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
DOI: 10.12677/pi.2025.146045, PDF, HTML, XML,    科研立项经费支持
作者: 黄敏洁*, 黄燕秋, 凌梦怡, 詹泽贤:广东江门中医药职业学院南药学院,广东 江门;方镕泽#:贵州中医药大学基础中医学院,贵州 贵阳
关键词: 黄芪–莪术肝癌机器学习网络药理学分子对接分子动力学模拟Astragali Radix-Curcumae Rhizoma Hepatocellular Carcinoma Machine Learning Network Pharmacology Molecular Docking Molecular Dynamics Simulation
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.
文章引用:黄敏洁, 黄燕秋, 凌梦怡, 詹泽贤, 方镕泽. 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[J]. 药物资讯, 2025, 14(6): 396-411. https://doi.org/10.12677/pi.2025.146045

1. 引言

肝癌是全球范围内高发的恶性肿瘤,分为肝细胞肝癌(hepatocellular carcinoma, HCC)、肝内胆管细胞肝癌(intrahepatic cholangiocarcinoma, ICC)以及混合型肝癌。其中,HCC占原发性肝癌病例的90% [1],HCC具有高度侵袭性和高死亡率的特点,在全球癌症发病率中排名第六,是全球癌症相关死亡的第三大原因[2] [3]。我国的新发HCC病例占全球近一半,对公共卫生构成重大挑战[1]。临床上治疗手段主要包括手术切除、肝移植、局部治疗如射频消融、经动脉化疗栓塞以及系统性治疗如靶向治疗、免疫治疗等[4]。然而这些治疗手段的疗效有限,如化疗药物毒副作用大,患者耐受性差,且易产生耐药性。靶向治疗虽能精准作用于特定靶点,但部分患者因肿瘤基因变异复杂,治疗效果欠佳,免疫治疗的响应率也有待进一步提高。与此同时,治疗费用高昂,给患者家庭和社会带来沉重经济负担。

在传统医学领域,黄芪–莪术的配伍在肿瘤治疗中具有较好的疗效。黄芪(Astragali Radix)性甘温,归脾肺经,具有补气升阳、益卫固表的作用;莪术(Curcumae Rhizoma)性辛温,归肝脾经,具有破血行气、消积止痛的功效。黄芪–莪术配伍源自张锡纯所著《医学衷中参西录》中的理冲汤,用药比例为1∶1 [5]。中医认为肝癌发病多因正气亏虚、瘀血凝滞,黄芪补气扶正,莪术破血消癥,二者配伍,相得益彰,起到破瘀不伤正、行气不留瘀的作用[6]。现代研究表明,黄芪–莪术药对及其活性成分在抗肝癌方面具有显著效果。黄芪中的黄芪甲苷可降低细胞表面PD-L1水平,发挥抗HCC作用,并通过miR-135b-5p/CNDP1途径消除免疫抑制。莪术中的莪术醇则能通过诱导细胞凋亡和周期阻滞,抑制肝癌细胞增殖和转移[7]。临床研究表明,黄芪–莪术药对能显著提高HCC患者术后生存率(HR = 0.73, 95% CI 0.61~0.87) [8],但黄芪–莪术药对治疗HCC的作用机制尚未明确。

基于此,本文聚焦于黄芪–莪术治疗肝细胞癌的机制探索,通过机器学习、网络药理学、分子对接、分子动力学模拟等方法全面阐释其分子机制,深入探讨黄芪–莪术治疗肝细胞癌的潜在机制,为肝细胞癌的治疗提供新思路和新策略。

2. 材料和方法

2.1. 黄芪–莪术的主要成分和作用靶点的筛选

在中医药系统药理学技术平台(TCMSP, http://tcmspw.com/tcmsp.php)检索“黄芪”“莪术”的化学成分,按照类药性(DL) ≥ 0.18和口服生物利用度(OB) ≥ 30%的条件进行初筛。随后,借助SwissADME (http://www.swissadme.ch/)进行二次筛选,选取口服生物利用度为“高”且至少满足3项药物相似性项目“是”的成分,进一步通过SwissTargetPrediction (http://www.swisstargetprediction.ch/)数据库获取黄芪–莪术筛选后化学成分的潜在作用靶点,并通过UniProt数据库对靶点名称进行统一规范,最终得到黄芪–莪术活性成分靶点。

2.2. HCC潜在靶点数据集的建立

本研究在人类基因组数据库(GeneCards)、在线人类孟德尔遗传数据库(OMIM)和药物和药物靶点信息数据库(Drugbank)数据库中以“Hepatocellular carcinoma”为关键词检索HCC相关基因靶点,以在GeneCards数据库大于中位数为筛选依据及OMIM、Drugbank数据库得到的相关基因靶点进行合并去重,得到HCC疾病相关靶点。将得到的相关靶点导入UniProt数据库进行标准化处理,建立HCC潜在靶点数据集。

2.3. 药物–疾病共同作用靶点的获取及网络的构建

通过取交集分析获得黄芪–莪术活性成分靶点与肝细胞癌潜在靶点的共同作用靶点。运用Cytoscape3.7.2软件构建“黄芪–莪术活性成分–HCC共同作用靶点”的相互作用网络图。

2.4. 蛋白相互作用(PPI)网络构建及潜在核心靶点的筛选

将共同作用靶点上传至String 12.0数据库(https://cn.string-db.org/),设定物种为“人源”,互作置信度阈值为0.700,隐藏孤立节点,建立初步PPI网络。将网络数据导入Cytoscape3.7.2软件,采用CytoNCA插件计算连接度、紧密度和介数中心性等拓扑参数,以各指标中位数为临界值进行核心靶点筛选,重构优化后的蛋白质相互作用网络,得到黄芪–莪术治疗HCC的潜在核心靶点。

