Adaptive collective motions: a hybrid method to improve conformational sampling with molecular dynamics and normal modes

Protein function is closely related to its structure and dynamics. Due to its large number of degrees of freedom, proteins adopt a large number of conformations, which describe a highly complex potential energy landscape. Considering the huge ensemble of conformations in dynamic equilibrium in solution, detailed investigation of proteins dynamics is extremely costly. Therefore, a significant number of different methods have emerged in order to improve the conformational sampling of biomolecules. One of these methods is Molecular Dynamics with excited Normal Modes (MDeNM) in which normal modes are used as collective variables in molecular dynamics. Here, we present a new implementation of the MDeNM method that allows a continuously controlled kinetic excitation energy in the normal mode space, while taking into account the natural constraints imposed either by the structure or the environment. These implementations prevent unphysical structural distortions. We tested the new approach on bacteriophages T4 lysozyme, Gallus gallus hen egg-white lysozyme and Staphylococcus aureus membrane-bound transglycosylase. Our results showed that the new approach outperformed free MD sampling and preserved the structural features comparatively to the original MDeNM approach. We also observed that by adaptively changing the excitation direction during calculations, proteins follow new transition paths preventing structural distortions.