Impact of force field polarization on the collective motions of proteins

ABSTRACT Correlated motions of proteins underpin many physiological mechanisms, such as substrate binding, signal transduction, enzymatic activity and allostery. These motions arise from low frequency collective movements of biomolecules and have mostly been studied using molecular dynamics simulations. Here, we present the effects of two different empirical energy force fields used for molecular dynamics simulations on correlated motions – the non-polarizable CHARMM 36m additive force field and the polarizable Drude force field. The study was conducted on two proteins, ubiquitin - a small protein with a well-described dynamic - and the nuclear receptor protein PPARγ. The ligand binding domain of PPARγ was of particular interest since its function is to regulate transcription through ligand and coregulator protein binding. It has been previously shown that a dynamical network of correlated motions ensures the transmission of information related to ligand binding. We present the results of classical MD simulations where we analyze the results in terms of residue fluctuations, correlation maps, community network analysis and Shortest Path Method analysis. We find that the RMS fluctuations tend to be greater and that the correlated motions are less intense when using the Drude force field than when using the non-polarizable all atom additive force field. Our results provide the first quantification of the impact of using a polarizable force field in computational studies that focus on collective motions.