signaling, so methods to probe the properties of proteins are essential for understand- ing signaling mechanisms. In this section, we will look at methods used to study the properties of purified proteins, including the quantitative analy- sis of their binding and enzymatic activities, and determination of their three-dimensional structure. Analytical methods can determine quantitative binding parameters The dissociation constant, on-rate, and off-rate for a binding reaction deter- mine both the likelihood that the interaction will occur in the cell and its dynamic behavior. Measuring these parameters accurately becomes ever more important as biologists strive to understand the dynamic properties of signal transduction networks and to develop quantitative computation- al models to describe their behavior. Here, we discuss a few of the most widely used methods for determining such parameters experimentally. Quantitative measures of binding are discussed in Chapter 2 [bound] [free] Figure 13.1 [bound] The dissociation constant (Kd) is often calculated from binding experi- ments in which a small amount of one protein (A) is incubated with vary- ing concentrations of a binding partner (B) and the binding reaction is allowed to achieve equilibrium. From the amount of A bound to B (its fractional occupancy) at different concentrations of B, a binding curve can be fitted to the data, and from this the dissociation constant can be cal- culated (see Figure 2.7). Such methods depend on being able to quantify precisely the amount of A that is bound to B. For binding in solution, this can often be determined by monitoring differences in the spectral proper- ties of A upon binding (for example, changes in fluorescence polarization when A is fluorescently labeled). In another experimental set-up, A is fixed to a solid surface, and B is tagged with a fluorescent or radioactive label. Binding reactions are then Scatchard analysis. the results of binding studies can be presented in the form of a scatchard plot. When [aB]/[B] (bound/free) is plotted versus [aB] (bound), in simple cases a straight line is obtained where the slope is –1/Kd and the intercept on the horizontal axis is the total number of binding sites for B. performed at different concentrations of B, and for each concentration the amount of B bound to A is measured directly after washing away unbound As above, such binding data can be used to determine the Kd. Although this approach is technically straightforward, one must keep in mind that one of the proteins is immobilized at relatively high concentrations on a solid surface, which can distort binding behavior. Computational fitting of binding data directly to the binding equation is used to calculate binding parameters with high precision. A convenient way to visualize binding data, however, is provided by the Scatchard plot (Figure 13.1). The basis for Scatchard analysis is the algebraic rear- rangement of the equation for equilibrium binding (see Equation 13.1) into the form of a straight line (y = mx + b), to obtain: [AB] = – [AB] + [A_total ] Equation 13.1 [B] Kd Kd If one plots [AB]/[B] (the amount of B that is bound to A, divided by the amount that remains unbound—that is, bound over free) on the verti- cal axis and [AB] (the bound fraction) on the horizontal axis for many different concentrations of B, in simple cases a straight line is obtained for which the slope is –1/Kd, and the horizontal intercept is [AB]max, or the total number of binding sites. The latter number can be quite useful, for example, when analyzing the binding of a labeled hormone to cells, where the total number of binding sites corresponds to the total number of receptors for the hormone in those cells. In some cases, the Scatchard plot may not yield a straight line, suggesting the existence of more than one class of binding site for B, each with a different affinity. In such situ- ations, the Scatchard plot can provide an estimate of the number of sites and Kd for each class of binding site. In practice, computational fitting of multiple datasets covering a wide range of concentrations of components is required for accurate determination of both Kd and the number of bind- ing sites. To measure the on-rate and off-rate for a binding reaction, binding and dissociation must be monitored in real time. This can be done using a surface plasmon resonance (SPR) instrument, which measures the change in the angle of reflection of light bouncing off a metallic surface as the mass on the surface is increased. In SPR, one binding partner is immobilized on a sensor chip in the apparatus, and solutions containing the other partner (or buffer alone) are injected and allowed to flow over the chip. Binding is then monitored over time, as a function of changes in the protein mass bound to the surface. By measuring initial rates of binding and dissociation at different concentrations of injected binding Figure 13.2 Determining binding parameters by surface plasmon resonance (SPR). protein a is fixed to a chip in the instrument; a solution containing its binding partner (protein B) is injected at the time indicated by the arrow, and the association of a and B is monitored. after a time, buffer alone (lacking protein B) is injected to allow the dissociation of B to be monitored. plots of binding versus time for three different concentrations of B are shown. initial rates of association or dissociation can be obtained from the slopes of the curves, and Kd can be calculated. partner, on-rates and off-rates can be calculated (Figure 13.2) as well as the overall dissociation constant for the interaction. One advantage of this approach is that none of the binding partners need to be labeled. SPR binding data must be interpreted with care, however, as one of the binding partners is immobilized on a surface, thus potentially distorting rates of binding and dissociation. It is also possible to determine the thermodynamic parameters [the changes in enthalpy (H) and entropy (S)] underlying a binding interac- tion. For example, rates of binding and dissociation obtained from SPR experiments at different temperatures can be used to derive the change in enthalpy (ΔH) and in entropy (ΔS) associated with binding. Isother- mal calorimetry (ITC) is another analytical method that provides direct information about thermodynamic parameters, in this case when both binding partners are in solution. In ITC, small amounts of one bind- ing partner are sequentially injected into a highly sensitive calorimeter containing a solution of the other binding partner. After each injection, the heat released or absorbed by the solution is measured precisely (Figure 13.3). The amount of heat absorbed or released, how this amount changes as more of the components are bound, the temperature of the system, and the concentrations of the components are used together to calculate Kd, ΔH, and ΔS. Michaelis–Menten analysis provides a way to measure the catalytic power of enzymes Many of the key proteins involved in signal transduction are enzymes, such as kinases and phosphatases, and their function in signaling often inject B buffer only time revolves around regulated changes in their activity. Thus it is essential to be able to measure enzyme function quantitatively in standard ways. The catalytic power of an enzyme is characterized by the degree to which it enhances the reaction rate over that of the uncatalyzed reaction. How- ever, assessing this activity can be relatively complex, as reaction rates are highly dependent on the concentrations of the enzyme and the sub- strates. For the simple case of an enzyme binding to a single substrate (S) and reacting to form product(s) (P), the reaction rate is described by the Michaelis–Menten equation (Equation 13.2), which describes the reac- tion velocity (V) as a function of enzyme (E) and substrate (S) concentra- tions. The derivation of the Michaelis–Menten equation can be found in any elementary biochemistry textbook. Figure 13.3 Determining binding parameters by isothermal calorimetry (ITC). small amounts of a solution of B are injected into a calorimeter containing a solution of a. changes in the heat of the solution are monitored after every injection. as more of a is bound by B, each subsequent injection results in a smaller change in heat because less free a is available for binding. Kd, Δh, and Δs can be derived from these data. The properties of enzymes and their role in signaling are de- scribed in Chapter 3 inject B 0 time Experimental biochemical analysis of an enzyme is typically carried out by measuring how the reaction rate varies as a function of substrate con- centration. This analysis is performed at several different enzyme concen- trations, and the data can be fitted to the Michaelis–Menten equation to determine the key kinetic parameters kcat and Km. For the reaction scheme: E + S k1 ES k–1 kcat E + P V = d[P] = k _cat [E]₀ [S] Equation 13.2 obs dt Km + [S] K = k–1 + kcat m k (a) (b) time 1 Figure 13.4a shows a schematic of typical experimental data used to determine kcat and Km. Each line on the graph represents an experiment conducted at a different initial substrate concentration, with the con- centration of the reaction product plotted as a function of time for each experiment. The initial slopes (d[P]/dt : change in product concentration as a function of time) of these lines equal the initial reaction velocities, V, which can then be plotted as a function of initial substrate concen- tration, [S], to yield the Michaelis–Menten plot (Figure 13.4b). A typical enzyme will show a hyperbolic plot, in which velocity initially increases linearly with increasing substrate concentration, but then asymptoti- cally approaches a maximal velocity, Vmax, when substrate approaches saturation. Intuitively, the two terms kcat and Km can be used to understand basic functional parameters of an enzyme: kcat reflects the maximum rate achievable in the presence of saturating substrate concentration, and Km is the substrate concentration at which half-maximal velocity is achieved (see Figure 13.4b). It is important to note that many signaling enzymes have multiple substrates—for example, a protein kinase uses both a pep- tide and ATP as substrates. In these cases, each substrate has its own dis- tinctive Km that reflects the concentration at which half-maximal velocity (c) [substrate] [substrate] is reached. This Km is typically measured and reported for conditions in which the other substrate is already saturating. Overall, the Michaelis–Menten equation tells us how a typical enzyme will behave at different substrate concentrations. If substrate concentra- tion is high (that is, much greater than Km), then the enzyme active site Figure 13.4 Quantifying enzyme catalysis by Michaelis–Menten analysis. (a) in a typical series of experiments, the formation of product over time is measured for different starting substrate concentrations and with a substoichiometric concentration of enzyme, [e]. each line represents an experiment performed at a different substrate concentration. (b) the initial rates (V) from part (a) are then plotted as a function of substrate concentration [s] to give the classical hyperbolic michaelis–menten plot. the maximal rate achieved at saturating substrate concentrations is defined as the Vmax. the parameter Km reflects the substrate concentration at which the rate is half-maximal. the catalytic rate constant, kcat, is equal to Vmax /[e]. (c) cooperative enzymes, where the substrate acts to allosterically activate the enzyme, deviate from michaelis–menten behavior and show a sigmoidal dependence of reaction velocity on substrate concentration. these types of enzymes are well suited for switchlike control. will be fully saturated with substrate and the rate equation simplifies to a form (Equation 13.3) that depends only on the total enzyme concentra- tion [E]₀: Vobs = Vmax = kcat [E]0 Equation 13.3 If the substrate concentration is low (that is, considerably lower than Km), then the enzyme will not be operating near saturation and the rate equa- tion simplifies to Equation 13.4: V = kcat [E] [S] Equation 13.4 obs 0 m kcat/Km is the apparent bimolecular rate constant for the enzymatic reac- tion. Because concentrations of substrates in a cell are often low, kcat/Km pro- vides a good estimate for how an enzyme performs in vivo. kcat/Km is often referred to as the specificity constant because the relative values of kcat/ Km for two different substrates reflect the degree of specificity of the enzyme for the two substrates. It is important to note that kcat/Km determines spe- cificity regardless of whether the substrate concentration is saturating or not; readers are referred to specialized biochemistry textbooks (see the ref- erence section at the end of the chapter) for a rigorous derivation. Not all signaling enzymes show simple Michaelian (hyperbolic) behav- ior. In particular, a key function of many signaling enzymes is to act as allosteric regulatory nodes. Briefly, allostery is the property of being able to exist in two or more structural states of differing activity. Many sign- aling enzymes may show a dramatic change in kcat and/or Km upon stim- ulation by an upstream signal, such as covalent modification or ligand binding. These changes in enzyme kinetic parameters are often caused by allosteric conformation changes induced by the upstream input. In cases where the substrate itself is an allosteric activator, cooperative activation of the enzyme is observed, which leads to significant deviations from sim- ple Michaelian behavior and a sigmoidal plot of V versus [S] (Figure 13.4c). Cooperative activation is often seen for oligomeric enzymes, where bind- ing of substrate to one subunit affects binding to the other subunits. Such cooperative enzymes have an important function—they make the enzyme output less like an analog signal and more like a digital signal (all or none). For a cooperative enzyme, there is a threshold substrate concentra- tion below which the rate of catalysis is very low, and above which the rate is near maximal; unlike a standard Michaelis–Menten enzyme, there is only a small range of substrate concentrations that leads to intermediate reaction velocities. Methods to determine and analyze protein conformation are central to the study of signaling The three-dimensional structure or conformation of signaling pro- teins is what determines their interactions and their catalytic activity. In addition, how their conformations change based on input signals, and how this alters their output functions, is at the heart of how they function as allosteric signal transmission devices. Thus, methods to determine and analyze protein conformation are central to the study of signaling. Here, after first describing the fundamental elements of protein structure, we outline the two major approaches used to deter- mine protein structure: x-ray crystallography and nuclear magnetic resonance (NMR). The structure of a protein can be described on a number of levels. Here, these different levels are illustrated using the serine/threonine kinase Pak1. Allosteric enzymes are described in Chapter 3 Figure 13.5 Primary and secondary structure. the amino acid sequence of the catalytic 249 αA αB β1 β2 SDEEILEKLRSIVSVGDPKKKYTRFEKICQGASCTVYTAMDVAT 292 domain of the serine/threonine kinase pak1 (human, amino acids 249 to the end) β3 in single-letter code. positions of major αC β4 secondary structure elements (α helices and β strands) are depicted above the sequence. (adapted from m. lei et al., C e ll 293 GQEVAIKQMNLQQQPKKELIINEILVMRENKNPNIVNYLDSYLV 336 102:387–397, 2000. With permission from elsevier.) 337 GDELWVVMEYLAGGSLTDVVTETCMDECQIAAVCRECLQALEFL β7 β8 380 381 HSNQVIHRDIKSDNILLGNDGSVKLTDFGFCAQITPEQSKRSTM 424 αEF αF 425 VGTPYNMAPEVVTRKAYGPKVDIWSLGINAIEMIEGEPPYLNEN 468 αG αH 469 PLRALYLIATNGTPELQNPEKLSAIFRDFLNRCLDMDVEKRGSA αI αJ 512 513 KELLQHQFLKIAKPLESLTPLIAAAEEATKNNH 545 The primary structure of the protein is the simple linear sequence of amino acids in the polypeptide chain (Figure 13.5). The secondary structure consists of local structural elements, predominantly α heli- ces and β strands (described further below). In Figure 13.5, the α heli- ces are depicted as cylinders and the β strands as block arrows. Each helix and strand is numbered by its order of appearance in the protein, starting at the N-terminus. The tertiary structure of the protein is how these α helices and β strands, and the loops that connect them, are folded together. In most proteins, this fold generates one or more globular domains (Figure 13.6). Finally, the quaternary structure refers to how αB Figure 13.6 Tertiary and quaternary structure. the pak1 structure is depicted here as a ribbon diagram, with α helices and β strands indicated schematically. the colors in the structure correspond to the colors on the secondary structure diagram above the sequence in figure 13.5. an n-terminal portion of pak1 (from residues 78–147) is also included here (green); residues 1–77 and 148–248 are not shown. the second subunit of the pak1 homodimer is shown to the right and is colored yellow. (adapted from m. lei et al., Cell 102:387–397, 2000. With permission from elsevier.) (b) (c) ino acid e chain bon rogen Figure 13.7 Common secondary structure elements: α helix and β sheet. (a, d) all the atoms in the polypeptide backbone are shown but the amino acid side chains are truncated and denoted by r. (b, e) the backbone atoms only are shown. (c, f) the shorthand symbols that are used to represent the α helix and the β strand in ribbon drawings of proteins. (adapted from B. alberts et al., molecular Biology of the cell, 5th ed. Garland science, 2008.) rogen ygen α helix (f) β sheet multiple subunits are arranged in a multiprotein complex. In this case, Pak1 exists as a dimer. In an α helix, the N–H group of every peptide bond is hydrogen-bonded to the C=O of a neighboring peptide bond located four peptide bonds away in the same chain (Figure 13.7a–c), thus forming a regular right-handed helix with 0.54 nm between turns. The side chains (R) of each amino acid project out from the axis of the helix. The individual polypeptide chains (strands) in a β sheet are held together by hydrogen-bonding between peptide bonds in different (adjacent) strands, and the amino acid side chains in each strand alternately project above and below the plane of the sheet (Figure 13.7d–f). In the example of a β sheet shown in Figure 13.7, adjacent peptide chains run in opposite (antiparallel) directions. Sheets with strands oriented in the same (parallel) direction are also possible. X-ray crystallography provides high-resolution protein structures Among methods used to analyze the structure of proteins and to detect changes in conformation, the highest resolution is provided by x-ray crystallography. In this method, highly concentrated solutions of puri- fied protein are induced to form crystals, in which the protein is arrayed in a regular lattice. These crystals are then exposed to an x-ray beam, and the resulting diffraction pattern can be analyzed to calculate a three- dimensional electron density map of the crystallized protein. The known amino acid sequence of the protein can be fitted into this map to provide a high-resolution structural model (often with a resolution of approxi- mately 2 Å). At this high resolution, one can identify the position of most individual atoms in the protein (as long as that region of the protein is well ordered in the crystal). A comparison of crystal structures of a protein in two different conformational states can reveal precisely which residues undergo movements (Figure 13.8). A potential disadvantage of x-ray crystallography is that it provides a static picture of the protein structure, because the protein must be immo- bilized in the crystal lattice for analysis. Even when a protein is capable of adopting multiple conformations, only a single conformation is likely to be observed in one crystal lattice. Furthermore, not all proteins can be crystallized successfully. Crystallization is particularly challenging in the case of proteins with transmembrane segments, such as channels and receptors, as these hydrophobic regions tend to aggregate in solu- tion. Nonetheless, in quite a few cases, researchers have been able to crystallize and determine the structure of a signaling protein in several distinct states (that is, bound to different allosteric ligands, or in distinct phosphorylation states). In such cases, one can determine with very high (a) swi Figure 13.8 X-ray crystal structures illustrate conformational changes. (a) X-ray crystal structures of human h-ras in the Gdp-bound (left) and Gtp-bound (right) conformations; switch i and ii regions are shown in green and bound nucleotide is shown in blue. (b) Graph shows differences in distances between cα atoms in the structures of Gdp-bound and Gtp-bound conformations, plotted as a function of residue number. the largest differences occur in the switch i and ii regions. (b, adapted from m.V. milburn et al., Science 247:939–945, 1990. With permission from aaas.) 10.0 8.0 6.0 4.0 2.0 20 40 60 80 100 120 140 160 residue number resolution precisely how the protein structure changes in each state and how this is linked to protein function. Nuclear magnetic resonance (NMR) can reveal the dynamic structure of small proteins The structure of small proteins can also be determined by nuclear magnetic resonance (NMR) spectroscopy. In this approach, a con- centrated solution of the protein sample is placed in a strong magnetic field, which leads to a difference in the energy of the two oppositely ori- ented spin states of atomic nuclei. Radiofrequency energy is then used to flip the spin states of individual atomic nuclei within the protein (this occurs at the resonance frequency). If an atom of a particular type is in a unique chemical environment, it will show a unique and altered reso- nance energy—also known as a “chemical shift.” An NMR spectrum of the protein will reveal all the unique nuclei of a certain type in the molecule (most often protons, though the stable isotopes 13C and 15N can be used to monitor carbon and nitrogen nuclei, respectively). Because nuclei that are close in space or bonded to one another can perturb each other’s chemical environment, the interactions between resonance peaks can be analyzed to reveal spatial relationships such as interatomic distances and bond angles. These pieces of structural information can be combined to gener- ate a three-dimensional model of the protein structure. Because there is no complete structural map, as in the case of x-ray crystallography, struc- tures determined by NMR are inherently of lower resolution. An advantage of NMR is that it can also reveal dynamic motions involved in conformational changes, because the protein is studied in aqueous solu- tion rather than crystallized form. In some cases, NMR can also monitor multiple conformations in the same sample and provide information on the fraction of the total in each population, even if the structure is not absolutely known. For example, one can monitor the change in chemical shifts corresponding to atoms that undergo a dramatic change in chemical environment in one conformation versus another, or before and after lig- and binding (Figure 13.9). Technical issues currently limit NMR analysis to relatively small proteins or domains. Electron microscopy can map the shape of very large protein complexes Although proteins are in general too small to observe directly by micro- scopy, for very large proteins or protein complexes it can be possible to map their overall shape using electron microscopy (EM). Because the wavelength of electrons is much shorter than that of light, the potential Calmodulin (no Ca2+) Calmodulin + high Ca2+ Figure 13.9 NMR spectroscopy reveals structural changes upon ligand binding. calmodulin, which has four ca2+-binding sites, was analyzed by two-dimensional nmr in the absence and presence of ca2+. resonance peaks that undergo shifts due to changes in local environment are highlighted by dotted arrows, from their positions in the absence of ca2+ (light blue) to their positions in high 11 9 104 108 112 116 7 11 9 104 108 112 116 7 ca2+ (pink). in this experiment, calmodulin was selectively labeled so only glycine residues appear in the nmr spectra. Knowledge of which peaks correspond to which residues (resonance assignments) is required to interpret the structural changes in an nmr spectrum. in nmr, chemical shifts are typically expressed in ppm (parts per million), based on differences in measured values compared to a reference value. (adapted from h. ouyang and h.J. Vogel, Biometals 11:213–222, 1998. With permission from springer science and 1H PPM 1H PPM Business media.) resolution of EM is much higher than for light microscopy. To obtain imag- es, purified protein samples are often immobilized in a thin layer of vitre- ous ice at very low temperatures, in a method called cryo-EM. Individual images are usually rather noisy, so images of many individual particles can be collected and analyzed by computational methods (single particle analysis) to generate an average image, which has higher resolution than each of the individual images. Some examples of EM-derived structures can be seen in Chapter 9, for the proteasome (Figure 9.9), the anaphase- promoting complex (APC) (Figure 9.10), and the apoptosome (Figure 9.22). While the resolution of electron micrographs is lower than for x-ray crys- tallography or NMR, this approach provides a way to visualize large-scale structures and conformational changes under relatively native conditions. Specialized spectroscopic methods can be used to study protein dynamics Other spectroscopic methods, such as circular dichroism or intrinsic fluo- rescence, can be used to monitor physical properties of proteins, such as the amount of α helix structure or the extent that aromatic residues are bur- ied in the hydrophobic core. Because such approaches monitor structural properties of proteins in solution, they can also be used to dynamically track changes in conformation. These methods, however, do not