Intraspecific Diversity and Pathogenicity of Bacillus thuringiensis Isolates from an Emetic Illness

This study describes an emetic food-borne intoxication associated with a Bacillus cereus group species and the characterization of the bacterial isolates from the incident in aspects of molecular tying, genetic factors, cytotoxicity, and pathogenic mechanisms relating to emetic illness. Through the polyphasic identification approach, all seven isolates obtained from food and clinical samples were identified as Bacillus thuringiensis. According to multilocus sequence typing (MLST) analysis, intraspecific diversity was found within the B. thuringiensis isolates. Four allelic profiles were found, including two previously known STs (ST8 and ST15) and two new STs (ST2804 and ST2805). All isolates harbored gene fragments located in the cereulide synthetase (ces) gene cluster. The heat-treated culture supernatants of three emetic B. thuringiensis isolates, FC2, FC7, and FC8, caused vacuolation and exhibited toxicity to Caco-2 cells, with CC50 values of 56.57, 72.17, and 79.94 µg/mL, respectively. The flow cytometry with the Annexin V/PI assay revealed both apoptosis and necrosis mechanisms, but necrosis was the prominent mechanism that caused Caco-2 cell destruction by FC2, the most toxic isolate.


Introduction
Bacillus cereus has long been recognized as a food-borne pathogen, one capable of producing two distinct types of toxins: diarrheagenic and emetic toxins [1,2]. Some other species that are closely related to B. cereus, known as the B. cereus group [2], were also occasionally found involved in food-borne outbreaks, such as B. thuringiensis and, more rarely, B. mycoides [3]. It is possible that some of the B. cereus group species might have been misidentified as B. cereus at the time of the outbreaks [3] or that B. cereus group species other than B. cereus have become emerging food-borne pathogens [4]. In addition, the potential of the B. cereus group species to cause food-borne illnesses has been increasingly recognized in recent years [5].

Morphological and Biochemical Characterization of Food-Borne Outbreak Isolates
All the B. thuringiensis isolates from food and the vomit samples were Gram-positive, spore-forming rods with unswollen sporangia, bearing roughly bipyramidal toxin crystals ( Figure 1). They could grow at 45 • C and below, but not at 48 • C or 50 • C. The API profiles of the isolates varied, but they all showed the typical biochemical reaction patterns for B. cereus/B. thuringiensis species (Table S1). All isolates were able to degrade starch and utilize salicin.
Cases of B. cereus emetic illness are usually associated with rice and pasta dishes [18][19][20][21][22][23], but the organisms involved in this small food-borne outbreak were isolated from the mackerel and mussels in the garlic butter sauce, not from the pasta. However, an extensive study by Messelhäusser et al. (2014) [24] showed that emetic toxin-producing strains belonging to the B. cereus group were not only found in foods with high carbohydrate content, such as pasta or rice, as was earlier found. These observations have implications for the breadth of sampling that should be carried out in the investigation of emetic outbreaks. The isolates had biochemical characteristics typical of B. cereus/B. thuringiensis, but it was interesting to note that the present isolates could degrade starch and utilize salicin, unlike many emetic B. cereus strains previously characterized [25][26][27]. This recalls the comment made by Thorsen et al. (2006) [28] that not all emetic toxin-producing B. cereus strains are unable to hydrolyze starch or produce acid from salicin. Cases of B. cereus emetic illness are usually associated with rice and pasta dishes [18][19][20][21][22][23], but the organisms involved in this small food-borne outbreak were isolated from the mackerel and mussels in the garlic butter sauce, not from the pasta. However, an extensive study by Messelhäusser et al. (2014) [24] showed that emetic toxin-producing strains belonging to the B. cereus group were not only found in foods with high carbohydrate content, such as pasta or rice, as was earlier found. These observations have implications for the breadth of sampling that should be carried out in the investigation of emetic outbreaks. The isolates had biochemical characteristics typical of B. cereus/B. thuringiensis, but it was interesting to note that the present isolates could degrade starch and utilize salicin, unlike many emetic B. cereus strains previously characterized [25][26][27]. This recalls the comment made by Thorsen et al. (2006) [28] that not all emetic toxin-producing B. cereus strains are unable to hydrolyze starch or produce acid from salicin.

