Environmental toxicant 2,3,7,8-tetrachlorodibenzo-p-dioxin induces non-obstructive azoospermia: New insights from network toxicology, integrated machine learning, and biomolecular modeling

Abstract

Objective: As industrial pollution intensifies, global male semen quality has been declining at a rate of 2.64 % per year in the 21st century. Among the various types of infertility, non-obstructive azoospermia (NOA) is the most severe and is closely associated with exposure to environmental toxins. The molecular mechanisms by which 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), a typical persistent organic pollutant, induces NOA have yet to be systematically elucidated.

Methods: This study employed the single-sample Gene Set Enrichment Analysis (ssGSEA) method to identify key toxicological pathways and constructed a diagnostic model based on 113 machine learning algorithms. By integrating Weighted Gene Co-expression Network Analysis (WGCNA) and single-cell analysis, we identified hub genes associated with the Sertoli Cell-Only Syndrome (SCOS) subtype. Finally, biomolecular modeling was conducted to validate the binding efficacy of the hub genes with TCDD.

Results: Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis based on the ssGSEA method indicated that TCDD may disrupt spermatogenesis by activating the Tumor Necrosis Factor (TNF) and Mitogen-Activated Protein Kinase (MAPK) signaling pathways while inhibiting the Vascular Endothelial Growth Factor (VEGF) signaling pathway, ultimately leading to NOA. Through the integration of machine learning techniques, 5 hub genes (AUC > 0.7) induced by TCDD and associated with NOA were identified: Androgen receptor (AR), Chromodomain Helicase DNA-Binding Protein 1 (CHD1), Discoidin Domain Receptor Tyrosine Kinase 2 (DDR2), Retinoic Acid Receptor-Related Orphan Receptor Alpha (RORA), and Glutamate Ionotropic Receptor AMPA Type Subunit 1 (GRIA1). WGCNA and single-cell analysis revealed that AR and DDR2 were specifically expressed in the testicular tissues of NOA patients and were closely associated with SCOS (p < 0.05). Immune infiltration analysis suggested that TCDD induces abnormal infiltration of various immune cells, indicating its close relationship with immune inflammatory responses (p < 0.05). Biomolecular modeling further demonstrated a strong binding affinity between AR and TCDD (∆G = -8.3 kcal·mol⁻¹, E_total = -37_.79 kcal·mol⁻¹), highlighting the critical role of AR in TCDD-induced NOA.

Conclusions: This study reveals the potential molecular mechanisms by which TCDD induces NOA, providing new targets for the development of diagnostic and therapeutic strategies.

Keywords: 2,3,7,8-Tetrachlorodibenzo-p-dioxin; Biomolecular Modeling; Machine Learning; Network Toxicology; Non-Obstructive Azoospermia.