Flexible framework of computing binding free energy using the energy representation theory of solution

Abstract

Host-guest binding plays a crucial role in the functionality of various systems, and its efficiency is often quantified using the binding free energy, which represents the free-energy difference between the bound and dissociated states. Here, we propose a methodology to compute the binding free energy based on the energy representation (ER) theory of solution that enables us to evaluate the free-energy difference between the systems of interest with the molecular dynamics (MD) simulations. Unlike the other free-energy methods, such as the Bennett acceptance ratio (BAR), the ER theory does not require the MD simulations for hypothetical intermediate states connecting the systems of interest, leading to reduced computational costs. By constructing the thermodynamic cycle of the binding process that is suitable for the ER theory, a robust calculation of the binding free energy is realized. We apply the present method to the self-association of \textit{N}-methylacetamide (NMA) in different solvents and the binding of aspirin to $β$-cyclodextrin (CD) in water. In the former case, the present method estimates that the binding free energy decreases as the solvent polarity decreases. This trend is consistent with the experimental finding. For the latter system, the binding free energies for the two representative CD-aspirin bound complexes, primary (P) and secondary (S) complexes, are estimated to be $-5.2\pm 0.1$ and $-5.03\pm 0.09~\mathrm{kcalmol^{-1}}$, respectively. These values are satisfactorily close to those from the BAR method [$-4.2\pm 0.2$ and $-4.1\pm 0.2\mathrm{kcal~mol^{-1}}$ for P and S, respectively]. Furthermore, the interaction-energy component analysis reveals that the van der Waals interaction between aspirin and CD dominantly contributes to the stabilization of the bound complexes, that is in harmony with the well-known binding mechanism in the CD systems.