Identification of Isolates Using 16S rRNA Gene Sequence Analysis
The 16S rRNA sequences of the B. thuringiensis isolates, two from food (FC1 and FC2; obtained from mussels and smoked mackerel in garlic butter sauce), and five from the vomit sample of one patient (FC6, FC7, FC8, FC9, and FC10), were determined. The 16S rRNA Blast results against strains in the NCBI database confirmed their identity as B. thuringiensis (Table 1).
The sequences of the B. thuringiensis isolates were also compared with those of three reference strains: B. thuringiensis IAM 12077 (or ATCC 10792, B. thuringiensis type strain); B. cereus ATCC 14579 (B. cereus type strain); and B. cereus F4810/72 (a reference B. cereus emetic strain, also known as strain B0358 in the Logan Bacillus Collection at Glasgow). Based on the 16S rRNA fragment, the FC isolates were also closely related to the B. thuringiensis and B. cereus strains tested.
B. thuringiensis is known as one of the closest species to B. cereus. These two species are closely related genetically, and their 16S rRNA sequence identities are more than 99% [29]. Polyphasic identification is therefore useful in this case. As expected from the Blast results, many of the B. cereus strains were also among the closest relatives. Because all FC isolates produced parasporal crystals, they were therefore confirmed to be B. thuringiensis by their genetic characteristics, which are also supported by the biochemical profiles.  The 16S rRNA sequences of the B. thuringiensis isolates, two from food (FC1 and FC2; obtained from mussels and smoked mackerel in garlic butter sauce), and five from the vomit sample of one patient (FC6, FC7, FC8, FC9, and FC10), were determined. The 16S rRNA Blast results against strains in the NCBI database confirmed their identity as B. thuringiensis (Table 1). The sequences of the B. thuringiensis isolates were also compared with those of three reference strains: B. thuringiensis IAM 12077 (or ATCC 10792, B. thuringiensis type strain); B. cereus ATCC 14579 (B. cereus type strain); and B. cereus F4810/72 (a reference B. cereus emetic strain, also known as strain B0358 in the Logan Bacillus Collection at Glasgow). Based on the 16S rRNA fragment, the FC isolates were also closely related to the B. thuringiensis and B. cereus strains tested.
B. thuringiensis is known as one of the closest species to B. cereus. These two species are closely related genetically, and their 16S rRNA sequence identities are more than 99% [29]. Polyphasic identification is therefore useful in this case. As expected from the Blast results, many of the B. cereus strains were also among the closest relatives. Because all FC isolates produced parasporal crystals, they were therefore confirmed to be B. thuringiensis by their genetic characteristics, which are also supported by the biochemical profiles.

