The estimation of the correct binding mode and affinity of a ligand into a target protein using computational methods is challenging. However, docking can introduce poses from which the correct binding mode could be identified using other methods. Here, we analyzed the reliability of binding energy estimation using the molecular mechanics-generalized Born surface area (MMGBSA) method without and with energy minimization to identify the likely ligand binding modes within docking results. MMGBSA workflow (a) outperformed docking in recognizing the correct binding modes of androgen receptor ligands and (b) improved the correlation coefficient of computational and experimental results of rescored docking poses to phosphodiesterase 4B. Combined with stability and atomic distance analysis, MMGBSA helped to (c) identify the binding modes and sites of metabolism of cytochrome P450 2A6 substrates. The standard deviation of estimated binding energy within one simulation was lowered by minimization in all three example cases. Minimization improved the identification of the correct binding modes of androgen receptor ligands. Although only three case studies are shown, the results are analogous and indicate that these behaviors could be generalized. Such identified binding modes could be further used, for example, with free energy perturbation methods to understand binding energetics more accurately.
Keywords: drug discovery; molecular modeling; receptor and ligands (agonist/antagonist); structure-based drug design.
© 2018 John Wiley & Sons A/S.