基于蛋白质功能位点相似性的药物重定位新方法的开发及应用

Since the R&D investment of new drugs has been increasing year by year and the risk of failure has also been increasing year by year, drug repositioning (or new uses for old drugs) has become a very important and adopted drug discovering strategy by both pharmaceutical companies and research institutes due to its drastically reduced R&D costs but increased success rate. Target prediction based on protein similarity has achieved a great success in predicting new indications of drugs as one of the core technologies of drug repositioning based on target-disease association. However, considering only the similarity of 1D sequence or 3D structure of proteins, it is easy to overlook targets that are not identical in sequence or fold, but share similar or even the same small molecule recognition sites. Therefore, the development of a method that can accurately predict the similarity of protein functional sites will provide an advantageous complementary tool for drug repositioning research based on target similarity. Although several related methods have been reported, the accuracy and speed of calculation have severely limited their application. In this study, we are aimed to develop an accurate and rapid method to evaluate the similarity of protein binding sites, which is sequence order-independent and also not dependent on the homology of 3D structures, and apply it to the actual drug repositioning research. The completion of this study will provide an effective tool for structure-based drug repositioning.

原创新药的研发投入逐年增高且失败风险逐年增大，因此，药物重定位（或老药新用）凭借其大幅度缩减的研发成本已经成为国内外制药企业及科研院所非常重视并采用的药物研发策略。基于蛋白质相似性的靶标预测作为基于靶标-疾病关联的药物重定位研究的核心技术之一，已在预测药物新适应症上取得了很大成功。然而，只考虑蛋白质一级序列或三维结构的相似性易忽略在序列或者结构上不同源，但却享有相似甚至相同小分子识别位点的靶标。因此，发展能准确预测蛋白质结合位点相似性的方法将为基于靶标相似性的药物重定位研究提供有利的补充工具。目前虽已有相关方法报道，但计算精度及计算速度严重限制了其发展和应用。本研究拟开发不依赖蛋白质一级序列和三维结构同源性的准确、快速的结合位点评价方法，并将其应用于实际的药物重定位研究之中。本研究的完成将为基于结构的药物重定位研究提供有效工具。

原创新药的研发投入逐年增高且失败风险逐年增大，因此，药物重定位（或老药新用）凭借其大幅度缩减的研发成本已经成为国内外制药企业及科研院所非常重视并采用的药物研发策略。基于蛋白质相似性的靶标预测是药物重定位研究的核心技术之一，然而，只考虑蛋白质一级序列或三维结构的相似性易忽略在序列或者结构上不同源，但却享有相似甚至相同小分子识别位点的靶标。基于本项目的资助，我们发展了一套全新的基于叠合的快速、准确的结合口袋三维相似性评价方法PocketShape，该方法打分函数的设计充分考虑了结合口袋骨架的几何形状、侧链的取向以及残基的理化性质，具有不依赖结合口袋序列计算的特点。PocketShape能高度识别相似性口袋和非相似性口袋（准确率>99%）；能准确地将含有83个酶结构的数据集分成12大类，且分类结果与真实的SCOP蛋白质家族分类高度一致；在基于结合口袋的反向虚拟筛选方面表现出色，在多个体系上都获得比其他方法高的真阳性富集率。PocketShape单次计算的运行时间达到秒级，计算精度和计算速度明显优于同类方法Trixp，方便进行大规模的结合口袋相似性比较。使用PocketShape开展应用研究，为分别靶向DHODH、DPP4、新冠病毒3CLpro和新冠病毒RdRp的多个候选药物分子预测新适应症。综上，PocketShape在基于蛋白质结合位点相似性的蛋白质功能分类及药物重定位研究方面表现出良好的应用前景。.基于本课题的资助，相关成果共发表SCI论文8篇，包括中科院一区论文3篇，二区论文4篇，发表在Acta Pharm. Sin. B（1篇）, J. Med. Chem.（2篇）等杂志上；获软件著作权1项；共申请发明专利2项，其中PCT专利2项，1项在国外6个国家和地区获得授权。

内容获取失败，请点击重试