Production of hemolysin BL by Bacillus cereus group isolates of dairy origin is associated with whole-genome phylogenetic clade

Abstract

Background: Bacillus cereus group isolates that produce diarrheal or emetic toxins are frequently isolated from raw milk and, in spore form, can survive pasteurization. Several species within the B. cereus group are closely related and cannot be reliably differentiated by established taxonomical criteria. While B. cereus is traditionally recognized as the principal causative agent of foodborne disease in this group, there is a need to better understand the distribution and expression of different toxin and virulence genes among B. cereus group food isolates to facilitate reliable characterization that allows for assessment of the likelihood of a given isolate to cause a foodborne disease. Results: We performed whole genome sequencing of 22 B. cereus group dairy isolates, which represented considerable genetic diversity not covered by other isolates characterized to date. Maximum likelihood analysis of these genomes along with 47 reference genomes representing eight validly published species revealed nine phylogenetic clades. Three of these clades were represented by a single species (B. toyonensis -clade V, B. weihenstephanensis - clade VI, B. cytotoxicus - VII), one by two dairy-associated isolates (clade II; representing a putative new species), one by two species (B. mycoides, B. pseudomycoides - clade I) and four by three species (B. cereus, B. thuringiensis, B. anthracis - clades III-a, b, c and IV). Homologues of genes encoding a principal diarrheal enterotoxin (hemolysin BL) were distributed across all, except the B. cytotoxicus clade. Using a lateral flow immunoassay, hemolysin BL was detected in 13 out of 18 isolates that carried hblACD genes. Isolates from clade III-c (which included B. cereus and B. thuringiensis) consistently did not carry hblACD and did not produce hemolysin BL. Isolates from clade IV (B. cereus, B. thuringiensis) consistently carried hblACD and produced hemolysin BL. Compared to others, clade IV was significantly (p = 0.0001) more likely to produce this toxin. Isolates from clade VI (B. weihenstephanensis) carried hblACD homologues, but did not produce hemolysin BL, possibly due to amino acid substitutions in different toxin-encoding genes. Conclusions: Our results demonstrate that production of diarrheal enterotoxin hemolysin BL is neither inclusive nor exclusive to B. cereus sensu stricto, and that phylogenetic classification of isolates may be better than taxonomic identification for assessment of B. cereus group isolates risk for causing a diarrheal foodborne disease.