The Molecular Mechanics Poisson-Boltzmann Surface Area (MM/PBSA) methods have been widely used in calculating protein-ligand binding affinities. However, further application of MM/PBSA method has been hindered due to its limitations in efficiency, accuracy and applicability. Here we report our efforts in the improvement and extension of the MM/PBSA method in several aspects. First we extended the linear PB solvers with periodic boundary conditions so they can be applied to systems that require periodic setup, such as membrane protein-ligand binding systems. We also implemented GPU-accelerated linear PB solvers utilizing Nvidia<sup>®</sup> CUDA to seek more efficiency and obtained a speedup ratio up to 50. Further development of GPU-accelerated MM/PBSA is also in progress. In addition, we made a thorough analysis over the use of different nonpolar solvation models, atomic radius sets, and solute dielectric constants, and the influence of numerical setup to assess their impacts to the agreement of MM/PBSA calculations with experiment. Finally we developed a new continuum membrane model with its location and thickness fine-tuned based on explicit-solvent MD simulations. Assisted with a two-step-two-probe grid labeling procedure to address the highly different accessibility in the membrane and water regions, the new continuum membrane model was able to identify water pores/channels automatically. The newly extended MM/PBSA method was applied to several membrane protein systems such as the human purinergic platelet receptor and human potassium channel as test cases, and was found to agree with experiment with standard MM/PBSA parameters.

Calculating protein-ligand binding affinities with MM/PBSA: Improvement and extension

254th National Meeting (2017)

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