Why ligand cross-reactivity is high within peptide recognition domain families? A case study on human c-Src SH3 domain.

Many important protein-protein interactions in eukaryotic signaling networks are mediated by peptide recognition domains (PRDs), which bind short linear sequence motifs in other proteins. However, high ligand cross-reactivity is observed within most PRD families, rendering a broad specificity for the family members. In the present study, we attempt to explore the molecular mechanism and physicochemical origin of PRD cross-reactivity. In the procedure, a structure-based method called atomic cross-nonbonded interaction analysis (ACNIA) is described to extract atomic-level nonbonded interaction information at domain-peptide interface and to correlate the information with peptide affinity based on a set of structure-solved, affinity-known protein-peptide complex samples compiled from numerous literatures and databases. The ACNIA-derived affinity predictor is tested rigorously with statistical validation approach, which is also demonstrated to be capable of perceiving slight structural change in the interface using three distinct panels of SH3-binding peptide data. Subsequently, with help of the affinity predictor we adopt the human c-Src SH3 domain, one of the most sophisticated PRDs, as a paradigm to investigate the ligand cross-reactivity within SH3 family. It is found that most of the family members have only few non-essential residue differences in their peptide-binding pockets, and thus exhibit a similar peptide recognition profile and high cross-reactivity. The cross-reactivity is even shared by different subclasses of SH3 domains. The findings suggest that inherent binding specificity is not the only factor to select appropriate binders for specific SH3 domains, and other aspects such as cellular context and the rest of the SH3-containing proteins may play important roles in reducing their ligand cross-reactivity.