Assessing the performance of the MM/PBSA and MM/GBSA methods. 6. Capability to predict protein–protein binding free energies and re-rank binding poses generated by protein–protein docking

Understanding protein-protein interactions (PPIs) is quite important to elucidate crucial biological processes and even design compounds that interfere with PPIs with pharmaceutical significance. Protein-protein docking can afford the atomic structural details of protein-protein complexes, but the accurate prediction of the three-dimensional structures for protein-protein systems is still notoriously difficult due in part to the lack of an ideal scoring function for protein-protein docking. Compared with most scoring functions used in protein-protein docking, the Molecular Mechanics/Generalized Born Surface Area (MM/GBSA) and Molecular Mechanics/Poisson Boltzmann Surface Area (MM/PBSA) methodologies are more theoretically rigorous, but their overall performance for the predictions of binding affinities and binding poses for protein-protein systems has not been systematically evaluated. In this study, we first evaluated the performance of MM/PBSA and MM/GBSA to predict the binding affinities for 46 protein-protein complexes. On the whole, different force fields, solvation models, and interior dielectric constants have obvious impacts on the prediction accuracy of MM/GBSA and MM/PBSA. The MM/GBSA calculations based on the ff02 force field, the GB model developed by Onufriev et al. and a low interior dielectric constant (εin = 1) yield the best correlation between the predicted binding affinities and the experimental data (rp = -0.647), which is better than MM/PBSA (rp = -0.523) and a number of empirical scoring functions used in protein-protein docking (rp = -0.141 to -0.529). Then, we examined the capability of MM/GBSA to identify the possible near-native binding structures from the decoys generated by ZDOCK for 43 protein-protein systems. The results illustrate that the MM/GBSA rescoring has better capability to distinguish the correct binding structures from the decoys than the ZDOCK scoring. Besides, the optimal interior dielectric constant of MM/GBSA for re-ranking docking poses may be determined by analyzing the characteristics of protein-protein binding interfaces. Considering the relatively high prediction accuracy and low computational cost, MM/GBSA may be a good choice for predicting the binding affinities and identifying correct binding structures for protein-protein systems.