Multiscale Simulation Of Intra-Protein Communication

We develop a new computational framework to model intra-protein communication. The configurations sampled in atomic molecular dynamics trajectories are used to compute bond lengths and force constants in an elastic network approximation of the distribution of protein structures. To go beyond the harmonic approximation, a key novelty is to compute model parameters in consecutive time windows with a user-specified size to follow the time evolution of the mechanical coupling network of protein conformation. In analogy to spectrogram of sound waves, sequential elastic network models calculated from atomic trajectories are termed the fluctuogram of protein dynamics. By analyzing and comparing the fluctuograms of Ca2+-bound and apo subtilisin, we illustrate that intermittent conformational changes and mechanical coupling variation are important mechanisms of intra-protein communication. We also show that the fluctuogram can be used to predict residues with high tendency to co-evolve by comparing with the results of statistical coupling analysis of a multiple sequence alignment. In addition to the strength of mechanical coupling, we found that the fluctuation of inter-residue force constants is also an important descriptor for co-evolution. Together, the results of this work (a) reveal the intermittent nature of conformational changes and the mechanical coupling variation, (b) show that intra-protein communication can proceed without a drastic change of protein structure and the pathways of which can be identified by the fluctuogram, and (c) support the theory that mechanically coupled residues tend to co-evolve.