Abstract
Diethyl phthalate (DEP) is a ubiquitous environmental endocrine-disrupting chemical (EDC). Epidemiological studies have suggested a potential association between DEP exposure and an increased risk of endometrial cancer (EC); however, its underlying molecular mechanisms remain largely unclear. Four GEO datasets were integrated, and differential expression analysis combined with weighted gene co-expression network analysis (WGCNA) was performed to identify candidate genes potentially linking DEP exposure to EC. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were conducted to explore relevant signaling pathways. Machine learning models, coupled with Shapley Additive Explanations (SHAP), were employed to prioritize key genes. Molecular docking and molecular dynamics (MD) simulations were used to assess the binding affinity between DEP and the identified targets. A series of in vitro experiments in EC cell lines were subsequently conducted to validate the biological effects of DEP. Nineteen overlapping DEP-EC genes were identified, predominantly enriched in the MAPK, cAMP, and cGMP-PKG signaling pathways. Among them, FOS, NR4A1, ADRA2C, JUN, and SLC6A2 were prioritized as core genes through machine learning and SHAP analysis. Molecular simulations confirmed stable binding between DEP and these targets. In vitro assays demonstrated that DEP exposure induces oxidative stress, significantly enhances ERK1/2 and AKT phosphorylation, upregulates Cyclin D1/CDK4 expression, promotes G1/S phase transition, and facilitates EC cell proliferation. These findings suggest that DEP may promote endometrial carcinogenesis by triggering oxidative stress-mediated signaling crosstalk and accelerating cell cycle progression. This study establishes a multi-layered methodological framework-from computational screening and machine learning to experimental validation-offering novel mechanistic insights into the carcinogenic potential of environmental endocrine disruptors such as DEP.