2.5. 机器学习对潜在核心靶点的筛选

本研究采用随机森林模型(RFC)、最小绝对收缩和选择算子(LASSO)回归以及支持向量机递归特征消除(SVM-RFE)三种机器学习算法,对数据集GSE113996中的表达矩阵进行深度筛选,以验证网络药理学所预测靶点的可靠性,从而弥补其可能存在的预测偏差。将机器学习靶点筛选结果与上述网络药理学筛选的核心靶点进行汇总,最终确定黄芪–莪术治疗HCC的潜在核心靶点集。

2.6. 核心靶点的验证

利用R软件以GSE101685作为验证集对“2.5”的潜在核心靶点集进行表达水平验证。将在验证集GSE101685差异性表达的靶点确定为核心靶点。

2.7. GO和KEGG功能富集分析

将黄芪–莪术活性成分与HCC的核心基因导入Metascape数据库,进行基因本体功能(GO)富集分析及京都基因与基因组百科全书(KEGG)通路分析,并通过微生信在线网站Bioinformatics (https://www.bioinformatics.com.cn/)对结果进行可视化展示。

2.8. 分子对接验证

选取“2.6核心靶点的验证”的8个核心基因HSP90AA、AKT1、XAF、IFI44L、SQLE、SPINK1、ALDH1L1、HPR和“黄芪–莪术活性成分–药物靶点”网络图中度值前十的活性成分Jaranol、(6aR, 11aR)-3-Hydroxy-9,10-dimethoxypterocarpan、isoflavanone、(R)-Isomucronulatol、Isorhamnetin、quercetin、Kaempferol、3,9,10-Trimethoxypterocarpan、bisdemethoxycurcumin、7-O-methylisomucronulatol进行分子对接验证。

从PubChem数据库(https://pubchem.ncbi.nlm.nih.gov/)获取“药物–活性成分–靶点”网络中度值较高化合物的二维结构,使用Chem3D软件转换为3D结构并进行能量最小优化,保存为mol2格式。自RCSB PDB数据库(https://www.rcsb.org/)获取核心蛋白受体结构,运用PyMOL 2.6软件去水和残基,通过AutoDockTools-1.5.6软件进行加氢处理,将配体与受体统一转换为*pdbqt格式,设置对接区域覆盖蛋白活性位点。最后利用AutoDock Vina软件进行分子对接计算,依据结合能和氢键数量选择最佳结合构象,通过PyMOL 2.6软件实现结果的可视化分析。

2.9. 分子动力学模拟

为更真实地模拟生物体内蛋白质与黄芪–莪术活性分子的相互作用过程,本研究采用YASARA 3.16 软件对“2.9”部分所得的结合能最佳的成分–靶点复合物进行分子动力学模拟分析。

3. 结果

3.1. 黄芪–莪术的主要活性成分及分子作用靶点

按照2.1的方法,筛选出黄芪–莪术的主要活性成分17个,其中黄芪15个、莪术3个,共同活性成分1个。利用TCMSP和Swiss Target Prediction数据库查询黄芪–莪术活性成分对应的所有基因靶点,去除重复靶点后共获得480个药物靶点。黄芪–莪术OB值前10的化合物基本信息见表1

Table 1. Basic information of the top 10 OB-valued compounds in Astragali Radix-Curcumae Rhizoma

1. 黄芪–莪术OB值前10的化合物基信息表

序号

分子ID

Pubchem ID

活性成分名称

OB (%)

DL

来源

靶点

1

MOL000398

160767

Isoflavanone

109.99

0.30

黄芪

98

2

MOL000940

5315472

Bisdemethoxycurcumin

77.38

0.26

莪术

16

3

MOL000378

15689652

(R)-2,3-Dimethoxy-6-(7-methoxychroman-3-yl)phenol

74.69

0.30

黄芪

100

4

MOL000392

5280378

Formononetin

69.67

0.21

黄芪

56

5

MOL000438

10380176

(R)-Isomucronulatol

67.67

0.26

黄芪

98

6

MOL000380

14077830

(6aR,11aR)-3-Hydroxy-9,10-dimethoxypterocarpan

64.26

0.42

黄芪

99

7

MOL000371

15689655

3,9,10-Trimethoxypterocarpan

53.74

0.48

黄芪

100

8

MOL000239

5318869

Jaranol

50.83

0.29

黄芪

100

9

MOL000354

5281654

isorhamnetin

49.6

0.31

黄芪

100

10

MOL000906

101603568

wenjine

47.93

0.27

莪术

16

3.2. HCC潜在靶点数据集

以“Hepatocellular carcinoma”为关键词在GeneCards、OMIM、Drugbank数据库检索HCC的潜在靶点,并以GeneCards数据库大于中位数为筛选依据及OMIM、Drugbank数据库得到的相关基因靶点汇总后去除重复值得到1576个疾病基因靶点。

3.3. 药物–疾病共同作用靶点的获取及网络的构建

黄芪–莪术与HCC交集靶点有162个,见图1。在网络图中共181个节点,涵盖2个药物节点、17个活性成分节点和162个靶点节点,菱形标识代表药物,圆形标识代表药物成分,方形标识代表共同作用靶点,节点间通过521条连接边构成相互作用网络,见图2