provide detailed atomic information about structural changes, only global shifts. Specifically modified proteins can be used to track conformational changes also. For example, if a protein is thought to undergo some type of hinge- bending motion in which the distance between two points on the pro- tein changes dramatically, then one can chemically attach spectroscopic probes to these points and monitor properties that change with distance. For example, fluorescent probes (fluorophores) can be attached to these points and the distance between them can be monitored by fluorescence resonance energy transfer (FRET), which is discussed in more detail below. Essentially, FRET measures how the excitation of one fluorophore modi- fies the fluorescence properties of a second nearby fluorophore. FRET is exquisitely sensitive to the distance between the two fluorophores, so it can detect when they move further than a few angstroms apart. If such FRET experiments are performed using fusions of genetically encoded fluorescent proteins, such as yellow fluorescent protein (YFP) and cyan fluorescent protein (CFP), then these molecules can be used as biosensors to report on conformational changes (for example, as a result of binding or post-translational modification) in living cells (Figure 13.10). (b) no FRET FRET Figure 13.10 ligand-induced conformational change [ligand] Detecting ligand-induced conformational changes by fluorescence resonance energy transfer (FRET). (a) diagrammatic depiction of the structure of a ligand-binding protein, showing positions of the fret donor (green) and fret acceptor (pink). Binding of ligand (orange circle) reduces the distance between the fret donor and acceptor, increasing fret efficiency. (b) fret efficiency plotted as a function of ligand concentration. mappinG protein interactions and localization Given the importance of protein–protein interactions in signal transduc- tion, it is no surprise that extraordinary effort has been expended to iden- tify binding partners for signaling proteins. Through characterizing such physical interactions, investigators have been able to work their way up and down signaling pathways and to forge connections between different pathways. In this section, we describe some of the methods used to identify specific physical interactions between proteins, and discuss methods for analyzing protein interactions and subcellular localization in living cells. Interacting proteins can be identified by isolating protein complexes from cell extracts One approach to identifying binding partners for a protein of interest is to isolate complexes containing that protein from cell lysates made by solubilizing cells with detergent. The most usual method of obtaining such complexes is by co-immunoprecipitation (co-IP). A specific antibody that binds the protein of interest is used to separate it and its associated proteins from all the other proteins in the extract. The antibody-bound pro- teins are then analyzed, usually by gel electrophoresis, which separates proteins on the basis of their size and/or charge. The separated proteins can then be visualized and identified by various methods (Figure 13.11). Binding partners are often identified by immunoblotting (also termed Western blotting), in which the proteins separated on the gel are trans- ferred (blotted) onto a membrane and then probed with labeled antibodies specific for possible candidate proteins (see also Figure 13.21c, later in this chapter). Although co-IP provides good evidence for the in vivo asso- ciation of two proteins, false-negative and false-positive results can occur. For example, two proteins that interact in vivo might not be detected by co-IP if the off-rate of the interaction is high, as the complex will dis- sociate during the multiple incubation and washing steps. On the other hand, the process of dissolving cells in detergent may allow two proteins to interact that might never encounter each other in the intact cell. A common variation on the co-IP assay involves the expression in cells of the protein of interest modified by an epitope tag. This is a small peptide sequence that is specifically recognized by a well characterized (usually monoclonal) antibody; the advantage is that the same antibody can be used to precipitate any protein tagged with that epitope, so that a differ- ent antibody is not needed for each protein. Various epitope tags and other types of so-called affinity tags have been developed for purifying proteins. Multiple distinct affinity tags can also be combined into one large tag, allowing sequential purification steps that greatly increase the specifi- city of the overall purification. In another variation on co-IP, one poten- tial binding partner is tagged with an epitope tag and the other with an enzyme such as luciferase. This enables sensitive, high-throughput, auto- mated detection of the enzyme-tagged binding partner after the immuno- precipitation of the partner bearing the epitope tag. Figure 13.11 Detecting binding partners by co-immunoprecipitation. a cell extract is incubated with an antibody (purple) that specifically binds a protein of interest ( o r a n g e ). the immune complexes are collected and washed, thus separating the protein of interest and any proteins that associate with it ( g r ee n ) from the mixture. the individual proteins in the isolated complexes are then resolved by gel electrophoresis and identified. immune complexes are collected proteins resolved by electrophoresis binding proteins A pull-down assay is frequently used as an alternative to co-IP. In this method, large amounts of a pure protein or a protein fragment are first produced, usually by expression in bacteria. The purified protein is cou- pled to tiny beads, which are then incubated with the cell extract. The beads, along with any proteins that have bound during the incubation, are then separated from the extract, washed, and analyzed to identify binding partners. This approach has the advantage over co-IP in that the physical state of the protein of interest can be controlled experimentally, but pull- down assays may detect interactions that do not occur under physiological conditions. Identifying the proteins that associate with the target protein in co-IP or pull-down experiments can be a challenge. Where a particular inter- action is suspected, the suspicion can be confirmed by immunoblotting; but often, little information is available to guide the investigator. In such cases, mass spectrometry (MS) can be used to identify the binding part- ners directly. Mass spectrometry is an analytic method that provides extremely accurate information on the atomic mass of molecules such as peptides (described more fully below; see Figure 13.23a). For a protein, the atomic masses of its proteolytic fragments often provide sufficient infor- mation to identify it. Other MS approaches allow a selected peptide to be further fragmented during analysis, providing direct information on the amino acid sequence of the peptide. MS is now the basis for many large- scale efforts to characterize protein interactions. (a) ( Binding partners can be identified by screening large libraries of genes The yeast two-hybrid (Y2H) assay is one example of various methods that have been developed to identify the binding partners of a protein by screening cDNA expression libraries—large collections of cDNAs reverse-transcribed from total cell mRNA. Such methods are particularly valuable because they can identify specific partners from literally millions of candidates without any prior information on their identity. In the Y2H assay, the cDNA for a protein or domain of interest is fused to a sequence encoding a DNA-binding domain, and the fusion protein is expressed in yeast. An expression library consisting of random fragments of cDNA fused to a transcriptional activation domain is then introduced into the yeast strain expressing the first fusion protein, such that an individual yeast cell expresses the first fusion along with one of the second fusions encoding a possible binding partner. If the two fusion proteins bind to each other in the cell, a functional transcriptional activator is assem- bled and transcription of a marker gene is induced (Figure 13.12). The marker gene is one that allows growth or induces a color change in yeast colonies grown on selective media. The cDNA of the binding partner can then be recovered and its identity determined by DNA sequencing. One important advantage of such an approach over in vitro methods is that binding must occur in the nucleus of a living cell to be detected, increasing the likelihood that those interactions are relevant under normal physi- ological conditions. Figure 13.12 The yeast two-hybrid assay. a protein of interest (green) is expressed in yeast as a fusion with a dna-binding domain (blue). other proteins are expressed as fusions with a transcription activation domain (orange). (a) When the two fusion proteins bind to each other, transcription of a selectable gene is induced (pink). (b) When the two fusion proteins do not bind to each other, no transcription is seen. Direct protein–protein interactions can be detected by solid-phase screening The apparent association of two proteins in co-IP and pull-down assays may, in some cases, be the result of indirect interaction—an unknown molecule may act as a bridge between the two. This can even be the case for interactions obtained by Y2H screening. Far-Western blotting is designed to visualize direct interactions only. In this approach, the pro- teins in a cell extract are separated by gel electrophoresis and transferred to a solid membrane, as in immunoblotting. The membrane is then incu- bated with a purified protein or protein domain that is labeled or tagged in some way so it can be visualized. If the labeled protein binds to any of the proteins in the extract, a labeled band (or bands) will be visible on the membrane after washing (Figure 13.13). Because proteins in the cell extract are usually completely denatured by boiling in detergent before gel electrophoresis, this approach is most useful for detecting interac- tions, such as protein–peptide interactions, that do not require the native, folded structure of the partner in the extract. Other types of solid-phase binding assays take advantage of microarray technology, which increases the ability to score many possible interactions at once in a single experiment. For instance, many recombinant proteins or protein lysates can be arrayed as tiny spots on a solid support, which can then be probed with labeled purified proteins or cell lysates. Such experimental approaches allow interactions to be probed on a proteome- wide scale, but can be subject to artifacts because one binding partner is immobilized at high concentration on a surface. Fluorescent protein tags are used to locate and track proteins in living cells Proteins are not uniformly distributed in the cell, and signal-induced changes in protein localization provide an important signaling mecha- nism. Furthermore, binding interactions in signaling are highly dynamic and can vary dramatically at different locations and times in the same cell. Although the experimental methods outlined above provide use- ful information on protein–protein interactions that might occur, it is obviously much more useful to know precisely when and where in the cell such interactions actually do occur or, more fundamentally, whether they occur at all during normal signaling. For this, we need to detect and quantify changes in localization and in specific protein–protein interac- tions in living cells. Advances in live-cell imaging have begun to make this possible. For many years, biologists have used antibodies labeled with fluorescent dyes to probe the subcellular localization of proteins (a method termed immunofluorescence). However, this approach requires cells to be killed by chemical fixation before analysis, and thus provides only a static pic- ture of protein localization. The discovery and exploitation of small fluo- rescent proteins has revolutionized the imaging of proteins in living cells. When a fluorescent protein is expressed in a cell as a genetically encoded fusion with a protein of interest, the subcellular localization and local concentration of the fusion protein can be followed over time by fluores- cence microscopy. The first such fluorescent protein to be widely used was green fluorescent protein (GFP), isolated from a jellyfish and named for the color of fluorescent light emitted upon excitation. The isolation of fluorescent proteins from other organisms and the creation of new vari- ants by mutation have broadened the spectrum of available colors, which now include cyan (CFP), yellow (YFP), and red (RFP), and have improved Interactions between proteins and short, linear peptides are discussed in Chapter 2 incubate with labeled protein probe Figure 13.13 Detecting binding partners by far- Western blotting. proteins in a cell lysate are separated by gel electrophoresis and transferred to a membrane, which is then incubated with a labeled protein. proteins on the membrane that directly bind the labeled protein can be visualized (pink). EYFP-CRKL mCherry-Actin merge Figure 13.14 Monitoring subcellular localization of fluorescently tagged proteins. mouse fibroblasts were engineered to express actin tagged with the red fluorescent protein mcherry, and the crKl adaptor tagged with the yellow fluorescent protein eyfp. serum-starved cells were photographed before (top) and 15 min after (bottom) treatment with the mitogen platelet-derived growth factor (pdGf), which causes actin cytoskeletal rearrangements including loss of actin cables (pink arrows) and formation of transient structures termed dorsal actin ruffles ( p i n k a rr o w h ea d s ). in starved cells, crKl is localized to focal adhesions, structures at the ends of actin cables that attach the cell to the underlying extracellular matrix. Upon pdGf stimulation, crKl is released from focal adhesions and relocalizes to dorsal actin ruffles. crKl and actin can each be visualized individually using different wavelengths of ultraviolet light; in the right panels, the two images have been merged to better visualize areas of co-localization (eyfp is colored g r ee n and mcherry is colored r e d in the merged images). (images courtesy of susumu antoku, University of connecticut health center.) the brightness, stability, and other desirable properties of the proteins. Using different fluorescent proteins, it is possible to track the location of multiple fusion proteins in the same cell (Figure 13.14). This means of visualizing proteins in a living cell provides a wealth of information that can be used to evaluate whether or not a specific protein–protein interaction is likely to occur. For example, if two fluo- rescent proteins are highly enriched in the same subcellular compart- ment, their local concentrations at that site will be high and they are much more likely to bind than if they were evenly distributed through- out the cell. The fact that each molecule emits a fixed amount of light makes it possible to calculate the absolute number of molecules and local concentration of each fluorescent species present. A caveat in this type of analysis is that, except in situations where the gene for the nor- mal endogenous protein can be replaced precisely with the fluorescent fusion version, it is likely that the expression level of the fluorescent protein will be different from that of its normal counterpart. Also, care must be taken to ensure that fusion to the fluorescent moiety does not alter the properties of the protein of interest. But at minimum, such experiments can provide valuable clues about whether a particular interaction is likely. Fluorescent proteins have also been exploited as a means of visualizing the intracellular locations of binding sites for proteins of interest. Small, modular protein-binding or lipid-binding domains can be fused to GFP or other fluorescent proteins, and changes in their subcellular localization, and thus in the local concentrations of their binding sites, are tracked after stimulation of the cells or some other manipulation of signaling pathways. Such fluorescent fusions exemplify a class of molecules known as biosensors, which are designed to detect and monitor changes in the signaling status of living cells (biosensors are described in more detail below). Such probes can provide information on the dynamics and sub- cellular localization of signaling events that would not be apparent from biochemical analysis of whole-cell lysates. Protein–protein interactions can be visualized directly in living cells Fluorescently tagged signaling proteins provide hints about whether a specific protein–protein interaction may occur in the course of signal- ing, but typically cannot demonstrate conclusively that the interaction actually occurs. One reason is that the resolution of light microscopy is generally limited by the wavelength of light to ~200 nm, many times the diameter of a typical protein (~5 nm). Thus, even if two proteins are shown to co-localize precisely by fluorescence microscopy, they may not physically interact. More specialized imaging methods have been developed, how- ever, that do allow the detection of direct, physical interactions of proteins in living cells. One such approach is based on fluorescence resonance energy trans- fer (FRET). FRET depends on the ability of one fluorescent molecule to excite a second fluorescent molecule with different excitation and emission spectra when the two are in close proximity (Figure 13.15). The strength of FRET diminishes with the sixth power of the distance between the two fluorescent molecules, so in practice it only occurs when the two mol- ecules are separated by less than 80 Å—that is, when they are physically associated. In a typical FRET experiment, one protein of interest will be expressed in cells as a CFP fusion and a second will be expressed as a YFP fusion. In the absence of FRET, excitation of CFP with short-wavelength laser light leads to emission of cyan fluorescence only (Figure 13.15a). Wherever the CFP fusion and YFP fusion proteins associate in the cell, however, some of the energy from CFP will be transferred to YFP, resulting in a decrease in cyan fluorescence and an increase in yellow fluorescence that can be detected through quantitative image analysis (Figure 13.16). Thus, where and when the two proteins associate can be observed in liv- ing cells. Three-color FRET experiments are also possible, opening up the prospect of monitoring the assembly of multiprotein complexes in living cells. FRET is technically rather challenging, however, and requires care- ful analysis to exclude possible artifacts. Another approach that is conceptually similar to FRET is called the protein-fragment complementation assay (PCA). In PCA, two pro- teins are expressed as fusions with complementary fragments of a reporter enzyme or fluorescent protein; only if the two proteins bind to each other can the two reporter fragments fold together correctly and assemble into (a) cyan fluorescence no fluorescence Figure 13.15 Detecting protein interactions by fluorescence resonance energy transfer (FRET). (a) protein a is expressed in the cell as a fusion with cyan fluorescent protein (cfp), and protein B is expressed as a fusion with yellow fluorescent protein (yfp). excitation of cfp with short-wavelength light (light pink arrow) gives cyan fluorescence, whereas yfp is not excited at this wavelength. (b) When the two fusion proteins are physically associated, some of the energy is transferred from cfp to yfp, resulting in quenching of cyan fluorescence and a corresponding increase in yellow fluorescence. (b) yellow fluorescence Figure 13.16 Visualizing protein binding in a living cell with FRET. the association between meK (a map kinase kinase) and its substrate erk (a map kinase) is visualized by fluorescence resonance energy transfer (fret). cultured cells (hela cells) expressing cfp–erk as the fret donor and meK–yfp as the fret acceptor were stimulated with epidermal growth factor (eGf), which activates the map kinase pathway, and fluorescence images were taken at different times (left panel). fret was calculated from these images and displayed with colors indicating the amount of association (right panel). the cfret (corrected fret) images show the net amount of erk–meK complex at each pixel. in contrast, the cfret/cfp images show the proportion of meK-bound erk versus free erk at each pixel. initially, most erk is bound to meK in the cytoplasm (red). after stimulation with eGf at time 0, a large proportion of erk dissociates from meK in the cytoplasm and migrates to the nucleus. cfp, cyan fluorescent protein; yfp, yellow fluorescent protein. (courtesy of michiyuki matsuda, Kyoto University.) EGF (min) 0 6 30 MEK–YFP CFP–Erk cFRET/CFP cFRET dissociation association (a) (b) GFP fluorescence a functional reporter. In a system based on fragments of GFP, for example, the two proteins by themselves are not fluorescent, but upon their associ- ation a functional, fluorescent GFP molecule is assembled (Figure 13.17). In principle, this is a highly sensitive method of detecting protein interac- tions, because the background of fluorescence in the absence of binding is essentially zero, and thus the dynamic range of the assay is much higher than for FRET. On the other hand, folding and activation of the fluores- cent protein is quite slow and is essentially irreversible, so the method is not ideal for detecting rapid and/or transient changes in binding. methods to pertUrB cell siGnalinG netWorKs and monitor cellUlar Figure 13.17 Detecting protein–protein interactions with the protein- fragment complementation assay (PCA). (a) protein a is expressed in the cell as a fusion with a nonfunctional fragment of green fluorescent protein (Gfp). protein B is expressed as a fusion with a complementary nonfunctional fragment of Gfp. neither fusion protein is fluorescent. (b) When protein a and protein B bind to each other, the two halves of Gfp fold together and reconstitute a functional, fluorescent molecule that can be observed in the cell. responses So far in this chapter, we have focused on methods to analyze the properties of individual signaling proteins and their interactions with substrates or binding partners. However, we are often interested in the behavior of larger signaling networks, which may involve complex and dynamic interactions between many different components. Ultimately, analyzing cellular signal- ing networks requires methods both to stimulate or perturb cells (to manip- ulate signaling inputs) and to analyze how such perturbations change the cells’ internal states and behaviors (to monitor signaling outputs). In this section, we first discuss the range of methods available to perturb signal- ing networks; we then go on to discuss methods that have been developed to monitor various signaling readouts within the cell, and to track their changes dynamically within populations as well as in individual cells. Genetic and pharmacological methods can be used to perturb networks Signaling is often studied in cells stimulated by a physiologically rele- vant input, such as the addition of a mitogen or hormone, or by subjecting cells to environmental stresses, such as shifting to a new temperature or to medium with altered solute concentration (osmolarity). However, in addition to these physiological inputs, there are other approaches to perturb and interrogate the underlying network (Figure 13.18). One can use genetic deletion mutations (knockouts) or RNA interference (RNAi)-mediated knockdown to eliminate or decrease the expression of key protein nodes within the signaling network, in order to identify those nodes that are critical for signal transduction (Figure 13.18b). Another traditional approach is to use pharmacological small-molecule inhibitors to map the role of a particular inhibitor target in the function of the path- way, as well as to analyze pathway relationships (Figure 13.18c). In some cases, natural products or synthetic chemicals can inhibit a particular signaling protein with high specificity, allowing for a very precise analy- sis. In other cases, inhibitors block a large class of signaling molecules in a relatively nonspecific way; for example, okadaic acid blocks the activity of several major classes of Ser/Thr protein phosphatases. Although less specific, the use of such molecules can still be very informative. More recently, it has been possible to use hybrid chemical genetic methods to generate systems that can be conditionally inhibited in a very specific way (Figure 13.18d). For example, a key conserved “gate- keeper” residue in the ATP-binding pocket of many protein kinases can be mutated, without altering its natural function. However, this mutation Figure 13.18 Methods to perturb cell signaling networks. (a) a simple cell signaling network is illustrated, in which binding of input ligand to a cell-surface receptor leads to signaling output. this network can be perturbed in a variety of ways. (b) it can be perturbed by decreasing or eliminating the expression of one of the components. (c) it can be perturbed by adding a small-molecule inhibitor of one of the components. (d) it can be perturbed using chemical genetics; a component of the network, such as a kinase, can be replaced with an engineered version that is uniquely inhibited by an atp analog. (e) strategy for design of a mutant kinase (green) that is engineered to be inhibited by an atp analog that cannot bind other atp- binding proteins, including the corresponding wild-type kinase (gray), because of steric incompatibility. here, the kinase is mutated to create a “hole” that can accommodate the extra “bump” on the atp analog. (a) INPUT INPUT INPUT INPUT analog-sensitive mutant kinase OUTPUT? mutation of gatekeeper residue “HOLE” ATP analog “BUMP” wild-type kinase analog sensitive mutant kinase INPUT INPUT INPUT Figure 13.19 Non-natural ways to activate a cell signaling pathway. (a) in this example, normal wild-type (Wt) signal output depends on the