UC San Diego

Implicit solvation provides a means of accelerating and improving the efficiency of computational biomolecular studies by eliminating explicit solvent degrees of freedom while still representing the effects of solvation. Evidence is provided supporting the importance of defining the implicit solvent-solute boundary such that solvent is excluded from spaces smaller than a water molecule. The pairwise analytical generalized Born (GB) model, a popular implicit solvent model, is extended to incorporate this property. Methods for conducting molecular dynamics simulations at a constant pH, rather than the traditional constant protonation state, are reviewed and a constant pH method employing a consistent GB-based Hamiltonian for conformational and protonation state sampling is developed. Even with the improved efficiency of implicit solvent, it is difficult to achieve sufficient sampling in molecular dynamics. This problem is addressed by accelerated molecular dynamics, a technique for accelerating sampling that requires no advance knowledge of the potential energy landscape is presented. Analysis of molecular dynamics data is aided by Interactive Essential Dynamics, a tool for visualization of principal component analysis results. Implicit solvent methods are applied to the computer-aided design of inhibitors for the zinc(II) proteases stromelysin-1 and anthrax lethal factor. Inhibitors with IC-50 of 100 nM and 14 micromolar are reported for stromelysin-1 and lethal factor, respectively. Use of the GB model developed here allows for accurate elucidation of the binding mode of the lethal factor inhibitor, while GB models that allow solvent in spaces smaller than a water molecule identify an incorrect binding mode