NMR Relaxation and Molecular Dynamics Simulations of Side Chain Dynamics in Proteins

Abstract

Molecular dynamics (MD) simulations and nuclear magnetic resonance (NMR) spin relaxation experiments have become increasingly powerful to study protein dynamics at atomic resolution due to steady improvements in physical models and computation power. Good agreement between generalized Lipari-Szabo (S2NH) order parameters derived from experiment and simulation is now observed for the backbone dynamics of a number of proteins. Unfortunately, the agreement for side chains, as e.g. probed by S2CH3 for methyl-containing side chains, is much poorer. In this work we discuss several issues with methyl side chains that need to be addressed to close the gap between NMR and MD. Accounting for protein tumbling is one very important factor to obtain a good agreement. In our hands, the application of improved water force fields with an appropriate way of including anisotropic overall protein tumbling improves the prediction of experimentally measured dynamic observables by MD simulations. We demonstrate these aspects for T4 lysozyme as a representative example. Our results guide the way for extracting from the NMR relaxation data the most accurate parameters that describe protein side chain dynamics and report on conformational entropy. Additionally, we reparametrized side chain dihedral angle energy barriers for methyl rotation in the Amber99SB∗-ILDN force field. We demonstrate that this leads to a much more realistic dynamics of methyl groups and is another important feature to close the gap between spectral densities from NMR relaxation and MD simulation.