binding of a cytosolic protein (purple) to another protein localized on the membrane ( g r ee n ). (b) this binding interaction can be experimentally induced if the two components are expressed as fusion proteins that bind to a small-molecule chemical heterodimerizer (blue diamond). in optogenetic systems, the association of the two components can be controlled by fusion to light-induced protein–protein interaction domains (left). light can also be used to activate specific membrane ion channels (right). such light-sensitive proteins have been isolated from plants and algae, but can be expressed in other cell types. An example of such a light- sensitive conformational change is illustrated in Figure 10.27 allows the modified kinase to bind to and be blocked by a bulkier ATP ana- log inhibitor that cannot fit into any endogenous kinase—the ATP analog has a “bump” which can only be accommodated by a kinase with an engi- neered complementary “hole” (Figure 13.18e). When this type of chemical- ly sensitive allele can be genetically introduced into a researcher’s system of interest (replacing the wild-type allele), then one can create a system in which a targeted kinase can be uniquely and selectively inhibited, in the absence of off-target effects on other proteins. Chemical dimerizers and optogenetic proteins provide a dynamic way to artificially activate pathways In addition to methods to inhibit specific nodes within a pathway, there are also methods to artificially activate signaling nodes (Figure 13.19). Most simply, a protein of interest can be overexpressed, or a constitutively active mutant can be introduced, in order to probe the consequences of increased activity. In the case of molecules that are activated by locali- zation or by binding to a specific partner (Figure 13.19a), it is possible to link these molecules to small-molecule-dependent heterodimerization domains (Figure 13.19b). In these cases, formation of an active complex or recruited state of a molecule can be induced by addition of a small- molecule artificial dimerizer. The advantage of this system is that a sin- gle, well-defined molecular event (the binding of two proteins of choice) can be rapidly induced in the absence of other confounding changes. The most commonly used examples of these artificial dimerization domains are derived from proteins that bind to the rapamycin family of immuno- suppressant drugs. Another way to artificially activate signaling nodes is through optoge- netic methods (Figure 13.19c). These take advantage of light-sensitive proteins and domains, originally found in plants or protists, that when illuminated with light of a specific wavelength, will undergo conforma- tional changes that result in functional changes. In some cases, light- sensitive domains show light-induced heterodimerization. By fusing the two optogenetically controlled partner domains to signaling proteins that are controlled by co-recruitment or localization, and expressing these in cells, one can create a system that is specifically activated by light. In some cases, this may allow localized activation only in the small region of the cell that is illuminated. Another type of optogenetic con- trol involves light-sensitive channel proteins, such as channel rhodopsins, which can be used to activate excitable cells such as neurons. This kind of optogenetically controlled system can be used to specifically activate signaling systems with complex temporal or spatial patterns. cDNA microarrays and high-throughput sequencing are used to monitor the transcriptional state of a cell Many signaling pathways ultimately lead to changes in transcription. Changes in the levels of specific mRNAs can be monitored by a number of methods, for example by quantitative polymerase chain reaction (qPCR). The most powerful insights, however, come from high-throughput meth- ods that provide a global overview of a cell’s transcriptional response (Figure 13.20). Microarray analysis allows one to quantify changes in microarray analysis (b) RNA seq control cells (unstimulated) + INPUT stimulated cells control cells (unstimulated) + INPUT stimulated cells RNA isolation mRNA mRNA reverse transcribe, label cDNA cDNA RNA isolation mRNA mRNA reverse transcribe, fragment cDNA cDNA red luorescent probes green luorescent probes next-generation high-throughput sequencing cluster analysis of genes that are up-regulated or down-regulated by stimulation Figure 13.20 Analysis of cellular gene expression output changes. (a) in microarray analysis, mrna from two different cell populations is isolated and reverse- transcribed into fluorescently tagged cdna. the two cdna samples are then mixed and hybridized to a panel of dna probes arrayed on a glass slide. data about the absolute amount of each cdna, and the relative amounts of each cdna in the two samples, are obtained. (b) in rnaseq, cdna from different cell samples is directly sequenced in a high-throughput dna sequencer. the abundance of different mrnas can be inferred from the abundance of sequencing reads that correspond to each mrna. cluster analysis is a bioinformatic method that groups or “clusters” cell samples and cdnas together based on their similarity in a series of quantitative values, such as the expression levels of different cdnas. for example, two genes with very similar expression patterns in different samples will cluster closely together. (image of illumina hiseq® 2500 system courtesy of illumina, inc. ©2012. all rights reserved.) transcription by measuring mRNA levels in the cell before and after stimulation. This method depends on hybridization of labeled cDNA from the cells to be analyzed to DNA or oligonucleotide microarrays printed on chips. Even more information is provided by direct high-throughput cDNA sequencing analysis (RNA-seq). In this approach, millions of independent cDNA fragments are sequenced for each sample, allowing the relative abundance of different messages to be determined. While fast, powerful, and comprehensive, these methods have two major limitations. First, they only provide information about transcriptional changes in the cell, and not about post-transcriptional changes, such as changes in pro- tein abundance or post-translational modifications. Second, they typically require material from a large number of cells to be pooled for analysis, and thus information about what is happening in individual cells is lost. Modification-specific antibodies provide a method to track post-translational changes Antibodies that specifically recognize particular post-translational modi- fications are powerful tools for characterizing the state of these modifica- tions in a cell that has been exposed to a particular history of inputs. For example, antibodies are available that recognize tyrosine-phosphorylated proteins, without particularly strong discrimination between distinct sites. But highly specific antibodies have also been generated against indi- vidual phosphorylated sites, for example the activation loop phosphoryla- tion sites in a particular kinase. These antibodies recognize phosphate in the context of a specific amino acid sequence (Figure 13.21). Such antibodies can be extremely useful in measuring the activity of particu- lar signaling pathways through methods such as immunoblotting or flow (a) (c) control cells (unstimulated) + INPUT stimulated cells quench and lyse cells run extract on denaturing protein gel Figure 13.21 Analysis of changes in phosphorylation using phosphospecific antibodies. (a) a phosphorylated peptide (in this case, the phosphorylated activation-loop sequence of the map kinase erk1) is used to immunize rabbits, and antibodies are prepared. (b) antibodies that bind to the phosphorylated peptide, but not to unphosphorylated peptide or to other phosphorylated sites, are isolated. (c) a hypothetical experiment in which lysates of control and stimulated cells are subjected to immunoblotting with the phosphospecific erk antibody. phospho-erk is revealed as two bands on the immunoblot (arrow), and the extent of phosphorylation is indicated by the intensity of the bands. binding no binding T Y no binding transfer to membrane probe with phosphospecific antibody 1 2 3 4 phospho-Erk immunoblot antibody unbound antibody bound Figure 13.22 Chromatin immunoprecipitation (ChIP). dna in chromatin is cross-linked to associated proteins and fragmented into small pieces. the fragmented chromatin is then incubated with antibodies to a specific protein or post-translational modification. in this example, an antibody to acetylated histone is used. antibody-bound chromatin is isolated, and protein is removed. the dna in this fraction is enriched in dna that was associated with chromatin that contained acetylated histones (pink). cytometry (see below). For example, antibodies recognizing the activat- ing phosphorylation sites or known substrates of Erk or Akt kinases can be used as readouts of Ras/MAPK or PI3K–Akt pathway activity, respec- tively. Similar antibodies have been generated recognizing other specific post-translationally modified sites, for example histones that have been methylated or acetylated at specific sites. Not only do such antibodies allow the quantification of changes in modifi- cation over time in different cell samples, they can also be used to purify proteins and other components associated with the modified proteins. For example, in the chromatin immunoprecipitation (ChIP) method, antibodies to specific modified histone sites are used to purify other pro- teins and genomic DNA associated with the modified histones. Analysis of the associated DNA provides a measure of which genes and sequences are associated with particular chromatin modifications (Figure 13.22). Protein-binding domains that recognize specific modified sites (reader domains) can also provide the basis for purifying their binding partners or quantifying changes in their abundance. For example, SH2 domains can be used in a manner similar to phosphospecific antibodies to detect and quantify their tyrosine-phosphorylated binding sites in cell lysates or fixed cells. When expressed in cells, fused to a fluorescent marker, modification-specific protein-binding domains (such as PH domains that can bind to specific phosphoinositol lipids) can also be used to probe changes in the amount or subcellular localization of their binding sites in living cells (discussed in more detail below). Mass spectrometry is the workhorse for identification of proteins and their modifications Advances in mass spectrometry (MS) methods and instrumenta- tion have dramatically improved our ability to identify proteins in complex mixtures. These approaches can be used to determine not only which proteins are post-translationally modified, but also the specific amino acids where the modification takes place. In a typical experiment (Figure 13.23), a mixture of proteins is digested into peptides using a (a) (b) control cells (unstimulated) + INPUT stimulated cells 12C/14N (light) Lys, Arg mix 13C/15N (heavy) Lys, Arg lyse cells digest with trypsin enrich phosphopeptides mass spectrometry m/z m/z peptide sequence protein identifcation Figure 13.23 Analysis of changes in protein abundance and post-translational modification using mass spectrometry (MS). in a typical experiment, cells are lysed and proteins fractionated by size using polyacrylamide-gel electrophoresis (paGe). Gel slices containing multiple proteins are digested using a protease such as trypsin, producing a set of peptides for each protein. the peptides are separated by liquid chromatography and ionized. in the initial ms spectrum, the intensity and mass-to-charge (m/z) ratio of each ion is measured. individual ions are then fragmented, resulting in an ms/ms spectrum that can be used to deduce the sequence of the peptide. these sequences are then compared to proteome databases, thus allowing the identification of proteins in the initial sample. (b) in silac, cells are grown in the presence of essential amino acids labeled with distinct isotopes. in this example, unstimulated cells are grown in 12c/14n (light) lysine and arginine, while stimulated cells are grown in 13c/15n (heavy) lysine and arginine. the samples are mixed and processed together, and the relative intensity of the light and heavy peaks in the ms spectrum can reveal changes in abundance. in this example, stimulation reduces the abundance of peptide 1, while peptide 2 is unchanged and peptide 3 increases upon stimulation. protease such as trypsin, which cleaves C-terminal to lysine or arginine. These peptides are separated by liquid chromatography and converted to ions in the gas phase, and the intensity and mass-to-charge (m/z) ratio of every ion is measured by the mass spectrometer. Individual peptides are then excited such that chemical bonds are broken, producing a series of fragment ions (referred to as the MS/MS or MS2 spectrum), usually cor- responding to the sequential loss of amino acids from the initial peptide. Thus, the fragmentation pattern of an ion essentially allows the peptide to be sequenced, and the corresponding protein can be identified by com- paring these sequences to proteome databases. Post-translational modifications can be identified by mass spectrometry because they change the mass of the peptide to which they are attached, and also because they produce characteristic fragment ions. For exam- ple, phosphorylated peptides produce a fragment ion at m/z 79 (HPO –). Phosphates attached to tyrosine are more stable to fragmentation, but produce a phosphotyrosine immonium ion with m/z 216. Ubiquitylated lysines are protected from digestion by trypsin, so the complement of pep- tides detected from a given protein is changed. In addition, treatment with trypsin leaves the C-terminal Gly-Gly motif from ubiquitin attached to the modified lysine. Post-translational modifications are often substoichiometric and transient, so the fraction of peptides that are modified is often fairly low. Therefore, it is usually beneficial to enrich for peptides that carry the modification of interest, particularly in studies seeking to identify large numbers of sites. Several methods are used (alone or in combination) to enrich for phos- phorylated peptides. Immobilized metal affinity chromatography (IMAC) exploits the affinity of metal ions such as Fe3+ for the negatively charged phosphate group. Similarly, TiO₂ or other metal oxides are often used. Another common enrichment strategy employs anti-phosphotyrosine antibodies, which can be used on intact proteins or on peptide mixtures. Pan-specific phosphoserine/phosphothreonine antibodies provide poor enrichment for mass spectrometry, but some motif-specific antibodies that recognize sequences phosphorylated by specific kinases have been suc- cessfully used. Analogous antibodies can be employed to enrich for other post-translational modifications. In addition to identifying sites of post-translational modification, it is useful to quantify changes that occur under different conditions. For example, this can help distinguish background phosphorylation (which remains stable) from pathway-specific phosphorylation (which increases or decreases upon stimulation). A number of techniques have been devel- oped to quantify the amount of protein or post-translational modification in different samples. In one method, referred to as stable-isotope labeling of amino acids in cell culture (SILAC), cells are grown in the presence of either natural 12C/14N (light) or 13C/15N-labeled (heavy) amino acids such as lysine and arginine. After labeling, the two cell cultures are mixed and processed together. Isotopic variants of a given peptide have the same chemical characteristics, so they co-fractionate, but they can be distin- guished because of their slight difference in mass. As a result, every pep- tide produces a pair of peaks in the mass spectrum. The intensities of the two peaks can then be used to determine the relative abundance of the protein or modification in the two samples (Figure 13.23b). A related tech- nique is iTRAQ (isobaric tag for relative and absolute quantification), in which peptides are chemically modified with tags that have the same mass but produce distinct fragmentation patterns. One advantage of iTRAQ is that it can simultaneously quantify up to eight different samples in a sin- gle experiment. Label-free techniques that allow absolute quantification (AQUA) have also been developed. For each peptide of interest, a vari- ant labeled with 13C and/or 15N isotopes is chemically synthesized, and a defined quantity is added as an internal standard to the mixture of pep- tides. The stoichiometry of post-translational modifications can be deter- mined by measuring the absolute concentrations of both unmodified and modified versions of the same peptide. The specific mass spectrometry approach taken depends on the purpose of the experiment. In some cases, the objective is to discover new sites of post-translational modification by identifying large numbers of peptides. These experiments are inherently irreproducible because not all peptides are selected for fragmentation. In experiments that focus on specific sig- naling pathways, it is useful to focus on pre-selected peptides in a tech- nique referred to as multiple reaction monitoring (MRM). Live-cell time-lapse microscopy provides a way to track the dynamics of single-cell responses A shortcoming of the methods described above for analyzing cell state is that they all require the lysis and pooling of large numbers of cells to provide sufficient material for analysis, and thus these methods have limited ability to yield single-cell information. Furthermore, limits on the throughput and speed of these analyses can prevent the researcher from obtaining a high-resolution dynamic picture of the timing of various responses. Single-cell information is often very important, since there can be a great deal of stochastic cell-to-cell variability in a response, due to, for example, slight variations in the number of signaling molecules or ribos- omes in each cell or its precise cell-cycle stage at any given time. Thus, when one looks at a large population of cells, important information about the nature of a signaling response could be lost. Time-lapse microscopy of live cells provides one approach to overcome these limitations (Figure 13.24). Imaging can be done in standard culture vessels, or cells may be loaded into a microfluidic device that traps cells within a viewing window for microscopy (Figure 13.24a). Such devices, depending on how they are constructed, allow the researcher to use vari- ous in-flow ports to change the media of the trapped cells in a highly con- trolled, dynamic way. For example, as shown in Figure 13.24b, the input signal molecule concentration can be rapidly shifted from low to high in a step function, allowing real-time monitoring of cellular responses after this sudden change in input. To monitor responses, one requires a live- cell reporter or biosensor: for transcriptional responses, this could be expression of a fluorescent reporter protein from a promoter; for post- translational modifications, this could be a FRET reporter of a specific phosphorylation event (described further below). With such reporters, one can track how each individual cell in the field of view responds to the input change, and can monitor variation in response within the popula- tion. In the example illustrated in Figure 13.24b, when two individual cells are tracked, there appears to be stochastic variability in the timing of when the reporter 1 response is initiated after the change in input. By contrast, there seems to be a tighter correlation of the timing of the reporter 2 response after the initiation of the reporter 1 response. Thus, through analysis of multiple individual cells, one can see how different aspects of a cellular response show high or low variability. Overall, time-lapse microscopy provides a powerful way to follow single- cell responses and the best way to follow the dynamics of these responses (Table 13.1). However, this approach is limited by the need for a live- cell reporter (for example, not all specific phosphorylation events have a (a) input [high] input [low] buffer microfluidic device viewing window outlet Figure 13.24 Live-cell time-lapse microscopy. cells can be imaged in a microfluidic chamber. cells are loaded into the device, and specific inlet valves are used to control media flow over the cells, allowing precise spatiotemporal control over the input stimulation that the cells experience. the control valves cell loading reporter 1 reporter 2 cell A field of cells in the device can be tracked by microscopy. (b) for each individual cell, multiple outputs can be monitored over time using different fluorescent reporters (g r ee n and p i n k in this example). the response dynamics of each individual cell in the field of observation can be tracked. in this example, the individual cell traces demonstrate that there is cell-to-cell variation in the time of reporter 1 activation, but reporter 2 is consistently activated after reporter 1. (microscope image courtesy of leica microsystems.) 0 min 30 min 60 min reporter 1 reporter 2 suitable reporter). It is also limited by throughput—although microfluidic devices help tremendously, usually experiments are limited to the analysis of tens to hundreds of cells, which is several orders of magnitude lower than what can be analyzed by methods like flow cytometry (described below). Biosensors allow signaling activity to be monitored in living cells Monitoring the activity of signaling pathways and networks at the single- cell level requires some way to convert changes in a biochemical activ- ity of interest into changes that can be visualized in the microscope. To accomplish this, a variety of biosensors have been developed, typically based on fluorescent molecules that can be directly visualized by fluores- cence microscopy. A very simple type of biosensor can be made by fusing a modification-specific modular binding domain to a fluorescent protein. Table 13.1 Comparison of single-cell analysis methods single-cell dynamics no single-cell dynamics medium throughput (hundreds of cells) high throughput (103–105 cells) requires live-cell reporter live-cell reporter not required information on subcellular localization little information on localization For example, GFP can be fused to a PH domain that recognizes a specific phosphoinositide species such as phosphatidylinositol 3,4,5-trisphosphate (PIP₃). When such a biosensor is expressed in cells, changes in the level of PIP₃ can be visualized as changes in the amount of GFP fluorescence on the membrane. Not only can such a biosensor monitor overall chang- es in the level of PIP₃ over time, but also can reveal specific subcellular locations where PIP₃ levels are particularly high or low. An example of the use of such PH domain-based sensors is provided in Figure 7.11. Other modular domains that recognize modified targets, such as reader domains that bind phosphorylated motifs (SH2 domains, for example), can simi- larly be used to monitor changes in the abundance or location of their binding sites in living cells. Another simple type of biosensor consists of cell-permeable fluorescent dyes whose spectral properties change based on their chemical environ- ment. For example, Fura-2 is a dye that binds calcium. The amount of fluorescence emitted when it is excited at particular wavelengths depends on whether calcium is bound or not. Thus, it is relatively straightforward to monitor the absolute concentration of calcium in cells, as well as the spatiotemporal dynamics of changes in intracellular calcium, in the pres- ence of this dye. Such a calcium-sensitive dye was used to visualize cal- cium waves in Figure 6.12. Other specific dyes can be used to monitor changes in different properties, such as the potential difference (voltage) across a membrane or local pH. More sophisticated biosensors can monitor changes in properties such as the activity of specific G proteins or protein kinases, or the concentration of small molecules such as cAMP. Such biosensors are typically based on FRET; changes in the conformation of the biosensor due to pathway activ- ity (G protein binding, phosphorylation, cAMP binding) are converted into changes in the FRET ratio, which can in turn be monitored by microscopy. For example, to monitor the activity of a protein kinase, a biosensor can be constructed that contains a specific substrate peptide sequence favored by that kinase, along with a modular protein-binding domain that binds to the substrate site only when it is phosphorylated. Thus, the conforma- tion of the biosensor will be quite different in the unphosphorylated state versus when it is phosphorylated; this change in conformation can be coupled to changes in FRET when the biosensor contains a FRET donor and FRET acceptor. Biosensors to monitor G protein activity often contain the G protein itself fused to an effector domain that binds the G protein only when it is active (in the GTP-bound state) (Figure 13.25). Thus, an intramolecular interaction between the two occurs only when the G protein has been activated by endogenous guanine nucleotide exchange factors (GEFs). Again, the conformational change can be monitored by changes in FRET. Expression of such FRET-based biosensors in cells allows the magnitude and location of the activity of interest to be moni- tored by microscopy (Figure 13.25b,c). Biosensors such as those described above are very powerful tools for inter- rogating pathway activity in individual cells in real time. Experimental results must be interpreted carefully, however, particularly in cases where the biosensor itself may perturb signaling. Often, relatively high levels of biosensor are needed for visualization, and this may result in inhibition of the pathway being monitored. For example, a GFP-PH domain fusion may prevent the binding of the endogenous PH domain-containing effectors to the membrane, or a kinase sensor may prevent normal levels of phosphor- ylation of the endogenous targets of the kinase. Furthermore, development of a useful