3.4. 黄芪–莪术治疗HCC的PPI网络及潜在核心靶点筛选

将获取到的162个药物–疾病共有靶点导入STRING 12.0数据库,下载“TSV文件”并导入Cytoscape3.7.2软件绘制PPI网络图,具体详见图3。网络图共有158个节点,1016条边。通过CytoNCA插件对核心聚类蛋白进行分析,选取蛋白质–蛋白质相互作用网络特征向量中心性、介度中心性、接近中心性均大于中位数的节点作为潜在核心靶点,潜在核心靶点网络图共有21个节点,133条边。Degree值排名前5位的分别为热休克蛋白90α家族A类成员(HSP90AA1)、表皮生长因子受体(EGFR)、丝氨酸/苏氨酸激酶1 (AKT1)、肉瘤细胞来源的蛋白激酶(SRC)、白介素-6 (IL6)。

Figure 1. Venn diagram of overlapping drug-disease common targets

1. 药物和疾病交集共同靶点韦恩图

Figure 2. Network diagram of Astragali Radix-Curcumae Rhizoma active ingredients and disease targets

2. 黄芪–莪术活性成分–疾病靶点网络图

Figure 3. PPI network of Astragali Radix-Curcumae Rhizoma for HCC treatment

3. 黄芪–莪术治疗HCC的PPI网络图

3.5. 机器学习方法筛选黄芪–莪术治疗HCC的潜在核心靶点

基于LASSO回归分析得到8个潜在核心靶点基因,如图4(A)图4(B),SVM-RF模型筛选得到16个潜在核心靶点,如图4(C)。随机森林RF模型筛选得到10个潜在核心靶点,如图4(D)。通过三种种机器学习最终确定6个黄芪–莪术治疗HCC的共同靶点:XAF、IFI44L、SQLE、SPINK1、ALDH1L1、HPR。

注:(A),LASSO回归模型中8个变量的系数曲线;(B),LASSO回归模型中转向参数(λ)选择的十重交叉验证;(C),SVM-RFE分析的均方误差经5倍交叉验证后曲线变化的精度;(D),随机森林RF分析的ROC曲线。

Figure 4. Map of potential core targets for Astragali Radix-Curcumae Rhizoma against HCC identified via machine learning

4. 机器学习筛选黄芪–莪术治疗HCC的潜在核心靶点图

3.6. 核心靶点的验证

将网络药理学筛选排名前5的靶点(HSP90AA1、EGFR、AKT1、SRC和IL6)和机器学习筛选得到的6个靶点(XAF、IFI44L、SQLE、SPINK1、ALDH1L1、HPR)在验证集GSE101685进行表达水平验证,发现HSP90AA1、AKT1、XAF、IFI44L、SQLE、SPINK1、ALDH1L1、HPR共8个核心基因均差异性表达,确定为核心靶点,见图5

Figure 5. Validation of core genes

5. 核心基因的验证

3.7. 黄芪–莪术治疗HCC的潜在核心靶点的GO/KEGG富集分析

Figure 6. GO enrichment analysis

6. GO富集分析

基于Metascape平台对筛选到的潜在核心靶点进行生物功能分析,以P < 0.01为筛选条件,GO富集分析共获得683个条目,包括595项生物过程(BP),25项细胞组分(CC),63项分子功能(MF)。选取各类别中P值最小的前10个条目,通过微生信在线网站可视化呈现(图6)。结果显示,在BP方面,黄芪–莪术的活性成分主要参与紫外线响应、光刺激响应、辐射响应;CC方面,主要涉及转录调控复合体、常染色质、小窝/质膜微囊;MF方面,主要影响与蛋白激酶结合、激酶结合,特异性地结合RNA聚合酶II(Pol II)相关DNA结合转录因子。

KEGG信号通路富集分析共获得142条显著富集信号通路。选取排名前20的信号通路进行可视化呈现(图7),绿色代表P值越大,红色代表P值越小,P值越小代表显著性越强,气泡体积越大表示该信号通路的基因数目越大。结果表面,黄芪–莪术治疗肝癌的关键信号通路可能与癌症信号相关通路、癌症中的蛋白聚糖、FoxO信号通路、癌症中的PD-L1/PD-1检查点通路及JAK-STAT信号通路关系密切,提示黄芪–莪术可能通过调控肿瘤发生和发展的关键通路发挥抗肝癌作用。此外,一些病毒感染信号通路如卡波西肉瘤相关疱疹病毒感染、乙型肝炎、人类巨细胞病毒感染、甲型流感等显著富集,提示黄芪–莪术可能通过调节病毒感染相关机制间接影响肝癌进程。另外,代谢及炎症相关通路如脂质与动脉粥样硬化、糖尿病并发症中的AGE-RAGE信号通路、TNF信号通路、HIF-1信号通表明其可能通过改善代谢紊乱和抑制炎症反应发挥治疗作用。内分泌抵抗、甲状腺激素信号通路、雌激素信号通路等也被显著富集,提示黄芪–莪术可能影响激素依赖型肿瘤的进展和耐药性。

Figure 7. KEGG pathway analysis

7. KEGG通路分析

3.8. 分子对接验证

通过PubChem数据库、Chem3D 软件、AutoDockTools软件和Vina软件获得分子对接结果,见表2图8。从分子对接的结合能判断成分与相关潜在靶点的结合程度,构像越稳定,结合能越低,当结合能< 1 kcal/mol,提示分子间存在基础结合活性;若结合能 < −5.0 kcal/mol,表明二者有较强结合力;当结合能 < −7.0 kcal/mol,则表示分子间形成了高度稳定的强相互作用。

分子对接结果显示,这80个组合均表现出良好的结合能力,结合能力最强的是核心靶点SQLE与活性成分bisdemethoxycurcumin及HSP90AA1与Isorhamnetin,对接模式见图9

Table 2. Results of molecular docking between compounds and targets

2. 化合物与靶点分子对接的结果

活性成分

结合能(kcal/mol)

