Ochratoxin A (OTA) is a naturally occurring toxin produced by molds found in common foods such as grains, nuts, wine, coffee, cheese, and processed meats. It is one of the most widespread food contaminants in the world, detected across many geographic regions and linked to significant health risks.
In 1993, the International Agency for Research on Cancer classified OTA as a Group 2B carcinogen, meaning it is possibly carcinogenic to humans. Laboratory studies in animals have confirmed its ability to damage kidneys, alter DNA, disrupt the cell cycle, and promote cancer through multiple biological pathways.
Clear cell renal cell carcinoma (ccRCC) is the most common and most deadly form of kidney cancer, accounting for about 85 percent of all renal cell carcinoma cases. It originates from the cells lining the kidney's filtering tubules and is often diagnosed at advanced stages when treatment becomes difficult.
While OTA is known to accumulate in kidney tissue and cause nephrotoxicity, the precise molecular mechanisms linking OTA exposure to ccRCC development have not been well understood. This study set out to systematically explore those potential links using computational biology tools.
The study used a network toxicology framework, which maps the connections between a toxic compound, its molecular targets, and a disease to understand complex relationships that single-target studies cannot capture. This is especially useful for multi-pathway toxins like OTA.
Potential OTA target genes were identified using three complementary databases: ChEMBL, SwissTargetPrediction, and SEA, yielding a final list of 232 predicted OTA target genes. ccRCC-related genes were identified from two public gene expression datasets by combining differential expression analysis and WGCNA network analysis, generating 3,224 cancer-related genes.
The intersection of OTA and ccRCC gene sets produced 56 shared target genes, which were then analyzed using Gene Ontology (GO) and KEGG pathway enrichment to understand what biological processes they participate in.
To identify which of the 56 genes are most important, the researchers built 113 machine learning prediction models using various algorithm combinations. The best-performing model was then interpreted using SHAP analysis, which reveals how much each gene contributes to the model's predictions in a way that is human-readable.
Finally, molecular docking was used to computationally test whether OTA can physically bind to the five core target proteins, providing a structural basis for the hypothesis that OTA interacts with these proteins to promote cancer.
The overlap between OTA targets and ccRCC-associated genes produced 56 shared target genes, representing a set of molecules that may be influenced by OTA in ways that are specifically relevant to kidney cancer development.
Gene Ontology analysis revealed that these shared genes are involved in processes critical to cancer biology: extracellular matrix (ECM) remodeling, including integrin binding and metallopeptidase activity, which affect how cancer cells invade surrounding tissue and spread.
KEGG pathway analysis further showed enrichment in cancer-relevant signaling pathways including the Rap1 pathway, TNF signaling, IgSF CAM interactions, focal adhesion, and leukocyte transendothelial migration. These pathways regulate how tumor cells migrate, evade immune detection, and recruit blood vessels.
To verify that these findings were specific to OTA rather than generic mycotoxin effects, researchers performed a negative control analysis using Ochratoxin B (OTB), a structurally similar compound that lacks OTA's chlorine atom. OTB overlapped with only 39 ccRCC genes versus 56 for OTA, and uniquely missed pathways like Rap1 and IgSF CAM signaling that are particularly relevant to ccRCC, confirming OTA's specificity.
Among 113 machine learning models tested, the glmBoost + RF (Generalized Linear Model Boosting followed by Random Forest) combination achieved the best performance, with an AUC of 0.999 in the training dataset and values between 0.967 and 1.000 in three independent validation cohorts. This means the model can distinguish tumor from normal tissue with near-perfect accuracy.
SHAP interpretability analysis identified five core genes driving the model: IGFBP3, ITGA5, PYGL, SLC22A8, and LTB4R. IGFBP3 and ITGA5 contributed the most to predictions, making them the principal driver genes in the OTA-ccRCC relationship.
IGFBP3 (Insulin-like Growth Factor Binding Protein 3) was significantly upregulated in ccRCC tissues and is known to promote tumor cell proliferation, migration, and invasiveness through the AKT/STAT3/MAPK-Snail signaling axis. OTA may amplify these cancer-promoting effects by elevating IGFBP3 expression.
ITGA5 (Integrin Alpha 5) helps cancer cells adhere to and remodel the extracellular matrix, promoting invasion and angiogenesis. Its high expression in ccRCC correlated with poor prognosis, and OTA binding to ITGA5 may enhance these aggressive cellular behaviors.