biosensor may require a great deal of protein engineering to optimize for low background, high sensitivity, and appropriate specificity. (a) inactive (low FRET) active (high FRET) GEF GAP (b) before + LPA R 2.0 1.0 (c) 1 1.6 1.4 1.2 0.8-5 0 5 10 15 time (min) Flow cytometry provides a method to analyze rapidly single-cell responses One of the most powerful and rapid methods for high-throughput analy- sis of single-cell responses is flow cytometry (Figure 13.26). Typically, a large population of cells is stimulated and the response is monitored using fluorescent reporters. In the case of a transcriptional response, this can be expression of a fluorescent protein, like GFP. In the case of a phos- phorylation-based response, for example, one can fix, permeabilize, and stain cells with a fluorescently labeled phosphospecific antibody (or other output-specific labeling reagent). In either case, cells are fixed at various times after stimulation and each sample is analyzed in the flow cytometer. Hydrodynamic focusing allows the cells to pass through the instrument in single file; lasers are used to excite each cell, and the amount of fluores- cence is measured by a detector. A large number of different reporters can be followed within each cell, if the instrument is equipped with multiple lasers and detectors. The data can then be processed to give a histogram of the distribution of cells with a given level of fluorescence within the population. An example of flow cytometry analysis is shown in Figure 13.26b, where a population of cells has been stimulated for 0, 30, 60, or 90 minutes, and the output of a fluorescent reporter measured (here, for example, it might be the amount of phospho-Erk after mitogen stimulation). In the exam- ple on the left, the cells respond in a graded fashion—the cell population gradually shifts with stimulation time to higher fluorescence values, and the cells pass through a stage of intermediate fluorescence. By contrast, in the example on the right, the cells respond in a bistable fashion—the cells exist only in a low or high fluorescence state, and it is the distribu- tion of cells in each of these states that changes with the time of stimu- lation. Although these two classes of responses (graded versus bistable/ all-or-none) are very different, and can lead to very different biological Figure 13.25 A fluorescence resonance energy transfer (FRET) biosensor for activation of Rho GTPases. (a) the nucleotide state of rho Gtpases is regulated by Gefs (guanine nucleotide exchange factors) and Gaps (Gtpase- activator proteins), with the Gtp-bound form being active and capable of interacting with downstream effectors. an example of a rhoa biosensor consists of full-length rhoa protein, a rhoa-binding domain derived from its effector pKn (brown), and a fret pair of fluorescent proteins [cyan fluorescent protein (cfp) and yellow fluorescent protein (yfp)]. these components are connected with linkers in such a way that the activation of rhoa in the sensor causes an intramolecular conformational change that increases fret efficiency. (b) representative ratio (R) images (fret/cfp) of an mcf-7 breast cancer cell are shown before and after addition of lysophosphatidic acid (lpa), a known activator of rhoa. By conducting time-lapse imaging of multiple cells, the temporal kinetics of rhoa activation can be estimated based on changes of ratios (data shown as mean of whole cells ± standard error of the mean). scale bar 5 μm. (images courtesy of taofei yin and yi Wu, University of connecticut health center.) stimulate cells with INPUT fix, stain with phospho-specific antibody (fluorescently labelled) to read OUTPUT flow cytometry graded response bistable response sample nozzle hydrodynamic focusing fluorescence emitted from cells detected OUTPUT (fluorescence) OUTPUT (fluorescence) laser light source forward- and side- scattered light detected Figure 13.26 Flow cytometry. (a) cell populations (either unstimulated or stimulated) are labeled with fluorescent antibodies or other fluorescent reporters of signal output; in this example, a phosphospecific antibody is used. the flow cytometer rapidly quantifies fluorescence for each cell; forward and side scatter are used to monitor the size and granularity of cells. (b) the number of cells and fluorescence level are plotted for each cell population. a graded response (left), in which individual cells can have intermediate levels of output, can be distinguished from a bistable response (right), where cells only can exhibit either low or high output. behaviors, it is important to note that these two responses could not be distinguished by a population-based assay such as an immunoblot using a phosphospecific antibody. The pooling of cells required for obtaining enough material for detection on the immunoblot would result in the loss of all single-cell information, and even the cells showing a bistable response would appear graded because of variation in the timing of when each cell shifted from the low to high fluorescence state. Overall, flow cytometry is a very powerful way to obtain single-cell infor- mation because of the very large number of cells (103–105) that can be analyzed in this way, allowing one to obtain excellent statistics from large cell populations (see Table 13.1). Another advantage of flow cytometry is that it does not require a live-cell reporter, and thus can be used, for example, to track simultaneously many different phosphorylation events using phosphospecific antibodies labeled with different fluorophores. A limitation of flow cytometry is that one cannot obtain single-cell dynamic data—that is, how one specific cell is responding at time A and then at time B. This is because in flow cytometry, one does not track the same cells, but follows a large population of cells by sampling their response distributions at each time. qUestions You are studying the development of a particular region of the mouse brain, and have genetic evidence that protein X is involved. From your search of the literature, you find that protein X has several identified domains and a fairly long region with no known mapped function. You suspect this region may be a new protein-binding domain, and in searching databases you think you find weak sequence similar- ity between this region and other proteins. Describe an experimental strategy to test your hypothesis and to find which proteins (or other biological molecules) bind to the putative domain. references Based on theoretical considerations, you believe that the modification of protein Y by enzyme X is operating at saturation in the cell (that is, the reaction is zero order). What information is needed to evaluate whether this is the case in the cells that you are studying? Provide a set of experiments to test your hypothesis. You find that stimulation of cells with a cytokine leads to rapid changes in the actin cytoskeleton. The receptor for the cytokine binds to pro- tein X upon activation, and you develop an optogenetic tool that allows you to recruit protein X to the plasma membrane. How might this tool help you dissect the mechanism of actin rearrangement? What are the advantages and disadvantages of studying the response to the normal stimulus (the cytokine) versus responses to the optogenetic tool? You are interested in cellular responses to low pH and perform a SILAC (stable-isotope labeling of amino acids in cell culture) experiment to look for changes in phosphorylation upon a change in pH. You find a phosphorylated peptide from protein Y that is rapidly phosphorylated on serine when the pH of the culture medium is lowered. Design a series of experiments to determine what fraction of the total protein Y in the cell is phosphorylated at this site, and to test whether phosphorylation of Y is important for inducing the transcription of pH-responsive genes. You have developed an inhibitory antibody that binds to a mitogen receptor and are in the early stages of clinical testing to see whether this antibody might provide a new way to stop the growth of certain tumors that have abnormally high receptor activity. You find that the antibody decreases growth of a tumor cell line by ~10%. Discuss how you might determine whether this represents a weak response by all cells, or a strong response by a small fraction of the cells. If the latter, how would you address why some cells respond and not others? references Biochemical and Biophysical analysis of proteins Branden C & Tooze J (1998) Introduction to Protein Structure, 2nd ed. New York: Garland Science. Hammes GG (2000) Thermodynamics and Kinetics for the Biological Sciences. New York: Wiley-Interscience. Klotz IM (1997) Ligand–Receptor Energetics: A Guide for the Perplexed. New York: John Wiley & Sons. Kuriyan J, Konforti B & Wemmer D (2012) The Molecules of Life: Physical and Chemical Principles. New York: Garland Science. Lei M, Lu W, Meng W, Parrini MC, Eck MJ, Mayer BJ, Harrison SC. Structure of PAK1 in an autoinhibited conformation reveals a multi-stage activation switch. Cell 2000; 102:387-397. Menten L & Michaelis MI (1913) Die Kinetik der Invertinwirkung. Biochem Z 49, 333–369. [Recent translation, and an older partial translation.] Rich RL & Myszka DG (2000) Advances in surface plasmon resonance biosensor analysis. Curr. Opin. Biotechnol. 11, 54–61. Voet D, Voet JG & Pratt CW (2013) Principles of Biochemistry, 4th ed. New York: Wiley. Winzor DJ & Sawyer WH (1995) Quantitative Characterization of Ligand Binding. New York: Wiley-Liss. mappinG protein interactions and localization Choudhary C & Mann M (2010) Decoding signalling networks by mass spectrometry-based proteomics. Nat. Rev. Mol. Cell Biol. 11, 427–439. Giepmans BNG, Adams SR, Ellisman MH & Tsien RY (2006) The fluorescent toolbox for assessing protein location and function. Science 312, 217–224. Golemis EA & Adams PD (eds) (2005) Protein–Protein Interactions: A Molecular Cloning Manual, 3rd ed. New York: Cold Spring Harbor Press. Jones RB, Gordus A, Krall JA & MacBeath G (2006) A quantitative protein interaction network for the ErbB receptors using protein microarrays. Nature 439, 168–174. Lippincott-Schwartz J & Patterson GH (2003) Development and use of fluorescent protein markers in living cells. Science 300, 87–91. Michnick SW (2003) Protein fragment complementation strategies for biochemical network mapping. Curr. Opin. Biotechnol. 14, 610–617. Wu JQ & Pollard TD (2005) Counting cytokinesis proteins globally and locally in fission yeast. Science 310, 310–314. methods to pertUrB cell siGnalinG netWorKs and monitor cellUlar responses Barrios-Rodiles M, Brown KR, Ozdamar B et al. (2005) High-throughput mapping of a dynamic signaling network in mammalian cells. Science 307, 1621–1625. Bishop A, Buzko O, Heyeck-Dumas S et al. (2000) Unnatural ligands for engineered proteins: new tools for chemical genetics. Annu. Rev. Biophys. Biomol. Struct. 29, 577–606. Fujioka A, Terai K, Itoh RE et al. (2006) Dynamics of the Ras/ERK MAPK cascade as monitored by fluorescent probes. J. Biol. Chem. 281, 8917–8926. Levskaya A, Weiner OD, Lim WA & Voigt CA (2009) Spatiotemporal control of cell signalling using a light- switchable protein interaction. Nature 461, 997–1001. Perez OD & Nolan GP (2002) Simultaneous measurement of multiple active kinase states using polychromatic flow cytometry. Nat. Biotechnol. 20, 155–162. Pertz O & Hahn KM (2004) Designing biosensors for Rho family proteins—deciphering the dynamics of Rho family GTPase activation in living cells. J. Cell Sci. 117, 1313–1318. Toettcher JE, Voigt CA, Weiner OD & Lim WA (2011) The promise of optogenetics in cell biology: interrogating molecular circuits in space and time. Nat. Methods 8, 35–38. Young JW, Locke JC, Altinok A et al. (2011) Measuring single-cell gene expression dynamics in bacteria using fluorescence time-lapse microscopy. Nat. Glossary 14-3-3 proteins A class of small proteins that bind specifically to target proteins that are phosphorylated on serine or threonine residues. 14-3-3 binding can regulate the activity, conformation, and/or subcellular localization of its targets. N -acetylation The transfer of an acetyl group to the terminal amino group of lysine residues in proteins. activation loop An important regulatory element for protein kinases that undergoes dramatic conformational changes between the inactive and active forms of the catalytic domain. The active conformation of the kinase is typically promoted by phosphorylation of the activation loop. S -acylation Reversible attachment of a fatty acid group to a protein via a thioester linkage; palmitoylation is an example of S-acylation. adaptation A property of a signaling system in which, in response to a signaling input, there is an initial burst of output and then output returns to basal levels even if the input stimulus remains. adaptive immune response Response of the immune system to a specific antigen that typically generates an immunological memory. adaptor A protein with multiple protein-binding domains that mediates the assembly of complexes of three or more components. Grb2, a canonical adaptor, binds to tyrosine- phosphorylated proteins such as growth factor receptors via its SH2 domain, and binds downstream effectors such as the Ras activator Sos via its SH3 domains. affinity Strength of a noncovalent binding interaction; the higher the affinity, the more likely two binding partners will exist in a complex. agonist A ligand that binds a receptor and stimulates its activity. AKAPs (A-kinase anchoring proteins) Scaffold proteins with specific binding sites for protein kinase A regulatory subunits and other proteins, and which localize to specific subcellular compartments such as the plasma membrane, mitochondrion, or centrosome. Akt A family of serine/threonine kinases activated by phosphoinositides that play a role in signaling pathways that control cell growth, proliferation and survival. Also known as protein kinase B (PKB). allosteric switch proteins Modular enzymes in which the activity of catalytic domains is coupled to conformational changes induced by upstream inputs. allostery The property of being able to exist in two or more structural states of differing activity. The equilibrium between these states is modulated by ligand binding or covalent modification. alpha helix (α helix) Common structural element in proteins in which a linear sequence of amino acids adopts a right-handed helical conformation stabilized by internal hydrogen bonds between backbone atoms. amphipathic Molecules containing both hydrophilic and hydrophobic portions. amplification— see signal amplification anaphase-promoting complex (APC) A large, multisubunit ubiquitin ligase complex that regulates cell- cycle progression by ubiquitylating specific proteins and targeting them for proteasomal degradation. angiogenesis The generation of new blood vessels. antagonists Compounds that bind to a receptor but which fail to evoke an activating response. antigen Any molecule or part of a molecule recognized by the variable antigen receptors of lymphocytes. antigen-presenting cell (APC) Cell that displays a foreign antigen complexed with MHC at its cell surface for presentation to T cells. apoptosis A highly programmed form of cell death initiated by specific signals that activate caspases and lead to characteristic biochemical and morphological changes. apoptosome A large cytosolic complex that serves as a scaffold for the recruitment and activation of caspases. avidity Increased apparent affinity of a molecule for its ligand due to the presence of multiple binding sites on both partners. Bcl2 family Family of proteins that either promote or inhibit apoptosis by regulating the permeability of mitochondria. beta sheet (β sheet) Common structural element in proteins in which individual polypeptide chains are held together by hydrogen bonds between peptide bonds in adjacent strands. Strands can run in the same or opposite directions. binding The relatively stable association of two components. binding isotherm Curve in which binding (fractional occupancy) of one component of a binding reaction is plotted as a function of concentration of the other component, under conditions of constant temperature. biosensor Molecular device, often a fluorescent small molecule or protein, that can be used to monitor specific changes in the physiological state of a cell. bistability A property of signaling systems that exist in either of two distinct output states, in the absence of stable intermediate states. bromodomain Modular protein domain that recognizes peptide motifs containing an acetylated lysine. buried surface area The total surface area (in Å2) at a macromolecular interface that is exposed to solvent in the uncomplexed proteins but is buried in the complex. Ca2+/CaM-dependent protein kinases (CaM-Ks) A family of serine/threonine kinases that are regulated by the binding of Ca2+-bound calmodulin (CaM). CAAX box motif A motif found at the C-terminus of proteins destined for prenylation. During prenylation, the C-terminal three residues (AAX, where A is an aliphatic amino acid and X is any amino acid) are proteolytically removed and the isoprenyl group is added to the cysteine by a thioether linkage. calmodulin (CaM) A small calcium-binding protein that confers calcium regulation on cellular signaling and effector molecules, including protein kinases and phosphatases, to which it binds. CaM— see calmodulin cAMP Cyclic nucleotide synthesized from ATP and used as a signaling mediator. cascade A signal transduction pathway in which multiple enzymes are linked in series, such that output of one enzyme directly or indirectly regulates the next enzyme in the cascade. caspases A specialized group of cysteine proteases, activated in apoptosis, which specifically cleave peptide bonds after aspartic acid residues in their protein targets. catalytic domain Part of an enzyme that is responsible for promoting a specific chemical reaction. catalytic rate constant ( k _cat ) The rate constant for the catalytic step carried out by the enzyme when saturated with substrate. caveolae Small, cup-shaped patches of plasma membrane enriched in cholesterol and the protein caveolin. CDK— see cyclin dependent kinase cDNA expression libraries Large collections of cDNAs reverse-transcribed from total cell mRNA. cDNA sequencing analysis (RNA-seq) High-throughput sequencing of cDNA fragments generated from a cell sample, providing detailed information about the expression levels of all mRNA species present in the sample. cell cycle The cyclical series of events encompassing duplication of the genomic DNA, mitosis, and cell division. cell lysate A cell extract usually made by solubilizing cells with detergent-containing buffer solution. cGMP Cyclic nucleotide synthesized from GTP and used as a signaling mediator. chemotaxis The directed movement of a cell either up or down a concentration gradient of a chemical stimulus. cholesterol A component of vertebrate membranes that is highly hydrophobic and has a rigid, polycyclic structure. It can profoundly affect membrane fluidity and lateral diffusion through its interaction with the fatty acid chains of other membrane lipids. chromatin immunoprecipitation (ChIP) A method in which antibodies to specific modified histones or chromatin- associated proteins are used to purify DNA and other proteins associated with that protein. Analysis of associated DNA provides a measure of which genes and sequences are associated with particular chromatin modifications or chromatin-associated proteins. chromodomain Modular protein domain that recognizes peptide motifs containing a methylated lysine. clathrin A structural protein that self-assembles into a hollow, spherical lattice surrounding a membrane vesicle during the process of clathrin-mediated endocytosis. coherent feed-forward— see feed forward, coherent co-immunoprecipitation (co-IP) Isolation of a protein and its binding partners from a cell extract through the use of an antibody specific for one of the proteins, usually bound to beads. co-IP— see co-immunoprecipitation combinatorial complexity For protein modifications, refers to the fact that multiple independent modifications lead to an exponential increase in the number of possible states of the protein. This greatly expands the coding capacity of the genome. conformation Three-dimensional shape of a molecule. cooperativity When binding of one ligand alters the binding affinity of an additional ligand (or additional ligands). Thermodynamically, cooperativity is observed when the free energy of two ligands binding simultaneously differs from the sum of the free energies of the two ligands binding individually. cyclic nucleotides Small signaling mediators synthesized from either ATP or GTP by cyclase enzymes. cyclin The regulatory subunit of a cyclin-dependent kinase (CDK). Cyclins are necessary for activity of the associated catalytic subunit, and also contribute to substrate specificity. Because cyclins are essential for activity of CDKs, modulation of cyclin abundance by transcription and regulated proteolysis is used to regulate CDK activity in the cell cycle. cyclin-dependent kinases (CDKs) Serine/threonine protein kinases that require binding to a regulatory cyclin subunit for activity. CDKs control progression through the cell cycle by phosphorylating specific substrate proteins. cytokine Polypeptide signaling molecule that participates in immune responses. Cytokines often act locally but can act systemically. cytokinesis Division of the cytoplasm of a eukaryotic cell into two daughter cells. cytosol Contents of a cell contained within the plasma membrane, excluding the nucleus of eukaryotic cells. DAG— see diacylglycerol death domain (DD) Modular domain that mediates homotypic interactions, such as between death receptors and downstream effectors. death effector domain Modular domain that mediates homotypic interactions, such as between the adaptor protein FADD and initiator caspases. death receptors A related family of transmembrane receptors that induce apoptosis when bound to their ligands. death-inducing signaling complex (DISC) Supra- molecular complex containing death receptors, adaptor proteins, and initiator caspases that induces apoptosis. desensitization The process by which receptors respond to sustained activation by becoming less responsive to input signal. deubiquitinase (DUB) A specialized type of protease catalyzing cleavage of the isopeptide bond between a lysine amino group and the C-terminus of ubiquitin, thereby removing the ubiquitin group from a protein. diacylglycerol (DAG) A membrane lipid consisting of a glycerol backbone linked to two fatty acid chains. DAG is generated when phospholipase C cleaves the phosphorylated head group from a phospholipid such as PIP₂. digital response— see switchlike response dissociation constant ( K _d ) Quantitative measure of binding affinity, reflecting the concentrations of free components and complex at equilibrium. The lower the dissociation constant, the higher the affinity of the interaction. distributive (multisite phosphorylation) Independent phosphorylation of multiple sites within a protein by the same kinase, where each phosphorylation is the result of a distinct binding event. docking sites In protein kinases, substrate-binding sites distant from the catalytic cleft that participate in determining substrate specificity. domain Compact unit of protein structure that is usually capable of folding stably as an independent entity in solution. E1 ubiquitin activating enzyme The first step of the protein-ubiquitylation reaction. Uses energy of ATP hydrolysis to couple the C-terminus of ubiquitin to a cysteine residue on the E1. E2 ubiquitin conjugating enzyme The second step of the protein-ubiquitylation reaction. Ubiquitin is transferred from the E1 to a cysteine on the E2 enzyme. E2 enzymes generally determine the linkage between ubiquitin units in a polyubiquitin chain. E3 ubiquitin ligase The third step of the protein- ubiquitylation reaction. The E3 binds to specific substrates and to E2 enzymes, facilitating transfer of ubiquitin from the E2 to the substrate. ECM— see extracellular matrix EF hand Conserved protein motif containing acidic residues that can chelate Ca2+. effector caspase A capase that is activated by cleavage by upstream initiator caspases. Effector caspases cleave cell proteins to execute the apoptotic program of cell death. Also known as executioner caspases. eicosanoids A large family of bioactive lipids derived from arachidonic acid that includes the prostaglandins and leukotrienes. Eicosanoids signal through G-protein- coupled receptors to regulate physiological processes such as inflammation. electron microscopy (EM) Use of microscope that uses a beam of electrons to create the image. endocrine Relating to hormones or the glands and tissues that secrete them. endocytosis A process in which a portion of the plasma membrane invaginates and pinches off into the cytosol, generating a free cytosolic vesicle or endosome. endosome A cytosolic vesicle generated by endocytosis. enthalpy A form of energy, equivalent to work, that can be released or absorbed as heat at constant pressure. entropy A measure of the disorder or randomness of a molecule or system. enzymes Proteins (or other biological macromolecules) that greatly increase the rate of a chemical reaction without altering the thermodynamic equilibrium between the reactants and the products. epitope tag Short peptide which, when fused to a protein, allows that protein to be specifically bound by an antibody to that peptide. ESCRT (endosomal sorting complex required for transport) machinery Series of large multiprotein complexes that mediates sorting of endocytosed proteins for recycling or lysosomal degradation. executioner caspase— see effector caspase exportin A transport protein that functions to ferry cargo proteins out of the nucleus. extracellular matrix The meshwork of proteins and carbohydrate that forms between cells in tissues; it can include fibronectin, collagen, vitronectin, and other components. extrinsic apoptotic pathway The apoptotic cell death pathway induced by extracellular ligands that