Table 1.
List of the 49 Bacillus thuringiensis-associated foodborne outbreaks in France from 2007 to 2017.
Fig 1.
Frequency of association of Bt and non-Bt isolates with eight specific types of food.
The analysis was performed from 59 representative Bt and 437 representative non-Bt isolates, collected from FBO investigations (i.e. 1 isolate per FBO, per dish/foodstuff and per genotypic and phenotypic profile). The frequencies correspond to the % of each type of isolate collected from a dish containing tomatoes (including cooked tomatoes but excluding tomato sauce), lettuce, raw vegetables, fruits (raw and cooked), starch products, meat, fish or seafood, and spices or dried herbs.
Table 2.
List of the 10 commercial Bt strains collected for the study from 19 commercially available products, distributed into 4 different subspecies: Aizawai, kurstaki, israelensis and morrisoni.
Table 3.
Phenotypic and genotypic characteristics of representative FBO- and commercial Bt isolates.
Fig 2.
Nhe and Hbl production by a representative panel of Bt isolates composed of FBO-Bt isolates from group A (n = 21) and commercial Bt strains (n = 10).
(A) Nhe production was expressed as Nhe absorbance (414nm), corresponding to the coloration intensity obtained with the BDE VIATM kit (3M-Tecra), according to the manufacturer’s recommendations. (B) Hbl dilution values correspond to the highest dilution for which Hbl remained detectable, using the immunoenzymatic kit BCET-RPLA (Oxoid), and according to the manufacturer’s recommendations. Dilution = 1 is the limit of detection of Hbl and corresponds to 2 ng/mL of Hbl components. The value indices obtained for FBO-Bt and commercial Bt were compared using the non-parametric Wilcoxon statistical test. ns = not significant.
Fig 3.
Dendrogram of representative FBO- and commercial Bt isolates based on M13-PCR typing data.
After migration of M13-PCR products onto 1% agarose gels, the acquired images were analyzed with Bionumerics 7.6. The dendrogram was constructed using Dice coefficients and UPGMA clustering, with tolerance and optimisation set at 1%. Three groups of isolates were defined (1 to 3) based on a calculated similarity level above 80%.
Fig 4.
K-mer based phylogeny of 234 Bc/Bt isolates.
The dataset was composed of 208 Bacillus thuringiensis isolates. Among them, 153 were collected from FBOs, 19 from pesticides, and 32 from the environment (soils or insects). Genomic sequences were obtained in this study after genomic DNA extraction and NGS for 172 isolates, or collected from the NCBI database for 62 isolates. Detailed information related to each isolate is listed in S4 Table. The phylogenic tree was obtained with an in-house script using Mash as distance estimation and neighbor-joining (NJ) as the clustering method. Tree visualization was performed with iTOL v5 [46]. The IDs of Bt strains isolated from biopesticides were highlighted according to their respective subspecies (green: morrisoni, blue: israelensis, yellow: kurstaki, and orange: aizawai). The genome IDs from the dominant clade was highlighted in grey.
Fig 5.
Core-genome phylogenic analysis of 154 Bc/Bt isolated from FBOs and commercial products.
The analysis was performed using the pipeline iVARCall2 [41], and HD73 as a reference Bt strain. For better visualization, the tree (B) was built from the main one (A), excluding the cluster e, composed of 17 distant isolates. Phylogenetic reconstructions were performed by maximum likelihood with K3P model of substitution based on IQtree [43] and visualization of trees was performed with iTOL v5 [46]. (B) The trees were re-rooted on a branch of cluster e (A) and on a branch of cluster d (B). Bootstraps from 80 to 100 were represented by proportionately sized circles. Based on their cluster distribution (a to e), isolates were associated with red, orange, yellow, green and blue color, respectively. (C) Distribution of intra- (i.e. a, b, c, d or e) and inter-cluster cluster (i.e. a/b, a/c, a/d, b/a, b/c, b/d, c/a, c/b, c/d, d/a, d/b, d/c, e/a, e/b, e/c, e/d) pairwise SNP distances. Distances were compared with Wilcoxon rank-sum test implemented in R. *** = p-value<0.001.
Fig 6.
Diagram depicting the distribution of Bc-associated samples characterized in this study.
From 250 toxic episodes, at least one Bt was isolated in the context of 49 toxic events (i.e. 20%). Based on the genetic relatedness of corresponding isolates, FBOs were classified into 4 clusters (a to d). Nine FBOs were associated the insecticide strain GC-91 within the cluster a, 11 with the insecticide strain ABTS-1857 within the cluster b, 13 with Bt kurstaki SA-12, PB-54 and EG-2348 within the cluster c, and 21 with SA-11 and ABTS-351 within the cluster d.