抗肿瘤药物设计虚拟筛选新算法和新流程研究

Virtual screening has been used in large-scale discovery of active compounds, and is one of the core technologies of innovative drug research. It can be divided into two categories: one is the molecular docking-based virtual screening, and the other is the ligand-based virtual screening which includes shape analysis and pharmacophore modeling etc. Previous work shows that the hit rate of target-based virtual screening largely depends on the quality of the receptor structure. Besides, although the design of in vitro inhibitors of a tumor protein from virtual database screening can reach a high success rate, a part of these anti-tumor leads could fail to inhibit the targeted protein in cells. Therefore, it is critical to reversely find the off-targets of these anti-tumor agents in vivo by using ligang-based screening. Moreover, the active compounds can also be further optimized to improve their biological activity and creativity. Here, we continue to develop and improve a new effective algorithm to determine and optimize the best receptor structures for target-based screening. The initial work of this algorithm has been recently published in the highly esteemed journal in chemistry (Journal of Chemical Information and Modeling). Furthermore, we plan to establish an efficient workflow process for drug discovery by using ensemble docking based on this algorithm, shape analysis, pharmacophore modeling, reverse-docking methods, scaffold hopping and structural optimization, combined with experimental tests. Finally, we will apply this new workflow to design the innovative anti-cancer small molecule inhibitors of the abnormal high expression of proteins such as RSK2, JNK and FYN from tumor cells and cancer stem cells. The conduction of this research project has the important social significance and application value in developing the China's homegrown anti-cancer drugs and breaking the monopolies of foreign pharmaceutical companies on cancer treatment.

虚拟筛选被大规模应用于活性化合物的发现，是创新药物研究的核心技术之一。它可分为基于分子对接的靶向虚拟筛选，以及包括形状分析、药效基团建模等方法在内的基于配体结构的虚拟筛选。前期工作表明靶向虚拟筛选的命中率高低取决于所用受体结构的质量。其次，通过靶向筛选的抗肿瘤抑制剂，可能在细胞内对靶标蛋白失活，因而需要基于配体结构筛选来逆向寻找这些抗癌剂的真实靶标。再次，对活性化合物还可进行结构优化设计以提高其生物活性和创新性。这里，我们在自己提出的靶向筛选受体结构高效优选新算法基础上应用系综对接，并结合形状分析、药效团建模等靶标逆向探查技术，以及化合物结构修饰改造和骨架跃迁等优化手段，拟建立一套高效的药物发现计算机辅助虚拟筛选和实验相配套的测试新流程。最后，将其应用到RSK2, JNK和FYN等肿瘤蛋白的抗癌小分子药物设计中去，为我国开发具有自主知识产权结构的一类抗癌新药，具有重要的社会意义和应用价值。

计算机虚拟筛选在发现和筛选药物活性化合物研究中具有重要意义，本项目的主要研究内容是应用系综对接，并结合形状分析和药效基团建模等基于配体相似性筛选方法，以及骨架跃迁和结构修饰改造等优化设计，建立一套高效的药物发现计算机辅助虚拟筛选和实验相配套的测试流程，并将其应用到肿瘤细胞和肿瘤干细胞异常高表达蛋白的抗癌小分子药物的设计中去。首先，我们对计算机辅助药物设计整个流程的计算方法、现有数据库和应用实例进行了全面的梳理，筛选了一些关键的软件和数据库，初步整理了一套从药靶发现、活性位点识别和验证，靶向虚拟筛选，计算钓靶和筛选先导化合物ADMET（药物的吸收，分配，代谢，排泄和毒性）预测的高效计算机辅助虚拟筛选流程，并发表了三篇高影响的系统性综述。其次，我们开展了靶向肿瘤细胞和肿瘤干细胞异常高表达蛋白的计算机辅助的抗癌小分子药物设计，并对筛选出来的先导化合开展了肿瘤细胞细胞毒活性、侵袭转移、周期凋亡等影响的体内外实验测试验证，发现了四个小分子抑制剂：靶向P90核糖体S6激酶2（RSK2）的小分子抑制剂补骨脂乙素，靶向肿瘤坏死因子受体相关因子6（TRAF6）的小分子抑制剂Bis(4-hydroxy-3,5-dimethylphenyl)sulfone，靶向蛋白酪氨酸激酶FYN原癌基因的小分子抑制剂阿莫地喹和靶向M2亚型糖酵解丙酮酸激酶PKM2的小分子抑制剂羟苄丝肼。再次，我们运用数据挖掘和生物信息学分析技术，针对宫颈癌、结肠癌和胰腺癌这些常见癌症患者临床样本和基因芯片数据，建立了可有效预测癌症预后miRNA模型和关键基因，为这些癌症的诊疗提供新的生物标志物和药物治疗靶点。我们共发表标注该课题项目基金号的国际学术论文22篇（其中影响因子超过3分的SCI期刊论文有17篇，超过5分的有9篇），申请专利7项，培养5名中青年教师、13名研究生和百余名创新科研本科生，主编国际知名SCI期刊2期，完成了预期的研究目标。