Proline is commonly thought of as a turn-promoting residue. Indeed, a large number of turn-inducing proline mimetics have been described, some of which have been synthetically incorporated into peptides. Sarcosine, also known as N-methyl glycine (CH3NHCH2COOH), is a commonly employed proline replacement. Indeed, the latter has been used in substrates for cy-clin-dependent protein kinases, enzymes that require a proline residue on its intended substrate to direct phosphorylation to an adjacent serine moiety. For example, Ando and coworkers showed that peptides containing the sequence from vimentin, Leu-Gly-Ser-Ala-Leu-Arg-Arg-Arg-amide, in which the alanine moiety is replaced by either proline or sarcosine, serve as enhanced substrates for the cdc2 protein kinase (V/K 600-fold for Pro-peptide and 90-fold for Sar-peptide versus the Ala-peptide) (Ando et al. 1993). These investigators subsequently examined the specificity of the cdc2 and cdk5 proline-directed protein kinases with respect to a variety of N-substituted proline replacements including, N-methyl alanine, N-methyl leucine, N-methyl valine, and the four (azetidine carboxylic acid) and six ring (piperi-dine carboxylic acid) analogs of proline (Ando et al. 1997). Perhaps not too surprisingly, the peptides containing the ring analogs of proline were among the best substrates for these protein kinases, with the proline-containing parent serving as the most efficient substrate. Nevertheless, the sarcosine derivative serves as a substrate as well, albeit somewhat more efficiently for the cdc2 versus cdk5 enzyme. The latter observation is significant since sar-cosine is the parent of N-substituted glycine derivatives. Peptides containing the latter ("peptoids") are readily prepared using standard solid-phase peptide synthesis conditions (Figliozzi et al. 1996; Burkoth et al. 2003). Indeed, recent improvements have reduced the coupling time to a few minutes (Olivos et al. 2002). Consequently, it is now possible to prepare large libraries of peptoid-based derivatives in an essentially automated fashion.
Lim and his colleagues have reported that N-substituted amino acids serve as proline replacements in peptide-based ligands targeting both SH3 and WW domains (Nguyen et al. 1998; Nguyen et al. 2000). These investigators demonstrated that the proline selectivity for these domains is due to a preference for N-substituted residues and not simply a function of the rigid cyclic structure that is unique to proline. The SH3 domain from Sem5 recognizes the motif PPPVPPR, whereas the WW domain from Yap prefers GTPPPPYTVG (where the "essential" prolines are underlined). The initial strategy employed a scanning approach in which both essential and nonessential prolines were replaced by either the Ca-substituted alanine or the N-substituted sarcosine. Replacement of the essential prolines with sar-cosine is well tolerated, whereas introduction of alanine at these sites is not. The three-dimensional structures of these protein recognition domains are known. The ligand binds as a polyproline type II left-handed helix, which contains three residues per turn. The binding grooves on these protein domains can accommodate substituents from the type II helix, but only in a closely packed fashion. The latter can be achieved by a motif in which a Ca-substituent is adjacent to an N-substituted residue. A small library was prepared that consisted of peptides containing 11 different N-substituted glycines at the two essential proline sites in an SH3-targeted ligand. These were screened against the SH3 domains from Src, Grb2, Crk, and Sem5.
The lead peptide for Grb2, 23, exhibits a KD of 40 nM, more than 100-fold better than that of the corresponding proline-containing derivative. A subsequently prepared library, containing N-substituted glycine residues with oxygen- and nitrogen-containing substituents, was also evaluated (Nguyen et al. 2000). The latter functionality was chosen for their ability to interact with polar residues that reside near the ligand-binding grooves of the SH3 domain. Compound 24 exhibits a 30 nM KD for the Grb2 SH3 domain and high selectivity versus Crk (300-fold) and Src (1,500-fold). A high-affinity li-gand with impressive selectivity was also identified for the Crk SH3 domain. Double proline-substituted derivatives were also prepared, but these ligands failed to show an additive improvement in affinity for their protein targets. The latter is likely because the N-substituted glycines at the two different "essential" proline sites were identified independently of one another. Introduction of an N-derivatized glycine at one position likely induces a conformational change between ligand and protein that influences the interaction at the second proline site. Nevertheless, these results demonstrate that "essential" residues cannot only be replaced with unnatural analogs, but that the latter can furnish dramatically enhanced potency and selectivity.
Was this article helpful?