香蕉皮防治2型DN作用机制的网络药理学研究
Network Pharmacology Study on the Mechanism of Banana Peel in Preventing Type 2 Diabetes Nephropathy

目的：为了系统全面地阐明羽扇豆酮原料药材香蕉皮防治2型DN的潜在分子机制。方法：通过中国知网、维普网、万方数据知识服务平台、Pubmed等文献数据检索香蕉皮的化学成分，根据文献对活性成分防治DN的报道，同时在TCMSP数据库中以口服生物利用度(OB ≥ 30%)和类药性分析(DL ≥ 0.18)参数，筛选香蕉皮的活性成分。从TCMSP、Swiss Target Prediction、PhamMapper数据库中获取活性成分的蛋白靶点，在OMIM、GeneCards数据库检索2型DN相关靶点，通过UniProt数据库将活性成分和疾病靶点转换为基因名称。取活性成分和疾病的交集靶点，利用STRING数据库建立蛋白互作网络(PPI)。在Cytoscape软件中对PPI网络进行分析，根据网络拓扑参数筛选出关键靶点，进行可视化，并建立“成分–靶点–疾病”网络。同时将关键靶点导入到Metascape数据库进行GO及KEGG富集分析，获取香蕉皮防治2型DN的生物过程及信号通路。最后，从PDB数据库获取靶点蛋白晶体结构，利用AutoDock软件对活性成分及靶点蛋白进行分子对接模拟验证，并将结果可视化。结果：通过OB、DL筛选及文献对化学成分防治2型DN的报道，获得9个活性成分分别为槲皮素、儿茶素、β-胡萝卜素、新叶黄素、羽扇豆酮、阿魏酸、芦丁、没食子酸、柚皮苷；并筛选得到共计960个活性成分靶点和1078个疾病靶点，通过取两者的交集共得到179个香蕉皮防治2型DN的潜在作用靶点，绘制得到蛋白互作网络图。在Cytoscape软件中根据网络拓扑参数进一步筛选出INS、ALB、AKT1、IL6、TNF、VEGFA、TP53、EGFR、CASP3、MAPK8、CXCL8、MAPK1等28个关键靶点。GO功能富集分析得到氧化应激、细胞对氧化应激的反应、DNA结合转录因子活性的正调控、细胞对活性氧的反应、细胞对脂多糖的反应等1178个条目，涉及生物过程1102条，细胞组成34条，分子功能45条。KEGG信号通路富集得到AGE-RAGE信号通路、IL-17信号通路、TNF-信号通路、Toll样受体信号通路、FoxO信号通路、MAPK信号通路、HIF-1信号通路、催乳素信号通路等130条信号通路。分子对接结果显示，除化合物芦丁、柚皮苷与靶蛋白ALB结合活性较差外，网络药理学所预测的其他活性成分与关键靶点INS、ALB、AKT1均具有较好的结合活性。结论：运用网络药理学方法初步揭示了香蕉皮防治2型DN的活性成分除羽扇豆酮外，还有多种活性成分在共同发挥作用，表明香蕉皮多成分、多靶点、多途径防治2型DN的整体调节作用特点，为香蕉皮活性成分与实验研究提供科学依据。
Objective: In order to comprehensively elucidate the potential molecular mechanism of banana peel, the raw material of lupenone, in preventing type 2 diabetes nephropathy (DN) systematically. Method: Chemical components of banana peel were retrieved through literature search from CNKI, VIP, Wanfang Data Knowledge Service Platform, Pubmed, etc. The active ingredients of banana peel were screened based on the literature reports on the prevention of DN by active ingredients and parameters such as oral bioavailability (OB ≥ 30%) and drug-likeness (DL ≥ 0.18) in TCMSP database. Protein targets of active ingredients were obtained from TCMSP, Swiss Target Prediction, and PhamMapper databases, and type 2 DN-related targets were searched in OMIM and GeneCards databases. Active ingredients and disease targets were converted into gene names using the UniProt database. The intersection targets of active ingredients and disease were taken, and a protein-protein interaction (PPI) network was established using STRING database. The PPI network was analyzed in Cytoscape software, and key targets were selected based on network topology parameters, visualized, and a “component-target-disease” network was established. Meanwhile, key targets were imported into the Metascape database for GO and KEGG enrichment analysis to obtain biological processes and signaling pathways of banana peel in preventing type 2 DN. Finally, protein crystal structures of