Molecular Typing of Isolates Using Multilocus Sequence Typing (MLST) Analysis
We further examined the genetic diversity and relatedness of the food poisoning isolates using multilocus sequence typing (MLST), which is highly discriminatory for  [30], seven housekeeping genes for the B. cereus group (glpF, gmk, ilvD, pta, pur, pycA, and tpi) were used in the analysis. A sequence type, ST26, to which B. cereus F4810/72 belongs, was also included in the analysis.
Each amplified fragment of the housekeeping genes was sequenced, and the processed sequences were submitted to the MLST database (see Section 4.2.2 for details). The dendrogram constructed using UPGMA with allelic profiles of STs is presented in Figure 2a. The phylogenetic analysis based on sequence alignment revealed that the B. thuringiensis isolates belonged to different clonal groups. The isolates FC1, FC6, FC8, FC9, and FC10 were arranged in one clade, and FC2 and FC7 were separated from this clade and were distinct from each other. All of the B. thuringiensis isolates were separated from B. cereus  Since B. thuringiensis strains have been used as biopesticides, there have been concerns raised about the use of this organism for this purpose due to the potential enterotoxin production of this species. Extensive use of these organisms, especially in the form of spores, may cause B. thuringiensis to be more widely spread in the environment and have more frequent implications for food-borne illnesses. Cases of diarrheal disease have been linked to biopesticide strains [34]. Recently, B. thuringiensis found in salad was suspected to be the cause of a food-borne outbreak in the EU [29]. Biggel et al. (2022) [34] suggested that many food isolates of B. thuringiensis had their origins in biopesticide Using the goeBURST algorithm, one group and two singletons were defined. FC1, FC6, FC7, FC8, FC9, and FC10 (Clonal Complex 8; CC8) were assigned to Group 1, and FC2 and F4810/72 were designated as singletons. Overall, the bacterial strains, including MLST typing analyzes a set of seven housekeeping genes. This method has been successfully used for different applications, including the typing of food-borne isolates [30]. It has also been used for typing of environmental strains of B. cereus group [31]. The established B. cereus MLST database (PubMLST; http://pubmlst.org/bcereus/info/primers.shtml, accessed on 1 December 2022) allows comparisons of a set of isolates to the existing sequence type and the identification of a new sequence type, under both of which techniques we found our strains belonged.
B. cereus F4810/72, a classical emetic strain that was used as a reference in this study, has the allelic profile ST26 [32]. This ST was also the most commonly identified ST for B. cereus strains [33]. The fact that the B. thuringiensis isolates belonged to different STs clearly confirmed that there was an intraspecific genetic diversity of strains involved in the emetic illness case recorded in this study. Because B. thuringiensis was not commonly present in food, and the occurrence of emetic strains with direct evidence of clinical symptoms was even more unusual (Hoffmaster et al., 2008) [33], this finding is worthy of note as it may indicate various sources of contamination or the possibility of gene transfer.
Since B. thuringiensis strains have been used as biopesticides, there have been concerns raised about the use of this organism for this purpose due to the potential enterotoxin production of this species. Extensive use of these organisms, especially in the form of spores, may cause B. thuringiensis to be more widely spread in the environment and have more frequent implications for food-borne illnesses. Cases of diarrheal disease have been linked to biopesticide strains [34]. Recently, B. thuringiensis found in salad was suspected to be the cause of a food-borne outbreak in the EU [29]. Biggel et al. (2022) [34] suggested that many food isolates of B. thuringiensis had their origins in biopesticide strains and belong to previously observed sequence types (i.e., ST8, ST15, ST16, or ST23) [35]. Similarly, our MLST results showed that many of the food-borne outbreak B. thuringiensis isolates belong to ST8 (four isolates from vomit) and ST15 (one food isolate). These sequence types, which might have been linked to biopesticide strains, have been frequently found in fresh vegetable samples [29,36,37]. Moreover, it is interesting to record the new ST genotypes found in this study and their association with emetic illness. Although we cannot be certain about the original source of the strains belonging to these STs, these data could be useful in future analyses of epidemiological data and assessments of the ecological and biological roles of B. thuringiensis isolates. Moreover, the new STs might indicate an emerging pathogenic strain and could be useful in the monitoring of food safety and the control of disease in relation to this rare emetic food-borne pathogen. Genetic factors that are associated with the Bacillus emetic traits, i.e., emetic toxin production capability or cereulide synthetase, were investigated in the B. thuringiensis isolates through PCR amplification of fragments of the NRPS/cereulide synthetase (ces) genes. The expected amplified regions in the ces gene cluster (based on the sequence of strain F4810/72, Genbank accession no. DQ360825.1 [38]) are shown in Figure 3a. The PCR results are shown in Figure 3b. Although the amplified gene fragments were achieved with high quantity and specificity with strain F4810/72, the primer pairs had different efficacies when applied to B. thuringiensis isolates. The primers BEF/BER, CER1/EMT1, and EM1-F/EM1-R yielded amplicons of expected sizes from all B. thuringiensis isolates, even though nonspecific amplifications were also observed in some reactions with primers BEF/BER and EM1F/EM1R. However, primers CesF1/CesR2 gave positive results only with FC2 (food isolate) and FC8 (clinical isolate).
sults are shown in Figure 3b. Although the amplified gene fragments were achieved with high quantity and specificity with strain F4810/72, the primer pairs had different efficacies when applied to B. thuringiensis isolates. The primers BEF/BER, CER1/EMT1, and EM1-F/EM1-R yielded amplicons of expected sizes from all B. thuringiensis isolates, even though nonspecific amplifications were also observed in some reactions with primers BEF/BER and EM1F/EM1R. However, primers CesF1/CesR2 gave positive results only with FC2 (food isolate) and FC8 (clinical isolate).  The four pairs of primers used in this investigation were evaluated by different researchers as being specific to B. cereus emetic strains or to the ces gene. Even though these The four pairs of primers used in this investigation were evaluated by different researchers as being specific to B. cereus emetic strains or to the ces gene. Even though these sets of primers were independently designed, their specific targets were at different sites on the ces gene. The BEF/BER and CER1/EMT primers specifically target cesA, whereas the EM1F/EM1R and CesF1/CesR2 primers specifically target cesB (Figure 3a [41], respectively, to be associated with cereulide-producing strains. The CesF1/CesR2 primers were described by Ehling-Schulz et al. (2005b) [42] as specific to the valine activation module of the cereulide synthetase gene, and disruption of this gene resulted in a non-cereulide-producing mutant. The differences in the amplification patterns observed in our study for the B. thuringiensis isolates and B. cereus F4810/72, as seen in Figure 3b, may be due to the variation of the nucleotides within and around these fragments.
Failure to amplify the ces fragment by CesF1/CesR2 primers in many B. thuringiensis isolates, apart from FC2 and FC8, is not unexpected. Kim et al. (2010) [43] observed that this pair of primers could yield a negative result with some emetic toxin-producing B. cereus strains. Our experience showed that this pair of primers required more specific reaction conditions than the others. On the other hand, failure to amplify the gene fragment may be a result of non-conserved DNA sequences around this region of the cesB gene in emetic toxin-producing Bacillus, especially with sequence diversity within members of the B. cereus group having been previously observed [44]. This observation in our study suggested that using these four pairs of primers offers a more sensitive option for identifying the presence of fragments of cesA and cesB, the two consecutive genes encoding CesA and CesB proteins that are required for cereulide production [45]. Further investigations into the genetic features of the isolates are desirable, and in this study, we explored the sequence of some of these gene fragments.

