Integration of network toxicology, machine learning and single-cell sequencing reveals the effects of soil pollutants on prostate cancer

BACKGROUND: Prostate cancer (PRAD) is a common malignancy in men, and exposure to soil pollutants may contribute to its development. And exposure to soil pollutant has been linked to its development, as well as to other diseases including cardiovascular disorders, neurological conditions, and additional cancers. METHODS: This study integrates network toxicology, machine learning, and advanced technologies to investigate the mechanisms through which soil pollutants affect prostate cancer. Network toxicology was used to identify targets and signaling pathways associated with soil pollutants. Target prediction of soil pollutants was performed through SwissTargetPrediction and the Similarity Ensemble Approach (SEA). Prostate cancer-associated target genes were obtained from the Genecards and OMIM databases. Protein-protein interaction analysis further identified key regulatory molecules. Then a predictive model was constructed with 101 machine learning algorithms, identified risk target molecules and assessed the relationship between soil pollutants and prostate cancer risk. Single-cell sequencing examined the expression profiles of these targets in prostate cancer cell types. Molecular docking simulations evaluated the binding affinities between pollutants and target proteins to explore potential molecular mechanisms. RESULTS: Ten soil pollutants were analyzed, including dibutyl phthalate (DBP), dioctyl phthalate (DOP), dimethyl phthalate (DMP), perfluorooctanoic acid (PFOA), perfluorooctanesulfonic acid (PFOS), methyl ethyl ketone (MEK), toluene (TOL), zearalenone (ZEN), ochratoxin A (OTA), and deoxynivalenol (DON). Integration of multiple databases identified 242 potential targets, and protein-protein interaction analysis revealed key regulatory molecules. A prognostic risk model identified 26 risk target molecules. Single-cell sequencing demonstrated their expression profiles in prostate cancer. Kaplan-Meier, univariate, and multivariate Cox regression analyses revealed independent prognostic significance for DNMT1 and MTOR. Molecular docking confirmed that most soil pollutants bind strongly to DNMT1 and MTOR, suggesting disruption of related signaling pathways. Furthermore, experiments were conducted to explore the relationship between soil pollutants and target proteins. CONCLUSIONS: Various analytical methods to elucidate mechanisms by which soil pollutants contribute to prostate cancer through interactions with key molecular targets DNMT1 and mTOR. These findings inform the prevention and treatment of environmentally related cancers and provide novel insights into the molecular connection between soil pollutants and prostate cancer, and laying the groundwork for preventive and intervention strategies targeting environmentally related cancers.