ABSTRACT
The eld of metagenomics, which is based on the analysis of DNA extracted from a community of microbes, has grown rapidly to address the challenge that the majority of micro organ isms are not yet culturable in the laboratory and that those that are amenable to culture express only a fraction of their encoded natural products. Screening of metagenomic DNA libraries has emerged as a promising method to access unculturable or unexpressed chemical diversity. The methods for natural product discovery from metagenomic libraries can be divided into two main conceptual frameworks: activity-based screening and sequence-based screening. The rst approach, activitybased screening, is similar to traditional bioassay-guided fractionation in that fragments of DNA are expressed in a heterologous host and assayed for production of a desired bioactivity. The alternative strategy, sequence-based screening, involves identifying fragments of DNA encoding biosynthetic genes based on sequence homology, followed by targeted heterologous expression and detection of the predicted small-molecule product. Both approaches rely on the tendency of microbial biosynthetic pathways to be clustered in a single region of the genome, and therefore a reasonable likelihood that all of the genes necessary for the production of a given natural product will be present in a contiguous fragment of DNA. Here we illustrate the two main strategies for discovery of natural products from metagenomic DNA libraries with case studies from the literature, including activity-based screening for the discovery of marine siderophores and sequence-based screening for the identication of the biosynthetic genes encoding the onnamides and polytheonamides. The gene clusters for the patellamides and ecteinascidins were recently found by direct sequencing of metagenomes. We highlight several of the practical challenges for efcient drug discovery from metagenomic libraries, including fractionation of complex communities, successful heterologous expression, and detection of the expressed products. We discuss current limitations to this approach that have resulted in the paucity of successes to date and make the case for a dramatic increase in the success rate of metagenomic approaches over the next 5 years. In addition, we explore some of the recently developed tools for metagenomic sequence assembly and detection of biosynthetic genes and make predictions about the future directions for in silico natural product discovery.
