Huang Ruili, Wallqvist Anders, Covell David G
Developmental Therapeutics Program, Screening Technologies Branch, Laboratory of Computational Technologies, National Cancer Institute-Frederick, Frederick, Maryland 21702, USA.
J Med Chem. 2006 Mar 23;49(6):1964-79. doi: 10.1021/jm051029m.
This paper examines two biological models of anticancer activity, cytotoxicity and hollow fiber (HF) activity, for chemotherapeutic agents evaluated as part of the National Cancer Institute's (NCI's) drug screening effort. Our analysis proposes strategies to globally assess compounds tested in the NCI's 60-cell (NCI60) in vitro anticancer screen in terms of structural features, biological activity, target specificity, and mechanism of action by data integration via our self-organizing maps of structural and biological response patterns. We have built statistical models to predict compound potency and HF activity based on physicochemical properties. Our results find that it is the combination of different structural properties that determines a compound's biological activity. A direct correlation is also found between compound potency and specificity, indicating that specific targeting, rather than promiscuous poisoning, gives rise to potency. Finally, we offer a strategy to exploit this relationship for future mining of novel anticancer candidates.
本文研究了作为美国国立癌症研究所（NCI）药物筛选工作一部分进行评估的化疗药物的两种抗癌活性生物学模型，即细胞毒性和中空纤维（HF）活性。我们的分析提出了通过我们的结构和生物反应模式自组织图进行数据整合，从结构特征、生物活性、靶标特异性和作用机制等方面对在NCI的60细胞（NCI60）体外抗癌筛选中测试的化合物进行全局评估的策略。我们建立了基于物理化学性质预测化合物效力和HF活性的统计模型。我们的结果发现，是不同结构性质的组合决定了化合物的生物活性。还发现化合物效力与特异性之间存在直接相关性，表明特异性靶向而非滥杀滥伤性中毒产生了效力。最后，我们提供了一种利用这种关系来挖掘新型抗癌候选物的策略。