HSP90AA1

AKT1

XAF

IFI44L

SQLE

SPINK1

ALDH1L1

HPR

Jaranol

−6.8

−7.3

−7.1

−6.9

−8.3

−5.9

−8.6

−7

(6aR,11aR)-3-Hydroxy-9,10-dimethoxypterocarpan

−6.7

−7.7

−6.6

−6.8

−7.3

−6.3

−7.7

−8.1

isoflavanone

−6.6

−7.7

−6.6

−6.7

−7.4

−5.9

−8.4

−7.1

(R)-Isomucronulatol

−7.3

−7.6

−6.1

−7.3

−6.7

−6

−8

−7

Isorhamnetin

−9.6

−8

−6.9

−7.3

−8.6

−6

−8.6

−7.8

quercetin

−7.6

−7.8

−7.5

−7.4

−8.3

−6.2

−8.3

−8.1

Kaempferol

−6.9

−7.9

−7.6

−7

−8.1

−6.1

−8.2

−7.6

7_O_methylisomucronulatol

−6.2

−7.3

−6.9

−7

−6.9

−5.5

−7.8

−6.5

3,9,10-Trimethoxypterocarpan

−6.8

−7.6

−6.7

−7

−7.2

−5.9

−7.8

−7

bisdemethoxycurcumin

−5.5

−7.3

−6.2

−8

−9.7

−5.9

−8.1

−7.6

Figure 8. Heatmap of molecular docking binding energy

8. 分子对接结合能热图

(A) SQLE与Bisdemethoxycurcumin (B) HSP90AA1与Isorhamnetin

Figure 9. Docking conformations of partial drug active ingredients with targets

9. 部分药物活性成分与靶点对接构象图

3.9. 分子动力学模拟

选取“3.8”中结合能最高的SQLE与Bisdemethoxycurcumin及HSP90AA1与Isorhamnetin组合进行分子动力学模拟。均方根偏差(RMSD)能够计算出模拟过程中的实时构象与最初构象之间的差异,是判断体系是否达到平衡状态的重要参考指标[9]。如图10(A)图10(B)所示,SQLE蛋白质链和SQLE-Bisdemethoxycurcumin复合物、HSP90AA1蛋白质链和HSP90AA1-Isorhamnetin复合物整体波动趋势稳定,且复合物的RMSD数值更低。提示蛋白质与黄芪–莪术分子结合后的稳定性有所提升。

均方根波动(RMSF)是一个反映蛋白质结构中具体到每一个氨基酸构象变化的指标[10]。如图10(C)图10(D)所示,SQLE-Bisdemethoxycurcumin复合物的1、23、41处的蛋白质构象变化大,分别对应甲硫氨酸、赖氨酸、丙氨酸,提示该位置残基柔性较大,容易与其他分子结合。同样,HSP90AA1-Isorhamnetin复合物的3、154、732处的RMSF值较大,残基柔性越高,对应谷氨酸、天冬氨酸、天冬氨酸,提示该区域可能是关键的功能或结合位点。

Figure 10. Results of molecular dynamics simulation

10. 分子动力学模拟结果

4. 讨论

本研究基于网络药理学、分子对接、机器学习及分子动力学模拟,确定了黄芪–莪术药对治疗HCC的药效成分和分子机制。黄芪–莪术排名前十的活性成分有刺芒柄花素(Jaranol)、(6aR,11aR)-3-Hydroxy-9,10-dimethoxypterocarpan、Isoflavanone、(R)-Isomucronulatol、异鼠李素(Isorhamnetin)、槲皮素(Quercetin)、山柰酚(Kaempferol)、7-O-methylisomucronulatol、3,9-di-O-methylnissolin、去甲氧基姜黄素(Bisdemethoxycurcumin)。Jaranol是黄芪的一种异黄酮类化合物,已被发现一系列抗肿瘤活性。研究表明,Jaranol是子宫内膜癌的活性化合物,通过促进雌激素受体β(ER)的表达发挥抗癌作用[11]。此外,Liu等人研究表明黄芪当归药对中的Jaranol是抗乳腺癌的最重要的成分,主要通过上调PIK3R1发挥抗肿瘤免疫作用[12]。(6aR,11aR)-3-Hydroxy-9,10-dimethoxypterocarpan是黄芪的紫檀烷类(pterocarpan)衍生物,具有广泛的生物活性。Feng D等人认为在低浓度(20 μM)下其能有效抑制黑色素合成[13]。Isoflavanone是黄芪中的异黄酮类成分,研究表明Isoflavanone可阻断细胞外信号通路如ERα依赖性通路抑制肿瘤血管生成,并通过下调cyclin B1和CDK2等蛋白阻碍癌细胞发展[14] [15]。(R)-Isomucronulatol是黄芪中的一种黄酮类活性成分。分子对接研究表明,(R)-Isomucronulatol与PI3K、AKT和mTOR蛋白具有显著结合亲和力,可能通过抑制TGF-β1刺激的HK-2细胞中PI3K/AKT/mTOR信号通路发挥抗肿瘤作用[16]。Isorhamnetin具有广泛的抗肿瘤作用,可通过调控PPARγ/PTEN/AKT通路抑制膀胱癌的肿瘤发生[17]。此外,Isorhamnetin通过抑制炎症因子如TNF-α、IL-6和调控Akt/MAPKs/Nrf2信号通路,抑制肝癌细胞的增殖和上皮–间质转化,同时激活PPAR-γ和自噬,抑制细胞周期进展[18]。Quercetin是黄芪的黄酮类化学物,通过多靶点、多通路发挥抗肿瘤作用。在口腔恶性肿瘤的治疗中,槲皮素通过诱导氧化应激和凋亡发挥抗癌作用[19]。在治疗鼻咽癌中,槲皮素能抑制上皮–间质转化和肿瘤迁移相关蛋白如N-cadherin,从而阻断转移[20]。在黑色素瘤中,槲皮素通过改变肿瘤免疫微环境提升免疫应答[21]。Kaempferol是一种黄酮类成分,在多种肿瘤模型中显示出显著的抗癌特性,其作用机制涉及调控PI3K/Akt、mTOR、Erk/MAPK等关键信号通路,通过抑制癌细胞增殖、诱导凋亡发挥抗肿瘤效果[21] [22]。Bisdemethoxycurcumin作为姜黄素类化合物之一,具有显著的抗肿瘤活性。在对激素依赖性肿瘤如卵巢颗粒细胞瘤和睾丸间质细胞瘤中,Bisdemethoxycurcumin对3β-HSD酶的抑制活性显著高于其他姜黄素类化合物,可能通过调节类固醇生成途径影响肿瘤生长[23]