Sequence Analysis of B. thuringiensis Amplified Gene Fragments
Selected gene fragments amplified from the DNA of the B. thuringiensis isolates using the primers BEF/BER, CER1/EMT1, and EM1-F/EM1-R were sequenced (the sequences are deposited in the NCBI BioProject database PRJNA188745). They were highly similar (with 98-99% identity) or identical to the ces gene found in B. cereus F4810/72 (Genbank accession no. DQ360825.1 [38] and the crs (another abbreviation for cereulide synthetase) gene from B. cereus no. 55 (Genbank accession no. AB248763.2 [46]). Fragments generated using primers CER1/EMT1 were also comparable (with 99-100% identity) to the B. cereus NRPS gene (Genbank accession no. AY331260.1 [40]). They also had 99-100% identity with the cesA gene found in toxigenic B. pumilus NR 19/5 and B. licheniformis NR 5106 (Genbank accession no. AM493712.1 and AM493711.1, respectively [47]). In addition, fragments yielded from primers EM1-F/EM1-R also had 96-97% identity to the cesB gene found in an emetic toxin-producing strain of B. pumilus (strain NIOB133; Genbank accession no. EU289221.1 [48]. The positions on the B. thuringiensis' ces gene, which were different from those on the ces gene of B. cereus and other Bacillus species, had mismatches, insertions, or deletions of nucleotides. From the biochemical and genetic characterization, it is clear that the outbreak isolates do not represent a single strain. The fact that they all harbored fragments that are parts of the ces gene cluster indicated that this gene may be distributed among B. thuringiensis strains in certain geographical locations [26,49]. Gene transfer is possible given that the ces gene in some strains is located on a megaplasmid [44,46,50] and mobile genetic elements (MGEs) have been found to be associated with the ces gene cluster [44]. More extensive studies on ces distribution in B. thuringiensis would help us understand how widely this feature is spread in the natural environment. It would also lead to a better assessment of its impact on food safety.

