Predicting protein dynamics from structural ensembles

TL;DR

This paper introduces LE4PD, a simulation-free coarse-grained method that predicts protein dynamics from structural ensembles, achieving high accuracy without extensive molecular dynamics simulations.

Contribution

The paper presents LE4PD, a novel Langevin formalism-based approach that accurately predicts protein dynamics using experimental NMR structures, bypassing computationally intensive simulations.

Findings

LE4PD achieves a correlation coefficient of 0.93 with NMR relaxation data.

The method accurately predicts dynamics across seven different proteins.

Predictions are consistent with molecular dynamics simulation results.

Abstract

The biological properties of proteins are uniquely determined by their structure and dynamics. A protein in solution populates a structural ensemble of metastable configurations around the global fold. From overall rotation to local fluctuations, the dynamics of proteins can cover several orders of magnitude in time scales. We propose a simulation-free coarse-grained approach which utilizes knowledge of the important metastable folded states of the protein to predict the protein dynamics. This approach is based upon the Langevin Equation for Protein Dynamics (LE4PD), a Langevin formalism in the coordinates of the protein backbone. The linear modes of this Langevin formalism organize the fluctuations of the protein, so that more extended dynamical cooperativity relates to increasing energy barriers to mode diffusion. The accuracy of the LE4PD is verified by analyzing the predicted dynamics across a set of seven different proteins for which both relaxation data and NMR solution structures are available. Using experimental NMR conformers as the input structural ensembles, LE4PD predicts quantitatively accurate results, with correlation coefficient \r{ho} = .93 to NMR backbone relaxation measurements for the seven proteins. The NMR solution structure derived ensemble and predicted dynamical relaxation is compared with molecular dy- namics simulation-derived structural ensembles and LE4PD predictions, and are consistent in the timescale of the simulations. The use of the experimental NMR conformers frees the approach from computationally demanding simulations.