通过网络药理学和三种机器学习方法,确定了在黄芪–莪术治疗HCC的八个核心靶点,包括HSP90AA1、AKT1、XAF、IFI44L、SQLE、SPINK1、ALDH1L1和HPR等。HSP90AA1作为分子伴侣蛋白,在肝癌细胞中高表达[24]。HSP90AA1通过结合并稳定脂代谢关键酶IDH1,促进肝癌细胞的脂质积累和化疗抵抗[25];研究发现,AKT1通过PI3K/AKT/mTOR信号通路促进肿瘤细胞增殖和存活[26];XAF1作为凋亡抑制因子XIAP的拮抗剂,其表达下调可导致肝癌细胞凋亡受阻[27]。IFI44L在干扰素信号通路中发挥重要作用,其异常表达与肝癌的免疫逃逸密切相关[28]。SQLE是胆固醇合成的限速酶,在肝癌组织中显著高表达,促进肿瘤微环境形成[29]。SPINK1作为丝氨酸蛋白酶抑制剂,可通过EGFR信号通路促进肝癌侵袭转移[30]。ALDH1L1参与叶酸代谢,其表达下调或缺失与肝癌恶性进展相关[31]。HPR (触珠蛋白相关蛋白)在肝癌微环境中高表达,可能通过调节铁代谢影响肿瘤生长[32]。这些核心靶点主要与Quercetin、Kaempferol、Isorhamnetin、Bisdemethoxycurcumin等化学成分关系密切。文献报道,Quercetin主要通过PI3K-Akt通路诱导细胞凋亡发挥抗肿瘤作用[33],Kaempferol通过抑制PI3K/AKT/mTOR通路促进非小细胞肺癌细胞[34]。而HSP90AA1、AKT1是PI3K-Akt信号通路的关键靶点,其异常激活与多种癌症的发生、发展、转移和治疗耐药性密切相关[35]。这些靶点与关键化合物共同构成了黄芪–莪术药对“多成分–多靶点–多通路”抗肝癌作用网络的基础。

GO富集结果表明,黄芪–莪术可能通过phosphorylation、response to decreased oxygen levels、protein kinase activity、protein tyrosine kinase activity等过程参与调控HCC,phosphorylation修饰在癌症信号转导中起关键作用,异常的激酶活性(如突变、扩增或组成型激活)可驱动肿瘤发生和进展,特别是在MAPK/ERK、JAK2/STAT3和PI3K/AKT/mTOR等通路中[36]-[38]。response to decreased oxygen levels是由于反映氧气的存在、缺乏或浓度的刺激而导致细胞或有机体状态或活动(在运动、分泌、酶产生、基因表达等方面)发生变化的任何过程,这个过程与氧化应激的发生和发展关系密切,通过氧化应激、表观遗传和干细胞特性维持促进恶性表型。不可忽视的是,缺氧诱导因子1α (HIF-1α)是缺氧反应的核心调控因子。HIF-1α通路激活驱动肿瘤增殖、血管生成和转移;这与KEGG富集结果具有较高的相同性。蛋白激酶活性和蛋白酪氨酸激酶活性在肿瘤发生发展中发挥着核心调控作用。蛋白激酶通过催化蛋白质底物的磷酸化修饰,调控PI3K/AKT/mTOR、MAPK/ERK等重要信号通路,影响肿瘤细胞的增殖、分化、凋亡和代谢过程[39] [40]

KEGG富集分析结果表明,黄芪–莪术可能通过Pathways in cancer、Hepatitis B、FoxO信号通路、PD-L1/PD-1检查点通路和HIF-1信号通路等多个关键通路发挥抗肝癌作用。研究发现,MAPK信号通路(Ras/RAF/MEK/ERK)是最明确的癌症通路之一,通过激活增殖基因同时抑制凋亡基因来驱动致癌过程[41]。乙型肝炎病毒(HBV)感染涉及多种信号通路的调控,PI3K-AKT和MAPK-ERK1/2通路,激活可抑制肝细胞核因子4α (HNF4α)的活性,从而抑制HBV复制[42]。FoxO信号通路中,黄芪–莪术的活性成分异鼠李素和槲皮素能够抑制PI3K/AKT信号,恢复FoxO转录因子的核定位,从而诱导肝癌细胞凋亡[43];在PD-L1/PD-1免疫检查点通路中,莪术中的去甲氧基姜黄素可能通过下调PD-L1,增强CD8+T细胞的抗肿瘤活性[44];而对于HIF-1信号通路,黄芪活性成分能竞争性结合HSP90AA1,阻断其与HIF-1α的结合,进而抑制肿瘤血管生成[45]。这些通路相互协同,FoxO与PD-L1/PD-1共同调控免疫逃逸,而HIF-1α与PI3K-Akt形成正反馈循环,黄芪–莪术通过多靶点干预打破这一促癌机制。

