Toward Practical, Rigorous Binding Affinity Calculations

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Ideal, rigorous binding affinity computations, based on statistical mechanics, would fully account for ligand and receptor (protein) flexibility, as well as non-additive effects, which cannot properly be included in faster, more approximate estimates. Such ideal calculations, however, push the limits of present-day computational resources, and therefore have had limited practical impact. Furthermore, rigorous affinity estimates also suffer from errors in the assumed forcefields, which may lead to inaccuracies even when well-designed, well- converged calculations are performed. This proposal aims to take important steps toward overcoming these obstacles, and thus to make rigorous affinity estimation a practical, reliable part of the modeler's toolkit. The issue of computational cost will be addressed with innovative, efficient methods; accuracy will be addressed by the use of both standard and polarizable forcefields; and, lastly, molecular flexibility will be addressed using novel end-point (non- "alchemical") methods and a new conformational sampling scheme. The estrogen receptor is a system which truly embodies all the challenges of affinity calculations, possessing a great diversity of ligands, some of which induce a large receptor conformational change. Beyond its far- reaching clinical importance, the estrogen-receptor is a key model system for understanding binding phenomena in nuclear hormone receptors. Our strategies for improving rigorous affinity calculations will be pursued in the estrogen receptor system, in a series of tasks of increasing complexity. First, estrogen receptor ligands will be studied in solution, then in simplified models of the receptor binding site, and finally the full system will be investigated. With a local experimental collaborator, we will attempt to engineer compounds of potential clinical importance. Successful computational approaches will be implemented in widely available software packages.