bind to death receptors on the cell surface FAK— see focal adhesion kinase far-Western blotting Method of detecting proteins that can bind directly to a protein of interest. Proteins in a cell lysate are separated by gel electrophoresis and transferred to a membrane, and the membrane is probed with a labeled purified protein of interest. feedback When the output from a signaling node follows a path of links that returns to regulate the original node. feed-forward Distinct paths fanning out from an upstream signaling node that reconverge on another downstream node. feed-forward, coherent A feed-forward loop with two divergent branches with the same overall sign. feed-forward, incoherent A feed-forward loop with two divergent branches with opposite signs. flow cytometry High-throughput analysis that assays the fluorescence and optical properties of many individual cells in a population. fluorescence resonance energy transfer (FRET) Method for detecting the physical proximity of two different fluorescent molecules on the basis of the nonradiative transfer of energy from one fluorescent molecule to the other. focal adhesion kinase (FAK) A nonreceptor tyrosine kinase that is activated upon integrin engagement. FAK plays a critical role in formation and turnover of focal adhesions. focal adhesions Highly complex cellular structures that couple sites of cell–matrix adhesion to intracellular actin cables or stress fibers. fractional occupancy The fraction of the total amount of A that is complexed with B for the reaction: A + B ↔ AB. free energy The energy that can be extracted from a system to drive reactions. free-energy barrier The difference in free energy between the initial state (ground state) of reactants and the high- energy transition state that must be passed through for the reaction to proceed to completion. The higher the free- energy barrier, the slower the reaction will be, even if the overall reaction is highly favored (ΔG is negative). FRET— see fluorescence resonance energy transfer G domain A 20 kD domain that binds guanine nucleotides and can adopt alternative conformations depending on whether GDP or GTP is bound. The small G proteins essentially consist of a single G domain, while the heterotrimeric G proteins contain a G domain in their α subunits. G protein Any of a large class of GTPases that act as molecular switches that are active when bound to GTP and inactive when bound to GDP. They may be heterotrimeric G proteins with α, β, and γ subunits, which typically signal from seven-transmembrane receptors, or small G proteins of the Ras superfamily. G proteins are also known as GTPases, GTP-binding proteins, or guanine-nucleotide- binding proteins. G-protein-coupled receptors (GPCRs) Any of a diverse class of cell-surface receptors with seven membrane- spanning segments that, upon activation, serves as a guanine nucleotide exchange factor to activate heterotrimeric G proteins. GAP— see GTPase-activator protein gated ion channels Ion channels that undergo regulated opening and closing in response to a stimulus such as ligand binding or change in membrane potential. GDF— see GDI displacement factor GDI displacement factor (GDF) A protein that facilitates the dissociation of a GDI from a GTPase, leading to delivery of the GTPase and insertion of its lipid group into the target membrane. GDI— see guanine nucleotide dissociation inhibitor GEF— see guanine nucleotide exchange factor gel electrophoresis Method of separating macromolecules based on their migration through a porous gel under an electric current. In most experimental conditions, smaller molecules migrate more rapidly than larger ones. GFP— see green fluorescent protein glycerophospholipid A phospholipid in which two of the hydroxyl groups of glycerol are linked to fatty acids and the third to a phosphate group. glycosylation The addition of carbohydrate groups (sugars) to proteins. In the case of cell-surface or secreted proteins, complex chains of carbohydrate are added in the endoplasmic reticulum and Golgi apparatus either to the hydroxyl groups of serine or threonine (O-glycosylation) or the amino group of asparagine (N-glycosylation). Single N-acetyl glucosamine groups can also be added to cytosolic proteins. glycosylphosphatidylinositol anchor (GPI anchor) A complex structure consisting of lipids and carbohydrates that is reversibly attached to some proteins to target them to the cell membrane. GPCRs— see G-protein-coupled receptors graded response— see linear response green fluorescent protein (GFP) Fluorescent protein first isolated from a marine jellyfish; it is widely used as a fusion partner to visualize a protein of interest in cells by fluorescence microscopy. ground-state energy The state of lowest energy for a molecule or system. growth factors Inducers of cell growth and cell bulk (the term is also sometimes used more broadly to include mitogens). GTPase— see G protein GTPase-activator protein (GAP) Protein that interacts with G proteins and accelerates their rate of GTP hydrolysis, leading to inactivation of the G protein. guanine nucleotide dissociation inhibitor (GDI) A protein that can bind specific G proteins and shield their lipid groups, thus sequestering the G protein in the cytoplasm. GDIs lock G proteins in their GDP-bound (inactive) state and prevent their localization to the membrane. guanine nucleotide exchange factor (GEF) Protein that interacts with G proteins and catalyzes the exchange of bound GDP for GTP, leading to activation of the G protein. half-life For a binding reaction, the time it takes for half of the complex to dissociate (or the time at which there is a 50% chance that an individual complex will have dissociated). hedgehog (Hh) signaling pathway Signaling pathway in which ligand binding causes a change in the processing of Gli, a protein that acts as a transcriptional activator, controlling developmental processes. heterotrimeric G protein A G protein composed of three different subunits: an α subunit with GTPase activity, and associated β and γ subunits. Exchange of bound GDP for GTP on the α subunit causes dissociation of the heterotrimer into a free α subunit and a βγ heterodimer; hydrolysis of the bound GTP causes reassociation of the subunits. histidine kinases protein kinases, primarily found in prokaryotes, that transfer the terminal phosphate group from ATP to one of their own histidine residues via a phos- phoramidate linkage. In two-component signaling systems, the phosphate is then rapidly transferred to the carboxyl group of an aspartate side chain on a response regulator protein. histone acetyl transferase (HAT) Enzyme that catalyzes the N-acetylation of lysine residues in histones (or other proteins). histone deacetylase (HDAC) Enzyme that catalyzes the removal of N-acetyl groups from lysine residues in histones (or other proteins). histones The major protein component of the nucleosomes, which package genomic DNA into chromatin. Histone modification is a major mode of regulating chromatin structure and thus gene expression at different sites on the genome. homeostasis The ability of living systems to adjust their behavior spontaneously to maintain a stable intracellular environment, despite varying environmental conditions. hormone A soluble signaling molecule that induces physiological effects at a distance by binding to a specific receptor present on target cells. hysteresis Describing a system where the input level at which the system switches between the two output states will be different if one is moving up from low input to high input, or moving down from high input to low input. immunoblotting Method of detecting and quantifying a protein of interest in a sample. Proteins are separated by gel electrophoresis and transferred to a membrane, and the membrane is probed with a specific antibody to the protein of interest (also termed Western blotting). immunofluorescence Technique in which fluorescently labeled antibodies are used to determine the location of their corresponding antigens in fixed cells or tissues. immunoreceptor tyrosine-based activating motif (ITAM) Two tyrosines, separated by approximately nine amino acids, that when phosphorylated serve to recruit and activate kinases of the ZAP-70 family. importin A transport protein that functions to ferry cargo proteins into the nucleus. incoherent feed-forward— see feed-forward, incoherent inflammation Physiological response to infection, allergens, or trauma involving localized swelling, redness, and pain. initiator caspase A caspase that is directly activated by apoptotic signals. They cleave and activate effector caspases. inositol 1,4,5-trisphosphate (IP ₃ ) A soluble second messenger molecule generated by cleavage of PIP₂ by phospholipase C. IP₃ binds to and activates calcium channels on the endoplasmic reticulum, thereby releasing intracellular calcium stores. input stimuli A substance or change in state that evokes a response in a cell. integrins A family of cell-surface adhesion receptors that bind to cell-matrix- or cell-surface-associated proteins, such as fibronectin, laminin, and fibrinogen. interaction domain Part of a protein that mediates interactions with other molecules. intracellular receptors Sensor molecules within cells that bind to signaling molecules and transmit a response. intrinsic apoptotic pathway The apoptotic cell death pathway induced by signals generated within the cell, such as stress responses. Involves permeablization of mitochondrial outer membranes and release of cytochrome c and other components that lead to assembly of the apoptosome. IP ₃ — see inositol 1,4,5-trisphosphate isothermal calorimetry (ITC) Analytical method using a highly sensitive calorimeter that provides thermodynamic parameters of the binding of two molecules. ITAM— see immunoreceptor tyrosine-based activating motif ITC— see isothermal calorimetry JAK–STAT pathway The signaling pathway activated by the cytokine/hematopoietin receptor family, in which receptor engagement leads to activation of JAK family tyrosine kinases and the subsequent phosphorylation and activation of STAT family transcription factors. juxtacrine signaling Signaling that involves direct contact between two adjacent cells or a cell and the extracellular matrix. karyopherin A class of transport proteins that functions to transport cargo proteins into or out of the nucleus; cargo binding is regulated by the Ran G protein. k _cat — see catalytic rate constant K d— see dissociation constant kinetochore Structure formed by proteins of the mitotic chromosome that attaches to the microtubules of the mitotic spindle. K m The Michaelis constant, which is the substrate concentration at which the reaction rate is one-half of Vmax. For many enzymes, Km is similar or equal to Kd, the affinity of the enzyme for the substrate. ligand Small molecule or macromolecule that recognizes and binds to a macromolecule. ligand-gated channel A membrane channel whose opening is controlled by the binding of specific ligands. linear response (graded response) A system where the response is proportional to the input signal. link The regulatory relationships between individual components (nodes) in a signaling network. Links are positive when the action of the upstream node on the downstream node results in activation, and negative when the result is repression/inhibition. lipid bilayer This consists of two layers of polar lipids, arranged in a sheet with the hydrophobic chains oriented inward and the polar head groups on the surface of the sheet facing the aqueous environment. lipid raft Local lipid domains, enriched in sphingomyelin and cholesterol, that are highly ordered and yet allow a high degree of lateral mobility. local concentration The effective concentration of a component at a specific site, such as in the immediate vicinity of a second component. logic gates Device that specifies output depending on the combination of two inputs. lysine acetyl transferase (KAT)— see histone acetyl transferase (HAT) lysine deacetylase (KDAC)— see histone deacetylase (HDAC) lysine demethylase (KDM) Enzyme that removes methyl groups from methylated lysine residues within a protein. lysine methyl transferase (KMT) Enzyme that transfers a methyl group from S-adenosyl methionine (SAM) to the terminal amino group of lysine residues within a protein. lysophospholipid A glycerophospholipid from which one of the two fatty acid chains has been cleaved. lysosome A membrane-enclosed intracellular compartment where the protein and lipid components of endocytosed vesicles are broken down enzymatically by proteases and lipases. MAP kinase (mitogen activated protein kinase) An important family of serine/threonine kinases that are activated by upstream signals and which phosphorylate targets such as transcription factors. MAP kinase is the third in a three-kinase series (the MAP kinase cascade). MAP kinase (mitogen-activated protein kinase) cascade A pathway module, consisting of three kinases that act in series, utilized in a remarkably wide variety of cellular responses. The essential components are a MAP kinase kinase kinase (MAPKKK), which phosphorylates and activates a MAP kinase kinase (MAPKK), which in turn phosphorylates and activates a MAP kinase (MAPK). MAP kinases often phosphorylate nuclear targets such as transcription factors. The three kinases of the cascade are often co-localized in a single multiprotein complex by scaffold proteins. mass spectrometry (MS) Analytical method that separates molecules such as peptides on the basis of their mass-to-charge ratio, providing extremely accurate information on their atomic masses. mechanistic target of rapamycin (mTOR) A protein serine/threonine kinase that acts as a master regulator of cell growth, survival, and metabolism. membrane channels Protein pores in a lipid membrane that allow passage of hydrophilic molecules such as ions through the membrane. N -methylation The transfer of methyl groups to amino groups of proteins. O -methylation The transfer of methyl groups to oxygen atoms of protein side chains such as that of glutamate. Important for regulating prokaryotic systems such as bacterial chemotaxis. Michaelis–Menten constant— see K _m Michaelis–Menten equation An equation describing the velocity of an enzymatic reaction as a function of enzyme and substrate concentration. microarrays High throughput analytic method in which many protein or nucleic acid samples are arrayed as tiny spots on a solid support, which can then be probed with labeled binding partners or cell lysates. mitogen Extracellular molecule that stimulates cell proliferation. mitogen-activated protein kinase— see MAP kinase mitosis Division of the nucleus of eukaryotic cells, such that each resulting nucleus contains one copy of each chromosome. modular domain Domain found in many different proteins in the same organism that confers a specific function or activity. molecular memory The conversion of a transient input into a permanent (or semipermanent) change in output. motif Conserved peptide sequence that is recognized specifically by interaction domains. MS— see mass spectrometry mTOR— see mechanistic target of rapamycin multivesicular body (MVB) Organelle to which receptor– ligand complexes are transported for sorting to either lysosomes or the cell surface. N -myristoylation Irreversible attachment of a myristoyl group to the N-terminal glycine of a protein via an amide linkage. negative feedback When the output from a signaling node follows a path of links that results in negative regulation of the original node. NES— see nuclear export signal network Linked system of multiple signaling molecules that regulate one another. network architecture The topology of a signaling network, defined by the specific links between molecular nodes and the sign of these links (positive or negative). NF-κB Transcription factor that is present in a latent form in the cytosol of unstimulated cells and which is translocated to the nucleus upon activation. nitric oxide (NO) a small diatomic gas that acts as a signaling mediator. It passively diffuses into responding cells and directly activates guanylyl cyclase. Nitric oxide plays an important role in regulating smooth muscle relaxation and blood vessel dilation. NLS— see nuclear localization signal node Individual signaling component in a signaling pathway or network. Notch A family of cell-surface receptors that are proteolytically cleaved upon ligand binding and that often participate in cell-fate determination during development. nuclear export signal (NES) A short modular peptide motif that mediates nuclear export by binding to an exportin. nuclear localization signal (NLS) A short modular peptide motif that mediates nuclear import by binding to an importin. nuclear magnetic resonance (NMR) spectroscopy Method for determining protein structure and conformation; based on the resonant absorption of electromagnetic radiation at a specific frequency by atomic nuclei in a magnetic field, due to flipping of the orientation of their magnetic dipole moments. nuclear pore complex A very large multiprotein complex that regulates the passage of macromolecules through the nuclear pore, and thus into and out of the nucleus. nuclear receptor (NR) superfamily Intracellular receptors for hydrophobic signaling molecules such as steroid hormones. The receptor–ligand complex acts as a transcription factor. nucleosome The basic unit of chromatin, consisting of eight histone subunits arranged in a disclike structure, around which is wrapped ~147 base pairs of DNA. A typical nucleosome contains two molecules each of histone 2A (H2A), histone 2B (H2B), histone 3 (H3), and histone 4 (H4). oncogene A gene that when mutated or disregulated can lead to the uncontrolled cell growth characteristic of cancer. Oncogenes can be activated by mutation (point mutation, deletion, or truncation) or by an increase in expression level. Oncogenes act dominantly (they exert their effect even when the normal version of the oncogene is present in the cell). optogenetic (methods) These take advantage of light- sensitive proteins and domains, originally found in plants or protists, that when illuminated with light of a specific wavelength, will undergo conformational changes that result in functional changes. The aim is to achieve local control over specific signaling inputs using light. orphan receptors Receptors for which the relevant physiological ligand(s) has not yet been identified. oscillation Fluctuation of output levels between states of high and low activity in a periodic manner. output responses The responses of a cell to signaling input. p53 A master regulator of cellular responses to a wide range of environmental stresses such as DNA damage. Depending on the specific stress and cellular context, p53 can induce momentary cell-cycle arrest, or permanent cell- cycle arrest and apoptosis. Prevents cells from replicating inappropriately and passing on damaged genomic DNA. Most frequently mutated gene in human cancers. S -palmitoylation Transfer of the 16-carbon fatty acid palmitic acid to cysteine residues of a target protein. Unlike most other lipid modifications, S-palmitoylation is relatively dynamic. paracrine Local signaling to nearby or adjacent cells. pathway A linear chain of interactions where the output of each node serves as the input for the next downstream node. PCA— see protein-fragment complementation assay PH domain Modular protein domain, many of which bind to specific phosphoinositol-derived lipids. phorbol esters Organic compounds that mimic the structure of diacylglycerol (DAG) and thereby promote the activation of PKC in vivo. phosphatidic acid (PA) A glycerophospholipid in which the head group consists only of phosphate. Can be generated either from other glycerophospholipids by the action of phospholipase D, or from diacylglycerol by the action of diacylglycerol kinase. phosphatidylinositol (PI) A membrane phospholipid with the six-membered sugar inositol as its head group. phosphatidylinositol 3-phosphate [PI(3)P] A membrane phospholipid in which the inositol head group is phosphorylated on position 3. phosphatidylinositol 3,4,5-trisphosphate (PIP ₃ ) A membrane phospholipid in which the inositol head group is phosphorylated on positions 3, 4, and 5. phosphatidylinositol 4,5-bisphosphate (PIP ₂ ) A membrane phospholipid in which the inositol head group is phosphorylated on positions 4 and 5. phosphatidylinositol 3-kinase (PI3K) A signaling enzyme that adds a phosphate to the 3 position of PIP2 to generate the membrane-bound second messenger PIP₃. phosphoinositides Phospholipids where the head group is the six-member sugar inositol. Phosphates can be added and removed from specific positions on the ring via the action of specific lipid kinases and phosphatases. phospholipase A ₂ (PLA ₂ ) An enzyme that cleaves at the sn-2 (middle) position of the glycerol backbone of a glycerophospholipid, generating a free fatty acid and a lysophospholipid. phospholipase C (PLC) An enzyme that cleaves the phosphorylated head group from a phospholipid, generating diacylglycerol and the phosphorylated head group. phospholipase D (PLD) An enzyme that cleaves the head group from a phospholipid, generating phosphatidic acid and the unphosphorylated head group. phosphorylation The transfer of the terminal phosphate group from ATP to proteins or other molecules. photoreceptor Cell or molecule that is sensitive to light. PI3K— see phosphatidylinositol 3-kinase PIP ₂ — see phosphatidylinositol 4,5-bisphosphate PIP ₃ — see phosphatidylinositol 3,4,5-trisphosphate PKA— see protein kinase A PKC— see protein kinase C podosome Actin-rich cell-surface protrusion that mediates adhesion and invasion in normal cells (a similar structure found in tumor cells is termed an invadopodium). polarity Functional or structural asymmetry in a cell. polyproline type II helix (PPII helix) A left-handed helical structure with three amino acids per turn that forms spontaneously in proline-rich peptides. positive feedback When the output from a signaling node follows a path of links that results in positive regulation of the original node. postsynaptic density (PSD) A specialized cellular substructure on the postsynaptic side of a neuronal junction, containing neurotransmitter receptors and other signaling proteins. post-translational modifications Covalent modifications of proteins that are added and removed by specific enzymes after synthesis of the protein. prenylation Irreversible attachment of either a farnesyl or geranylgeranyl lipid group to a protein via a thioether linkage. primary cilium A specialized filamentous organelle constructed from microtubules. primary structure The simple linear sequence of amino acids in the polypeptide chain. In many cells it acts as a signaling center. priming (multisite phosphorylation) Phosphorylation of a substrate by one kinase makes it a better substrate for phosphorylation by a second, distinct kinase. processive (multisite phosphorylation) Phosphorylation of a protein on mutiple sites by an enzyme that remains associated with its substrate. proline hydroxylation Hydroxylation of the 4 position of the proline ring, generating 4-OH-proline. Proline hydroxylation of the transcription factor HIF-1α is an important component of the oxygen-sensing mechanism in metazoans. prolyl cis-trans isomerization A switch in the conformation of a proline residue caused by rotation around the peptide bond. Spontaneously, this reaction proceeds very slowly, but it can be speeded up greatly through the action of peptidyl prolyl cis-trans isomerase (PPIase). proteases The enzymes that cleave the peptide bonds of proteins. proteasome A large multiprotein structure that mediates the proteolysis of cytosolic proteins. It consists of a hollow cylinder, lined on the inside with proteases, capped at each end by a “lid” structure that controls access to the proteolytic machinery of the inner chamber. protein arginine methyl transferase (PRMT) Enzyme that transfers a methyl group from S-adenosyl methionine (SAM) to arginine residues within a protein. protein kinase Enzyme that covalently modifies proteins by the addition of a phosphate group. protein kinase A (PKA) A protein serine/threonine kinase that is activated by cyclic AMP. protein kinase C (PKC) A family of related serine/ threonine protein kinases, whose activation is variously dependent on calcium and diacylglycerol (DAG). protein phosphatase Enzyme that removes phosphate groups from proteins. protein trafficking The targeted transport of proteins from one subcellular location to another within the cell, generally by membrane vesicles. protein-fragment complementation assay (PCA) Technique for detecting protein interactions in living cells, based on the reconstitution of a functional reporter molecule (such as green fluorescent protein) when two proteins bind to each other. proteolysis The cleavage of the peptide bonds that link individual amino acid residues in a protein. PTP (protein tyrosine phosphatase or phospho- tyrosine phosphatase)— see tyrosine phosphatase pull-down assay A protein or protein fragment is produced in quantity and bound to beads, then used to detect binding partners in cell extracts. quaternary structure The arrangement of multiple subunits within a protein complex. Ras A small G protein that functions as a central regulator of cell proliferation and differentiation. Ras was first discovered as a viral oncogene, and it is frequently mutated and activated in human cancers. receptor A protein that, when bound to a specific signaling molecule (ligand), undergoes a change in activity that transmits a signal. receptor tyrosine kinases (RTKs) Single-pass trans- membrane receptors with intracellular protein tyrosine kinase domains. regulated intramembrane proteolysis (RIP) Sequential proteolytic cleavage of a membrane protein, first by ADAM- mediated ectodomain cleavage, followed by further processing within the membrane by the γ-secretase complex. regulator of G protein signaling protein (RGS protein) Protein that binds to the free GTP-bound α subunit of a heterotrimeric G protein and stimulates its GTPase activity. response regulator (RR) The