Dissecting Ligand Binding Sites : A Layer at a Time

Abstract

Biological phenomena at a molecular level are a consequence of complex interplay of interactions between ligands and their associated protein targets. Such interactions are localized to defined protein regions termed as binding sites (residues within ∼4.5A from ligand). Numerous cases exist wherein proteins of diverse folds bind to similar ligands. We undertake a study with an objective to quantify similarities in binding sites, that utilize the structural information of a binding site at atomistic detail, in order to examine extension of such similarities to additional regions surrounding the binding site. We find that 68% and 22% of high scoring comparisons among single-domain and multi-domain proteins respectively, exhibit similarity at the binding site that extrapolates to regions beyond it (an additional 4.5A from binding site). Such similarities in binding sites as well as their extension to adjoining regions are analysed along with residue-wise correspondences and pairwise structural alignment of binding sites. Sequence identity and ligand topology are found to be independent of the extent of binding site similarity. Statistical validation of these findings re-iterates their relevance by means of p-value calculation as well as decoy site generation.Occurrences of similarity beyond the binding site are also further investigated using Protein Side-chain Networks (PScN). Such networks are created wherein each node represents amino acid residues, and an edge is constructed based on interaction strength between side chain of non-covalently bonded residues. PScNs lead to the identification of densely connected modules i.e cliques and communities in regions of binding site and surrounding areas. Thus, a combinatorial approach capturing atomistic details of binding site architecture and global perspectives of ligand-binding by PScNs could aid in providing valuable insights to current understanding of molecular recognition events.