Predictive food microbiology tools for Bacillus cereus in dairy products

Bacillus cereus sensu lato is a group of spore-forming bacterial pathogens commonly found in various food types, including dairy products. The aim of this PhD project was to (i) evaluate the prevalence of B. cereus subgroups with different characteristics in dairy products; (ii) develop extensive growth and growth boundary models; and (iii) develop sporulation models; to improve the safety of dairy products by understanding the characteristics and behavior of B. cereus.

A comprehensive investigation of B. cereus s. l. in dairy products was performed. Out of 71 examined products, 49% were contaminated with B. cereus, with dairy powders (83%), pasteurized cheese whey (43%), and cheeses (42%) showing the highest prevalence and the highest level of contamination (up to 100 CFU/g). 81 strains, including isolates from dairy processing facilities, dairy products, and reference strains were characterized. Five panC groups (II, III, IV, VI, and VIII) were identified among these dairy B. cereus isolates. Almost all (97%) of the isolates carried one or more toxin genes, with 42% demonstrating lactose fermentation. In addition, growth rates of isolates at low and high temperatures (10 and 45°C), low pH of 5.1, and high concentrations of NaCl (6%) were determined. Significant variability in growth rates both between and within panC groups was observed. Tolerance to these growth conditions, the toxin gene profile, and lactose fermentation abilities, were used as criteria for selection of isolates. Subsequently, two distinct bacterial cocktails, one for mesophilic B. cereus (containing six isolates) and another for psychrotolerant B. cereus (containing seven isolates), were selected. These carefully selected strain cocktails were used for developing predictive models and conducting challenge tests to evaluate bacterial growth and sporulation kinetics in dairy products.

Using the two selected cocktails of strains, two extensive cardinal parameter models for predicting growth and growth boundaries of mesophilic and psychrotolerant B. cereus subgroups were developed. Each model incorporated the combined inhibitory effect of 11 environmental factors, including temperature, pH, water activity (NaCl concentration), organic acids (acetic, benzoic, citric, lactic, and sorbic acids), and phosphate salts (orhto-, di-, and tri-phosphates). The models were developed using 344 and 303 maximum specific growth rates (μ_max) for mesophilic and psychrotolerant B. cereus subgroups, respectively. These growth responses were determined using both a standard laboratory medium (BHI broth) and a dairy-specific liquid medium (Ultra-filtrated (UF) permeate from whey). Both mesophilic and psychrotolerant B. cereus cocktails showed greater acid tolerance in the UF permeate compared to standard BHI broth. This was evidenced by a lower minimum pH for growth (pH_min) and higher minimum inhibitory concentrations of lactic acid. Mesophilic B. cereus showed higher tolerance to citric acid as well. The cardinal model parameters were conservatively selected from both media, aiming to capture the widest growth range for each environmental factor.

These models were further extensively evaluated for predicting growth and growth boundaries of mesophilic and psychrotolerant B. cereus in dairy products. This evaluation included 66 and 67 new challenge tests for mesophilic and psychrotolerant isolates, respectively. Additionally, 139 and 109 growth responses from published studies were also included in models’ evaluation. The mesophilic model demonstrated good performance, with bias-/accuracy-factor values of 1.12/1.50 and 80% correct predictions in new challenge tests, and 91% correct predictions for literature data. The psychrotolerant model showed similarly strong results, with 84% correct predictions in new challenge tests. While facing substantial limitations in predicting growth at temperatures below 6°C, the model had a good performance at temperatures above 6°C, achieving 91% correct predictions for growth/no-growth responses from literature data.

The selected cocktails of strains were further used to investigate spore formation kinetics of B. cereus in dairy solutions and milk under diverse environmental conditions. Primary and secondary models were proposed to simultaneously predict growth and spore formation kinetics. The minimum temperatures for spore formation of both mesophilic and psychrotolerant B. cereus were close to their theoretical minimum growth temperatures. The maximum spore yield was strongly matrix-dependent, with environmental conditions showing no clear systematic effect on the sporulation yield. Moreover, sporulation was initiated during the stationary phase, suggesting that high cell concentrations are crucial for triggering spore formation in dairy matrices.

The developed and validated growth and growth boundary models offer an important advancement in predicting the growth responses of B. cereus subgroups. By considering storage temperature and a wide range of dairy product characteristics, including the combined effect of organic acids and phosphate salts, these models provide unbiased predictions. These new models can support the evaluation and management of the two B. cereus subgroups in various dairy products. The sporulation models provide valuable insights for evaluating how processing and storage conditions impact B. cereus spore formation in dairy products, ultimately contributing to more effective food safety strategies.