Molecular dynamics (MD) simulations of proteins cannot efficiently equilibrate water occupancy in buried cavities, leading to slow convergence of thermodynamic averages. This becomes particularly problematic for protein-ligand binding free energy calculations when the ligand is in a buried site, as standard MD cannot account for changes in hydration when a ligand is either changed alchemically or fully decoupled from the site. Several prior works have addressed this problem by integrating grand canonical Monte Carlo (MC) steps with MD simulations, so that water molecules can be added to, or removed from, random locations in the system. However, the grand canonical method adds algorithmic complexity, particularly as it leads to changes in the numbers of water molecules in the course of a simulation.<br/>We have addressed this challenge by integrating standard MC translational water moves with MD. These moves allow water to exchange with buried sites during a simulation. For efficiency, the algorithm avoids moves within the bulk, instead focusing on ones that can transfer water between bulk and the protein interior. A steric grid allows avoidance of moves that would lead to obvious steric clashes, and a fast energy calculation is used as a pre-filter to reduce the number of expensive full energy calculations.<br/>We tested this new MD/MC method as follows. First, we showed that the potential energy distributions it generates match those from control simulations using only MD. Second, we used MC/MD for hydrating and dehydrating buried cavities. The first test system for this purpose contains a “wall” of carbons preventing water from moving between two regions of the simulation box. While MD cannot equilibrate this system, MC/MD correctly equalizes the densities of water across the two regions. A second test system contained a frustrated water molecule trapped inside a buckyball. Using MC/MD, the water went back to the surrounding bulk water. In addition, MC/MD was applied to two biological systems, the Major Urinary Protein and Cytochrome c Peroxidase. For both systems, a known buried cavity was correctly hydrated using MC/MD, while MD failed to sample this hydration site.<br/>We anticipate that a key application of this technology will be to accelerate protein-ligand binding free energy calculations involving buried binding sites. The method will be available in the widely disseminated AMBER18 simulation package.

Simulating water exchange to buried binding sites

2019 Spring National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.