分子对接和分子动力学模拟分析证实了黄芪–莪术药对治疗肝细胞癌的多靶点作用机制。研究显示,80个成分–靶点组合均具有良好的结合活性,其中SQLE-Bisdemethoxycurcumin和HSP90AA1-Isorhamnetin复合物表现出最强的结合能力(结合能分别为−9.7 kcal/mol和−9.6 kcal/mol)。为评估分子对接结果的可靠性,选取结合能较低的SQLE-Bisdemethoxycurcumin和HSP90AA1-Isorhamnetin两组复合物开展分子动力学模拟研究。结果显示,两个复合物体系在动态过程中保持较好的构象稳定性、结构紧密度及持续的相互作用,表明其在生理环境下能够形成稳定的结合模式。该结果为阐明黄芪–莪术通过SQLE胆固醇合成通路和HSP90AA1分子伴侣系统发挥抗肝癌作用提供了分子基础。

综上所述,通过网络药理学和机器学习揭示了黄芪–莪术治疗HCC的关键化合物,核心基因和相关途径,结合分子对接和分子动力学模拟验证了关键化合物和核心基因之间的关系,探索了黄芪–莪术治疗HCC的潜在治疗机制。然而,网络药理学及机器学习等过于依赖人工智能算法及大数据,对于关键的调控目标和途径机制仍需要在动物或细胞实验中进行验证。

