Mode Localization in the Cooperative Dynamics of Protein Recognition

Abstract

The biological function of proteins is encoded in their structure and expressed through the mediation of their dynamics. But how do the random thermal motions of a folded protein in the picosecond timescale propagate to become emergent biological processes? Our approach is a novel theoretical method, the Langevin Equation for Protein Dynamics (LE4PD), which takes the structural ensemble of a protein as input and projects the dynamics analytically. Both simulation-derived and experimental NMR conformers are the input structural ensembles for the LE4PD. The model is solved in a set of diffusive modes which span a vast range of timescales from the picosecond to the microsecond regime. Local fluctuations initiate biologically relevant pathways as they cooperatively enhance the dynamics in specific regions in the protein. The slowest, most collective motion localizes directly to highly conserved regions involved in binding partner recognition and active-site regulation. The picture that emerges is a dynamically heterogenous protein where biologically active regions provide energetically-comparable conformational states that can be trapped by a reacting partner. Starting from the static structural ensemble of a protein, the LE4PD predicts where specific regions in the protein three-dimensional structure become dynamically active at a given timescale and how allosteric dynamics are enhanced or suppressed upon binding. We analyze this mechanism as we calculate the dynamics of monomeric and dimerized HIV protease, and free Insulin Growth Factor II Receptor (IGF2R) domain 11 and its IGF2R:IGF2 complex, and other examples. The diffusive mode rendition precisely indicates the position inside the primary sequence of these energetically-guided local fluctuations, allowing us to predict which parts of the protein will lead the kinetics of biologically relevant processes.