Proteome Mining as a Predicitve Tool of Drug Toxicity

描述（由申请人提供）： 在这项提案中，我们将利用化学蛋白质组学，代谢组学和信息学的独特融合，在发现过程的最早阶段开发一种高度预测药物毒性的检测方法。在实现这一目标的过程中，我们将专门降低药物开发后期更昂贵阶段的铅消耗率。药物发现的传统方法开始于针对单个酶/受体靶标或疾病细胞系的小分子文库的高通量筛选。重要的是，这两种测定都不能预测与其他脱靶蛋白的相互作用，这些脱靶蛋白在药物开发过程的后期更昂贵的阶段，特别是在动物和人类研究中，总是表现为毒性和不良反应。我们的技术称为蛋白质组挖掘，能够在可逆亲和阵列中定量捕获包含数百个潜在药物靶标（包括治疗和毒性靶标）的靶向蛋白质组，并针对药物样分子进行筛选。从该筛选中获得的知识可用于鉴定所有脱靶倾向，并用于驱动迭代化学以同时提高效力和选择性。作为概念验证，我们将利用蛋白质组靶向嘌呤。嘌呤利用酶是人类基因组中最常表达的酶，并且含有经典药物靶标（例如二氢叶酸还原酶和HMG CoA还原酶）以及尖端靶标（蛋白激酶和应激诱导蛋白）。重要的是，对当前OMIM数据库的检查显示，嘌呤利用酶可能也代表了药物毒性基因组的显著比例，其中超过1500种不同的酶被鉴定与代谢中的先天性缺陷相关。为了验证我们的技术，我们将检查几种已建立的药物及其代谢产物的选择性概况，这些药物及其代谢产物在动物和人类中表现出良好的副作用。使用质谱分析中最先进的技术进行的定向代谢组学分析将用于验证在蛋白质组挖掘中鉴定的潜在毒性靶点确实在体内受到抑制。

DESCRIPTION (provided by applicant): In this proposal we will utilize a unique blend of chemoproteomics, metabolomics and informatics to develop a highly predictive assay of drug toxicity at the earliest stages of the discovery process. In achieving this goal we will specifically reduce lead attrition rates that often occur in later more expensive stages of drug development. Traditional approaches to drug discovery begin with high throughput screens of small molecule libraries against single enzyme/receptor target or disease cell line. Importantly, neither assay can predict interactions with other off target proteins that invariable manifest themselves as toxicities and adverse reactions in later more expensive stages of the drug development process, in particular, in animal and human studies. Our technology called proteome mining, enables a targeted proteome containing many hundreds of potential drugs targets (includes therapeutic and toxicity targets) to be quantitatively captured in reversible affinity arrays and screened en masse against drug like molecules. Knowledge derived from this screen can be used to identify all off target liabilities and used to drive iterative chemistry to improve potency and selectivity simultaneously. As a proof of concept we will target the purine utilizing proteome. Purine utilizing enzymes are the most frequently expressed enzymes in the human genome and contain both classical drug targets (e.g. dihydrofolate reductase and HMG CoA reductase) as well as cutting edge targets (protein kinases and stress induced proteins). Importantly, inspection of the current OMIM database reveals that purine utilizing enzymes potentially also represent a significant proportion of the drug toxicity genome, with over 1500 distinct enzymes being identified with associations with inborn errors in metabolism. To validate our technology we will examine the selectivity profiles of several established drugs and their metabolites exhibiting well characterized side effects in animals and humans. Directed metabolomic profiling using state of the art techniques in mass spectrometry will be used to verify that potential toxicity targets identified in proteome mining are indeed inhibited in vivo.