基金项目

广东江门中医药职业学院科学研究项目课题([科] JMZYYKY20232005)。

NOTES

*第一作者。

#通讯作者。

参考文献

[1] Wei, L., Wang, Z., Jing, N., Lu, Y., Yang, J., Xiao, H., et al. (2022) Frontier Progress of the Combination of Modern Medicine and Traditional Chinese Medicine in the Treatment of Hepatocellular Carcinoma. Chinese Medicine, 17, Article No. 90. [Google Scholar] [CrossRef] [PubMed]
[2] Zhu, Z.H., Xu, X.T., Shen, C.J., et al. (2023) A Novel Sesquiterpene Lactone Fraction from Eupatorium chinense L. Suppresses Hepatocellular Carcinoma Growth by Triggering Ferritinophagy and Mitochondrial Damage. Phytomedicine, 112, Article 154671.
[3] Liu, Z., Ma, H. and Lai, Z. (2023) The Role of Ferroptosis and Cuproptosis in Curcumin against Hepatocellular Carcinoma. Molecules, 28, Article 1623. [Google Scholar] [CrossRef] [PubMed]
[4] Asgharzadeh, F., Moradi Binabaj, M., Fanoudi, S., C. Cho, W., Yang, Y., Azarian, M., et al. (2024) Nanomedicine Strategies Utilizing Lipid-Based Nanoparticles for Liver Cancer Therapy: Exploring Signaling Pathways and Therapeutic Modalities. Advanced Pharmaceutical Bulletin, 14, 513-523. [Google Scholar] [CrossRef] [PubMed]
[5] 侯秋苑, 殷蕾. 理冲汤的药理作用及临床应用的研究进展[J]. 中外医疗, 2022, 41(29): 185-188+194.
[6] 鲍宁, 陈子超, 赵春芹, 等. 黄芪-莪术药对及其活性成分抗肝癌作用机制研究进展[J]. 中草药, 2023, 54(15): 5101-5111.
[7] Ma, Y., Li, Y., Wu, T., Li, Y. and Wang, Q. (2023) Astragaloside IV Attenuates Programmed Death-Ligand 1-Mediated Immunosuppression during Liver Cancer Development via the miR-135b-5p/CNDP1 Axis. Cancers, 15, Article No. 5048. [Google Scholar] [CrossRef] [PubMed]
[8] Wang, J. (2024) Traditional Chinese Medicine in Post-Resection Liver Cancer Therapy: A Meta-Analysis of Efficacy. American Journal of Translational Research, 16, 5763-5775. [Google Scholar] [CrossRef] [PubMed]
[9] Sabaghan, S., Srivastava, R., Yadav, P., Kumari, M., Soni, R., Beri, S., et al. (2025) Exploring Ketones in Chrysopogon Zizanioides: A Computational Molecular Dynamic Approach to C-Met Modulation. Molecular Biotechnology. [Google Scholar] [CrossRef] [PubMed]
[10] Alomair, L., Mustafa, S., Jafri, M.S., Alharbi, W., Aljouie, A., Almsned, F., et al. (2022) Molecular Dynamics Simulations to Decipher the Role of Phosphorylation of SARS-CoV-2 Nonstructural Proteins (NSPS) in Viral Replication. Viruses, 14, Article 2436. [Google Scholar] [CrossRef] [PubMed]
[11] Zhang, Q. and Huang, X. (2021) The Modulatory Properties of astragalus Membranaceus Treatment on Endometrial Cancer: An Integrated Pharmacological Method. Peer J, 9, e11995. [Google Scholar] [CrossRef] [PubMed]
[12] Liu, H., Lian, L., Hou, L., Liu, C., Ren, J., Qiao, Y., et al. (2024) Herb Pair of Huangqi-Danggui Exerts Anti-Tumor Immunity to Breast Cancer by Upregulating PIK3R1. Animal Models and Experimental Medicine, 7, 234-258. [Google Scholar] [CrossRef] [PubMed]
[13] Feng, D.N., Fang, Z.X. and Zhang, P.Z. (2022) The Melanin Inhibitory Effect of Plants and Phytochemicals: A Systematic Review. Phytomedicine, 107, Article 154449.
[14] Ul Hassan, M.H., Shahbaz, M., Imran, M., Momal, U., Naeem, H., Mujtaba, A., et al. (2025) Isoflavones: Promising Natural Agent for Cancer Prevention and Treatment. Food Science & Nutrition, 13, e70091. [Google Scholar] [CrossRef] [PubMed]
[15] Ferriere, F., Aasi, N., Flouriot, G. and Pakdel, F. (2024) Exploring the Complex Mechanisms of Isoflavones: From Cell Bioavailability, to Cell Dynamics and Breast Cancer. Phytotherapy Research, 39, 957-979. [Google Scholar] [CrossRef] [PubMed]
[16] Zhan, K., Chen, S., Ji, L., Xu, L., Zhang, Y., Zhang, Q., et al. (2024) Network Pharmacology to Unveil the Mechanism of Astragali Radix in the Treatment of Lupus Nephritis via PI3K/AKT/mTOR Pathway. Scientific Reports, 14, Article No. 25983. [Google Scholar] [CrossRef] [PubMed]
[17] Peng, Z., Yisheng, S., Lei, S., et al. (2023) Effect of Isorhamnetin on Carbonic Anhydrase IX Expression and Tumorigenesis of Bladder Cancer by Activating PPARγ/PTEN/AKT Pathway. Tissue and Cell, 82, Article 102048.
[18] Sarkar, S., Das, A.K., Bhattacharya, S., Gachhui, R. and Sil, P.C. (2023) Isorhamnetin Exerts Anti-Tumor Activity in DEN+CCL4-Induced HCC Mice. Medical Oncology, 40, Article 188. [Google Scholar] [CrossRef] [PubMed]
[19] Saadh, M.J., Ahmed, H.H., Chandra, M., Al-Hussainy, A.F., Hamid, J.A., Mishra, A., et al. (2025) Therapeutic Effects of Quercetin in Oral Cancer Therapy: A Systematic Review of Preclinical Evidence Focused on Oxidative Damage, Apoptosis and Anti-Metastasis. Cancer Cell International, 25, Article No. 66. [Google Scholar] [CrossRef] [PubMed]
[20] Li, T. and Li, Y. (2023) Quercetin Acts as a Novel Anti-Cancer Drug to Suppress Cancer Aggressiveness and Cisplatin-Resistance in Nasopharyngeal Carcinoma (NPC) through Regulating the Yes-Associated Protein/Hippo Signaling Pathway. Immunobiology, 228, Article 152324. [Google Scholar] [CrossRef] [PubMed]
[21] Li, X., He, X., Lin, B., Li, L., Deng, Q., Wang, C., et al. (2024) Quercetin Limits Tumor Immune Escape through PDK1/CD47 Axis in Melanoma. The American Journal of Chinese Medicine, 52, 541-563. [Google Scholar] [CrossRef] [PubMed]
[22] Ma, X.J., Zhang, X.Y., Wang, X., et al. (2023) The Role of Kaempferol in Gynaecological Malignancies: Progress and Perspectives. Frontiers in Pharmacology, 14, Article 1310416.
[23] Qiao, X., Ye, L., Lu, J., Pan, C., Fei, Q., Zhu, Y., et al. (2023) Curcumin Analogues Exert Potent Inhibition on Human and Rat Gonadal 3β-Hydroxysteroid Dehydrogenases as Potential Therapeutic Agents: Structure-Activity Relationship and in Silico Docking. Journal of Enzyme Inhibition and Medicinal Chemistry, 38, Article 2205052. [Google Scholar] [CrossRef] [PubMed]
