目的 基于柴胡总皂苷（total saikosaponins of Bupleuri Radix，TSS）肝组织暴露成分与肝损伤表型的关联，筛选柴胡肝毒性质量标志物（hepatotoxicity quality marker，HQM），并探讨其潜在作用机制。方法 采用不同剂量TSS干预db/db小鼠，检测血清生化指标[丙氨酸氨基转移酶（alanine aminotransferase，ALT）、天冬氨酸氨基转移酶（aspartate aminotransferase，AST）、总胆固醇（total cholesterol，TC）、三酰甘油（triglyceride，TG）]及肝组织病理变化。采用超高效液相色谱-四极杆飞行时间质谱（UPLC-QTOF-MS）构建选择反应监测离子对数据库，结合超高效液相色谱-三重四极杆串联质谱（UPLC-TSQ-MS/MS）拟靶向分析表征柴胡药材、TSS及肝组织中的相关成分。以ΔALT为主要肝损伤表型、ΔAST为辅助参考指标，采用偏最小二乘回归、灰色关联分析、Pearson和Spearman相关分析评价肝组织暴露成分与肝损伤的关联，并结合网络药理学、机器学习、分子对接及分子动力学模拟分析潜在作用靶点。结果 TSS可显著降低db/db小鼠血清TC和TG水平（P＜0.05、0.01），但高剂量TSS同时升高ALT并加重肝组织损伤。拟靶向分析共鉴定了TSS中39个肝组织暴露成分，偏最小二乘回归分析初步筛选得到4个候选成分；经多方法交叉评价，6''-O-acetyl-saikosaponin a和6''-O-acetyl-saikosaponin d与肝损伤表型的关联证据较为一致，被列为柴胡HQM的优先候选成分。机器学习筛选获得磷酸丙糖异构酶1（triosephosphate isomerase 1，TPI1）、细胞周期蛋白B1（cyclin B1，CCNB1）、检查点激酶1（checkpoint kinase 1，CHEK1）、促红细胞生成素肝细胞受体A2（erythropoietin-producing hepatocellular receptor A2，EPHA2）、丝氨酸/苏氨酸激酶1（serine/threonine kinase 1，AKT1）、极光激酶A（aurora kinase A，AURKA）和细胞周期蛋白依赖性激酶2（cyclin-dependent kinase 2，CDK2）7个核心靶点，富集分析提示其主要涉及细胞周期、磷脂酰肌醇3-激酶（phosphatidylinositol 3-kinase，PI3K）/蛋白激酶B（protein kinase B，Akt）及p53等信号通路。分子对接和分子动力学模拟支持上述候选HQM与相关核心靶点之间可能形成相对稳定的结合构象。结论 通过拟靶向技术表征“柴胡药材-TSS-肝组织暴露”成分传递过程，结合肝损伤表型关联分析、机器学习和分子模拟，将6''-O-acetyl-saikosaponin a和6''-O-acetyl-saikosaponin d列为柴胡HQM的优先候选成分，为其后续确证及柴胡肝毒性物质基础研究提供依据。

Objective To screen hepatotoxicity quality marker (HQM) of Chaihu (Bupleuri Radix) based on the associations between liver-exposed components of total saikosaponins of Bupleuri Radix (TSS) and liver injury phenotypes, and to investigate their potential mechanisms. Methods db/db mice were administered different doses of TSS, serum biochemical indicators [alanine aminotransferase (ALT), aspartate aminotransferase (AST), total cholesterol (TC), triglycerides (TG)] and pathological changes in liver tissue were detected. A selected reaction monitoring ion-pair database was established using ultra-performance liquid chromatography coupled with quadrupole time-of-flight mass spectrometry (UPLC-QTOF-MS). Pseudotargeted analysis using ultra-performance liquid chromatography coupled with triple quadrupole tandem mass spectrometry (UPLC-TSQ-MS/MS) was then performed to characterize relevant components in Bupleuri Radix, TSS and liver tissue. ΔALT was used as the primary liver injury phenotype and ΔAST as a supplementary reference indicator, partial least squares regression (PLSR), grey relational analysis (GRA), Pearson correlation analysis and Spearman correlation analysis were employed to evaluate the associations between liver-exposed components and liver injury. Network pharmacology, machine learning, molecular docking and molecular dynamics simulations were further integrated to investigate potential targets. Results TSS significantly reduced levels of TC and TG in serum of db/db mice (P < 0.05, 0.01), but high-dose TSS also increased ALT and aggravates liver tissue damage. Pseudotargeted analysis identified 39 TSS-derived components exposed in liver tissue, and four candidate components were preliminarily selected using PLSR. Following cross-evaluation using multiple analytical methods, 6''-O-acetyl-saikosaponin a and 6''-O-acetyl-saikosaponin d showed relatively consistent associations with liver injury phenotypes and were therefore prioritized as candidate HQMs of Bupleuri Radix. Machine learning identified seven core targets, including triosephosphate isomerase 1 (TPI1), cyclin B1 (CCNB1), checkpoint kinase 1 (CHEK1), erythropoietin-producing hepatocellular receptor A2 (EPHA2), serine/threonine kinase 1 (AKT1), aurora kinase A (AURKA) and cyclin-dependent kinase 2 (CDK2). Enrichment analysis indicated that these targets were mainly involved in the cell cycle, phosphatidylinositol 3-kinase (PI3K)/protein kinase B (Akt) and p53 signaling pathways. Molecular docking and molecular dynamics simulations supported the possibility that the candidate HQMs could form relatively stable binding conformations with the corresponding core targets. Conclusion Pseudotargeted analysis was employed to characterize the transmission of constituents along the “Bupleuri Radix-TSS-liver exposure” chain. By integrating phenotypic association analysis of liver injury, machine learning and molecular simulations, 6''-O-acetyl-saikosaponin a and 6''-O-acetyl-saikosaponin d were prioritized as candidate HQMs of Bupleuri Radix. These findings provide a basis for their subsequent validation and for elucidating the material basis underlying the hepatotoxicity of Bupleuri Radix.

R282.6

国家自然科学基金资助项目（82274197）；国家自然科学基金资助项目（81773997）；山东省自然科学基金资助项目（ZR2024LFG005）