57, adjusted, P<0.001): Culture supernatants from growth at 8, 15 and 37 °C were analysed in a Vero cell cytotoxicity assay (Table 1). The initially observed feature was the full cytotoxicity shared by both bacterial species after culturing at 15 °C. At 8 °C, two of the four B. weihenstephanensis strains showed high cytotoxicity, while the other two were negative or very low. After growth at 37 °C, no cytotoxic activity was seen from the four B. weihenstephanensis strains, while two of three B. cereus strains were TSA HDAC cost highly cytotoxic (80–100% range) and one strain (B. cereus NVH 1230-88)

was low (in the 20–50% range). The production of adequate levels of three-component protein cytotoxins Nhe and/or Hbl results in cytotoxicity in the Vero cell assay (Lund & Granum, 1996; Prüss et al., 1999; Dietrich et al., 2005). Components of those toxins were sought in the tested bacterial culture supernatants using commercially available antibody-based kits (Table 2). The L2 component of Hbl was detected at all three temperatures in all strains, except in this website B. cereus strain NVH 0075-95, which does not contain the operon encoding Hbl (Granum et al., 1996). On the other hand, the Nhe component NheA was detected

in all strains after growth at 15 °C, while when grown at 37 °C, NheA was only found in one of the four B. weihenstephanensis strains. One B. weihenstephanensis strain did not produce NheA at 8 °C. The boar spermatozoa assay was used to screen the Bacillus spp. strains for the production of cereulide (emetic B. cereus toxin) under standard conditions. Only one of the seven Bacillus strains (a B. cereus strain, NVH 1230-88) was weakly positive in this assay (results not

shown). However, screening by PCR for the genes encoding cereulide was negative for all the strains (results not shown). In vivo virulence results from experiments performed on G. mellonella larvae with the seven bacterial strains are presented in Fig. 1, which shows kinetics of induced mortality through the observation period following larval infection by two routes and at both temperatures. In order to give a general picture of virulence and because the exact dose repeats PAK5 were not possible to obtain, larval mortalities are shown as raw data, and B. cereus and B. weihenstephanensis strains are each grouped together. The results reveal variation in virulence level among strains (individual strain results not shown), even in the same group, but differences are particularly marked for mortality speed related to temperature between B. cereus and B. weihenstephanensis. Thus, using our linear regression model for in vivo experimental data, a significant difference was found in virulence between B. weihenstephanensis and B. cereus species (P<0.001), observed mainly at 37 °C (Fig. 1). Indeed, the psychrotolerant strains mostly showed negligible virulence both 24 and 90 h after infection, while B. cereus strains were generally highly virulent at 37 °C.