Computational assessment of mutations of protein cores

Protein mutations are known to change both protein stability and function. Many computational methods have been developed to predict these changes, but often with limited success. We investigate how mutations to residues in protein cores produce changes in packing fraction, side chain conformations, and backbone motion. To investigate the extent of mutation-induced backbone motion, we created two datasets of protein crystal structures. One dataset contains high-resolution crystal structures of mutated proteins and the crystal structure of the associated wildtype proteins. The backbone root-mean-square-deviation (RMSD) of these structures is compared to a duplicate dataset that contains pairs of crystal structures of identical proteins. We also investigate the change in packing fraction of mutated residues. Finally, we show that the hard-sphere plus stereochemical constraint model is able to recapitulate the side chain dihedral angle combinations for buried amino acids in protein cores, protein-protein interfaces, and mutated proteins. We find that the side chain dihedral angle prediction accuracy decreases as the residue solvent accessible surface area (SASA) increases. This reveals the dominance of steric interactions in determining side chain conformations for buried residues and a relative increase in the strength of other interactions for solvent exposed residues. These results further our understanding of how protein mutations affect protein structure and will inform future computational methods of predicting protein stability changes due to mutations.