Cytotoxicity of Emetic B. thuringiensis Isolates to Caco-2 Cells
According to the clinical symptoms and the results from the genetic characterization presented above, we presumed that some of the B. thuringiensis isolates would display similar biological activity to that of emetic B. cereus, which can produce the emetic toxin cereulide, a heat-stable cyclic peptide [51,52]. Because of the molecular nature of the emetic toxin, which has a very low molecular weight (1.2 KDa), it is difficult to detect it using an immunological method or a simple analytical chemistry method [53]. Therefore, biological assays, especially cytotoxicity assays, have been extensively used for the detection of the emetic toxin [54][55][56].
In this study, the cytotoxicity of the heat-treated culture supernatants of the outbreak isolates was tested on human colorectal adenocarcinoma (Caco-2) cells. The isolates tested included B. thuringiensis FC2, FC7, and FC8, which represented sequence types ST2804, 2805, and ST8, respectively. A preliminary cytotoxicity test showed that the heat-treated supernatants of the B. thuringiensis isolates were cytotoxic to Caco-2 cells (a non-emetic strain control, B. cereus DSM 4384, did not show cytotoxic activity in the same experiment) ( Figure S1). Further investigation of the degree of cytotoxicity (expressed as half-maximal cytotoxic concentration, CC 50 ) of isolates FC2, FC7, and FC8 showed that the reduction of Caco-2 viability induced by the heat-treated culture supernatants of these isolates was in a dose-dependent manner (Figure 4a). According to the CC 50 values, the FC2 isolate showed a higher degree of cytotoxicity to Caco-2 cells than FC7 and FC8 (Figure 4a). Vacuolation was observed in Caco-2 cells treated with the heat-treated supernatants from all three isolates (Figure 4b).
( Figure S1). Further investigation of the degree of cytotoxicity (expressed as half-maximal cytotoxic concentration, CC50) of isolates FC2, FC7, and FC8 showed that the reduction of Caco-2 viability induced by the heat-treated culture supernatants of these isolates was in a dose-dependent manner (Figure 4a). According to the CC50 values, the FC2 isolate showed a higher degree of cytotoxicity to Caco-2 cells than FC7 and FC8 (Figure 4a). Vacuolation was observed in Caco-2 cells treated with the heat-treated supernatants from all three isolates (Figure 4b).  It is difficult to identify the isolate(s) responsible for the illness, as the MLST results (Section 2.2.2) indicated a diversity among the outbreak isolates. The cytotoxicity results suggested a possible role of FC2 in the emetic illness, as it was more toxic to Caco-2 cells than the other isolates, although the contribution of the other isolates to the illness may not be excluded. It is difficult to identify the isolate(s) responsible for the illness, as the MLST results (Section 2.2.2) indicated a diversity among the outbreak isolates. The cytotoxicity results suggested a possible role of FC2 in the emetic illness, as it was more toxic to Caco-2 cells than the other isolates, although the contribution of the other isolates to the illness may not be excluded.

Annexin V/PI Apoptosis Detection
To clarify the cell death pathway caused by the heat-stable toxin in the supernatants from each bacterial isolate, we performed annexin V/PI staining to discriminate the apoptosis and necrosis mechanisms of cell death through flow cytometry. Caco-2 cells were treated for 48 h with the cell-free supernatants of FC2, FC7, and FC8 at a concentration of 50% (v/v) (which had 56.62, 62.70, 64.91 µg/mL total protein, respectively). Skim milk and the culture supernatant of F4810/72 were used as negative and positive controls, respectively.
From the results, B. cereus F4810/72 appeared to cause both apoptotic cell death (14.85 ± 1.49%) and necrotic cell death (5.91 ± 0.32%) (Figure 5a,b). For cell death through the apoptotic mechanism, 6.07 ± 0.81% was detected as early apoptosis and 8.78 ± 0.69% as late apoptosis. Virtanen et al., (2008) [57] have reported that cereulide from B. cereus F4810/72 increased necrotic cell death in porcine fetal Langerhans and beta-cells within 2 days. Similarly, Hoornstra et al. (2013) [58] also observed necrotic cell death caused by cereulide from B. cereus F4810/72 in other mammalian cells (PBMC, HaCaT, PK-15, and L-929). Our testing of the effect of heat-stable toxins in the culture supernatant of B. cereus F4810/72 with Caco-2 cells also pointed to cell destruction through necrosis. However, apoptosis seemed to be a major mechanism of cell destruction in Caco-2 cells.
As for the B. thuringiensis FC2, FC7, and FC8 isolates, the heat-treated cell-free supernatants caused cell death through both the apoptosis and necrosis pathways, as observed with the reference emetic strain B. cereus F4810/72. However, necrosis was a major pathway in the FC2 isolate, which had the highest rate of necrosis cell death (23.47 ± 2.92%) among the three B. thuringiensis isolates tested (Figure 5b). For FC7 and FC8, cell death rates through the necrosis mechanism were similar to those of apoptosis, which was mainly early apoptosis. These results suggest that the B. thuringiensis isolates from this emetic outbreak, although they could cause a similar symptom to that typically recognized as being caused by B. cereus, might have some differences in the toxin entity from that of B. cereus F4810/72. natants caused cell death through both the apoptosis and necrosis pathways, as observed with the reference emetic strain B. cereus F4810/72. However, necrosis was a major pathway in the FC2 isolate, which had the highest rate of necrosis cell death (23.47 ± 2.92%) among the three B. thuringiensis isolates tested (Figure 5b). For FC7 and FC8, cell death rates through the necrosis mechanism were similar to those of apoptosis, which was mainly early apoptosis. These results suggest that the B. thuringiensis isolates from this emetic outbreak, although they could cause a similar symptom to that typically recognized as being caused by B. cereus, might have some differences in the toxin entity from that of B. cereus F4810/72. The bar graphs show the percentages (as means ± SD) of cells treated with the culture supernatants from the test strains in the early apoptosis, late apoptosis, and necrosis phases, which were statistically significantly different (***, p < 0.005) from the control (untreated cells).

