Integrating network toxicology, machine learning, and single-cell sequencing to reveal the FASN-mediated role of phenolic endocrine disruptors in water in promoting prostate cancer

Abstract

Background Phenolic endocrine-disrupting chemicals (EDCs) like nonylphenol (NP) and octylphenol (OP) are widespread water pollutants. Their estrogen-like properties are suspected contributors to prostate cancer, but their precise molecular mechanisms remain unclear. Methods We employed a multidimensional framework to investigate this link. Potential NP/OP targets were predicted using SwissTargetPrediction, SEA, and CTD databases and cross-referenced with prostate cancer-associated genes from GeneCards and OMIM. Differential expression analysis of the GSE46602 dataset (36 tumor vs. 14 benign samples) identified candidate genes, which were refined to core genes using Least Absolute Shrinkage and Selection Operator (LASSO) and Support Vector Machine-Recursive Feature Elimination (SVM-RFE) algorithms. Their diagnostic power was evaluated via an Artificial Neural Network (ANN) model and validated in The Cancer Genome Atlas (TCGA) cohort. Single-cell RNA sequencing data from six prostate cancer samples (GSE137829) were analyzed to reveal cell-type-specific expression patterns. Molecular docking and molecular dynamics (MD) simulations assessed binding stability between pollutants and target proteins. Results We identified 143 overlapping genes between NP/OP targets and prostate cancer-associated genes, significantly enriched in lipid metabolism and prostate cancer pathways (adjusted P < 0.05). Dual-algorithm screening identified four core genes (ENPP2, FASN, PTGS2, and CHRM1). Among them, Fatty Acid Synthase (FASN) exhibited the best diagnostic performance in the TCGA validation cohort (AUC = 0.800), outperforming PTGS2 (0.783), ENPP2 (0.621), and CHRM1 (0.605), and was significantly overexpressed in prostate cancer tissues (|log2FC| > 1, adjusted P < 0.05). Single-cell analysis across seven annotated cell types revealed specific FASN overexpression in epithelial cells, with expression progressively upregulated along pseudotime disease trajectories. Gene Set Variation Analysis (GSVA) demonstrated significant activation of oncogenic pathways — including PI3K-AKT-mTOR, androgen response, and early estrogen response — in FASN-high epithelial cells. Molecular docking confirmed favorable binding of NP and OP to FASN (binding affinities of −6.0 and −6.1 kcal/mol, respectively), and MD simulations showed that both complexes reached stable equilibrium with RMSD fluctuations below 0.3 nm. Molecular Mechanics/Generalized Born Surface Area (MM/GBSA) calculations further yielded binding free energies of −19.70 kcal/mol (NP-FASN) and −17.24 kcal/mol (OP-FASN). Conclusion This study computationally identifies FASN as a potential molecular hub that may link phenolic EDC exposure to prostate cancer. Our bioinformatic analyses suggest a hypothetical mechanism involving pollutant-driven disruption of lipid metabolic reprogramming via FASN, potentially activating a pro-oncogenic network, which warrants future experimental validation.