MM/PB(GB)SA benchmarks on soluble proteins and membrane proteins

Shiyu Wang; Xiaolin Sun; Wenqiang Cui; Shuguang Yuan

doi:10.3389/fphar.2022.1018351

MM/PB(GB)SA benchmarks on soluble proteins and membrane proteins

Front Pharmacol. 2022 Dec 1:13:1018351. doi: 10.3389/fphar.2022.1018351. eCollection 2022.

Authors

Shiyu Wang^{1

2

3}, Xiaolin Sun^{1

3}, Wenqiang Cui^{1

3}, Shuguang Yuan^{1

3

4}

Affiliations

¹ Research Center for Computer-Aided Drug Discovery, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China.
² College of Chemical Science, University of Chinese Academy of Sciences, Beijing, China.
³ AlphaMol-SIAT Joint Laboratory, Shenzhen, China.
⁴ Faculty of Chemistry, University of Warsaw, Warsaw, Poland.

Abstract

Predicting protein-ligand binding free energy rapidly and accurately remains a challenging question in modern drug discovery. Molecular mechanics/Poisson-Boltzmann (Generalized Born) surface area (MM/PB(GB)SA) has emerged as an essential tool for accelerating cost-efficient binding free energy calculation. This study presents benchmarks with three membrane-bound protein systems and six soluble protein systems. Different parameters were sampled for different benchmarks to explore the highest accuracy. These include ligand charges, protein force fields, extra points, GB models, nonpolar optimization methods, internal dielectric constants and membrane dielectric constants. Comparisons of accuracy were made between MM/PB(GB)SA, docking and free energy perturbation (FEP). The results reveal a competitive performance between MM/PB(GB)SA and FEP. In summary, MM/PB(GB)SA is a powerful approach to predict ligand binding free energy rapidly and accurately. Parameters of MM/PB(GB)SA calculations, such as the GB models and membrane dielectric constants, need to be optimized for different systems. This method can be served as a powerful tool for drug design.

Keywords: FEP; MM/PB(GB)SA; binding energy; membrane protein; soluble protein.