One Plus One Makes Three: Triangular Coupling of Correlated Amino Acid Mutations

Abstract

Correlated mutations have played a pivotal role in the recent success in protein fold prediction. Understanding nonadditive effects of mutations is crucial for altering protein structure, as mutations of multiple residues may change protein stability or binding affinity in a manner unforeseen by the investigation of single mutants. While the couplings between amino acids can be inferred from homologous protein sequences, the physical mechanisms underlying these correlations remain elusive. In this work we demonstrate that calculations based on the first-principles of statistical mechanics are capable of capturing the effects of nonadditivities in protein mutations. The identified thermodynamic couplings cover the short-range as well as previously unknown long-range correlations. We further explore a set of mutations in staphyloccocal nuclease to unravel an intricate interaction pathway underlying the correlations between amino acid mutations.

Divergence in Dimerization and Activity of Primate APOBEC3C

“…Energy decomposition analysis indicates the dimer interaction is destabilized by approximately 770 kcal/mol ( Table S1 ). This destabilization is greater than the sum of the individual variants, consistent with what has been previously observed with multiple mutants 34 .…”

Divergence in dimerization and activity of primate APOBEC3C

“…Energy decomposition analysis indicates the dimer interaction is destabilized by approximately 770 kcal/mol (Table S1). This destabilization is greater than the sum of the individual variants, consistent with what has been previously observed with multiple mutants (Werner, Gapsys, & de Groot, 2021).…”

Accurate protein stability predictions from homology models

“…In our work, we only considered single amino acid substitutions, though the ability to model multiple substitutions is important for protein engineering purposes. Previous results suggest that it is difficult to predict non-additive effects of multi-mutants even when using crystal structures as input to for example FoldX [30] , [31] , whereas more computationally demanding free energy calculations achieve better accuracy [31] . Thus, a first step is to achieve higher accuracy for multi-mutants using the faster computational methods, and then examine the accuracy using homology models as input.…”