Function changing mutations in glucocorticoid receptor evolution correlate with their relevance to mode coupling.

Nonlinear effects in protein dynamics are expected to play role in function, particularly of allosteric nature, by facilitating energy transfer between vibrational modes. A recently proposed method focusing on the non-Gaussian shape of the configurational population near equilibrium projects this information onto real space in order to identify the aminoacids relevant to function. We here apply this method to three ancestral proteins in glucocorticoid receptor (GR) family and show that the mutations that restrict functional activity during GR evolution correlate significantly with locations that are highlighted by the nonlinear contribution to the near-native configurational distribution. Our findings demonstrate that the analysis of nonlinear effects in protein dynamics can be harnessed into a predictive tool for functional site determination.