second (effector) component of two-component systems in bacteria and simple eukaryotes. After activation of a histidine kinase by incoming signals, phosphate is transferred to aspartate side chains of the response regulator, inducing conformational changes. Many response regulators are DNA-binding proteins that regulate transcription. RGS protein— see regulator of G protein signaling protein RNA-seq— see cDNA sequencing analysis robustness The ability of a biological system or network to maintain its performance under a wide range of conditions and to be relatively insensitive to perturbations of the system. scaffold proteins Proteins that bind multiple proteins, such as enzymes and their substrates, involved in a single process. Scatchard analysis Method of quantifying the binding of a labeled, soluble analyte to an immobilized binding partner. Binding data can be plotted in the form of a straight line, the slope of which reveals the K _d of the interaction. This graph is called a Scatchard plot. SCF complex A multisubunit ubiquitin ligase complex consisting of an E3 ubiquitin ligase, a Skp1 adaptor, a cullin, and an F-box specificity factor. Among other roles, the SCF complex targets phosphorylated substrates for degradation during cell cycle transitions. second messengers— see small signaling mediators secondary structure Local, regular structural elements in proteins, primarily α helices and β strands. serine/threonine kinase A protein kinase that phosphorylates serine or threonine residues on its substrates. serine/threonine phosphatases A protein phosphatase that dephosphorylates phosphoserine or phosphothreonine residues on its substrates. SH2 domain Modular protein domain that binds to tyrosine-phosphorylated peptides. SH3 domain Modular protein domain that binds to proline-rich peptides that adopt a specific helical secondary structure. signal amplification A property of signaling mechanisms where a single activated enzyme molecule can generate many molecules of product, thereby amplifying the output of the system. signal integration The integration of multiple signaling inputs to produce a single output. sister chromatids A pair of identical chromosomes generated by DNA replication during S phase. One of the pair is segregated to each daughter cell during mitosis. small G protein The large and diverse family of G proteins that consist entirely of a G domain, in the absence of other domains or subunits. Small G proteins are distinct from the other major class of G proteins, the heterotrimeric G proteins. small signaling mediators Small, highly diffusible molecules that carry signaling information within cells. soluble guanylyl cyclase (sGC) An enzyme that converts GTP into cGMP and pyrophosphate; the primary intracellular effector of nitric oxide signaling. specificity The degree of selectivity for one partner or class of partners relative to a set of competing interactions. sphingomyelin An abundant sphingolipid consisting of ceramide linked to a phosphorylcholine head group. A major component of the outer leaflet of the plasma membrane, it can be metabolized to generate a variety of bioactive lipids. specificity constant A measure of the efficiency of an enzyme for a particular substrate, defined as kcat/Km. SPR— see surface plasmon resonance Src family kinases A group of nonreceptor tyrosine kinases that regulate processes such as adhesion and lymphocyte activation. standard free energy (ΔG°) The change in free energy associated with formation of a compound, under standard conditions of concentration, temperature, and pressure STAT A family of transcription factors that are rapidly activated by cytokine and growth factor receptors. Phosphorylation by JAK family kinases leads to dimerization, nuclear localization, and DNA-binding activity. state machines Devices that can exist in multiple discrete states, and which can change their state in response to specific instructional inputs. stoichiometry the relative amounts of each of the individual components of a molecule or complex, for example the number of each type of subunit present in a multimeric complex. substrate specificity The selectivity of an enzyme to react with certain substrates over others. The degree of specificity of an enzyme for two substrates can be described quantitatively by the relative values of kcat/Km for two different substrates. surface plasmon resonance (SPR) Analytical method for determining quantitative binding parameters. Uses an instrument that can monitor over time the extent and kinetics of binding and dissociation of a macromolecule to a second macromolecule fixed to a surface. switch I and II regions Regions of a G protein that alter their conformation upon binding to the γ-phosphate of GTP. switchlike response A nonlinear or all-or-none response in which the system does not respond to a stimulus until it reaches a threshold value, at which point the system responds with maximum output. synthetic biology The use of natural biological components to build and engineer novel functional systems. tertiary structure Three-dimensional folded structure of a polymer chain such as a protein or RNA. Toll-like receptors (TLRs) A family of receptors that bind pathogen-specific ligands and indirectly activate the NF-κB family of transcription factors. transcription factor A protein or protein complex that binds to a gene near its promoter and regulates transcription of that gene. transition state The species of highest free energy either in a reaction or a step of a reaction. transmembrane receptor A transmembrane protein that binds to extracellular ligands and transmits information into the cell by ligand-induced changes in conformation and/or enzymatic activity. transphosphorylation The phosphorylation of one kinase molecule by another in a dimer or multiprotein complex. tumor suppressor Gene product that helps prevent the formation of cancer by antagonizing cell proliferation and survival pathways. two-component system A common signaling system in prokaryotes, also found in some simple eukaryotes such as yeast. Two-component systems are composed of a histidine kinase linked to a receptor and a response regulator. Activation of the histidine kinase by the receptor leads to its autophosphorylation on histidine. Phosphate is then transferred to aspartate on the response regulator, leading to conformational changes that transmit the signal. tyrosine kinase A protein kinase that phosphorylates tyrosine residues on its substrates. tyrosine phosphatase A protein phosphatase that dephosphorylates phosphotyrosine residues on its substrates. ubiquitin A 76-residue protein that is enzymatically added via its C-terminus to lysine side chains in a target protein, generating an isopeptide bond linking the two. Additional ubiquitin units can be added to lysine side chains or the N-terminus of each ubiquitin, generating long chains (polyubiquitylation). ubiquitin-binding domain (UBD) One of several structurally distinct small binding domains and motifs that bind specifically to ubiquitin, usually to a hydrophobic patch of ubiquitin centered around Ile44. ultrasensitivity When a relatively small change in input leads to a much larger than proportional change in output. voltage-gated ion channel Ion channel across a membrane that only allows the passage of ions in response to a change in voltage across the membrane. Western blotting— see immunoblotting Wnt signaling pathway A conserved pathway regulating key developmental events. Activation of canonical Wnt pathway results in transcription mediated by β-catenin. Transduction mechanism involves protein phosphorylation and regulated protein degradation. x-ray crystallography Technique for determining the three-dimensional arrangement of atoms in a molecule based on the diffraction pattern of x-rays passing through a crystal of the molecule. Y2H— see yeast two-hybrid assay yeast two-hybrid (Y2H) assay Molecular genetic technique for finding proteins that interact with a protein or protein fragment of interest. zero-order ultrasensitivity When an increase in the amount or activity of the forward modifying enzyme leads to an ultrasensitive increase in the amount of the modified target protein. Occurs when both forward and reverse enzymes are fully saturated with substrate. zymogen An inactive precursor form of an enzyme (usually a protease), which must be processed by proteolytic cleavage in order to be activated. Taylor & Francis Taylor & Francis Group https://taylorandfrancis.com Index Note: abbreviations following page numbers are: B, box; F, figure; and T, table. Prefixes are ignored in the alphabetical sequence – thus β-Catenin will be found under the letter C. 14-3-3 proteins 247T control of nuclear transport 120 dimerization 255, 259 prevention of dephosphorylation 100 recognition of phosphorylated sites 99, 254–255, 254F regulation of kinase activity 267, 267F regulation of subcellular localization 255, 255F 19S regulatory particle 225F, 226 26S proteasome 225F, 226 53BP1 329 A Abl association with membrane 125 oncogenic fusion proteins 269–270, 270F Accessory domains/subunits regulating kinase substrate specificity 57, 57F, 58 regulating phosphatase activity 64–65 Acetylation 86F, 87 chemical effects 89, 89F histone see under Histone(s) p53 96 protein domains recognizing 255– 256, 256F N-Acetylation 86F, 87 N-Acetyl glucosamine (GlcNAc) 87, 88F Actin cytoskeleton integrin-mediated adhesion 192– 193, 192F regulation by phospholipase D 168 studying changes in 358F Actin-related protein 2/3 (Arp2/3) complex 284, 284F Activation loop, protein kinase 53F, 54 phosphorylation 54–55, 55F, 56 transphosphorylation 181–182, 182F Active state 14–15, 15F S-Acylation 123T, 124, 124F ADAM-10 221, 222F, 224 ADAM-17 221, 224 ADAM metalloproteases 220, 221–222 domain structure 221F Eph–ephrin signaling 221, 222F Notch signaling 197 regulated intramembrane proteolysis 224, 224F ADAMTS proteins 221F Adaptability 2–3 Adaptation 5, 296–299, 296F, 298F visual system 312–313 Adaptive immune response 333–344 Adaptor proteins 264, 264F Adenosine triphosphate see ATP Adenylyl cyclase 136, 140–141, 140F, 141F β-Adrenergic receptor kinase (β-ARK; GRK2) 211–212 β-Adrenergic receptors 186, 213F Affinity 25 classification 32–33 cooperative binding and 35–37, 37F effect of local concentration 31, 31F effect of multiple binding sites 30–31, 31F factors determining ideal 31–32 functional constraints 32–33, 33F independent modulation 34–35, 34F range 32–33 see also Dissociation constant Affinity tags 355 Agonists 180 A-kinase anchoring proteins (AKAPs) 143, 150–152, 151F Akt kinases activation by membrane recruitment 126–127, 127F mTOR pathway 169 PH domain function 261 All-or-none responses see Switchlike responses Allosteric conformational changes 47–49, 349 as consequence of binding 22–23 diversity of types 48–49, 49F lipid-modified proteins 125 as mechanism of signal transmission 44–45 regulation 48, 48F Allosteric regulation 48, 48F Allosteric switch proteins 266–268, 266F engineered 270–272, 271F signaling functions 279–281, 280F α helices 350, 350F armadillo repeats 245, 246F effects of phosphorylation 50–51, 51F gated ion channels 199–200, 199F structure 351, 351F transmembrane receptors 180, 180F Alzheimer’s disease 197, 224 Amino acid sequence 350, 350F Amphipathic molecules 156, 157 Amplification, signal see Signal amplification Amyloid β 224 Amyloid precursor protein (APP) 197, 224, 224F Analytical methods 345–354 Anaphase-promoting complex (APC) 227, 228–230, 328 regulation by phosphorylation 228–230, 229F structure 227F, 228 Xenopus cell cycle 300F, 301 AND gates, biological 282, 282F coherent feed-forward loops 292– 293, 293F cyclin-dependent kinases 284 modular signaling proteins 284– 285, 284F transcriptional promoters 285–286, 285F AND NOT gate, biological 283F Androgens 94, 94F Angiogenesis 205 Ankyrin (ANK) repeats 247T NF-κB proteins 230, 230F Antagonists 180 ANTH/CALM domain 247T Antibodies avidity of binding 30–31, 31F recognizing post-translational modifications 364–365, 364F specificity of phosphotyrosine binding 25, 25F Antigen 333 antibody binding 31, 31F see also Peptide–MHC complexes Antigen-presenting cells 333, 335, 339 AP-1 285F, 286, 337 Apaf1 240, 241F APC see Anaphase-promoting complex Apc2 227F APC protein 194, 194F Apoptosis 95, 232–240 extrinsic pathway 235–237, 236F intrinsic pathway 238–241, 239F irreversibility 218 morphological features 232–233, 233F phosphatidylserine (PS) signaling 159 receptor signaling via proteolysis 197–199, 198F signals initiating 233, 234 Apoptosis inducing factor (AIF) 240 Apoptosome 234, 238, 240–241, 241F AQUA (absolute quantification) 367–368 Arachidonic acid (AA) 162, 172 metabolites 172–174, 173F Arf proteins 68, 68T Arginine finger 74, 74F, 75 Arginine methylation 109 Armadillo repeats 245, 246F ARM repeat 247T Arp2/3 complex 284, 284F Arrestins GPCR desensitization 211, 211F, 212, 213F hedgehog signaling 196F, 197 rhodopsin desensitization 313 ASC 237F, 238 Aspartate phosphorylation 102–103, 103F Assemblies, protein see Complexes, protein Association constant (K _a) 27 ATM/ATR kinases 329 ATP analogs 361F, 362 driving phosphorylation/ dephosphorylation 46–47, 47F Atrial natriuretic peptide (ANP) receptor 189, 190F Aurora-B kinase 282, 283F Autocrine signaling 340 Automatic door analogy 276, 277–278, 277F Avidin–biotin interaction 30 Avidity 30–31, 31F Axin 194, 194F, 195 B B7 335, 341 Bacteria chemotaxis see Chemotaxis, bacterial two-component systems 102–103, 102F BAD 238, 238F, 255 BAK 238, 238F, 239 BAR domains 247T, 262, 262F BAX 238, 238F, 239, 240F Bcl2 238, 238F, 240 Bcl2 family proteins 238, 238F activation 239–240, 240F anti-apoptotic 238–239, 239F, 240 induction of apoptosis 239–240, 239F, 240F pro-apoptotic 238–239, 239F Bcl homology domains see BH1–BH4 domains Bcl-X_L 238, 238F, 255 BCR–Abl fusion protein 269–270, 270F BEACH domain 247T Bem1 79F, 80 β sheets 351, 351F β strands 350, 350F WD repeats 245, 246F Beta blockers 180 BH1–BH4 domains 238–239, 238F, 247T BH3-only proteins 238, 238F activation 239–240 induction of apoptosis 239, 239F, 240F BID 236, 238, 238F, 240 BIM 238, 238F, 239–240 Binding 21, 22–30 affinity see Affinity cooperative see Cooperativity direct and indirect consequences 22–23 effects of post-translational modifications 89–90, 90F interaction surfaces 23–24, 23F K _d see Dissociation constant parameters, determination 345– 347, 346F, 347F specificity see under Specificity thermodynamics 27–28, 28F Binding isotherm 26, 26F, 288 Binding probability 26 Binding rate constant see On-rate Binding sites, total number of 346, 346F Biochemical analysis 345–354 Bioinformatics, identifying protein domains 244–245, 245F Biophysical analysis 345–354 Biosensors 369–370 calcium-binding fluorescent dyes 147, 370 fluorescent protein fusions 359, 369–370 FRET-based 354, 370, 371F BIR domains 235, 247T Bistable systems 299, 301 flow cytometric analysis 371–372, 372F Blood coagulation see Coagulation, blood BRCA1 106 BRCT domain 247T Bromodomains 247T, 250F, 255, 256F BTB/POZ domain 247T Buried surface area 23, 23F C C1 domain 247T C2 domain 247T, 254 CAAX box 124 Caenorhabditis elegans 7 programmed cell death 232 vulval development 8–9, 8F Calcineurin (PP2B) 60, 62F T cell signaling 285F, 286, 337 Calcium (Ca2+) 139T biosensors 147, 370 clotting cascade 219, 220F influx into cells 147 protein kinase C activation 145, 145F signaling 136, 145–149, 146F downstream effectors 136, 147–148, 148F negative feedback control 295, 295F speed 137, 137F T-cell activation 285F, 286 visual transduction cascade 310, 313 waves 148–149, 149F Calcium (Ca2+)-binding proteins 147 Calcium (Ca2+)/calmodulin 148, 148F control of cAMP degradation 141 nitric oxide release 140 see also Calmodulin Calcium/calmodulin-dependent protein kinase (CAMK) 59F, 148 II isoform (CaMKII) 265F Calcium (Ca2+) channels 146–147, 146F Calcium (Ca2+-ATPase) pumps 146F, 147 Calmodulin (CaM) 146F, 147–148, 148F fused Ca2+ biosensors 147 NMR spectroscopy 353F see also Calcium (Ca2+)/calmodulin cAMP see Cyclic AMP Cancer biology 6F, 7–9 modular domain rearrangements 269–270, 270F Carbon monoxide (CO) 205 CARD domains 238, 240, 247T Cascades, signaling enzyme 75–80 signal amplification 294, 294F ultrasensitive switchlike responses 290, 291F Casein kinase 1 (CK1) 59F hedgehog signaling 195, 196F Wnt signaling 194, 194F, 195 Caspase-activated DNase (CAD) 235 Caspases 232–240 activation by death receptors 197– 199, 198F domain structure 233, 234F effector (executioner) 233, 234– 235, 234F extrinsic apoptotic pathway 235– 237, 236F inflammasome activation 237–238, 237F inflammatory 234 initiator 198, 233–234, 234F intrinsic apoptotic pathway 240, 241F regulation of activation 233–235, 234F substrate specificity 235 Catalysis, enzymatic 44–47 driving reactions in one direction 46–47, 47F quantitative analysis 347–349, 348F thermodynamic mechanisms 45–47, 45F Catalytic domains 243 allosteric switch proteins 266–268, 266F GAPs 71–72, 72F GEFs 71–72, 72F, 73–74, 73F protein kinases 52–53, 52F protein phosphatases 60, 64, 65F structure 244 Catalytic rate constant (k _cat) 46, 348–349, 348F β-Catenin armadillo repeat domain 245, 246F Wnt signaling 194–195, 194F Caulobacter crescentus 103 Caveolin-mediated endocytosis 128, 128F inhibiting TGFβ signaling 129– 130, 130F Cbl 105, 263F oncogenic forms 211 phosphotyrosine-binding domains 252, 252F receptor down-regulation 209, 210F T-cell receptor degradation 341 CD3 338 CD4 co-receptor 190, 335, 336 CD8 co-receptor 335 CD28 co-stimulatory receptor 335, 341 CD45 336 Cdc4, WD40 repeats 246F Cdc20 complex with APC (APCCdc20) 228– 230, 229F, 328 robust oscillator 300F, 301 spindle assembly checkpoint 330 Cdc24 79F, 80 Cdc25 63 cell cycle regulation 283F, 300, 301F, 328 targeting for proteolysis 228, 329 Cdc25 homology domain 72–73 Cdc42 284, 284F, 285 CDCP1 254 Cdh1–APC complex (APCCdh1) 227F, 228, 229F, 230 Cdk1 328 Cdk7 112 CDKs see Cyclin-dependent kinases cDNA expression libraries 356 cDNA microarrays 363–364, 363F cDNA sequencing analysis (RNA- seq) 363F, 364 Cell(s) adhesion to extracellular matrix 192–193, 192F environmental cues received by 178–179, 178F extracts, isolation of proteins 355– 356, 355F polarity 115 as state machines 276, 277F see also Living cells Cell–cell adhesion 189 Cell–cell signaling complexes 264– 265, 265F Cell cycle 323–331 checkpoints 329–330 irreversible commitment 218, 324 molecular network 325 oscillations 299–301, 300F phases 324 phase transitions 324, 325–328 ubiquitylating complexes controlling 226–230, 328 Cell death necrotic 233, 233F programmed see Apoptosis Cell division 323 Cell lysate 355 Cell motility, small G proteins controling 68, 68F Cell signaling 1–19 biological functions 6–11 challenges 4–5, 5T diversity of inputs and outputs 3–4, 3F hierarchical organization 17–18, 17F, 278 information processing perspective 2–3, 2F molecular basis 11–15 spatial and temporal scales 9–11 Ceramidase (CDase) 170, 171F, 172 Ceramide 139T, 144 downstream targets 172 metabolism 170, 171F, 172 structure 157 Ceramide 1-phosphate (C1P) 170, 171F cGMP see Cyclic GMP Channels, membrane 179 gated see Gated ion channels Chaperones, affinity and specificity of binding 33, 33F CH domain 247T CheA/CheB/CheR/CheY 298F, 299 Chemical dimerizers 362, 362F Chemotaxis, bacterial 102 negative feedback adaptation 297–299 receptors 182 Chk2 329 Cholesterol 157 chemical structure 158F membrane fluidity and 159, 159F Choline 168 Chromatin 107–108 nuclear receptor binding 206–207 remodeling 108–109, 110–112, 111F structure 108, 108F Chromatin immunoprecipitation (ChIP) 365, 365F Chromodomains 247T, 250F, 255–256, 256F Chromo-shadow domain 247T Chromosomal translocations 269–270, 270F Chronic myelogenous leukemia (CML) 269–270 Cilium, primary 195, 196F CK1 see Casein kinase 1 Claritin® (loratadine) 180 Clathrin-mediated endocytosis 128, 128F GPCR desensitization 212 mediating TGFβ signaling 129–130, 130F receptor down-regulation 209 Cluster analysis 363F Coagulation, blood 219–220, 223 clotting cascade 220F physiological role 219, 219F Cofilin 223F Coherent feed-forward loops 288B, 292–293, 293F phospholipase D 165B, 170 Coiled coil (CC) domains 247T Co-immunoprecipitation (co-IP) 355, 355F Combinatorial complexity 93 Complexes, protein 22 cooperative binding 35, 35F dynamic molecular assemblies 38 electron microscopy 353–354 isolation from cell extracts 355– 356, 355F variability of stability and homogeneity 37–38 see also Dimerization; Oligomerization; Protein–protein interactions Concentrations, protein 14 binding affinities and specificities and 32–33, 33F local see Local concentration Cones (photoreceptor) 307 Conformation, protein 13, 22, 349–351 methods of determining/ analyzing 352–354 see also Protein structure Conformational changes 13, 44–45, 47–49 allosteric see Allosteric conformational changes binding-induced 22–23, 23F diversity of mechanisms 48–49, 49F ligand–transmembrane receptor binding-induced 180–181, 181F, 182 light-induced 310–311 methods of determining/ analyzing 352–354 multiple protein substates 47–48, 47F phosphorylation-induced 50–52, 51F post-translational modification- induced 89, 90F regulating nuclear import of STATs 120–121, 121F regulation of stability 48, 48F Cool temperatures, receptors responding to 202 Cooperativity 35–37 cAMP binding to protein kinase A 142 coupled SH2 domains 252, 253F enzyme activation 348F, 349 functional consequences 36–37 homotypic/heterotypic 37, 37F inducing switchlike behavior 37, 289–290, 289F membrane binding 124, 125, 125F molecular mechanisms 36, 36F negative 35 positive 35–37, 35F Cortactin 223F CRKL adaptor 358F Cryo-electron microscopy (cryo-EM) 354 Csk 244F, 336 C-terminal peptide motifs, recognition by PDZ domains 258–259, 259F CTLA-4 341 CUE domain 247T Cul1 227, 227F Cullins 227, 227F, 228 Cyan fluorescent protein (CFP) 354, 357–358, 359, 359F Cyclic AMP (cAMP) 136, 139T, 140–143 AKAP-mediated regulation of signaling 150–152, 151F diffusion rate 137, 137F downstream targets 136, 141–142, 141F regulation of protein kinase A 142–143, 143F synthesis and degradation 140– 141, 140F, 141F Cyclic AMP-dependent protein kinase see Protein kinase A Cyclic AMP (cAMP) phosphodiesterase 140–141, 140F AKAP interactions 150–152, 151F Cyclic GMP (cGMP) 139T, 140–142 downstream targets 141F, 142 nitric oxide signaling 204, 205F synthesis and degradation 140– 141, 140F, 141F visual transduction cascade 309, 310–311 Cyclic GMP (cGMP)-dependent protein kinases (cGK) 142 Cyclic GMP (cGMP) phosphodiesterase 140–141, 140F, 141F, 142 Cyclic nucleotide binding domains (CNB) 141, 142, 143F Cyclic-nucleotide-gated (CNG) channel 309, 310 Cyclic nucleotides 139T, 140–143, 140F Cyclin-dependent kinase (CDK) inhibitors 325 proteolytic destruction 228, 327 Cyclin-dependent kinases (CDK) 324–328 allosteric signal integration 283– 284, 283F phylogenetic relationships 59F regulation of catalytic activity 53F, 54, 325 substrate specificity 57, 58 targeting proteins for destruction 94, 94F, 226–227, 228F Xenopus cell cycle 300F, 301 Cyclins 324–328 allosteric activation 53F, 54 docking sites 58 regulation of CDK activity 325 ubiquitin-mediated regulation 226, 228–230, 229F Cyclooxygenase (COX) 173, 173F Cys-loop superfamily of ion channels 199F, 202–203 Cytochrome c apoptosome assembly 240, 241F induction of release 238, 239F Cytokine receptors activation of coupled tyrosine kinases 190, 191F nuclear import of activated STATs 120–121 Cytokines 178, 237–238 Cytosol 4 Cytotoxic (killer) T cells 333, 334 D Death domain (DD) 193, 198, 235, 247T Death effector domain (DED) 198, 198F, 235, 247T Death-inducing signaling complex (DISC) 198, 198F, 234 activation of apoptosis 235 Death receptors activation of apoptosis 235–237, 236F signaling via proteolysis 197–199, 198F Delta, regulated intramembrane proteolysis 224, 224F Delta-Serrate-Lag2 (DSL) 198F Dendritic cells 333, 335 DEP domain 247T Dephosphorylation membrane lipids 163 protein see Protein dephosphorylation Desensitization, receptor 208–209, 208F Deubiquitinases (DUBs) 105, 226 Developmental biology 6F, 7 DIABLO/Smac 235, 240 Diacylglycerol (DAG) 139T, 143, 144–145 biophysical properties 161 biosynthesis 144, 144F, 162F, 163 phosphorylation 163 regulation of protein kinase C activation 144–145, 145F T cell activation 337 Diacylglycerol (DAG) kinases 163 Diffusion membrane fluidity and 159 within membranes 160–161, 160F small signaling mediators 137– 138, 137F Digital responses see Switchlike responses Dimerization chemical induction 362, 362F protein interaction domains 259– 260, 259F, 260F transmembrane receptors 181– 184, 181F Disheveled (DVL) 194F, 195 Disorder/order transitions 49, 49F Dissociation constant (Kd) 26–30 binding energy and 27–28 calculation methods 346, 347, 347F effect of multiple binding sites 30–31, 31F factors determining ideal 32 functional constraints 32–33, 33F local concentration effects 31 physiological examples 27T rates of binding and dissociation and 28–30, 29F see also Affinity Dissociation rate constant see Off-rate Distributive protein phosphorylation 100–101, 101F, 290 DNA damage 106–107, 329 degradation during apoptosis 232 Doc1 227F Docking sites protein kinases 57, 57F, 58 protein phosphatases 60 Domains, protein 24, 243–273 bioinformatics for identifying 244– 245, 245F catalytic see Catalytic domains combinations (multidomain proteins) 262–272 allosteric switch proteins 266– 268, 266F combinatorial diversity 262–263, 263F as scaffold proteins 263–266 containing smaller repeats 245, 246F interaction see Interaction domains, protein recognition functions 246–249 recombinations 269–272 engineered 270–272 leading to cancer 269–270, 270F structure 244, 245F Dose–response curves 288–289, 289F Double-strand breaks 106–107 DR4 236, 236F DR5 236, 236F Drk 9 Drosophila melanogaster 7–8, 195 DSL (Delta-Serrate-Lag2) 198F Dual-specificity phosphatases (DSPs) 50, 61F, 63, 64 Dynamic molecular assemblies 38 E E1 ubiquitin activating enzymes 104, 104F E2 ubiquitin conjugating enzymes 104, 104F, 105, 106F E3 ubiquitin ligases 104, 104F, 105–106, 106F cell cycle regulation 227, 227F see also Cbl; RING E3 ligases Ectodomain shedding 221 EDG receptors 172 EF-hand domains 247T Ca2+-binding