SLC22A8, a kidney detoxification transporter, was the only core gene significantly downregulated in ccRCC. Its reduced expression may impair the kidney's ability to clear OTA and other toxins, creating a microenvironment that is more permissive to DNA damage and malignant transformation.
Single-cell RNA sequencing analysis of kidney tissue samples identified ten distinct cell populations including T cells, CD8+ T cells, macrophages, endothelial cells, NK cells, monocytes, B cells, adipocytes, hepatocytes, and epithelial cells.
IGFBP3 and ITGA5 were predominantly expressed in endothelial cells, the cells lining blood vessels. This suggests that OTA may influence tumor progression not by directly transforming kidney tubule cells but by disrupting the blood vessel cells that support tumor growth and facilitate metastasis.
PYGL and LTB4R were expressed mainly in monocytes and macrophages, immune cells that patrol tissues and can either fight or support tumors depending on their activation state. OTA exposure may skew these cells toward a tumor-supporting, inflammatory phenotype.
SLC22A8 showed the highest expression in cells likely representing proximal tubular epithelial cells, consistent with its known role as a kidney-specific organic anion transporter. Its loss in ccRCC tissue suggests that OTA may directly impair the kidney's detoxification capacity, furthering the toxic environment conducive to cancer.
Molecular docking was performed to test whether OTA physically interacts with the protein products of the five core genes. All five proteins showed strong predicted binding affinity for OTA, with binding energies all below -5.0 kcal/mol, a commonly used threshold for significant molecular interaction.
The strongest predicted binding was to PYGL at -10.5 kcal/mol, followed by SLC22A8 at -9.8 kcal/mol and LTB4R at -9.8 kcal/mol. ITGA5 showed binding energy of -9.2 kcal/mol and IGFBP3 at -8.0 kcal/mol. All values indicate that OTA can dock stably within protein binding pockets.
For genes whose proteins promote tumor growth when upregulated (IGFBP3, ITGA5, PYGL, LTB4R), OTA binding may disrupt normal protein regulation and shift signaling toward cancer-promoting states. For the protective gene SLC22A8, OTA binding likely inhibits its detoxification function, compounding its loss of expression.
The researchers emphasize that these docking results are computational predictions and not yet experimentally confirmed under physiological conditions. The findings are best understood as hypothesis-generating evidence that warrants further laboratory testing.
The researchers propose that OTA promotes ccRCC through a multi-pathway mechanism involving oxidative stress and DNA damage, disruption of cell signaling, epigenetic alterations, and direct protein interactions. No single mechanism accounts for OTA's carcinogenic potential; rather, all these pathways likely work together.
Key signaling pathways including Rap1, TNF, and IgSF CAM were uniquely enriched in the OTA-ccRCC gene overlap but not in either ccRCC genes or OTA targets alone. This suggests OTA specifically potentiates these oncogenic pathways, particularly those governing cell migration, adhesion, and immune evasion.
PYGL's role in metabolic reprogramming is particularly notable. As a rate-limiting enzyme in glycogen breakdown, PYGL helps cells survive under hypoxic conditions typical in tumors. OTA stress may upregulate PYGL, fueling a metabolic shift that helps damaged cells survive and eventually become malignant.
The proposed conceptual framework organizes the five core genes into four functional modules: ECM remodeling and angiogenesis (ITGA5, IGFBP3), metabolic reprogramming (PYGL), immune microenvironment modulation (LTB4R), and detoxification and transport (SLC22A8). Disruption across all four modules may be required for OTA to drive cancer development.
This study provides the first systematic computational evidence that OTA may specifically interact with molecular pathways central to ccRCC development, going beyond simple toxicity to suggest a potential role in kidney cancer initiation and progression.
The five core genes identified, particularly IGFBP3, ITGA5, PYGL, SLC22A8, and LTB4R, represent candidate biomarkers for OTA-associated kidney cancer risk. Future studies could test whether patients with higher OTA exposure show altered expression of these genes in their kidney tissue.
Importantly, the authors emphasize that all findings in this study are computational and hypothesis-generating. They do not prove a direct causal relationship between OTA consumption and kidney cancer in humans. Experimental validation using cell cultures, animal models, and ultimately human cohort studies is needed.
For patients and families, this research underscores the importance of food safety measures and reducing mycotoxin exposure, as well as the growing recognition that environmental factors may influence kidney cancer risk alongside genetic predispositions and lifestyle factors. Regulatory limits on OTA in food products remain an important public health consideration.