Ochratoxin A (OTA) is a food-borne mycotoxin linked to liver toxicity and possible carcinogenesis; its connection to hepatocellular carcinoma (HCC) remains not fully elucidated. This study employs an integrated computational strategy combining public target databases, transcriptome profiling, machine learning algorithms, immune infiltration analysis, molecular docking, and molecular dynamics simulation to identify potential targets and pathways linking OTA exposure with HCC.
Overlap analysis identified 214 genes associated with both OTA toxicity and HCC, enriched in biological processes such as signal transduction, apoptosis, metabolism, and immune modulation. Analysis of the GSE36376 transcriptomic dataset revealed 443 differentially expressed genes in HCC, with 13 overlapping the OTA-HCC target list. Further feature selection prioritized five candidate genes: CYP3A4, KIFC1, AKR1C3, CA2, and TTR.
Expression patterns showed KIFC1 and AKR1C3 upregulated, while CYP3A4, CA2, and TTR were downregulated in HCC. These candidates demonstrated significant discriminatory ability with area under the curve (AUC) values ranging from 0.866 to 0.958. Molecular docking predicted favorable interactions between OTA and these targets, with CYP3A4 exhibiting the strongest binding (docking energy −10.8 kcal/mol). Molecular dynamics simulations of the OTA-CYP3A4 complex indicated stable binding dynamics.
The findings suggest that OTA may influence HCC development through pathways related to metabolism, genomic instability, and the immune microenvironment. This computational framework provides a reproducible approach to generate hypotheses for future experimental validation to clarify the molecular mechanisms by which OTA exposure contributes to hepatocarcinogenesis.
Hepatocellular carcinoma (HCC) is the most common primary liver cancer and a major cause of cancer mortality worldwide, with a poor prognosis and low 5-year survival. Etiological factors vary geographically; chronic viral hepatitis, aflatoxin exposure, alcohol use, and metabolic disorders are important contributors. Environmental toxins like OTA, a mycotoxin produced by Penicillium and Aspergillus species, contaminate diverse foods due to poor storage, leading to nephrotoxicity and hepatotoxicity.
Epidemiological evidence increasingly links OTA exposure with HCC risk, though the underlying molecular mechanisms are not fully elucidated. Traditional toxicology approaches focusing on single targets are limited in explaining the complex toxicity pathways. Emerging network toxicology and machine learning methods can integrate compound, target, pathway, and disease data to explore potential mechanisms systematically. Prior studies implicate xenobiotic metabolism, cell cycle regulation, reductase activities, carbonic anhydrases, and transport proteins in HCC pathogenesis, offering context to interpret computational findings.
This study integrates network toxicology, transcriptomic data analysis, machine learning-based feature selection, immune cell infiltration estimation, and molecular modeling to identify and prioritize candidate genes and pathways potentially mediating OTA-induced hepatocarcinogenesis.
The 2D molecular structure and SMILES representation of OTA were obtained from the PubChem database for subsequent analyses.
OTA-associated target genes were collected from ChEMBL, SEA, and SwissTargetPrediction databases, focusing on Homo sapiens. Duplicate entries were removed and gene names standardized using UniProt.
Genes related to HCC were retrieved from GeneCards (selection based on Relevance score > 5) and OMIM databases to compile the disease-related gene set.
Venn diagram analysis identified overlapping genes shared by OTA toxicity and HCC-associated gene sets.
The intersecting genes were converted to Entrez IDs and analyzed for Gene Ontology (biological process, cellular component, molecular function) and KEGG pathway enrichment using clusterProfiler in R. Significant enrichment was determined at p-value and q-value < 0.05.
The GSE36376 dataset, consisting of expression profiling by array for HCC samples and controls, was downloaded from the GEO database. Differentially expressed genes (DEGs) were identified using thresholds of p < 0.05 and |log2 fold change| > 1.
LASSO regression and support vector machine recursive feature elimination (SVM-RFE) were applied to prioritize key genes among overlapping DEGs. Immune cell infiltration analysis and immune-related association studies were performed, followed by molecular docking of OTA with prioritized targets and molecular dynamics simulations to evaluate binding stability.
Two hundred fourteen genes were identified at the intersection of OTA-related targets and HCC-associated genes. Functional enrichment revealed involvement in signal transduction pathways, apoptotic regulation, metabolic processes, and immune response.
Transcriptome analysis of GSE36376 found 443 DEGs, with 13 overlapping the OTA-HCC gene set. Feature selection methods narrowed candidates to five genes: CYP3A4, KIFC1, AKR1C3, CA2, and TTR. Expression profiling showed upregulation of KIFC1 and AKR1C3, and downregulation of CYP3A4, CA2, and TTR in HCC samples. These genes had strong predictive accuracy for HCC status (AUCs 0.866–0.958).
Molecular docking indicated energetically favorable OTA binding to these proteins, notably with CYP3A4 (−10.8 kcal/mol). Molecular dynamics simulations supported stable OTA-CYP3A4 complex formation during 20–100 ns trajectories, confirming binding plausibility.
This systems-level analysis integrates multiple bioinformatic and computational approaches, highlighting candidate targets and molecular pathways potentially linking OTA exposure to HCC pathogenesis. The identified genes encompass metabolic enzymes, mitotic regulators, reductases, carbonic anhydrases, and transport proteins previously implicated in liver cancer biology.
The predicted interactions with OTA and expression changes suggest that OTA may contribute to HCC by disrupting xenobiotic metabolism, promoting genomic instability, and modulating the tumor immune microenvironment. Further experimental validation in biological models and independent datasets is required to confirm these mechanistic hypotheses and to explore therapeutic or preventive implications.
This study demonstrates the utility of computational integrative analyses to generate testable hypotheses into environmental toxin-associated carcinogenesis processes, providing groundwork for future translational research in HCC risk and intervention.