proteins 147, 148 SH2 domain interactions 252, 252F EH domain 247T Eicosanoids 139T, 143, 162, 172–174 Electron microscopy (EM) 353–354 Endocrine glands/tissues 178 Endocrinology 6F, 7 Endocytosis 90, 127–128 mechanisms 128, 128F Notch signaling 197, 198F phospholipid–protein interactions 261–262, 262F receptor down-regulation 209–211, 210F, 212 receptor signaling after 128–130, 130F role of ubiquitylation 106, 128 signaling cascades 79–80, 79F EndoG 240 Endoplasmic reticulum, Ca2+ storage 146F, 147 Endosomes 128 FYVE domain proteins 261–262 signaling 130 Engineered signaling proteins 270– 272, 271F Enthalpy of binding 28, 347, 347F ENTH domain 247T Entropy of binding 28, 347, 347F Environment ability to respond to changes in 2–3 types of signals from 178–179, 178F Enzymes (signaling) 12, 13, 43–81 allosteric changes 22–23, 44–45, 47–49, 349 allosteric regulation 48, 48F as allosteric switches 266, 266F binding to substrates 22 cascades 75–80 cooperative activation 348F, 349 cooperative switches 289–290, 289F dose–response curves 288–290, 289F, 290F hallmark 44, 44F principles of catalysis 44–47 quantifying catalytic power 347– 349, 348F signal amplification 45, 294, 294F Epac 73, 141–142, 141F Ephrin (Eph) receptors 187F, 221, 222F Epidermal growth factor receptor (EGFR) ADAM-mediated cleavage of ligands 221, 222 domain structure 187F internalization 129, 209, 210F signaling 8F, 16, 187 Epinephrine 4 Epitope tags 355 ErbB4/HER4 224, 224F Erbin 259F Erk MAP kinase 77 Fos interactions 293, 293F methods for studying 360F, 364F PDGF signaling 317, 321 subcellular localization 121–122, 121F T cell signaling 285F, 286, 340, 343 see also Raf–MEK–Erk kinase pathway ESCRT complexes receptor down-regulation 209, 210F ubiquitin-binding domains 106 Euchromatin 108 EVH1 domains 246–248, 247T, 249F Evolution, recombination of protein domains 262–263 Exportins 118–119, 119F Extracellular matrix (ECM) 178 integrin-mediated adhesion 192– 193, 192F regulated proteolysis 222–223, 223F Eye development, Drosophila 7–8, 8F vertebrate 308 F Factor V 219 Factor X 219, 220F Factor XIII 219, 220F FADD 198, 198F, 235, 236–237, 236F Fan-in network architecture 287B, 287F Fan-out network architecture 287B, 287F Farnesyl groups 124 Far-Western blotting 357, 357F Fas 198–199, 198F activation of apoptosis 235–236, 236F Fas ligand (FasL) 235, 236F Fatty acids phospholipids 156–157, 157F enzymatic cleavage 162, 162F saturated 159F unsaturated 156, 159, 159F F-box domain 247T F-box proteins 227–228, 227F, 228F FCH domain 247T FCP phosphatases 61F, 62 Feedback 4, 5, 287B cell cycle phase transitions 325–328 see also Negative feedback loops; Positive feedback loops Feed-forward 288B see also Coherent feed-forward loops; Incoherent feed-forward loops FERM domain 247T FF domain 247T FH2 domain 247T FHA domain 247T Fibrin 219, 219F, 220F Fibrinogen 219, 220F Fibroblast growth factor (FGF) receptor 187, 187F Fibroblast growth factor receptor substrate 2 (FRS2) 187 Fibroblasts 315 detection of wounds 318 molecular control of proliferation 317, 319–320 responses to wounding 316 termination of proliferative response 321 FLIP 236 Flow cytometry 369T, 371–372, 372F Fluorescence resonance energy transfer (FRET) biosensors based on 354, 370, 371F detecting conformational changes 354, 354F detecting protein interactions 359, 359F, 360F visualizing Ca2+ changes 147 Fluorescent dyes, calcium- binding 147, 370 Fluorescent protein tags 357–359, 358F, 369–370 Focal adhesion kinase (FAK) 191F, 192 Focal adhesions 193, 358F Fos 293, 293F FOXO 255, 255F Fractional occupancy 26, 346 Free energy 27–28, 28F barrier, effect of enzymes 45, 45F conformational substates 47F, 48, 48F FRET see Fluorescence resonance energy transfer Frizzled 194F, 195 Fura-2 147, 370 Fus3 78F, 79 FYVE domains 247T, 261–262, 262F G G₁/S transition 324, 326–327 G₂/M transition 324, 328 GADS 258, 258F, 336 Gα subunit 68–69 activity cycle 69, 69F control of activation 75 families and their effectors 69, 70T GPCR signaling 184, 185F GAPs see GTPase-activator proteins GAT domain 247T Gated ion channels 179, 179F, 199–204 ligand-induced conformational changes 181 structure 199–200, 199F Gβ subunit 68–69 Gcn5 111, 250F GDI displacement factors (GDFs) 75, 125, 126F GDIs (guanine nucleotide dissociation inhibitors) 75, 125, 126F G domain 67, 67F, 68 GDP see Guanosine diphosphate GEFs see Guanine nucleotide exchange factors GEL domain 247T Gel electrophoresis 355, 355F Gene expression changes in 9–11, 9F methods of analyzing 363–364, 363F Genetic deletion mutations (knockouts) 361 Geranylgeranyl groups, modification by 124, 126F Gγ subunit 68–69 GK domain 247T GlcNAc (N-acetyl glucosamine) 87, 88F GLIC 203F Gli transcription factors 195, 196F, 197 Glucocorticoid receptor (GR) 206–207, 207F GLUE domain 247T Glutamate 308 Glycerophospholipids 156–157, 157F Glycogen synthase kinase 3 (GSK-3) 100 hedgehog signaling 195, 196F Wnt signaling 194, 194F, 195 Glycosylation 87, 88F N-Glycosylation 87 O-Glycosylation 87 Glycosylphosphatidylinositol (GPI) anchor 123, 123T, 124F GoLoco motifs (or domains) 75 GPCRs see G-protein-coupled receptors G-protein-coupled receptor kinases (GRKs) 211–212, 212F rhodopsin desensitization 313 G-protein-coupled receptors (GPCRs) 69, 184–186 desensitization 211–212, 211F, 212F diversity 71, 71F, 184 as drug targets 180 GEF activity 69F, 71, 184 ligand-induced conformational changes 180–181 protease-activated receptors 223– 224, 224F recycling (resensitization) 212 signaling 184–186, 185F specificity 150, 151F, 185 speed, magnitude and duration 186 G proteins 8, 43, 65–69 biosensors 370, 371F classification 67 as conformational switches 65, 65F downstream signaling 69 heterotrimeric 67, 68–69 activation of cAMP 140–141, 141F activity cycle 69, 69F control of activation 71, 75 GPCR signaling 184, 185F limited numbers 69, 184–185 phosphatidylinositol 3-kinase activation 167 phospholipase C activation 165 protein kinase C activation 145F structure 67, 67F molecular basis of conformational change 66–67, 66F regulation of activity 65–66, 65F, 70–75 signaling cascades regulating 79–80, 79F small see Small G proteins structure 67F switch I/II regions 66–67, 66F Graded (linear) responses 288, 289F, 290F flow cytometric analysis 371, 372F GRAM domain 247T Granzyme B 235 Grb2 domain structure 263F flexibility of adaptor function 264, 264F homologs 8F, 9 PDGF signaling 317, 319 receptor down-regulation 209, 210F SH2 domain 250F, 251, 251F Sos recruitment 73, 127F Green fluorescent protein (GFP) 357–358 biosensors 167F, 370 negative feedback loops 295–296, 295F, 302F protein-fragment complementation assay 360, 360F Grip1 265F GRIP domain 247T GRK2 211–212 GRKs see G-protein-coupled receptor kinases Ground-state energies 45 Growth factors 178 Growth hormone, human (hGH) 23F, 32, 33F Growth hormone receptor, human 23F, 32, 33F GRP1 249F GTP see Guanosine triphosphate GTPase-activator proteins (GAPs) 65F, 66, 70–73 domains, small G proteins 68T, 72, 72F GPCR signaling 186 heterotrimeric G proteins 75 mechanisms of action 74, 74F mechanisms of regulation 72–73, 72F small G proteins 71–73, 71F, 72F Guanine nucleotide dissociation inhibitors (GDIs) 75, 125, 126F Guanine nucleotide exchange factors (GEFs) 65F, 66, 70–74 control of membrane association 125, 126F different organisms 8F, 9 domains 68T, 72, 72F heterotrimeric G proteins 69F, 71, 184 mechanism of action 73–74, 73F mechanisms of regulation 72–73, 72F rhodopsin function 310 small G proteins 71–75, 71F, 72F Guanosine diphosphate (GDP) 65–66, 65F Guanosine triphosphate (GTP) 65–66, 65F γ-phosphate group 66–67, 66F Guanylyl cyclase 140F, 141 membrane (mGCs) 189, 190F soluble (sGC) 204–205, 205F visual transduction cascade 309, 311, 313 GYF domain 247T H Half-life, biological complex 29, 30 Hck 56F Heat, receptors responding to 202 HEAT repeat 247T HECT domain 247T HECT domain E3 ligases 105 Hedgehog (Hh) signaling 195–197, 196F Helper T cells 333, 334 Hemoglobin, conformational changes 49 Heptahelical receptors 184 Heterochromatin 108 Hierarchical organization, information-processing systems 17–18, 17F, 277–278, 278F HIF-1α 205, 206, 206F High-throughput sequencing 363– 364, 363F Histamine H1 receptor antagonists 180 Histidine kinase 58, 102–103, 102F Histidine phosphorylation bacteria 101–103, 103F eukaryotes 103–104, 104F Histone(s) 107–112 acetylation 87 protein domains recognizing 256F regulation of chromatin structure 109, 111, 111F androgen-mediated changes 94, 94F antagonistic modifications 282, 283F methylation 87, 109–110, 110F protein domains recognizing 109–110, 255–256, 256F regulation of chromatin structure 110, 111F, 112 nucleosome structure 108, 108F post-translational modification 108–110 Histone acetyl transferase (HAT) 87, 109 chromatin modification 111, 111F nuclear receptor interactions 207, 207F Histone deacetylase (HDAC) 87, 109 chromatin modification 111F, 112 nuclear receptor interactions 207, 207F Homeostasis 4 Homology domains 244–245 Hormones 4, 7, 178 HP1 antagonism by Aurora-B kinase 282, 283F chromodomain 250F, 255–256, 256F Hrs FYVE domain 261–262 UIM domain 106, 107F Hsp90 206, 207F Hydrophobic motif pocket (PIF pocket) 57 Hyperpolarization, light-induced 308 Hypoxia 205–206, 206F Hysteresis 301, 302F I IAP antagonists 240 IAPs (inhibitor of apoptosis proteins) 235 IκB NF-κB regulation 230–232, 231F Toll-like receptor signaling 193, 194 IKK (IκB kinase) complex 107 NF-κB regulation 230–232, 231F Toll-like receptor signaling 193– 194, 193F tumor necrosis factor receptor signaling 236, 236F Imatinib 269–270 Immobilized metal affinity chromatography (IMAC) 367 Immune response, adaptive 333–344 Immune synapse 190, 339 Immune tyrosine inhibitory motifs (ITIMs) 341 Immunoblotting 355 Immunofluorescence 357 Immunology 6F, 7 Immunoreceptor tyrosine-based activating motifs (ITAMs) phosphorylation 338 SH2 domain interactions 37, 252, 253F T cell activation 336 ZAP-70 recognition 190 Importins 118–119, 119F Inactive state 14–15, 15F Incoherent feed-forward loops 288B adaptive responses 296F, 297 Infections, adaptive immune response 333, 334 Inflammasome 234, 237–238, 237F Inflammation 172 Inflammatory mediators 172–174, 173F Information processing 2–3, 2F, 275–303 inputs and outputs 3–4, 3F integrating multiple inputs 281– 286, 281F modifying strength or duration of output 294–301 molecular currencies 11–15, 12T, 13F responding to strength or duration of input 286–293 Information-processing devices/ systems 276–281 hierarchical organization 17–18, 17F, 277–278, 278F input detection 278–279, 279F proteins as 278–279, 279F, 280F state machines 276, 277F Inhibitor titration 291F, 292 Inositol trisphosphate (IP₃) 139T, 143, 144–145 biophysical properties 161 biosynthesis 144, 144F, 163 control of protein kinase C activation 144–145, 145F T cell activation 337 Inositol trisphosphate (IP3)-gated Ca2+ channels (IP₃ receptors) activation by Ca2+ 146F, 147 propagation of Ca2+ waves 148– 149, 149F Inputs, signal 3–4, 3F, 275–276 detection 278–279, 279F generating changes in state 11, 11F integrating multiple 281–286, 281F responding to strength or duration of 286–293 Inside-out signaling 192–193 Insulin 4 Insulin-like growth factor-1 (IGF-1) receptor 187F Insulin receptors 7 activation 182 domain structure 187F signaling 99, 187 Insulin receptor substrate 1 see IRS1 Insulin receptor tyrosine kinase (IRK) 54–55, 55F Integration, signal 5, 281–286, 281F Integrins 190–193, 191F, 192F Interaction domains, protein 243, 246–268, 247–248T combination 262–268 allosteric switch proteins 266– 268, 266F diversity 262–263, 263F as scaffold proteins 263–266 detection of modified binding sites 365–366 dimer and oligomer formation 259–260, 259F, 260F recognition functions 246–262 phospholipids 260–262 post-translational modifications 249–257, 250F unmodified peptide motifs or proteins 257–260 structure 244, 244F Interleukin-2 (IL-2) autocrine and paracrine signaling 340 expression by T cells 285–286, 285F, 337 Internal state, monitoring 4 Intracellular domain (ICD) 224, 224F Intracellular receptors 179 Invadopodia 222, 223F Ion channels, gated see Gated ion channels IQ domain 247T IRAKs 193, 193F Irreversible systems 301, 302F IRS1 PTB domain 249F, 253–254, 254F scaffold function 99, 187 ISG15 104 Isotherm, binding 26, 26F, 288 Isothermal calorimetry (ITC) 347, 347F ITAMs see Immunoreceptor tyrosine- based activating motifs ITIMs (immune tyrosine inhibitory motifs) 341 iTRAQ 367 J JAKs (Janus kinases) 120–121, 121F, 190 JAK–STAT pathway 120, 190 JNK MAP kinase pathway 77, 77F, 78F Juxtacrine signaling 221 K Ka (association constant) 27 Kalirin 265F Karyopherins 118–119, 119F KCa3.1 potassium (K+) channel 103– 104, 104F kcat (catalytic rate constant) 46, 348–349, 348F kcat/Km (specificity constant) 349 KcsA potassium channel 200F Kd see Dissociation constant KHD (kinase homology domain) 189, 190F Killer T cells 333, 334 Kinetochores, detection of unattached 330 Km (Michaelis–Menten constant) 46, 348–349, 348F Knockout/knockdown 361, 361F koff see Off-rate kon see On-rate KSR 79 Kv channels see Voltage-gated potassium channels L Lac repressor 302F LAT 266, 336–337 Lck phosphorylation of adjacent receptors 339 Shp1 phosphorylation 342–343 T cell receptor interactions 190, 191F, 336, 338 Lectins 25 Leukemia 269–270 Leukotrienes 172, 173, 173F Ligand-gated calcium (Ca2+) channels 146 Ligand-gated ion channels 202–204 desensitization 209 open and closed conformations 203–204, 203F overall topology 199F, 203 Ligands 21, 177 allosteric regulation 48, 48F effect of local concentration on binding 31, 31F polymeric, avidity 30–31, 31F receptor internalization and 129 Light adaptation 312–313 mechanism of detection 310–311 signal amplification 311 signal transduction cascade 309 visual processing 307–314 Light-gated domains engineered 271, 271F regulating plant protein kinases 267, 267F LIM domain 247T Linear responses 288, 289F, 290F Links, network 287B, 287F Lipid(s) 155–174 membrane 156–159 bilayer formation 156F, 157–158 chemical structure 156–157, 157F, 158F effects on physical properties 158–159, 159F enzymes modifying 161–163, 162F signaling molecules from see Lipid-derived signaling mediators modification of proteins 87–88 mediating membrane association 123–124, 123T, 124F reversible membrane association and 125, 126F Lipid bilayer 156, 156F, 157–158 organization states 159, 159F Lipid-binding domains 24 mediating membrane associations 124–125 phosphoinositide signaling 163, 164F, 166 Lipid-derived signaling mediators 143–145, 161–174 biosynthesis 161–163, 162F classes 139T, 143–144 difficulties in studying 161 diversity of biophysical properties 161 major signaling pathways 164–174 phosphorylation/ dephosphorylation 163, 164F Lipid kinases 163 Lipid-modifying enzymes 161–163 complexity of regulation 165B Lipid phosphatases 163 Lipid rafts 159, 159F Liposomes 156F 5-Lipoxygenase (5-LO) 173, 173F Living cells biosensors 369–370, 371F localization and tracking proteins 357–359, 358F time-lapse microscopy 368–369, 369F visualizing protein–protein interactions 359–360, 360F Local concentration 14, 115–116 effect of membrane localization on 116–117, 117F effect on binding 31, 31F Localization, subcellular see Subcellular localization Logic gates 281–282, 282F Loratadine (Claritin•) 180 LOV domains 267, 267F Rac recombination 271, 271F LRP 194F, 195 LRR repeat 248T LUBAC 107 Lysine acetyl transferase (KAT) 87 Lysine deacetylase (KDAC) 87 Lysine demethylases (KDMs) 87, 109 Lysine methylation 109 Lysine methyl transferases (KMTs) 87, 109 Lysophosphatidic acid (LPA) 161, 163 Lysophosphatidylcholine 173, 173F Lysophospholipid 162–163 Lysosomes 106, 129, 225 M Macrophage-colony-stimulating factor (M-CSF) receptor 187F Macropinocytosis 128, 128F Mad1 330 Mad2 330 Mae 260, 260F Major histocompatibility complex (MHC)–peptide complexes see Peptide–MHC complexes Mammalian target of rapamycin see mTOR MAP kinase (MAPK) 76, 76F control of nuclear localization 121– 122, 121F see also Erk MAP kinase MAP kinase cascades 16, 76–79, 76F chimeric engineered scaffolds 270, 271F in different eukaryotes 8–9, 8F, 77, 78F in mammalian cells 77, 77F scaffold proteins 22, 77–79, 78F switchlike activation 290 T cell activation 337, 340 MAP kinase kinase (MAPKK) 76, 76F MAP kinase kinase kinase (MAPKKK) 76, 76F MAP kinase phosphatases (MKPs) 321 Mass spectrometry (MS) 356, 366– 368, 366F Matrix metalloproteases (MMPs) 220, 221F, 222–223 MBT repeat 248T Mdm2 95F, 96 Mechanistic target of rapamycin see mTOR Medicine 6–7, 6F MEK 77, 77F control of Erk2 localization 121, 121F visualizing Erk association 360F see also Raf–MEK–Erk kinase pathway Membrane(s) 155–161 curvature binding by BAR domains 262, 262F effects of phosphatidic acid 168– 169, 168F fluidity 159, 159F functional domains 116, 116F, 159, 160–161 lipid bilayer 156, 156F, 157–158 lipid components 156–157, 157F, 158F localization of proteins to 116, 122–127 coupling with activation 126– 127, 127F effect on local concentration 116– 117, 117F mechanisms 122–125, 123F reversibility 125, 126F molecular interactions within 160– 161, 160F physical properties and lipid composition 158–159, 159F trafficking, modulating signaling 127–131 see also Plasma membrane Membrane channels 179 see also Gated ion channels Membrane guanylyl cyclases (mGCs) 189, 190F Membrane lipids see Lipid(s), membrane Membrane proteins 122–123, 123F Membrane-type matrix metalloproteases (MT-MMPs) 221F Memory, molecular 301, 302F Mena 249F Metalloenzymes 60–62, 62F Metalloproteases, extracellular 220– 224, 221F Metaphase-to-anaphase transition 228, 324, 328 Metastasis, tumor 222 Methods, study 345–373 Methylation, protein 86F, 87 histone see under Histone(s) protein domains recognizing 109– 110, 255–256, 256F N-Methylation 86F, 87, 109–110, 110F O-Methylation 87 MH1/MH2 domains 248T MHC–peptide complexes see Peptide– MHC complexes Micelles 156, 157 Michaelis–Menten analysis 347–349, 348F Michaelis–Menten constant (K _m) 46, 348–349, 348F Michaelis–Menten equation 347–348 Microarrays 357, 363–364, 363F Microfluidic devices 368, 369F Mitochondria, intrinsic apoptotic pathway 238–241 Mitochondrial outer membrane permeabilization (MOMP) 238, 239, 239F, 240–241 Mitogen-activated protein see MAP Mitogens 178 Mitosis 4, 324 regulation 328 ubiquitin-mediated regulation 228–230, 229F MIU domain 248T, 257F Modular protein domains 243, 244–249 see also Domains, protein Modular proteins, signal integration 284–285, 284F Molecular currencies, cell signaling 11–15 Molecular memory 301, 302F Monoubiquitylation 104 Motifs, peptide 246 integrating multiple post- translational modifications 282– 283, 283F modified see Post-translational modifications unmodified, recognition by protein domains 257–260 mRNA, microarray analysis 363–364, 363F mTORC1 169, 169F, 170 mTORC2 169, 169F, 170 mTOR pathway, regulation by phospholipase D 165B, 169–170, 169F Multidomain proteins see under Domains, protein Multiple reaction monitoring (MRM) 368 Multivesicular body 209, 210F Mutual inhibition, lipid-modifying enzymes 164–165, 165B Myc 317 MyD88 193, 193F MYPT1 62F N-Myristoylation 123–124, 123T, 124F control of membrane association and 125 Myt1 284 N NADPH oxidase, allosteric regulation 268, 268F NALP1 inflammasome 237F, 238 Nck 284F, 285, 337 NDPK-B 103–104, 104F Necrosis, cell 233, 233F Nedd8 104 Negative discrimination, modulating affinity and specificity 35, 35F Negative feedback loops 287B adaptive responses 296F, 297 bacterial chemotaxis 297–299, 298F controling output level and precision 294–296, 295F double bistable systems 302F cell-cycle transitions 327–328 oscillators 299, 300F, 301 PDGF signaling 321 T cell signaling 341–343 visual system 312–313 NEMO NF-κB regulation 231, 231F recognition of polyubiquitin linkages 107, 107F Nerve growth factor (NGF) receptor 187F Networks 17 architectures 286, 286T, 287–288B artificial activation methods 362– 363, 362F detecting sustained input 292–293, 293F methods for studying 360–372 perturbation methods 361–362, 361F switchlike activation 290–292, 291F see also Feedback; Feed-forward Neuregulin-1 224 Neurobiology 6F, 7 Neurons Ca2+ signaling 147 mechanisms of signaling 137 retrograde signaling 130, 131F Neurotransmission 202–204 Neurotrophins, retrograde signaling 130, 131F Neutrophils, NADPH oxidase 268, 268F NFAT 285F, 286, 337 NF-κB 107, 230–232 canonical pathway of activation 230–231, 231F family 230, 230F noncanonical pathway of activation 231–232, 231F regulation of apoptosis 236–237, 236F T cell activation 337 Toll-like receptor signaling 193– 194, 193F Nicotinic acetylcholine receptor (nAChR) 203 Nitric oxide (NO) 139T activation of guanylyl cyclase 141, 141F signaling 204–205, 205F speed of signaling 137 transfer across plasma membrane 179 Nodes, signaling 11, 11F, 287B, 287F Nonlinear responses see Switchlike responses Noonan syndrome 267 NOR gate, biological 284 Notch nuclear localization 90, 122, 122F proteolytic processing 197, 198F, 224 signaling pathway 122, 197, 198F Notch extracellular domain (NECD) 197 Notch intracellular domain (NICD) 122, 122F, 197, 198F, 224 Noxa 238, 238F, 239 Nuclear export signals (NES) 118, 119F Nuclear factor κB see NF-κB Nuclear localization signals (NLS) 118, 119F Nuclear magnetic resonance (NMR) spectroscopy 353, 353F Nuclear pore complex 117–118 Nuclear receptor (NR) superfamily 206–208, 207F Nucleosome 108, 108F Nucleus localization of molecules to 116, 117–122 transport to/from, control of 118–122 Nutrients, as signals 3 NZF domain 248T O Off-rate (koff) 26, 29–30, 29F measurement 346–347, 347F Okadaic acid 62, 361 Olfaction 185 Oligomerization protein interaction domains 259– 260, 259F, 260F transmembrane receptors 181– 184, 181F Oncogenes 7, 8F, 67–68, 269–270 On-rate (kon) 25, 29–30, 29F measurement 346–347, 347F Opsin 310 Optogenetic systems 362–363, 362F OR gate, biological 284F, 285 Orphan receptors 206 Oscillation 299–301 Outputs, signal 3F, 276 changes in state generating 11, 11F modifying strength or duration 5, 294–301 Oxygen (O2) signaling 205–206, 206F P p21 228, 329 p27 228, 283F, 284 p38 MAP kinase pathway 77, 77F p53 95–97 post-translational modifications 95, 95F regulation of activity 95F, 96–97 p65 230, 230F p100/p52 activation pathway 231–232, 231F domain structure 230, 230F p105/p50 activation pathway 231–232, 231F domain structure 230, 230F p300/CBP complex 95F, 96, 111 Pak1 349–351, 350F S-Palmitoylation 88, 124 Ras isoforms 130–131 Paracrine signaling 204, 340 PAS domain 248T Patched (Ptc) 195–197, 196F Pathways, signaling 15–16 artificial activation 362–363, 362F see also Networks PB1 domain 248T Pbs2 78, 78F PCAF/Gcn5 111, 250F PDGF see Platelet-derived growth factor PDK1 kinase Akt activation 126–127, 127F PH domain 261 PDZ domains 248T recognition of C-terminal motifs 258–259, 259F scaffold proteins with 264–265, 265F Peptide–MHC complexes 333, 335, 336 binding properties 33, 33F discrimination of non-antigenic 342–343 Peptide–protein interactions 24, 24F Peroxisome proliferator activated receptors (PPARs) 174 Pharmacological small-molecule inhibitors 361, 361F PH domains 24, 248T biosensors based on 370 GPCR desensitization 211, 212F identification by database analysis 245, 245F mediating membrane association 124, 127, 127F phosphoinositide recognition 260– 261, 261F recognition functions 246–248, 249F Philadelphia chromosome 269, 270F Pho4 119–120, 120F Pho80/85 119–120, 120F Phorbol esters 145 Phosphatidic acid (PA) 168–170 biosynthesis 144, 162F, 163, 168 effect on membrane curvature 168–169, 168F regulation of mTOR pathway 169– 170, 169F Phosphatidylcholine (PC) 156, 158F enzymatic cleavage 162F, 168 position in membrane 159 Phosphatidylethanolamine (PE) 156, 158F, 159 Phosphatidylinositol (PI) 143, 156 chemical structure 158F, 164F position in membrane 159 Phosphatidylinositol 3,4,5-trisphosphate (PI(3,4,5)P₃) 164 Akt regulation 126–127, 127F biosynthesis 143, 164, 167 chemical structure 164F dephosphorylation 167 Phosphatidylinositol 3-kinase (PI3K) 126, 127F isoforms 167 p85 adaptor subunit, SH2 domains 251 PDGF signaling 319 phospholipase C interactions 164– 165, 165B Phosphatidylinositol 3-phosphate (PI(3)P) binding by FYVE domains 261– 262, 262F biosynthesis 166F, 167 subcellular localization 167, 167F TGFβ receptor internalization 129, 130F Phosphatidylinositol 4,5-bisphosphate (PI(4,5)P₂) 125 enzymatic cleavage 143, 144, 144F, 164 as membrane-localized binding site 164 N-WASP regulation 284–285, 284F phospholipase Cδ interaction 261, 261F phosphorylation/ dephosphorylation 164, 167 subcellular localization 167, 167F Phosphatidylinositol 4-phosphate (PI(4)P) 166F, 167, 167F Phosphatidylinositol 4-phosphate 5-kinases (PIP5K) 167 Phosphatidylinositol 5-kinase (PI5K) 165B Phosphatidylinositol 5-phosphatases 167 Phosphatidylinositol-specific phospholipase C (PI-PLC) 164–165, 165B Phosphatidylserine (PS) 156, 158F, 159 Phosphodiesterase 6 (PDE6) GTPase activating protein (GAP) activity 313 signal amplification 311 speed of response 312 visual transduction cascade 309, 310–311 Phosphodiesterases (PDEs) 140F, 141 AKAP interactions 150–152, 151F see also Cyclic AMP phosphodiesterase; Cyclic GMP phosphodiesterase Phosphoinositides 164–167 biosensors 167F, 370 as membrane-localized binding sites 117, 164, 166–167 phosphorylation/ dephosphorylation 163, 164, 166 protein domains recognizing 260–262 recognition by PH domains 260– 261, 261F as source of signaling mediators 139T, 143, 144F, 164–166, 166F subcellular distribution 166–167, 167F Phospholipase(s) 161–163, 162F Phospholipase A₂ (PLA₂) 162–163, 162F, 172–174 cytosolic α isoform (cPLA2α) 172– 173, 173F Phospholipase C (PLC) 162F, 163 β isoform (PLC-β) 144, 165 δ isoform (PLCδ), PH domain 261, 261F γ isoform (PLC-γ) 144, 165–166 activation by ZAP-70 266 domain structure 263, 263F SH2 domain 251, 251F T cell activation 336–337 hydrolysis of PIP₂ 143, 144F, 164 phosphatidylinositol-specific (PI-PLC) 164–165, 165B protein kinase C activation 145, 145F regulation of activity 164–166, 165B Phospholipase D (PLD) 163, 168–170 isoforms 168 mTOR pathway regulation 165B, 169–170 phosphatidylcholine cleavage 162F, 168 phosphatidylinositol 5-kinase interactions 165B regulation of membrane curvature 168–169, 168F responses to environmental signals 170 Phospholipids bilayer