Conformational Ensembles from Experimental Data and Computer Simulations

Conformational Ensembles from Experimental Data and Computer Simulations

Poster Abstracts

92-POS Board 12 Refining Binding Free Energies of Docked Complexes by Sampling Configurations During Molecular Dynamics Simulations Ai Shinobu , Kazuhiro Takemura, Akio Kitao. Institute of Molecular and Cellular Biosciences, University of Tokyo, Tokyo, Japan. Molecular docking is a widely used computational tool for predicting protein-protein complex structures. Evaluating relative energies of generated complex models is a crucial step in the process of selecting near-native structures. A method previously developed in our group, evaluates the binding free energy of a complex (ΔG C ) as the sum of its conformational energy and solvation free energy, the latter is calculated using the energy representation method (the ERMOD method) which requires short molecular dynamics (MD) simulations of the initial and final states. Using this method, ΔG C values were calculated for a series of docked complex models, producing lower values for models that resemble the crystal structure. This shows that including an explicit solvation term is important when evaluating energies of complexes. In the current study, we aim to refine relative ΔG C values for a set of protein-protein complex models by subjecting them to all-atom MD simulations in explicit water. Multiple configurations produced by the simulations are carefully selected and used as new configurations for which ΔG C values are recalculated. Our results show that solvation free energies decrease for the configurations collected during MD, the conformational energies increase, to a lesser degree, resulting in an overall decrease in ΔG C . This suggests that proper hydration of the system achieved by MD simulations is a crucial step for generating native-like complex structures. Moreover, the calculated ΔG C values are lower for MD configurations which are closer to the crystal structure, making this procedure useful for selecting the native-like configurations.

129