Conclusions
In this study, we report a food-borne emetic illness associated with B. thuringiensis and present the results of the investigation on the diversity and pathogenicity of the isolates from this food-borne outbreak incident. The biochemical and genetic profiles suggested an appreciable diversity among the isolates. The multilocus sequence typing (MLST) analysis revealed that the B. thuringiensis isolates did not represent one strain. Some isolates belonged to the existing sequence types (ST8 and ST15) that have been The bar graphs show the percentages (as means ± SD) of cells treated with the culture supernatants from the test strains in the early apoptosis, late apoptosis, and necrosis phases, which were statistically significantly different (***, p < 0.005) from the control (untreated cells).

Conclusions
In this study, we report a food-borne emetic illness associated with B. thuringiensis and present the results of the investigation on the diversity and pathogenicity of the isolates from this food-borne outbreak incident. The biochemical and genetic profiles suggested an appreciable diversity among the isolates. The multilocus sequence typing (MLST) analysis revealed that the B. thuringiensis isolates did not represent one strain. Some isolates belonged to the existing sequence types (ST8 and ST15) that have been linked to biopesticide strains, whereas some belonged to two new STs (designated ST2804 and ST2805). All food and clinical isolates harbored gene fragments located in the cereulide synthetase (ces) gene cluster, which indicated the likelihood that an emetic toxin of a similar nature to cereulide might have been implicated in the illness. The heat-stable toxin in the supernatant of B. thuringiensis FC2, FC7, and FC8, which represented ST2804, ST2805, and ST8, respectively, reduced the viability of Caco-2 in a dose-dependent manner. According to the CC 50 values, the FC2 isolate derived from the food involved in the illness showed a higher degree of cytotoxicity to Caco-2 cells than did FC7 and FC8. The flow cytometry with Annexin V/PI staining revealed that cell destruction by all three food-poisoning B. thuringiensis isolates occurred through both apoptosis and necrosis pathways, but for FC2, necrosis was likely the main mechanism that caused Caco-2 cell destruction.

Bacterial Strains/Isolates and Identification
The isolates from the food-borne outbreak comprised two from food: FC1 and FC2 (obtained from the leftover pasta sauce: mussels and smoked mackerel in garlic butter sauce) and five from the vomit of one patient: FC6, FC7, FC8, FC9, and FC10. They were isolated on Polymyxin Pyruvate Egg-Yolk Mannitol Bromothymol-Blue Agar (PEMBA), which was incubated at 37 • C for 48 h. The isolates were examined using phase-contrast microscopy and biochemically characterized using the API 20E and API 50 CHB systems (BioMérieux, Marcy-l'Étoile, France). Growth at 40 • C, 42 • C, 45 • C, 48 • C, and 50 • C, was also observed in Tryptone Soy Broth (Oxoid, Basingstoke, UK) cultures incubated in water baths. B. cereus F4810/72 (also known as strain B0358 in the Logan Bacillus Collection at Glasgow) was used as a reference emetic toxin-producing strain. All bacterial cultures were maintained on Tryptone Soy Agar (TSA) (Oxoid, Basingstoke, UK), stored at 4 • C, or in lyophilized form.  (Lane, 1991) [61]. The PCR products were purified using a PCR purification kit (Sigma-Aldrich, Darmstadt, Germany). Sequencing was performed by Celemics (Seoul, Republic of Korea). The low-quality bases in the sequences were trimmed using Bioedit software version 7.05.3 [62]. The processed sequences were subjected to Basic Local Alignment Search Tool (BLAST) analysis against the 16S rRNA sequences in the National Center for Biotechnology Information (NCBI) database to find the closest relatives.

