The interactions that govern the behavior of biomolecules in crowded environments range from repulsive volume exclusions to non-specific transient binding and molecular recognition. Entropic volume exclusion effects and their impact on biomolecular behavior are reasonably well understood. However, the interactions that result in transient binding or molecular recognition include electrostatics, hydrogen bonds, van der Waals forces and solvent-mediated interactions. Each interaction type features a unique angular and distance dependence, which impedes the development of general theories. Instead, computer simulations can be used to study individual systems but models with a high level of detail, e.g., all-atom simulations in an explicit solvent, are associated with significant computational costs.
Here, we aim to complement fully atomistic simulations with a multiscale modeling approach that uses all-atom simulations of individual biomolecules to construct detailed protein-protein interaction models. We then use these interaction models in computationally efficient meso-scale simulations of complex biomolecular environments.
Our work is focused on solvent-mediated interactions that describe changes in the solvation free energy as a function of biomolecular coordinates. We construct solvent-mediated interaction potentials from spatially resolved solvation free energy maps obtained from all-atom simulations with a previously developed method (3D-2PT). These potentials are then combined with continuum electrostatics and Lennard-Jones interactions to enable simulations on the meso-scale using a previously developed multi-conformation Monte Carlo (mcMC) algorithm. Our computationally efficient meso-scale simulation models allow us to study the impact of critical parameters such as system composition, packing density and ionic strength on biomolecular processes, which is currently not feasible with all-atom simulations in explicit solvents.

Matthias Heyden

Arizona State University

Appearances