Study on spore-forming Bacillus species involved in bread spoilage, contamination risk evaluation and bio-preservation tool

Contamination and persistence of Bacillus spp. spores in cereal products is a real concern for industrial production. The microbial eco-system of raw materials used for bread-making and in particular spore-forming bacteria play a role in the development of "bread rope spoilage". Heat-resistant spores can survive the baking process and under favourable conditions spore germination occurs and vegetative cells cause the rope spoilage, a deterioration process of bread texture. The knowledge of species occurring in bread and ingredients could help to evaluate the risk of contamination of the final product and its spoilage. Therefore the aim of this thesis was to evaluate the microbial contamination risk with predictive microbiology by investigating the spore-forming bacterial species involved in bread spoilage and studying the influence of bread formulation on the growth behaviour of natural sporeformers. A high bacterial diversity characterized raw materials. In total 176 isolates were collected and characterized: 13 bacterial species belonging to Bacillus (10) and Paenibacillus (3) genera were identified by sequencing of 16S rRNA, gyrA or gyrB genes. Bacillus amyloliquefaciens was identified as the species frequently associated with the ropy spoilage of bread (Chapter 2) The effects of wheat bran and Lactobacillus brevis based-bioingredient (LbBio) on microbiological and physico-chemical quality of yeast leavened wheat bread were investigated. Results indicated that bread formulation can influence the behaviour of naturally-occurring bacteria during bread storage and the consequent spoilage. Moreover, LbBio counteracted the negative effects of bran and allowed to obtain an enriched fibre bread with specific volume and soft crumb comparable to bread without bran (Chapter 3). To evaluate the contamination risk with the predictive microbiological tool "Sym'Previus", wheat bread containing or not bran or LbBio, was challenged with a selected B. amyloliquefaciens ISPA-S109.3 isolated from grain. Growth cardinal and spore thermal resistance parameters of the selected strain were determined and used, together with kinetic data from challenge tests, to evaluate the contamination risk of bread during storage when some physico-chemical factors were modified. The cardinal parameters of the ISPA-S109.3 strain were compared with those of the reference strain B. amyloliquefaciens ATCC8473 and allowed to obtain the growth/no growth boundaries for each strain in the studied food matrix. Considering the physico-chemical properties of white wheat bread, the storage temperature and the cardinal growth parameters of the B. amyloliquefaciens strains, a high probability of growth (> 90%) in bread was expected for strain ISPA-S109.3, while for the reference strain ATCC8473 this probability lowered to the 50%. Results from the spore thermal resistance experiments indicated a higher resistance for the reference strain ATCC8473 (D*: 0.765 min) with respect to the ISPA-S109.3 strain (D*: 0.027 min) even if the last one survived to the bread cooking process leading after three days storage to a bacterial count of 5.65 - 6.54 log CFU/g in white wheat bread and in wheat bran bread, respectively. The predictive microbiological tool Sym'Previus allowed to estimate the risk of bread contamination by studying the bacterial behaviour of a spore-forming strain which survived the cooking process, grew in bread during storage and finally caused the rope spoilage in bread. The simulation tool allowed to estimate a contamination density different from both bread types; in particular wheat bran bread resulted to be in all probability contaminated by the B. amyloliquefaciens strain ISPA-S109.3 respect to the white wheat bread. To determine whether predictions provide good description of growth in bread, simulation results were compared with additional experimental data obtained changing physico-chemical parameters obtaining a low (<1 log unit) discrepancy between observed and predicted log N (Chapter 4). Further experimental data should be collected to validate definitively and assess the reliability of the model.