formation 156F chemical structure 156–157, 158F enzymatic modification 162–163, 162F position in membrane 159 protein domains recognizing 260– 262, 262F Phosphorylation membrane lipids 163 protein see Protein phosphorylation Phosphoserine/phosphothreonine 14-3-3 protein interactions 254– 255, 254F protein domains recognizing 254 Phosphotyrosine motifs, scaffold proteins with multiple 265–266, 266F other domains recognizing 252– 254, 254F recognition by SH2 domains 249–252 specificity of antibodies binding 25, 25F Photoreceptor cells 307–314 adaptation 312–313 light-induced response 308 mechanism of light detection 310–311 signal amplification 311 signal transduction cascade 309 speed of response 312 Phox proteins 268, 268F PICK1 265F PIDDosome 234 PIF pocket 57 PINS protein 75 Plants activation of kinases by light 267, 267F MAP kinase cascades 77, 78F serine/threonine kinases 188 Plasma membrane 115 information transfer across 177–214 see also Transmembrane signaling localization to 116–117 effect on local concentration 116– 117, 117F Plasmin 220 Platelet-activating factor (PAF) 173, 173F Platelet-derived growth factor (PDGF) 315–322 actin cytoskeletal rearrangements 358F secretion in response to injury 316 signaling 317–321 Platelet-derived growth factor receptors (PDGFR) 317 domain structure 187F down-regulation 321 ligand-induced dimerization 181 mechanism of activation 318 responses to autophosphorylation 319 tyrosine kinase activity 181 Platelets 223, 315, 316 Pleckstrin homology domains see PH domains Podosomes 222 Polarity, cell 115 Polo-box domain 248T Polyacrylamide-gel electrophoresis (PAGE) 366F Polyhomeotic 259 Polyproline type II (PPII) helix 257– 258, 258F Polyubiquitylation 104–105, 105F Positive discrimination, modulating affinity and specificity 34–35, 34F Positive feedback loops 287B bistable systems 301, 302F cell-cycle transitions 326–328 leading to switchlike activation 291F, 292 lipid-modifying enzymes 165B, 168 robust oscillators 299, 300F, 301 T cell activation 339–340, 342–343 Postsynaptic density 264–265, 265F Post-translational modifications (PTMs) 13, 85–113 chemical effects 88–90, 89F, 90F control by writer/eraser/reader systems 90–92 interplay between 93–94, 93F, 94F methods of detection 364–368, 364F, 365F, 366F multiplicity of sites and types 92–93, 92F p53 95–97, 95F peptide motifs integrating multiple 282–283, 283F protein interaction domains recognizing 249–257, 250F regulating membrane associations 123–125 reversibility 217–218 role in subcellular localization 90, 90F, 117, 123–125 simple functional groups 86–87, 86F speed of signaling 92 sugars, lipids and other proteins 87–88, 88F switching between distinct states 93–94, 93F transmitting spatial information 92 Potassium (K⁺) channels KCa3.1 103–104, 104F KcsA 200F Kv (voltage-gated) see Voltage- gated potassium channels PP1 phosphatase 60, 62F PP2B phosphatase see Calcineurin PPII (polyproline type II) helix 257– 258, 258F PPM phosphatases 60, 61F, 62 PPP phosphatases 60, 61F, 62 Prenylation 123T, 124, 124F membrane association and 125 Primary cilium 195, 196F Primary structure 350, 350F Priming, of phosphorylation 100, 101F Processive protein phosphorylation 100–101, 101F, 290 Programmed cell death see Apoptosis Prokaryotes protein kinase-like genes 58 see also Bacteria Proline cis-trans isomerization 88, 88F Proline hydroxylase domain (PHD) proteins 205, 206F Proline hydroxylation 86F, 87 Proline-rich peptide motifs, recognition 257–258, 258F Promoters, transcriptional, signal integration 285–286, 285F Propeller structure 245, 246F Prostacyclin 173, 173F Prostaglandins 172, 173, 173F Protease-activated receptors (PARs) 223–224, 224F Proteases 88, 218–224 classification 218 receptors coupled to 197–199 receptors indirectly activating 193–197 substrate specificity 218 see also Proteolysis Proteasome 225–226, 225F cell cycle regulation 227, 228F NF-κB regulation 231–232, 231F p53 targeting to 95F, 96 protein targeting to 94, 94F, 226 Protein(s) biochemical and biophysical analysis 345–354 degradation see Proteolysis dynamics, spectroscopic analysis 354, 354F identification 356, 366–368 internalization of cell surface 127– 128, 128F membrane 122–123, 123F methods for studying 345–373 modular architecture 243–273 as signaling devices 279–280, 280F structure see Protein structure trafficking 90 Protein arginine methyl transferases (PRMTs) 87, 109 Protein-complex allosteric switches 280F, 281 Protein complexes see Complexes, protein Protein dephosphorylation 60 energetics driving 47, 47F see also Protein phosphatases Protein domains see Domains, protein Protein-fragment complementation assay (PCA) 359–360 Protein histidine phosphatase (PHPT- 1) 104, 104F Protein interactions 21–40 intermolecular see Protein–protein interactions intramolecular 249 methods of mapping 355–360 Protein kinase(s) 12, 43, 52–58 activation loop see Activation loop, protein kinase allosteric switches 266 biosensors 370 C-helix 53F, 54 dimerized/oligomerized receptors 181–182, 182F families 58, 59F inhibitors 97 key conserved catalytic residues 52–53, 53F pairing with phosphatases 45, 45F plant, regulation by light 267, 267F reaction energetics 46–47, 47F regulation of catalytic activity 53F, 54–56, 55F, 56F as state machines 276, 277F structure 52–53, 52F substrate specificity 99–100 binding interactions regulating 56–58, 57F phosphospecific binding modules and 100, 100F transmembrane receptors covalently linked to 189–193 transmembrane receptors indirectly activating 193–197 see also Serine/threonine kinases; Tyrosine kinases; specific kinases Protein kinase A (PKA) activation by cAMP 136, 141F AKAP interactions 150–152, 151F allosteric switch function 266 cooperative switchlike activation 290 coupling Rac activation to 271, 271F diffusion rate 137F hedgehog signaling 195, 196F key catalytic residues 53, 53F myristoylation 123–124 phylogenetic relationships 59F regulation by cAMP binding 142– 143, 143F Protein kinase B (PKB) see Akt kinases Protein kinase C (PKC) activation 144–145, 145F AKAP-mediated regulation 152 δ isoform (PKCδ) 254 Protein kinase G (PKG) 142 Protein kinase-like (PKL) genes 58 Protein phosphatases 12, 60–65 classes and reaction mechanisms 60–64, 61F pairing with kinases 45, 45F reaction energetics 47, 47F regulation of activity 64–65 substrate specificity 99–100 Protein phosphorylation 97–104 allosteric regulation 48, 48F amino acid residues involved 50, 50F chemical effects 89, 89F coupling with protein interactions 97–99 distributive 100–101, 101F, 290 E3 ligase affinity and 105–106 energetics 46–47, 47F induced conformational changes 50–52, 51F methods of studying 364–366, 364F, 367, 371 at multiple sites, diverse modes 100–101, 101F p53 96 pairing of kinases and phosphatases 45, 45F as post-translational modification 86, 86F priming 100, 101F processive 100–101, 101F, 290 protein domains recognizing 249–255 regulatory role 43, 49–52 Akt activation 126–127, 127F cell cycle 227, 228–230, 229F nuclear localization 119–121, 120F, 121F protein kinase activity 54–56, 55F, 56F substrate specificity 100, 100F see also Protein kinase(s); Protein phosphatases Protein–protein interactions 12, 21–40 cellular and molecular context 30–38 coupling with phosphorylation 97–99 effects of post-translational modifications 89–90, 90F identifying binding partners 356 methods of detection/ mapping 355–360 properties 22–30 regulation by ubiquitylation 256–257 solid-phase screening 357, 357F see also Complexes, protein Protein stability allosteric regulation 48, 48F post-translational modifications and 90, 90F Protein structure 349–351 methods of determining/ analyzing 352–354 primary 350, 350F quaternary see Quaternary structure secondary 350, 350F tertiary see Tertiary structure Protein tyrosine phosphatases (PTPs) 61F, 62–64 acting as receptors 64, 189, 189F cysteine (Cys)-based 62–64 modular domain structure 64–65, 65F oxidative regulation 63 reaction mechanism 63, 63F Proteolysis 13, 88, 217 cell-cycle phase transitions 326–328 irreversibility 217–218, 218F nuclear localization of Notch 122, 122F post-translational modifications and 90, 90F receptors coupled to 197–199 receptors indirectly activating 193–197 regulated 217–242 regulated intramembrane (RIP) 224, 224F Prothrombin 219, 220F PSD93 265F PSD95 264–265, 265F PTB domains 99, 248T binding phosphotyrosine sites 252–254, 254F recognition functions 246–248, 249F PTEN 166F, 167 PTP1B 63, 63F PTP domains 64 PTPs see Protein tyrosine phosphatases Pull-down assays 356 PUMA 238, 238F, 239 Pumilio repeat 248T PWWP domain 248T PX domain 248T Pyrin (PYD) domains 237F, 238 Q Quaternary structure 350–351, 350F phosphorylation-induced changes 51–52, 52F rearrangements 49, 49F R Rab control of membrane association 125, 126F GEF binding 73F signaling cascade regulating 79–80, 79F subfamily 68, 68T RabGDI 125, 126F Rac1-related G proteins 68, 68F Rac fusion proteins, engineered 270– 271, 271F Rad9 329 Raf in different organisms 8, 9 PDGF-mediated activation 320 Raf–MEK–Erk kinase pathway 8F, 77, 77F EGF receptor signaling 16 regulation by 14-3-3 protein 267, 267F scaffold protein 78F, 79 see also Erk MAP kinase; MEK RalA 170 Ran proteins 68, 68T control of nuclear transport 118– 119, 119F GEF binding 73F Rap1/Rap2 141–142, 141F Rap80 106–107, 107F Rapamycin 169F, 170 Raptor 169, 170 Ras 67–68 activation by Sos 126, 127F different organisms 8–9, 8F family proteins 68, 68T mechanisms of regulation 72–73, 73F, 74, 74F palmitoylation 130–131 PDGF signaling 317, 319, 320 prenylation 124 structure 67F subcellular localization and signal outputs 130–131, 131F T cell activation 337 H-Ras 130–131, 352F K-Ras 130–131 N-Ras 130–131 RasGAP 263F RasGRP 337, 340 Ras–MAP kinase pathway 8F, 16, 77 see also Raf–MEK–Erk kinase pathway Rbx1 227, 227F Rcc1 118 Reaction intermediates, enzyme- catalyzed reactions 45F, 46 Receptor protein tyrosine phosphatases (PTPs) 64, 189, 189F Receptors 177 desensitization 208–209, 208F GPCRs 211–212, 211F dimerization/oligomerization 181– 184, 181F advantages 182–184, 183F facilitating kinase activity 181– 182, 182F down-regulation 128, 208–212 methods 208–209, 208F role of ubiquitylation 209–211, 210F as drug targets 180 internalization as means of down-regulation 208, 208F mechanisms 128, 128F modulating signal transduction 128–129 retrograde signaling in neurons 130, 131F TGFβ signaling output after 129–130, 130F intracellular 178 ligand-induced conformational changes 180–181, 181F, 182 transmembrane 179, 179F, 180–199 activating both kinases and proteases 193–197 coupled to proteolysis 197–199 covalently linked to protein kinases 189–193 with intrinsic enzymatic activity 186–189 membrane-spanning segments 180–181, 180F with multiple membrane- spanning segments 180–181, 181F with single membrane-spanning segment 181–182, 181F transduction strategies 180–184 up-regulation 128 Receptor tyrosine kinases (RTKs) 8, 9, 186–187 activation 181–182, 182F, 187 autophosphorylation 97, 98F as dynamically regulated scaffolds 265–266 major families 187F recruitment of downstream effectors 97–99, 98F Red fluorescent protein (RFP) 357– 358, 358F Regulated intramembrane proteolysis (RIP) 224, 224F Regulators of G protein signaling (RGS) proteins 75, 186 Regulatory particle, 19S 225F, 226 c-Rel 230, 230F RelA 230 RelB activation pathway 231–232, 231F domain structure 230, 230F Rel homology domain 230, 230F Repeated sequences, protein domains 245, 246F Response regulator (RR) 102, 102F, 103 Retina, vertebrate 308 Retinal 310 Retrograde signaling 130, 131F RGS (regulators of G protein signaling) proteins 75, 186 RH (RGS homology) domain 211, 212, 212F, 248T Rheb 165B, 169, 169F, 170 Rho-associated kinase (ROCK) 235 Rhodopsin desensitization 313 light-induced conformational change 310–311 signal amplification 186, 311 signaling 185F, 309, 310–311 speed of signaling 186, 312 RhoGAPs 71–73, 71F, 72F RhoGDI 125 RhoGEFs 71–73, 71F, 72F Rho proteins 68, 68T biosensor 371F cell motility 68, 68F enzymes regulating 71–73, 71F, 72F membrane association 125 Rictor 169, 170 RING domain 248T RING E3 ligases 105 caspase inhibitors 235 cell cycle regulation 227, 227F, 228 control of NF-κB activation 231 RNA interference (RNAi) 361 RNA polymerase 108–109, 111–112, 111F Robust oscillators 299, 300F Rods 307–314 see also Photoreceptor cells Ryanodine receptors 148 S Saccharomyces cerevisiae (budding yeast) control of nuclear transport 119– 120, 120F MAP kinase cascades 77, 78F scaffold proteins 78–79 SCF complex 227F SAM domains 248T Abl activation, in leukemia 269– 270, 270F oligomerization 259–260, 259F, 260F SARA 129, 130F, 262 SBF 326, 327 Scaffold complexes caspase activation 234, 234F T cell activation 336–337 Scaffold proteins 22, 23F, 263–266 chimeric, engineered 270, 271F determining substrate specificity 99–100 G protein signaling 79F, 80 MAP kinase cascades 22, 77–79, 78F mechanisms of action 78–79, 79F with multiple phosphotyrosine motifs 265–266, 266F PDZ domain-containing 264–265, 265F phosphorylation by receptor tyrosine kinases 98F, 99, 187 small-mediator signaling 150–152, 151F Scatchard analysis 346, 346F SCF complex cell cycle control 227–228, 228F NF-κB regulation 230, 231, 231F structure 227, 227F Schizosaccharomyces pombe (fission yeast), Ras localization and signaling 131, 131F Secondary structure 350, 350F Second messengers see Small signaling mediators γ-Secretase 197, 198F, 224, 224F Securin 228–230, 229F, 328 Sem-5 8–9 Separase 328 Serine phosphorylation 50, 50F inducing protein interactions 98, 99 see also Phosphoserine/ phosphothreonine Serine/threonine (Ser/Thr) kinases 50 docking sites 57, 57F families 58, 59F, 61F plant, activation by light 267, 267F receptors with intrinsic activity 187–188, 188F substrate specificity 57, 57F, 58 Serine/threonine (Ser/Thr) phosphatases 50 families 60–62, 61F holoenzyme complexes 60, 62F reaction mechanisms 60, 62F regulation 64–65 Seven-transmembrane receptors (7-TMRs) 184 SH2 domain-containing proteins combinatorial diversity 263, 263F recruitment by receptor tyrosine kinases 97–98, 98F, 99 SH2 domains 248T, 249–252 peptide binding 24, 24F, 249–252, 250F adjacent domains influencing 252, 252F cooperativity 37, 37F, 252, 253F recognition function 246 specificity 251–252, 251F receptor tyrosine kinases 97–98, 98F regulation of kinase activity 55, 56, 56F regulation of phosphatases 65F, 100 regulation of substrate specificity 57, 100 structure 249–250, 250F SH3-containing proteins 244F SH3 domains 24, 244F, 248T recognition of proline-rich motifs 258, 258F regulation of kinase activity 55, 56F regulation of NADPH oxidase 268, 268F regulation of substrate specificity 57 specific binding to proline 35, 35F Shank 265F Shc 253–254 Sheddases see ADAM metalloproteases SHIP 167 Shp1 342–343 Shp2 65F Sic1 327 Signal amplification 5, 294, 294F cleavage of zymogens 218 enzymes 45, 294, 294F GPCRs 186 small signaling mediators 138 T cell activation 339–340 visual transduction system 311 Signal amplitude increasing see Signal amplification responses to 288–292, 289F Signal duration measuring 292–293, 293F modifying 294, 296–299, 296F Signal integration 5, 281–286, 281F Signals, diversity of 3–4, 3F SILAC 366F, 367 Sildenafil (Viagra•) 141, 141F Single-cell analysis flow cytometry 371–372, 372F methods compared 369T time-lapse microscopy 368–369, 369F Skp1 227, 227F Skp2 227F, 228 SLP-76 GADS SH3 domain binding 258, 258F T cell receptor signaling 336–337 transient SH2 binding sites 266 Smac/DIABLO 235, 240 SMAD2 129, 130F SMAD3 129, 130F SMAD4 129, 188, 188F SMAD7 129, 130F SMAD proteins 188, 188F R-SMAD proteins 188, 188F Small G proteins (small GTPases) 67–68 control of cell motility 68, 68F mechanisms of membrane association 125 phosphoinositide signaling and 167 regulation of activity 71–74, 71F additional mechanisms 75 GEF and GAP domains 71–72, 72F mechanisms of GEF/GAP action 72–74, 73F, 74F signaling cascades regulating 79–80, 79F structure 67, 67F subcellular locations and signal outputs 130–131, 131F subfamilies 68, 68T Small-molecule inhibitors, pharmacological 361, 361F Small signaling mediators (second messengers) 13, 135–152 classes 139, 139T complex spatiotemporal pattern generation 138–139, 138F mechanisms of downstream effects 136–137 properties 135–139 range of physical properties 140 scaffold proteins 150–152, 151F signal amplification 138 specificity and regulation 149–152, 150F speed of signaling 137–138 synthesis–degradation balance 136, 136F see also specific mediators Smell, sense of 185 Smoothened (Smo) 195–197, 196F, 212 SMURF ubiquitin ligases 129–130, 130F SNARE domain 248T SOCS box 248T Solid-phase screening 357, 357F Soluble guanylyl cyclases (sGC) 204– 205, 205F Sonic hedgehog (Shh) 195, 196F Sos activation of Ras 126, 127F different organisms 8, 9 Grb2 function 264, 264F mechanisms of regulation 72–73, 73F PDGF signaling 317, 319–320 T cell activation 337, 340 Spatial information, transmission of 92 Spatial organization/localization 10 see also Subcellular localization Specificity binding 25, 25F cooperative binding and 35–37, 37F factors determining ideal 31–32 functional constraints 32–33, 33F independent modulation 34–35, 34F quantification 28, 28F cellular signaling 4–5 enzymic reactions 22, 23F Specificity constant (k _cat/K _m) 349 Spectroscopic analysis, protein dynamics 354, 354F Sphingolipids 157, 170–172 Sphingomyelin 170–172 chemical structure 157, 158F lipid rafts 159 metabolites 143–144, 170–172, 171F Sphingomyelinase (SMase) 170, 171–172, 171F Sphingosine generation/metabolism 144, 170, 171F, 172 phosphorylation 163, 170 Sphingosine 1-phosphate (S1P) 139T, 170–171, 172 biophysical properties 161 biosynthesis 144, 163, 170, 171F downstream signaling 172 Sphingosine 1-phosphate receptors (S1PR) 172 Sphingosine kinases (SKs) 171F, 172 Spindle assembly checkpoint 229, 330 SPRY domain 248T Src 14 domain structure 263F SH2 domain 250F, 251, 251F Src family kinases (SFK) 14–15, 15F as allosteric switches 266 integrin signaling 191F, 192 phosphorylation-mediated regulation 55–56, 56F Src homology domains see SH2 domains; SH3 domains Stable-isotope labeling of amino acids in cell culture (SILAC) 366F, 367 Standard free energy (ΔG°) 27–28 START domain 248T START phase, cell cycle 324, 326 State, changes in 11–15, 11F coupling between 14–15, 14F, 15F generating pathways and networks 15–18 types 12–14, 12T, 13F State machines 276, 277F hierarchical organization 277–278, 278F Statins 124 STAT proteins control of nuclear import 120–121, 121F cytokine receptor signaling 190, 191F domain structure 263F Ste5 membrane localization 125 scaffold function 78, 78F, 79 Ste7 78F, 79 Ste11 78–79, 78F Ste20 79F, 80 STE family kinases 59F Steroid hormones 206 receptors 206–208, 207F transfer across plasma membrane 179 Sterols, membrane 157 Stoichiometry 37–38 Stress responses activation of apoptosis 238, 239–240 sphingolipid signaling 172 Subcellular localization 13, 115–132 changes induced by binding 22, 22F important sites 116, 116F localization switches 280F, 281 mechanisms of control 117–127 methods of mapping 357–359, 358F modulating signaling 127–131 post-translational modifications and 90, 90F, 117, 123–125 regulation by 14-3-3 proteins 255, 255F regulation by PH domains 260– 261, 261F transmitting information 116–117, 117F Substrates concentrations, Michaelis–Menten analysis 348–349, 348F enzyme binding to 22 SUMO 104 Supramolecular activation complex (SMAC) 339 Surface plasmon resonance (SPR) 346–347, 347F Surfaces, protein–protein interactions 23–24, 23F, 24F SWIRM domain 248T Switches, molecular 11, 11F allosteric switch proteins see Allosteric switch proteins bistable 301, 302F G proteins 65–66, 65F kinase/phosphatase systems 54 post-translational modifications and 93–94, 93F Switchlike responses 288, 289F cooperativity leading to 37, 289–290, 289F networks yielding 290–292, 291F Synapses, scaffold proteins 264–265, 265F Synaptojanin 167 Synaptotagmin 147 Synthetic biology 270–272 bistable systems 301, 302F T TAB2/TAB3 193, 193F TACE metalloprotease 197, 198F, 221 TAK1 193–194, 193F Talin 192–193, 192F T cell receptors (TCRs) 333 activation of coupled tyrosine kinases 190, 191F down-regulation 341 interaction with peptide–MHC complexes 335 phosphorylation initiating activation 338 signaling network 336–337 ZAP-70 interaction 252, 253F T cells 333–344 activation 335–343 antigen presentation to 335 discrimination of non-antigenic peptides 342–343 effector 333, 334 immune synapse formation 339 interleukin-2 expression 285–286, 285F, 337 naive 333, 334 TEL 259 TEL–Abl oncogene 269, 270F Temperature conformational changes and 47–48 receptors responding to 202 Tertiary structure 350, 350F phosphorylation-induced changes 51–52, 52F rearrangements 49, 49F Tet repressor (TetR) 302F TGFβ see Transforming growth factor β Thermal breathing 48 Thermodynamics binding interactions 27–28, 28F determination of parameters 347, 347F enzymatic catalysis 45–47, 45F G protein regulation 66 Threonine phosphorylation 50, 50F inducing protein interactions 98, 99 see also Phosphoserine/ phosphothreonine Thrombin activation of protease-activated receptors 223–224, 224F proteolytic activation 219, 220F Thrombolysis 219F, 220 Thromboxane 173, 173F Thrombus formation 219, 219F Time-lapse microscopy, live cells 368– 369, 369F Time scales 10 signaling by second messengers 137–138, 137F TIR domain 193, 248T Tissue inhibitors of metalloproteases (TIMPs) 222–223 Tissue plasminogen activator (t-PA) 220 TKB domain 252, 252F T lymphocytes see T cells Toll-like receptors (TLRs) 193–194, 193F TPR repeat 248T TRADD 236–237, 236F TRAF-2 236, 236F TRAF-6 193, 193F TRAF domain 248T TRAF proteins 231 TRAIL 197, 235, 236, 236F TRAIL receptors 197–198, 236, 236F Transcription changes in 9–10 chromatin remodeling 108–109, 110–112, 111F monitoring changes in 363–364, 363F Transcription factors 109 signal integration 285–286, 285F Transducin 185F, 309, 310 inactivation 313 signal amplification 311 speed of response 312 Transforming growth factor β (TGFβ) receptors 187–188 modes of internalization 129–130, 130F SARA binding 262 signaling 188, 188F wound healing 315 Transistor 276, 277F Transition state, stabilization by enzymes 45–46, 45F Translocations, chromosomal 269– 270, 270F Transmembrane signaling 177–214 membrane-permeable 204–208 three general strategies 179–180, 179F see also Gated ion channels; Receptors, transmembrane Transphosphorylation 181–182, 182F β-TrCP 194, 194F, 228 TRP (transient receptor potential) channels 199F, 202 TRPV1 vanilloid receptor 202 TSC complex (TSC1/TSC2) 169, 169F TUB domain 248T Tudor domains 109–110, 248T, 256, 256F Tumor necrosis factor α (TNFα) 197, 235 proteolytic processing 221 signaling pathway 236–237, 236F Tumor necrosis factor receptor (TNFR) 197–198, 236–237, 236F Two-component systems bacteria 102–103, 102F eukaryotes 103 Two-hybrid assay, yeast 356, 356F Tyrosine (Tyr) kinases 50, 58, 59F effect of receptor clustering on activation 182–183, 183F oncogenic fusion proteins 269–270, 270F phosphatidylinositol 3-kinase activation 167 phospholipase C activation 165–166 receptor see Receptor tyrosine kinases receptors covalently linked to 189–193 substrate specificity 57, 58 Tyrosine (Tyr) phosphatases 50, 62–64 classes 61F, 62–64 modular domain structure 64–65, 65F reaction mechanism 63, 63F Tyrosine phosphorylation 50, 50F inducing protein interactions 97–99, 98F T cell activation 338 see also Phosphotyrosine U UBA domain 248T, 257F UBAN domain 107, 107F Ubiquitin 88, 104 enzymes adding and removing 104–105, 104F polyubiquitin formation 104–105, 105F Ubiquitin-binding domains (UBDs) 104, 256–257 recognition of ubiquitin-mediated signals 106–107, 107F structural features 256, 257F Ubiquitin ligase complexes cell cycle regulation 226–230, 328 see also E3 ubiquitin ligases Ubiquitin-like (UBL) peptides 104 Ubiquitylation 88, 104–107 cell cycle control 226–230 domains recognizing see Ubiquitin- binding domains functional consequences 106–107, 107F machinery 104–105, 104F methods of detection 367 NF-κB regulation 231, 231F p53 96 proteasome targeting 94, 94F, 226 receptor down-regulation 209–211, 210F regulation of protein–protein interactions 256–257 role in endocytosis 106, 128 selection of substrates 105–106, 106F UEV domain 248T UIM domains 248T, 250F ubiquitin interactions 106–107, 107F, 257F Ultrasensitive responses 289F, 290, 290F oscillators 299, 300F signaling cascades 290, 291F zero order 290–292, 291F Unsaturated fatty acids 156, 159, 159F V Vanadate 63, 64F Vanilloid receptor 202 Vascular endothelial growth factor (VEGF) receptor 187F Vav 263F, 337 Vesicles, membrane organization of lipids into 156F, 158 regulation of intracellular trafficking 168–169, 168F VH1-like phosphatases see Dual- specificity phosphatases VHL-β protein 250F VHL domain 248T VHL protein 205–206, 206F VHS domain 248T Viagra• (sildenafil) 141, 141F Vision, vertebrate 307–314 Vmax 348, 348F Voltage-gated calcium (Ca2+) channels 146 Voltage-gated potassium (Kv) channels 200–202 gating mechanism 201–202, 201F selectivity filter 200–201, 200F topology 199F Von Hippel-Lindau syndrome 206 Vps27 250F Vps34 167 W N-WASP, signal integration 284–285, 284F WASP proteins, signal integration 284–285, 284F WD40 repeat domains 248T binding modified histones 256, 256F diversity of functions 246 F-box proteins 227 structure 245, 246F WDR5 256 Wee1 cell cycle regulation 283F, 284, 300F, 301, 328 targeting for proteolysis 228 Western blotting 355 Whi5 326, 327 Wnt signaling pathway 194–195, 194F Wound healing 315–322 Writer/eraser/reader systems 11F, 12, 80 G protein-regulating enzymes 65F, 66 histone modifications 109–110 kinase/phosphatase systems 45, 45F, 249 linking together multiple 94, 94F phosphoinositide signaling 166 photoreceptor signaling 311 post-translational control machinery and 90–92, 91F WW domain 248T X Xenopus laevis embryos 299–301, 300F X-ray crystallography 352–353, 352F Y Yan 260, 260F Yeast see Saccharomyces cerevisiae; Schizosaccharomyces pombe Yeast two-hybrid (Y2H) assay 356, 356F Yellow fluorescent protein (YFP) 354, 357–358, 358F, 359, 359F Ypt1 126F Z ZAP-70 coupled SH2 domains 37, 37F, 252, 253F phospholipase C-γ activation 266 T cell activation 190, 336, 338 Zero-order ultrasensitivity 290–292, 291F Zymogens 218