基于机器学习和网络药理学的黄芪–莪术治疗肝癌的机制探讨

Based on Machine Learning and Network Pharmacology: Investigation into the Mechanisms of Astragali Radix-Curcumae Rhizoma in Treating Hepatocellular Carcinoma

摘要: 目的：利用网络药理学、机器学习、分子对接和分子动力学模拟探究黄芪–莪术治疗肝癌(HCC)的作用机制。方法：运用TCMSP筛选黄芪–莪术的活性成分，再通过SwissTargetPrediction获取活性成分的靶点，运用Cytoscape3.7.2构建“中药–活性成分–靶点”网络图。通过GeneCards、DrugBank、OMIM获取疾病靶点汇总得到HCC疾病靶点，将疾病靶点与黄芪–莪术活性成分靶点取交集，从而得到黄芪–莪术治疗HCC的潜在靶点。借助String数据库及Cytoscape软件绘制蛋白间相互作用网络，筛选出潜在核心靶点后利用3种机器学习方法最终确定核心基因。使用Metascape对潜在核心靶点进行基因主体(GO)功能富集和京都基因与基因组百科全书(KEGG)信号通路富集分析。采用AutoDock软件进行分子对接，YASARA软件进行分子动力学模拟。结果：收集黄芪–莪术17个活性成分，480个药物活性靶点，肝癌靶点有1576个。将药物–疾病靶点取交集，共有162个交集靶点，构建机器学习模型筛选出HSP90AA、AKT1、XAF、IFI44L、SQLE、SPINK1、ALDH1L1、HPR等为核心交集靶点。GO和KEGG富集分析结果显示黄芪–莪术治疗肝癌主要与炎症、免疫、代谢有关。分子对接结果表明，黄芪–莪术的Bisdemethoxycurcumin与SQLE、Isorhamnetin与HSP90AA1均具有结合能力。分子动力学模拟进一步验证了黄芪–莪术的关键成分与SQLE、HSP90AA1具有良好的结合性。结论：黄芪–莪术可能下调SQLE表达阻断了胆固醇合成依赖的肿瘤生长、HSP90AA1分子伴侣系统等治疗肝癌。

Abstract: Objective: To explore the mechanism of action of Astragali Radix-Curcumae Rhizoma in the treatment of hepatocellular carcinoma (HCC) using network pharmacology, machine learning, molecular docking and molecular dynamics simulation. Methods: The active components of Astragali Radix-Curcumae Rhizoma were screened by TCMSP, and the targets of the active components were obtained through SwissTargetPrediction. The “Chinese medicine-active component-target” network diagram was constructed by Cytoscape3.7.2. The disease targets of HCC were obtained through GeneCards, DrugBank and OMIM. The intersection of the disease targets of HCC and the targets of the active components of Astragali Radix-Curcumae Rhizoma was taken to obtain the potential targets of Astragali Radix-Curcumae Rhizoma in the treatment of HCC. A protein-protein interaction (PPI) network was constructed using the STRING database and visualized with Cytoscape. Following the identification of potential core targets from this network, three machine learning methods were employed to determine the final core genes. Metascape was used to perform gene ontology (GO) functional enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) signaling pathway enrichment analysis on the potential core targets. Molecular docking was performed using AutoDock software, and molecular dynamics simulation was performed using YASARA software. Results: A total of 17 active components and 480 drug targets of Astragali Radix-Curcumae Rhizoma were collected, and there were 1576 HCC targets. The intersection of drug and disease targets was 162, and the machine learning model screened HSP90AA, AKT1, XAF, IFI44L, SQLE, SPINK1, ALDH1L1, HPR, etc. as core intersection targets. GO and KEGG enrichment analysis results showed that the treatment of HCC by Astragali Radix-Curcumae Rhizoma was mainly related to inflammation, immunity and metabolism. Molecular docking results indicated that Bisdemethoxycurcumin of Astragali Radix-Curcumae Rhizoma had binding ability with SQLE, and Isorhamnetin had binding ability with HSP90AA1. Molecular dynamics simulation further verified that the key components of Astragali Radix-Curcumae Rhizoma had good binding with SQLE and HSP90AA1. Conclusion: Astragali Radix-Curcumae Rhizoma may treat HCC by down-regulating the expression of SQLE and blocking cholesterol synthesis-dependent tumor growth and the HSP90AA1 molecular chaperone system.