Allostery is a ubiquitous process allows for protein function changes due to ligand binding. There are two complementary protein allosteric models: conformation-driven and dynamics-driven allostery. Developing allosteric drugs targeting key proteins in signaling pathways is becoming important revenues in drug discovery. However, the understanding of protein allosteric mechanisms is the major bottleneck in this effort. To make breakthrough, fully understanding allostery that is highly correlated with protein dynamics is inevitable. Recently, we developed a novel simulation method, named as rigid residue scanning (RRS), to systematically identify key residues for allostery. In this study, we employed RRS method to explore the conformation-driven allosteric mechanism of Vivid (VVD) protein from Light-Oxygen-Voltage (LOV) domains in circadian clock system. In RRS method, an efficient rigid body dynamics integrator was applied to carry out MD simulations to systematically probe the effect of removing individual residue internal degrees of freedom on the whole protein dynamics. It is revealed that the rigid residue perturbation caused VVD to reach hidden conformational space, which could relate to the regulation of VVD allostery and function. As a separate attempt, we applied two machine learning methods, decision tree and neural network models to build prediction models for dynamics-driven allostery of protein PDZ2. Both machine learning models showed promising results as differentiating structurally similar but allosterically different bound and unbound states of PDZ2.<br/>

Novel computational methods to explore protein allostery: Rigid residue scan and machine-learning methods

255th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.