INSR as a Potential Target of Bisphenol F-Induced Type 2 Diabetes: Causal Inference and Mechanistic Elucidation via Mendelian Randomization, Multi-Omics Integration, Single-Cell Transcriptomics, and Molecular Dockin

Background Bisphenol F (BPF), a widely adopted substitute for bisphenol A, is a suspected endocrine-disrupting chemical; however, its causal relationship with type 2 diabetes mellitus (T2D) and the underlying molecular mechanisms remain inadequately characterized. Methods Two-sample Mendelian randomization (MR) using European GWAS data evaluated the causal effect of BPF exposure on T2D and eight common diabetic complications. Differential expression analysis and weighted gene co-expression network analysis (WGCNA) were applied to GSE76894 to identify T2D-associated genes. Three machine learning algorithms—Random Forest, XGBoost, and LASSO regression—were integrated to identify the core gene, with diagnostic performance validated in GSE64998. Single-cell RNA sequencing data (GSE255566) were analyzed using Seurat for immune cell annotation and CellChat for intercellular communication profiling. Molecular docking and 100 ns molecular dynamics (MD) simulation were conducted to characterize BPF–INSR interaction. Results MR analysis demonstrated a significant positive causal association between BPF exposure and T2D (OR = 1.006, 95% CI: 1.000–1.013, P = 0.040), with sensitivity analyses confirming robustness. All three machine learning algorithms converged on INSR (insulin receptor) as the sole core gene; INSR was significantly upregulated in T2D and achieved an AUC of 0.92 in the training set and 0.82 in the independent validation set. ScRNA-seq analysis identified five peripheral blood immune cell types, revealing a marked reduction in neutrophil proportion alongside monocyte expansion in T2D; INSR exhibited the highest expression in monocytes. CellChat analysis identified the IGF1–INSR ligand–receptor pair as a central intercellular communication axis, with monocytes functioning as the primary signaling hub and monocyte–neutrophil communication showing the strongest interaction intensity. Molecular docking confirmed stable BPF binding within the INSR ligand-binding pocket (ΔG = − 6.7 kcal/mol) at key residues Tyr232, Met56, and Ala227, with MD simulation corroborating complex stability over 100 ns. Conclusion This integrative multi-omics study establishes a genetic causal link between BPF exposure and T2D and identifies INSR as the core molecular target mediating BPF-induced metabolic and immune dysregulation, providing a mechanistic framework and a candidate biomarker for environmentally driven insulin resistance.