Typing of Isolates Using Multilocus Sequence Typing (MLST) Analysis
The primers for the housekeeping genes (glpF, gmk, ilvD, pta, pur, pycA, and tpi) were used for multilocus sequence typing (MLST) according to Yang, et al. (2017) [30] (ilvD2 was also used if the ilvD locus failed to find an allelic profile). The sequencing of the amplified genes was performed by Macrogen (Seoul, Republic of Korea).
Sequences were analyzed using BioEdit software version 7.1.9 (Abbott, Carlsbad, CA, USA). All allelic profiles and sequence types (STs) detected in the study were assigned according to the online MLST database for B. cereus (https://pubmlst.org/organisms/ bacillus-cereus, accessed on 1 December 2022). Grouping tools were used to establish the molecular type, clonal complexes (CCs), singletons, and the relationship between established groups. All STs were displayed with BURST [63], using the characteristic profiles for each isolate as the input data. Minimal expansion trees were also generated using the geoBURST algorithm [63] performed with Phyloviz 2.0 [64] to define CCs and display models. A phylogenetic tree of the related sequences of the seven loci of the housekeeping genes was generated with UPGMA based on the method of Kimura-2parameter [65] and performed with 1000 bootstrap replicates in Mega X [66].

DNA Sequencing and Sequence Similarity Analysis
Purified PCR products from selected B. thuringiensis isolates amplified using primers CER1/EMT1, BEF/BER, and EM1-F/EM1-R were sequenced using an Applied Biosystems genetic analyzer (performed by First Base Laboratory, Selangor, Malaysia). The nucleotide sequences of the PCR fragments were analyzed for their similarities to sequences published in the NCBI database using the Blast program. In addition, they were compared to the sequence of reference strains using the Blast 2 Sequences program.

Preparation of Heat-Treated Supernatants from Bacterial Isolates
The overnight bacterial cultures (0.1% (v/v)) were inoculated in 50 mL of 2.5% skim milk medium (SMM) (Oxoid, Basingstoke, UK). The cultures were incubated for 24 h at 25 • C in an orbital shaking incubator (shaker speed: 150 rpm). After incubation, the supernatant portion of each bacterial culture was collected after centrifugation at 3000× g for 10 min, and the cell pellet was discarded. The supernatant was heated in an autoclave for 15 min at 121 • C, left to cool, and filtered through a 0.2 µm membrane filter [56].

Cytotoxicity Assay
The cytotoxicity of the heat-treated supernatant from the bacterial isolates was tested using the MTT cytotoxicity assay [56] and the PrestoBlue cytotoxicity assay [69]. In brief, Caco-2 cells were plated at 10 5 cells/well in 96-well plates for 48 h before the experiment. The Caco-2 cells were treated with 100 µL of the supernatant from the bacterial isolates prepared at various concentrations for 48 h. For the PrestoBlue cytotoxicity test, after 48 h of treatment, the PrestoBlue™ reagent (Thermo Fisher Scientific, Waltham, MA, USA) was added to measure cell viability, and the cells were incubated for 30 min at 37 • C in a 5% CO 2 incubator. Absorbance was read at 570/600 nm using a Bio Tek Synergy HTX microplate reader (Agilent Technologies, Santa Clara, CA, USA). The percentages of cell viability in relation to that of the non-treatment control were calculated using the following equation.

Annexin V/PI Assay
Apoptotic stages were evaluated by conducting an Annexin V/PI assay and analyzed using a flow cytometer. Caco-2 cells were seeded at 5 × 10 4 cells/mL in 6-well culture plates, using 2 mL per well, and treated with the supernatant from the bacterial isolates at 50% concentration or 1/2 dilution (having total proteins of 56.62 µg/mL, 62.70 µg/mL, 64.91 µg/mL for FC2, FC7, and FC8, respectively). After 48 h of treatment, the Caco-2 cells were collected and stained with Annexin V/PI, according to the Annexin V/PI kit protocol (ImmunoTool, Friesoythe, Germany). Briefly, the cell pellets were trypsinized and resuspended in 100 µL of ice-cold 1 × binding buffer. Annexin V-APC and PI were added to the tubes at the dilution of 1:100. The cell suspension was then mixed and incubated for 15 min in the dark. Samples were analyzed on a BD Accuri C6 Flow Cytometer and FlowJo software (BD Biosciences, San Jose, CA, USA).

Supplementary Materials:
The following are available online at: https://www.mdpi.com/article/ 10.3390/toxins15020089/s1, Table S1: Positive biochemical characteristics of the emetic outbreak B. thuringiensis isolates tested using the API 20 and API 50 CH galleries, Figure S1: Caco-2 cell viability after treatment with heat-treated culture supernatants of emetic food-borne outbreak isolates.