Implicit treatment of nonpolar solvation revisited

BIOL 34

Jianhan Chen, jianhanc@scripps.edu, Department of Molecular Biology, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037 and Charles L. Brooks III, Molecular Biology, TPC6, Scripps Research Institute, 10550 N. Torrey Pines Rd, La Jolla, CA 92037.
Proper treatment of solvent effects is essential in computational studies of protein structure and dynamics. While accurate and efficient computation of the electrostatic solvation free energy is now possible, particularly by the generalized Born (GB) model, implicit modeling of the nonpolar component has not enjoyed similar progress. It has been primarily estimated from the solvent-exposed surface area (SA) using a single phenomenological surface tension coefficient. More sophisticated treatment might involve decomposition of the nonpolar solvation into a cavity formation term and a solvent-solute dispersion interaction term. However, such a decomposition suffers from limited numerical accuracy and difficulties in parameterization. In this work, we explore a simpler model where the SA model is used in combination with functional group dependent surface tension coefficients. Such models have the advantages of simplicity, transparency and ease of parametrization. Specifically, explicit solvent simulations of amino acid side chain analog hydration free energy, pair-wise interactions and trimer interactions are used for identifying the appropriate parameters in the context of the GB implicit solvent. Furthermore, extensive folding and unfolding simulations of a range of helical peptides and beta hairpins are carried out to maintain a proper balance among polar solvation, nonpolar solvation and intramolecular interactions.