Pharmacophore- based virtual screening, 3D- QSAR, molecular docking approach for identification of potential dipeptidyl peptidase IV inhibitors

Pharmacophore modeling, molecular docking, and in silico ADME studies have been carried out to determine the binding mode and drug likeliness profile of Pyrrolidine derivatives as Dipeptidyl peptidase IV inhibitors. A five point pharmacophore model (AAADH_1) was generated using 96 compounds with IC₅₀ values ranging from 1.8 to 13000 nM. A statistically significant 3D-QSAR model was generated from the pharmacophore hypothesis. The model had a high correlation coefficient (R² - 0.92), cross validation coefficient (Q² - 0.776) and F value (F - 144) at 6 component Partial least square factor. Pearson r of 0.7124 indicated greater degree of confidence on the model. The accuracy and predictability of the generated model were checked by Enrichment factor, Receiver operating characteristic curves, area under curve, Boltzmann-enhanced discrimination of Receiver operating characteristic and the Robust initial enhancement. To identify novel and potent Dipeptidyl peptidase IV inhibitors, virtual screening was performed using the ligand and database screening. Considering the potent hit molecules on the basis of pharmacophore virtual screening, we have designed new molecules and further subjected to see the interaction with protein. The catalytic domain of Dipeptidyl peptidase IV enzyme in complex with Vildagliptin (PDB Code: 6B1E) was obtained from protein data bank with resolution 1.77 A°. Compound 75 showed the better binding (dock score -7.966) with protein than standard drug vildagliptin (Dock Score -6.554). The hits obtained on virtual screening of the database have provided new chemical starting points for design and development of novel Dipeptidyl peptidase IV inhibitory agents. Communicated by Ramaswamy H. Sarma.

本研究通过药效团建模（Pharmacophore modeling）、分子对接（molecular docking）与计算机辅助ADME研究（in silico ADME studies），系统探究了吡咯烷衍生物作为二肽基肽酶IV（Dipeptidyl peptidase IV, DPP-4）抑制剂的结合模式与成药性特征。研究以96个半最大抑制浓度（IC₅₀）范围为1.8~13000 nM的化合物，构建了含五个药效特征的药效团模型AAADH_1。基于该药效团假说，我们构建了统计学意义显著的三维定量构效关系（3D-QSAR）模型；该模型在6个偏最小二乘因子下，展现出较高的相关系数（R²=0.92）、交叉验证系数（Q²=0.776）及F值（F=144）；皮尔逊相关系数r为0.7124，表明该模型具有较高的可信度。本研究通过富集因子（Enrichment factor）、受试者工作特征曲线（Receiver operating characteristic curves, ROC）、曲线下面积（area under curve, AUC）、ROC曲线玻尔兹曼增强判别法（Boltzmann-enhanced discrimination of Receiver operating characteristic）及稳健初始增强法（Robust initial enhancement），对所构建模型的准确性与可预测性进行了验证。为筛选新型强效DPP-4抑制剂，本研究采用配体虚拟筛选与数据库虚拟筛选相结合的方式开展虚拟筛选。基于药效团虚拟筛选得到的高活性命中分子，我们设计了全新的化合物，并进一步考察其与靶蛋白的相互作用。本研究从蛋白质数据库（Protein Data Bank, PDB）获取了与维格列汀（vildagliptin）共晶的DPP-4酶催化结构域，该复合物的PDB码为6B1E，其分辨率为1.77埃。分子对接结果显示，化合物75与靶蛋白的结合得分为-7.966，优于阳性对照药维格列汀的对接得分（-6.554）。本研究通过数据库虚拟筛选得到的命中化合物，为新型DPP-4抑制剂的设计与开发提供了全新的化学起点。本文由Ramaswamy H. Sarma通讯。