Large-scale evaluation of dynamically important residues in proteins predicted by the perturbation analysis of a coarse-grained elastic model

We have performed a large-scale evaluation of the predictions of dynamically important residues by a variety of computational protocols including three based on the perturbation and correlation analysis of a coarse-grained elastic model. This study is performed for two lists of test cases with >500 pairs of protein structures. The dynamically important residues predicted by the perturbation and correlation analysis are found to be strongly or moderately conserved in >67% of test cases. They form a sparse network of residues which are clustered both in 3D space and along protein sequence. Their overall conservation is attributed to their dynamic role rather than ligand binding or high network connectivity.By modeling how the protein structural fluctuations respond to residue-position-specific perturbations, our highly efficient perturbation and correlation analysis can be used to dissect the functional conformational changes in various proteins with a residue level of detail. The predictions of dynamically important residues serve as promising targets for mutational and functional studies.Protein conformational dynamics [1,2] is critically involved in many biochemical processes ranging from catalysis [3] to allostery [4-9] and signal transduction [10]. Protein dynamics spans a wide range of time scales (from picoseconds to seconds or minutes). Biologically relevant conformational motions of proteins are often collective (for example in the form of hinge-bending or shearing motions between rigid domains, see [11]). These highly coordinated motions are thought to involve a network of key amino acid residues that couple spatially separated functional sites [9]. The conservation and variation of protein functions are likely underscored by the conservation and co-evolution of these dynamically important residues. The existence of a sparse network of allosterically coupled residues in various proteins has been revealed by the statistical coupling (or correlated mutation) a