Modeling Hydrophobic and Hydrophilic Interactions

Columbia Univ New York Morningside, New York NY

Investigators

Linked publications & trials

Abstract

DESCRIPTION (provided by applicant): The importance of hydrophobicity in protein folding is universally recognized. A particularly exciting phenomenon is the sudden large scale de-wetting transition that occurs as two large hydrophobic solutes are brought together. But are biological systems like proteins prone to such drying transitions? We have recently demonstrated that the wild-type Melittin tetramer indeed exhibits a fast de-wetting transition, and that spontaneous drying is sensitive mutation of certain hydrophobic residues. A similar study of the BphC enzyme revealed that the collapse is not induced by a drying transition. In this proposal, we aim to (a) investigate sensitivity to the topology of the hydrophobic groups;(b) identify key sequences associated with drying;(c) design bioinformatics tools to identify protein candidates for drying transitions;(d) investigate whether wild-type proteins are optimized for de-wetting;and (e) design algorithms for speeding up protein folding by reducing the attractive forces between protein and water. Hydrophobic interactions are also expected to be closely tied to enzymatic modulation, which is regulated by the interaction of a small molecule (ligand) and a protein. Initial results indicate that concave binding pockets lead to increased hydrophobicity and, thus, solvated protein-ligand complexes are very sensitive to surface topology. We are currently undertaking a detailed study using all atom molecular dynamics to verify these initial results, which suggest adding terms for hydrophobic enclosure to implicit solvent models. Another major impediment to rational drug design is the lack of realistic force fields. In a detailed QM/MM study, we have recently demonstrated the significance of induced atomic charges when a peptide undergoes conformational changes or moves into different environments. Correcting for this effect was shown to significantly improve the predicted binding affinities of ligands to proteins. We propose to develop a second generation polarizable force field that incorporates induced charges in addition to induced dipoles. This project will integrate two of our existing force fields, which account for fluctuating charges and fluctuating dipoles separately. Perhaps most significantly, it appears that the critical bottleneck in high-resolution protein prediction is the lack of adequate conformational sampling. It is a high priority of this proposal to develop improved sampling methods for biological systems such as proteins in aqueous solution.