Chicken embryos are a valuable model for the selection of Bacillus subtilis for probiotic purposes.

Bacillus subtilis (BS) has been used as an excellent probiotic; however, some BS strains seem to be opportunist pathogens or do not present inhibitory effects in the pathogenic bacteria, so the characterization of BS strains for use in animals is mandatory. This study aimed to select nonpathogenic strains of BS, which can inhibit Salmonella spp., avian pathogenic Escherichia coli (APEC), and Campylobacter jejuni (CJ) using a chicken embryo as a model. We tested nine (9) strains of BS isolated from several sources (named A to I) in in vitro by tests of mucin degradation activity, haemolytic activity, apoptosis, and necrosis in fibroblasts from chickens. After the in vitro test, we tested the remaining seven (7) strains (strains A to G) in a chicken embryo (CE) as an in vivo model and target animal. We inoculated 3 log CFU/CE of each strain via allantoic fluid at the 10th day postincubation (DPI). Each treatment group consisted of eight CEs. At the 17th DPI we checked CE mortality, gross lesions, CE weight, and whether BS strains were still viable. To perform the cytokine, total protein, albumin, and reactive C protein analysis, we collected the CE blood from the allantoic vessel and intestine fragments in the duodenum portion for histomorphometric analysis. After the results in CEs, we tested the inhibition capacity of the selected BS strains for diverse strains of Salmonella  Heidelberg (SH), S. Typhimurium (ST), S. Enteritidis (SE), S. Minnesota (SM), S. Infantis (SI), Salmonella var. monophasic (SVM), APEC and C. jejuni. After the in vitro trial (mucin degradation activity, haemolytic activity, apoptosis, and necrosis), we removed two (2) strains (H and I) that showed β-haemolysis, mucin degradation, and/or high apoptosis and necrosis effects. Although all strains of BS were viable in CEs at the 17th DPI, we removed four (4) strains (A, B, D, F) once they led to the highest mortality in CEs or a high albumin/protein ratio. C. jejuni inoculated with strain G had greater weight than the commercial strain, which could be further used for egg inoculation with benefits to the CE. From the tests in CEs, we selected the strains C, E, and G for their ability to inhibit pathogenic strains of relevant foodborne pathogens. We found that the inhibition effect was strain dependent. In general, strains E and/or G presented better or similar results than commercial control strains in the inhibition of SH, ST, SI, APEC, and two (2) strains of CJ. In this study, we selected BS strains C, E and G due to their in vitro and in vivo safety and beneficial effects. In addition, we emphasize the value of CE as an in vivo experimental model for assessing BS's safety and possible benefits for poultry and other animals.