目的 基于多组学孟德尔随机化（Mendelian randomization，MR）框架，评估肠道菌群、免疫细胞、血清代谢物与特发性肺纤维化（idiopathic pulmonary fibrosis，IPF）的因果关联，筛选核心因果基因并预测干预中药及化合物，为IPF早期诊断与中药干预提供遗传学依据。方法 提取肠道菌群、血清代谢物及免疫细胞的遗传工具变量（instrumental variables，IVs），通过两样本MR分析其与IPF的因果关联；注释IVs邻近基因获得“菌群-免疫-代谢”因果基因集，整合IPF转录组数据进行差异表达基因（differentially expressed genes，DEGs）分析和加权基因共表达网络分析（weighted gene co-expression network analysis，WGCNA）筛选候选因果基因；经机器学习构建诊断模型并筛选核心基因，分析免疫细胞浸润与单细胞RNA测序，通过CTD和ITCM数据库预测干预中药及化合物，并进行分子对接验证。结果 MR分析鉴定出2个肠道菌群分类群（乳杆菌科为保护因素，变形菌门为风险因素）、19个免疫细胞表型（9个保护因素、10个风险因素）及26种血清代谢物（12个保护因素、14个风险因素）与IPF存在显著因果关联（P＜0.05）。整合分析获得21个候选因果基因，机器学习筛选出5个核心基因：衔接蛋白复合体5亚基μ1（adaptor protein complex 5 subunit mu 1，AP5M1）、性别决定区Y框转录因子11（SRY-box transcription factor 11）、锌指MIZ型包含蛋白2（zinc finger and MIZ-type containing 2，ZMIZ2）、脆性组氨酸三联体（fragile histidine triad，FHIT）和细胞周期蛋白依赖性激酶6（cyclin dependent kinase 6，CDK6），均具潜在诊断价值。免疫浸润显示IPF组织中活化肥大细胞和单核细胞增多，M0/M2巨噬细胞及活化自然杀伤细胞减少；单细胞分析显示FHIT在多种细胞中富集程度最高。筛选出150种潜在干预中药，高频中药包括桑白皮、月季花、延胡索等；活性成分包括山柰酚、柚皮素、汉黄芩素、血根碱等。分子对接证实核心化合物可与靶点稳定结合。结论 揭示了肠道菌群、免疫细胞、血清代谢物与IPF的因果关联，筛选出AP5M1、SOX11、ZMIZ2、FHIT、CDK6作为IPF的潜在候选诊断标志物；预测出山柰酚、柚皮素、非洲防己碱等活性化合物及桑白皮、月季花等中药的潜在干预价值，为IPF诊疗提供遗传学证据与理论支撑。

Objective To evaluate the causal associations of gut microbiota, immune cells and serum metabolites with idiopathic pulmonary fibrosis (IPF) based on the multi-omics Mendelian randomization (MR) framework, screen core causal genes, and predict intervening traditional Chinese medicine (TCM) and compounds, so as to provide genetic evidence for the early diagnosis and TCM intervention of IPF. Methods Genetic instrumental variables (IVs) of gut microbiota, serum metabolites and immune cells were extracted, and their causal associations with IPF were analyzed by two-sample MR. Adjacent genes of IVs were annotated to obtain the "microbiota-immunity-metabolism" causal gene set. Integrated with IPF transcriptome data, candidate causal genes were screened by differentially expressed genes (DEGs) analysis and weighted gene co-expression network analysis (WGCNA). Diagnostic models were constructed by machine learning to screen core genes. Immune cell infiltration and single-cell RNA sequencing profiles were analyzed. Potential intervening TCM and compounds were predicted via Comparative Toxicogenomics Database (CTD) and Integrated Traditional Chinese Medicine Database (ITCM), followed by molecular docking verification. Results Two-sample MR identified 2 gut microbiota taxa (Lactobacillaceae as protective factor, Proteobacteria as risk factor), 19 immune cell phenotypes (nine protective, ten risk) and 26 serum metabolites (12 protective, 14 risk) that were significantly causally associated with IPF (P < 0.05). A total of 21 candidate causal genes were obtained by integrated analysis, and 5 core genes were screened by machine learning: adaptor protein complex 5 subunit mu 1 (AP5M1), SRY-box transcription factor 11 (SOX11), zinc finger and MIZ-type containing 2 (ZMIZ2), fragile histidine triad (FHIT) and cyclin dependent kinase 6 (CDK6), all with potential diagnostic value. Immune infiltration showed increased activated mast cells and monocytes, and decreased M0/M2 macrophages and activated natural killer cells in IPF tissues. Single-cell analysis revealed the highest enrichment of FHIT in multiple cell types. A total of 150 kinds of potential intervention Chinese medicine were screened. The high frequency Chinese medicine included Sangbaipi (Mori Cortex), Yuejihua (Rosae Chinensis Flos), Yanhusuo (Corydalis Rhizoma), etc. The active ingredients included kaempferol, naringenin, wogonin, sanguinarine, etc. Molecular docking confirmed that the core compound could stably bind to the target. Conclusions This study reveals the causal associations of gut microbiota, immune cells and serum metabolites with IPF, and identifies AP5M1, SOX11, ZMIZ2, FHIT and CDK6 as potential candidate diagnostic biomarkers for IPF. It also predicts the potential intervention value of active compounds such as kaempferol, naringenin and palmatine, as well as TCM including Morus alba and Rosa chinensis, providing genetic evidence and theoretical support for the diagnosis and treatment of IPF.

Q811.4;R285

中国科协青年科技人才培育工程博士生专项计划项目；国家自然科学基金青年科学基金项目（82505509）；湖北省自然科学基金资助项目（2023AFD173）；湖北省中医药管理局中医药科研项目（ZY2025M026）；四川省自然科学基金青年基金项目（2025ZNSFSC1853）；中国博士后科学基金第75批面上资助（2024MD753905）；成都市卫生健康委员会-成都中医药大学委校院联合创新基金（WXLH202403266）