4

The objective of peptide synthesis is usually to prepare a single enantiomer. However, the amino acid enantiomers used in the synthesis are not always chirally stable during manipulation. Isomerization may occur at any step of a synthesis, be it the preparation of amino acid derivatives, coupling, or deprotection. The isomerization involves removal of the proton at the α-carbon atom of a residue, followed by a shift of the double bond of the adjacent carbonyl to the α-carbon, giving the enol form and eliminating the chirality. Reprotonation at the α-carbon atom generates the two possible configurations (Figures 4.1-4.5). The process is referred to as enolization and can be initiated by either an acid or base, with the latter more often being the culprit. Usually, though not always, it occurs at a residue that is substituted at both the amino and carboxyl groups. The affected residue may be a single one unattached to other residues or one that is incorporated into a chain (Figures 4.1-4.3), and in either case it may form part of a small cyclic structure that is usually the oxazolone (Figures 4.4, 4.5). The tendency for enolization to occur depends on the nature of the three substituents on the α-carbon atom, electron-withdrawing moieties on the β-carbon, and C-1 of the residue favoring the ionization. The least-encountered situation is that of acid-catalyzed enolization (Figure 4.1). Here enolization is initiated by protonation of the oxygen of the carbonyl, which induces migration of the electrons to neutralize the oxocation. Regeneration of the carbonyl from the enol produces the two isomeric forms of the residue. Examples of isomerization by this mechanism are the generation of D-isomers during the hydrolysis of peptides or proteins and during the exposure of N-substituted-N-methylamino acids to hydrobromic acid in acetic acid (see Section 8.14). The same protonation is involved in

the deprotection of functional groups by acidolysis (see Section 3.5,) but, fortunately, no enolization occurs during cleavage by the commonly used acidolytic reagents.1-5

The most frequent cause of isomerization is direct removal of the α-proton by a base (Figure 4.2). The resulting unshared electron pair migrates to generate an equilibrium mixture of the carbanion and the oxoanion, with the double bond shifted to the α-carbon, thus eliminating the chirality. In the unique case of cysteine derivatives, however, the proton never gets completely detached but is carried from one side of the asymmetric center to the other by the base in a process known as isoracemization. Reprotonation generates the two configurations of the residue. Three different situations in which isomerization is caused by base-catalyzed enolization are encountered: first, the base is contained in the same molecule, with the classical case being the enantiomerization that occurs at activated Nα-alkoxycarbonylhistidine derivatives unprotected at the π-nitrogen of the imidazole ring of the side chain (see Section 4.3). Second, the base is introduced inadvertently, such as when the enantiomerization occurs at the activated residue of an acyl azide or activated ester that has been left in the presence of a tertiary amine. Particularly sensitive residues, because of the side chain electron-withdrawing groups, are

S-benzylcysteine, β-cyanoalanine, and β-carboxy-substituted aspartic acid. Third, the base is added intentionally. Here, isomerization can occur at the implicated residue such as that produced during the saponification of amino acid or peptide esters and during the aminolysis of Fmoc-proline chloride or aspartic acid activated at the β-carboxyl group (see Section 4.19). The isomerization also can occur at a distant residue such as that produced at the esterified, including resin-linked, carboxy-terminal cysteine or serine residues of a chain by the tertiary amine used to detach Fmoc-groups or to initiate onium salt-mediated coupling reactions (see Section 8.1) during chain assembly. Regardless, the residues most susceptible to basecatalyzed enolization are those in which the proton is abstracted from an α-or βcarbon that does not have an ionizable proton on the functional group linked to it, examples being PgN(CH3)- or -CO2R. Base-catalyzed enolization also occurs at a residue whose N-Cα-C=O atoms constitute part of the ring of an oxazolone (see Section 4.4) or a piperazine-2,5-dione (see Section 6.19), especially if prolyl is incorporated into the latter.3,6-13

It had been established by midcentury that N-alkoxycarbonylamino acids do not isomerize during coupling. However, there emerged the puzzling observation that

Nα-substituted histidines, whether side-chain protected or not, underwent considerable enantiomerization during diimide-mediated reactions. Even Boc-L-histidine coupled by the azide method gave enantiomerically impure products. The results were attributed to intramolecular base-catalyzed proton abstraction and enolization (see Section 4.2). At the time, the position on the ring of substituents was not known, as evidenced by the designation of the popular substituent im-benzyl, which was ambiguous. Moreover, it was not helpful that the nitrogen atoms of the imidazole ring of histidine were designated in different ways by biochemists and chemists; namely, as 1,3 and 3,1. At least discussion was facilitated by the recommendation of the pertinent nomenclature committees that the nitrogen atom nearest to the chain (δ) should be designated pros (“near,” abbreviated π) and the nitrogen atom farthest from the chain (ε) be denoted tele (“far,” abbreviated τ). The α,β,γ,δ,ε designations of atoms (Figure 4.3) are those employed by x-ray crystallographers. There followed the suggestion by D. F. Veber that the side reaction might be caused by the prosnitrogen atom of the imidazole ring, which remained unsubstituted in the derivatives. It is now known that substitution on the imidazole of histidine invariably occurs at the less hindered tele-nitrogen, and that isomerization indeed is caused by abstraction of the α-proton by the pros-nitrogen of the ring if it is left unprotected. The latter was demonstrated unequivocally by experiments with tele-and pros-substituted benzoylmethyl (phenacyl) derivatives of Cbz-histidine, with the latter being obtained from Cbz-His(τTrt)-OMe. pros-Substitution prevented isomerization during coupling; the tele-substituted derivative generated 30% of epimeric peptide (Figure 4.3). Further study established that for practical reasons, the pros-benzyloxymethyl derivative is the preferred derivative for couplings. Thus, substitution of the pros-nitrogen effectively suppresses enantiomerization during the coupling of Nα-alkoxycarbonylhistidines. Furthermore, use of the pros/tele nomenclature eliminates the ambiguity of the previous 1,3/3,1 designations for the nitrogen atoms of the ring.14-18

Oxazolones are readily formed from activated N-acylamino acids or peptides (see Sections 1.7, 1.9, and 2.23). Because of the strong tendency of the double bonds of the oxazolone to form a conjugated system, the α-proton is very labile and thus sensitive to base. The latter causes enolization, which eliminates the chirality (Figure 4.4). The lability of the α-proton is governed by electronic and conjugative effects at C-2, an electron donor such as methyl stabilizing the proton in contrast with phenyl, and steric effects at C-4, with the isopropyl of valine impeding release of the proton relative to isobutyl of leucine and then benzyl of phenylalanine. The effect of the latter is apparently anomalous and has been explained on the basis that the oxazolone from phenylalanine adopts a unique conformation resulting from stacking of the two rings, and this facilitates removal of the proton. As an example, the rates of racemization of the 2-phenyl-5(4H)-oxazolones from phenylalanine, leucine, and valine decrease in a ratio of 34:17:1 in dichloromethane. In tetrahydrofuran, in the presence of one equivalent of N-methylpiperidine, about 4 hours are required to

racemize Z-glycyl-L-valine. The rate of isomerization also depends dramatically on the nature of the base as well as the polarity of the solvent, being as much as 50 to 100 times greater in dimethylformamide than in dichloromethane. In the latter, Nmethylmorpholine isomerized the oxazolone from a dipeptide more slowly than triethylamine, but in dimethylformamide it was the opposite, with the slowest isomerization occurring in the presence of diisopropylethylamine.