An integrated computational framework for assessing toxicity mechanisms of benzo[a]pyrene diol epoxide-induced interstitial lung disease

Benzo[a]pyrene diol epoxide (BPDE), a key carcinogenic metabolite of benzo[a]pyrene, has been linked to respiratory damage, yet its role in interstitial lung disease (ILD) remains elusive. This study systematically investigated the mechanisms of BPDE-induced ILD by integrating network toxicology, machine learning, and molecular docking. Potential targets of BPDE were retrieved from ChEMBL, STITCH, and SwissTargetPrediction, while ILD-related genes were identified by differential expression analysis and weighted gene co-expression network analysis (WGCNA) from public transcriptomic datasets. We identified 79 shared targets between BPDE and ILD, with enrichment analyses revealing significant involvement in xenobiotic response, ion channel activity, and pathways such as PI3K-Akt, IL-17, and TNF signaling. Machine learning (SVM-RFE, LASSO, and Random Forest) pinpointed five core genes (EPHB3, GRIA1, NR4A2, RAF1, VEGFA) with high diagnostic accuracy. Single-gene GSEA revealed their collective roles in disrupting DNA repair, metabolism, and immune regulation. Immune infiltration analysis linked these core genes to distinct alterations in both innate and adaptive immunity. Molecular docking predicted strong BPDE-core protein binding. Finally, we proposed an adverse outcome pathway (AOP), from BPDE-induced core gene dysregulation, through activation of key signaling and immune dysregulation, to persistent lung injury and fibrosis. These findings reveal a multi-mechanistic framework for BPDE-induced ILD, and highlight novel core genes with diagnostic and therapeutic potential, laying the groundwork for further investigation into environmental pollutant-driven lung fibrosis.