Molecular Features of PRD

The signaling domains that recognize PRMs that have been identified so far seem to have evolved specifically to exploit the special features of the amino acid proline. The signaling domains known to interact with PRMs are the SH3 (Varmus et al. 1989), the WW (Bork and Sudol 1994), the EVH1 (Niebuhr et al. 1997), the GYF (Nishizawa et al. 1998, Freund et al., 1999) and the UEV domain (Garrus et al., 2001; Pornillos et al., 2002a). In addition, profilin was found to bind to PRMs (Tanaka and Shibata 1985), aside from its property as an actin-interacting protein involved in cytoskeletal assembly (Carlson et al., 1977). This set of adaptor domains may not be complete, for example, the binding domain of prolyl-4-hydroxylase could be an additional member of the PRD family (Myllyharju and Kivirikko 1999). From the studies of these domains, several features of PRM recognition have emerged from the structures of PRD:PRM complexes (Musacchio et al. 1994a; Yu et al. 1994; Macias et al. 1996; Prehoda et al. 1999; Pornillos et al. 2002a; Freund et al. 2002). A set of conserved, often aromatic amino acid residues defines the respective fold families, and the aromatic side chains form at least one stacked pair of an aromatic cradle that represents the proline-binding pocket. The structural comparison shows that while GYF and WW domains converge on a single pocket, SH3, EVH1 and UEV mediate PRM recognition via two neighboring surface depressions. A detailed structural comparison has been described in a number of excellent reviews and will not be discussed here (Mayer 2001; Zarrinpar et al. 2003a; Li 2005; Kay et al. 2000; Sudol 1996; Ball et al. 2005; Kofler and Freund 2006). Despite the convergence in ligand binding, it has to be noted that the folds of the various domains are quite different. They all represent stable scaffolds that are compatible with the formation of the aromatic cradle, and nature has exploited these scaffolds early on in evolution since all of the domains exist in S. cerevisiae. It is interesting in this respect that scaffold amplification is observed for SH3 and WW domains, which contribute to the assembly of many protein complexes in higher eukaryotes. In contrast, the occurrence of GYF domains has remained relatively constant. For example, there are three characterized GYF domains containing proteins each in S. cerevisiae and H. sapiens. The anticipated function of GYF domains in splicing-associated processes is therefore likely to be conserved throughout evolution, while several functional clusters are associated with SH3 and WW domains.

Whereas most members of the PRDs converge on the recognition of at least two prolines in the PRM, there is considerable variety in the accommodation of residues flanking the proline-rich core motif. Interestingly, variations occur within the respective fold families as well as between folds. Consequently, the array of lig-ands bound by one fold family member might be more similar to the member of another fold family than to a member of its own family. For example, an early finding suggested that the SH3 domain of Abl and the WW domain of FBP11 bind to similar PRM in vitro (Bedford et al. 1997), and evidence is mounting that such "shared targets" exist in vivo (Kofler and Freund 2006). How then is specificity obtained in PRM recognition? A certain degree of specificity is often introduced by charge complementarity. Charged residues flanking the proline-binding pocket within the domain preferably select ligands with opposite charge. For example, the Fyn-SH3 domain contains a very acidic RT loop (a feature shared by most SH3 domains) that leads to the preferred ligand signatures RxxPxxP and PxxPxR, dependent on the two pseudosymmetric orientations of the peptide (Cesareni et al. 2002). Similarly, the GYF domain of CD2BP2 favors positive-charged ligands over neutral or negatively charged ligands. Correspondingly, Fyn-SH3 and CD2BP2-GYF were shown to bind to the same positively charged PRM in the CD2 cytoplasmic tail with low, but comparable affinity (Freund et al. 2002). While sequence recognition outside the proline pocket(s) still allows a whole set of different peptides to bind to the individual domains, it clearly reduces sequence space considerably by deselection based on steric hindrance and charge repulsion. The power of negative selection was shown in a paper from the Lim group which showed that within a given organism an SH3 domain with almost exclusive ligand specificity could be observed (Zarrinpar et al. 2003b). However, such specificity can normally not be observed at the level of recognition rules alone since other powerful factors, such as compartmentalization, avidity and multiple additional interactions contribute to specificity in vivo. However, molecular promiscuity also seems to be inherent to many PRD and thereby allows the recruitment of the respective proteins to different target sites within the cell. A common feature of most PRDs is that their N and C termini are located relatively close in space, thereby allowing the domains to be integrated into the respective full-length proteins with minimal disruption of the overall protein structure.

0 0

Post a comment