Single nucleotide polymorphisms associated with cytotoxicity of Bacillus cereus group strains in Caco-2 cells

Bacillus cereus sensu lato (s.l.) encompasses strains with diverse impacts, ranging from foodborne illness and anthrax to beneficial applications in agriculture and industry. While the risk of anthrax and emetic intoxication can be reliably predicted by the presence of specific virulence genes, predicting diarrheal foodborne illness risk based solely on enterotoxin gene presence has proven unreliable. In this study, we evaluated cytotoxicity against Caco-2 human gut cells using a diverse collection of B. cereus s.l. isolates representing all eight panC phylogenetic groups and conducted genomic analyses to identify predictive markers of cytotoxicity. Isolates from panC groups I, IV, and V exhibited significantly higher cytotoxicity compared to other groups, although individual isolates from other panC groups have also been linked to illness. Logistic and random forest regression models revealed that while the presence of enterotoxin genes was a sensitive indicator of cytotoxicity, it lacked specificity. Logistic regression analysis identified 21 nonsynonymous single nucleotide polymorphisms (SNPs) within enterotoxin (Nhe and Hbl) gene sequences that were more effective predictors of cytotoxicity, providing higher specificity with comparable sensitivity. These SNPs achieved accuracy and precision values exceeding 0.7. Random forest models highlighted the importance of panC group, enterotoxin gene SNPs, and the presence of the full hbl operon as key predictors of cytotoxicity. The strong sensitivity, specificity, and biological relevance of these SNPs position them as promising markers for improving strain-based risk assessment of B. cereus s.l. IMPORTANCEEnterotoxin genes have been associated with B. cereus sensu lato (s.l.) diarrheal foodborne illness; however, their mere presence in a genome is an unreliable predictor of an isolates cytotoxicity towards human gut epithelial cells. To improve food safety risk assessment, more specific markers of cytotoxicity are required. In this study, we identified nonsynonymous SNPs within the coding sequences of the enterotoxins Nhe and Hbl. These SNPs offer potential targets for rapid molecular tests to identify B. cereus s.l. isolates with an elevated food safety risk due to their capacity to inflict cytotoxic damage on human gut epithelial cells. Implementation of such markers upon validation could improve consumer safety while reducing food waste.