Molecular Recognition of Proteins and Ligand Design

描述(由申请人提供)：该研究的主要目标是开发和使用计算工具来对蛋白质的结构、活性和抑制进行定量预测。针对癌症、关节炎、细菌感染、睡眠障碍和帕金森氏症的具体应用，人们正在为药物设计开发复杂的软件。这项研究包括在蛋白质建模方法和软件开发方面的基本进展，以及在蛋白质-配体结合、抑制剂设计、蛋白质折叠和酶反应方面的应用。原子水平的计算机模拟是三个复杂性水平的主要工具：使用GenMol程序的评分函数方法，该程序可以快速建立蛋白质-配体复合体的组合库，以及使用蒙特卡罗统计力学(MC)进行扩展线性响应(ELR)和严格的自由能微扰(FEP)计算。MC模拟是在显式水分子存在的情况下对蛋白质-配体复合体进行的，包括对体系的所有自由度进行采样。由此产生的详细的结构和能量信息有助于解释当配体的结构或蛋白质序列被修改时结合亲和力的变化。反过来，这些知识形成了设计高亲和力、蛋白质选择性配体的基础。为了改进方法的预测能力，训练和测试将使用具有模拟系列已知活性数据的约2000个蛋白质-配体复合体的数据库。蛋白质-配基结合和药物-前导优化的具体广泛研究针对几种蛋白质，包括CDK2激酶、环氧合酶、DNA旋转酶和脂肪酸酰胺水解酶。通过应用QikProp程序改进了药物设计，该程序通过评估药物相关属性(包括水的溶解性、细胞的渗透性、血脑屏障的渗透性和血清蛋白结合)来分析输入有机分子的药物相似性。其他活动包括发展改进的“力场”来描述分子内和分子间的能量学，研究酶反应的机理，以及使用新的MC采样方法和连续介质溶剂模型预测多肽在水溶液中的结构。

DESCRIPTION (provided by applicant): The primary goal of the research is to develop and use computational tools to make quantitative predictions on the structure, reactivity and inhibition of proteins. Sophisticated software is being created for drug design with specific applications to cancer, arthritis, bacterial infections, sleep disorders, and Parkinson's disease. The research includes fundamental advances in the development of methods and software for modeling proteins and features applications on protein-ligand binding, inhibitor design, protein folding, and enzymatic reactions. Atomic-level computer simulations are the primary tool at three levels of complexity: a scoring function approach with the GenMol program, which can rapidly build combinatorial libraries of protein-ligand complexes, and extended linear response (ELR) and rigorous free-energy perturbation (FEP) calculations using Monte Carlo statistical mechanics (MC). The MC simulations are performed for the protein-ligand complexes in the presence of explicit water molecules and involve sampling of all degrees of freedom for the systems. The resultant detailed structural and energetic information helps elucidate variations in binding affinities as either the structure of the ligand or the protein sequence is modified. In turn, this knowledge forms the basis for the design of high-affinity, protein-selective ligands. In order to refine the predictive abilities of the methods, training and testing will use a database of ca. 2000 protein-ligand complexes with known activity data for analog series. Specific extensive studies of protein-ligand binding and drug-lead optimization are targeted for several proteins including CDK2 kinase, cyclooxygenases, DNA gyrase, and fatty acid amide hydrolase. The drug design is enhanced by application of the QikProp program, which analyzes the drug-likeness of input organic molecules through estimation of pharmaceutically relevant properties including aqueous solubility, cell permeabilities, blood-brain barrier permeability, and serum protein binding. Other activities are the development of an improved "force field" for the description of intra and inter-molecular energetics, examination of the mechanisms of enzymatic reactions, and prediction of the structures of polypeptides in aqueous solution using new MC sampling methods with a continuum solvent model.