Mechanistic insights into aristolochic acid-induced hepatocellular carcinoma: a multi-dimensional analysis integrating network toxicology, machine learning, and molecular dynamics simulation

Abstract

Background

Aristolochic acids (AA) are naturally occurring carcinogens found in traditional herbal medicines derived from Aristolochia species. This study explores the potential link between AA and hepatocellular carcinoma (HCC), aiming to uncover key molecular targets driving AA-induced hepatocarcinogenesis.

Methods

Toxicogenomic databases were used to identify AA-related toxicological profiles and targets, which were integrated with HCC-associated gene datasets. Protein-protein interaction networks were constructed, followed by enrichment analyses. A multi-machine learning framework was applied to prioritize candidate genes. Bioinformatic validation included immunohistochemistry, survival analysis, and immune infiltration profiling. Compound-protein interactions and structural stability were assessed via molecular docking and molecular dynamics simulations.

Results

We identified 2378 overlapping genes between AA and HCC. Enrichment analyses revealed significant activation of proteasome, cell cycle, and antigen processing pathways. PSMB4 was prioritized as a core gene, overexpressed in HCC tissues, associated with poor prognosis (HR = 1.69, p < 0.05), and strongly correlated with regulatory T cell and M2 macrophage infiltration. Docking and MD simulations confirmed strong and stable binding (−8.7 kcal/mol) between AA and PSMB4.

Conclusion

This study establishes a potential mechanistic connection between AA and HCC using an integrated network toxicology and machine learning, offering novel insights into toxicant-driven hepatocarcinogenesis and potential therapeutic targets.