Physics-based computational framework for absolute binding affinity estimation

by Kumar, Vivek Govind; Moradi, Mahmoud

The protein-ligand binding affinity quantifies the binding strength between a protein and its ligand. Computer modeling and simulations can be used to estimate the binding affinity or binding free energy using data- or physics-driven methods or a combination thereof. Here, we discuss a purely physics-based sampling and anal. framework based on mol. dynamics (MD) simulations. The novel framework presented here has similarities to the stratification strategies previously discussed in the literature. Stratification-based binding free energy estimators generally restrain different rotational and translational degrees of freedom of both the ligand and protein in order to reach a faster convergence. Our proposed methodol. uses bias-exchange umbrella sampling (BEUS) simulations with a minimal use of restraining or without any addnl. restraints to accurately estimate the absolute binding free energies. We instead use nonparametric reweighting to estimate the corrections needed to compensate for the differential sampling introduced by not-fully-converged BEUS simulations. Our novel strategy presented here uses a simplified and general scheme that can be easily tailored for any system of interest with different levels of protein/ligand conformational flexibility. We estimate the binding affinity of human fibroblast growth factor 1 (hFGF1) to heparin hexasaccharide based on the available crystal structure of the complex as the initial model. We use five different versions of the proposed generalized method to compare against the exptl. determined binding affinity obtained from isothermal calorimetry (ITC) experiments Our results indicate that all five versions of the method predict the absolute binding affinity reasonably with varying degrees of accuracy. One may also examine the relative importance of different degrees of freedom in sampling. It turns out that taking into account the sampling of the orientation of the ligand plays a crucial role in reaching an accurate estimate for absolute binding affinity. The general approach presented here provides a flexible scheme for designing practical and accurate binding free energy estimation methods for proteins and ligands of varying flexibility within a purely physics-based computational scheme.