[24] Santos, L.D.S., Silva, V.R., de Castro, M.V.L., Dias, R.B., Valverde, L.D.F., Rocha, C.A.G., et al. (2023) New Ruthenium-Xanthoxylin Complex Eliminates Colorectal Cancer Stem Cells by Targeting the Heat Shock Protein 90 Chaperone. Cell Death & Disease, 14, Article 832. [Google Scholar] [CrossRef] [PubMed]
[25] Wang, S., Cheng, H., Huang, Y., Li, M., Gao, D., Chen, H., et al. (2024) HSP90a Promotes the Resistance to Oxaliplatin in HCC through Regulating Idh1-Induced Cell Competition. Biochimica et Biophysica ActaMolecular Cell Research, 1871, Article 119680. [Google Scholar] [CrossRef] [PubMed]
[26] Pecoraro, C., Scianò, F., Carbone, D., et al. (2024) Synthesis and Biological Evaluation of a New Class of Azole Urea Compounds as Akt Inhibitors with Promising Anticancer Activity in Pancreatic Cancer Models. Bioorganic Chemistry, 153, Article 107959.
[27] Liu, B., Liu, R., Li, W., Mao, X., Li, Y., Huang, T., et al. (2022) XAF1 Prevents Hyperproduction of Type I Interferon Upon Viral Infection by Targeting IRF7. EMBO Reports, 24, e55387. [Google Scholar] [CrossRef] [PubMed]
[28] Zhou, Y., Wang, J., Chen, Y., Lin, W., Zhou, R., Zhao, L., et al. (2025) NRIR Promotes Immune Escape in Hepatocellular Cancer by Regulating IFNγ-Induced PD-L1 Expression. Journal of Advanced Research. [Google Scholar] [CrossRef] [PubMed]
[29] Zou, Y.H., Zhang, H.Y., Bi, F., et al. (2022) Targeting the Key Cholesterol Biosynthesis Enzyme Squalene Monooxygenase for Cancer Therapy. Frontiers in Oncology, 12, Article 938502.
[30] Man, K., Zhou, L., Yu, H., Lam, K., Cheng, W., Yu, J., et al. (2023) Spink1-Induced Tumor Plasticity Provides a Therapeutic Window for Chemotherapy in Hepatocellular Carcinoma. Nature Communications, 14, Article No. 7863. [Google Scholar] [CrossRef] [PubMed]
[31] Sasaki, M., Yamamoto, K., Ueda, T., Irokawa, H., Takeda, K., Sekine, R., et al. (2023) One-Carbon Metabolizing Enzyme ALDH1L1 Influences Mitochondrial Metabolism through 5-Aminoimidazole-4-Carboxamide Ribonucleotide Accumulation and Serine Depletion, Contributing to Tumor Suppression. Scientific Reports, 13, Article No. 13486. [Google Scholar] [CrossRef] [PubMed]
[32] Li, Y., Chen, Y., Zhang, Y., Fang, Y., Wu, L., Zhao, Y., et al. (2024) Integrating Multi-Omics Techniques and in Vitro Experiments Reveals That GLRX3 Regulates the Immune Microenvironment and Promotes Hepatocellular Carcinoma Cell Proliferation and Invasion through Iron Metabolism Pathways. Frontiers in Immunology, 15, Article 1496886. [Google Scholar] [CrossRef] [PubMed]
[33] Zhou, B.C., Yang, Y., Pang, X.M., et al. (2023) Quercetin Inhibits DNA Damage Responses to Induce Apoptosis via SIRT5/PI3K/AKT Pathway in Non-Small Cell Lung Cancer. Biomedicine & Pharmacotherapy, 165, Article 115071.
[34] Wang, R., Deng, Z.Y., Zhu, Z.M., et al. (2023) Kaempferol Promotes Non-Small Cell Lung Cancer Cell Autophagy via Restricting Met Pathway. Phytomedicine, 121, Article 155090.
[35] Li, Y., Dong, M., Qin, H., An, G., Cen, L., Deng, L., et al. (2025) Mulberrin Suppresses Gastric Cancer Progression and Enhances Chemosensitivity to Oxaliplatin through HSP90AA1/PI3K/AKT Axis. Phytomedicine, 139, Article 156441. [Google Scholar] [CrossRef] [PubMed]
[36] Piersma, S.R., Valles-Marti, A., Rolfs, F., Pham, T.V., Henneman, A.A. and Jiménez, C.R. (2022) Inferring Kinase Activity from Phosphoproteomic Data: Tool Comparison and Recent Applications. Mass Spectrometry Reviews, 43, 725-751. [Google Scholar] [CrossRef] [PubMed]
[37] Zhang, X., Xu, H., Bi, X., Hou, G., Liu, A., Zhao, Y., et al. (2021) SRC Acts as the Target of Matrine to Inhibit the Proliferation of Cancer Cells by Regulating Phosphorylation Signaling Pathways. Cell Death & Disease, 12, Article 931. [Google Scholar] [CrossRef] [PubMed]
[38] Fatma, M., Parveen, S. and Mir, S.S. (2025) Unraveling the Kinase Code: Role of Protein Kinase in Lung Cancer Pathogenesis and Therapeutic Strategies. Biochimica et Biophysica ActaReviews on Cancer, 1880, Article 189309. [Google Scholar] [CrossRef] [PubMed]
[39] Jiang, M., Zhang, K., Zhang, Z., Zeng, X., Huang, Z., Qin, P., et al. (2025) PI3k/AKT/mTOR Axis in Cancer: From Pathogenesis to Treatment. MedComm, 6, e70295. [Google Scholar] [CrossRef] [PubMed]
[40] Wilmerding, A., Bouteille, L., Caruso, N., Bidaut, G., Etchevers, H.C., Graba, Y., et al. (2022) Sustained Experimental Activation of FGF8/ERK in the Developing Chicken Spinal Cord Models Early Events in Erk-Mediated Tumorigenesis. Neoplasia, 24, 120-132. [Google Scholar] [CrossRef] [PubMed]
[41] Yuan, J., Dong, X., Yap, J. and Hu, J. (2020) The MAPK and AMPK Signalings: Interplay and Implication in Targeted Cancer Therapy. Journal of Hematology & Oncology, 13, Article 113. [Google Scholar] [CrossRef] [PubMed]
[42] Yu, Y., Hu, H., Zhang, Y., Zhang, Z., Ying, S., Dong, S., et al. (2025) Hedyotis Chrysotricha Aqueous Extract Inhibits Hepatitis B Surface Antigen and Viral Replication via Hepatocyte Nuclear Factor 4α Regulation. Phytomedicine, 143, Article 156726. [Google Scholar] [CrossRef] [PubMed]
[43] Tu, S. and Qiu, Y.Q. (2023) Molecular Subtypes and Scoring Tools Related to Foxo Signaling Pathway for Assessing Hepatocellular Carcinoma Prognosis and Treatment Responsiveness. Frontiers in Pharmacology, 14, Article 1213506.
[44] Zheng, J., Liu, W., Wang, X., Li, H., Wang, Z. and Ai, Z. (2025) Curcumin Enhances Anti-Tumor Immunity in Anaplastic Thyroid Carcinoma by Elevating CD8+ T Cell Function and Downregulating the Akt/mTORC1/STAT3/PD-L1 Axis. PathologyResearch and Practice, 269, Article 155898. [Google Scholar] [CrossRef] [PubMed]
[45] Jiang, H., Zhou, R., An, L., Guo, J., Hou, X., Tang, J., et al. (2023) Exploring the Role and Mechanism of Astragalus Membranaceus and Radix Paeoniae Rubra in Idiopathic Pulmonary Fibrosis through Network Pharmacology and Experimental Validation. Scientific Reports, 13, Article No. 10110. [Google Scholar] [CrossRef] [PubMed]