Enteropathogenic Potential of Bacillus thuringiensis Isolates from Soil, Animals, Food and Biopesticides

Despite its benefits as biological insecticide, Bacillus thuringiensis bears enterotoxins, which can be responsible for a diarrhoeal type of food poisoning. Thus, all 24 isolates from foodstuffs, animals, soil and commercially used biopesticides tested in this study showed the genetic prerequisites necessary to provoke the disease. Moreover, though highly strain-specific, various isolates were able to germinate and also to actively move, which are further requirements for the onset of the disease. Most importantly, all isolates could grow under simulated intestinal conditions and produce significant amounts of enterotoxins. Cytotoxicity assays classified 14 isolates as highly, eight as medium and only two as low toxic. Additionally, growth inhibition by essential oils (EOs) was investigated as preventive measure against putatively enteropathogenic B. thuringiensis. Cinnamon Chinese cassia showed the highest antimicrobial activity, followed by citral, oregano and winter savory. In all tests, high strain-specific variations appeared and must be taken into account when evaluating the hazardous potential of B. thuringiensis and using EOs as antimicrobials. Altogether, the present study shows a non-negligible pathogenic potential of B. thuringiensis, independently from the origin of isolation. Generally, biopesticide strains were indistinguishable from other isolates. Thus, the use of these pesticides might indeed increase the risk for consumers’ health. Until complete information about the safety of the applied strains and formulations is available, consumers or manufacturers might benefit from the antimicrobial activity of EOs to reduce the level of contamination.


Introduction
Bacillus thuringiensis is a Gram-positive and spore-forming bacterium, which is used worldwide as biopesticide due to its capability to produce insecticidal proteins [1][2][3]. The two main parasporal toxins are Cyt (cytolytic) and Cry (crystal) toxins [4]. In addition to that, B. thuringiensis produces other entomopathogenic factors such as vegetative insecticidal proteins (Vip) and secreted insecticidal proteins (Sip) [1,5,6]. As a result of its distinct types of toxins, B. thuringiensis is virulent to a wide range of insect orders and species, e.g., Coleoptera, Diptera and Lepidoptera, as well as nematodes and even plant and animal pathogenic microorganisms [2,7,8]. A clear advantage is the host-specific mode of action of the mentioned toxins, which do rarely harm non-target species. Table 1. Overview of the 24 B. thuringiensis strains used in this study. Origin of the isolates: A: animal, F: food, P: pesticide, S: soil, U: undefined. +: positive PCR result for toxin gene. −: negative PCR result for toxin gene. *: NheB production determined in sandwich enzyme immunoassays (EIAs) after growth under simulated intestinal conditions according to Jessberger et al. [15]. **: Cytotoxicity towards CaCo-2 cells after growth under simulated intestinal conditions according to Jessberger et al. [15].

Cell Lines and Culture Conditions
CaCo-2 cells from DSMZ (German Collection of Microorganisms and Cell Cultures, Braunschweig, Lower Saxony, Germany) were cultivated in RPMI 1640 medium (with stable glutamine) supplemented with 10% foetal bovine serum (Biochrom AG, Berlin, Germany). For cultivation, 80 cm 2 culture flasks and a humidified incubator (37 • C and 7% CO 2 ) were used. Cells were passaged 1:6 every three to four days. For differentiation, cells were incubated for 14 days. Medium was changed every three to four days.

PCR Analyses
Generally, amplification was performed with a total volume of 50 µL in a Biometra thermocylcer (Analytik Jena, Jena, Thuringia, Germany). The reaction mixtures contained 5 µL 10× buffer, 4 µL MgCl 2 (25 mM), 1 µL dNTP suspension (10 mM each), 0.5 µL Taq DNA polymerase, 2.5 µL of each primer (10 µM) and 2 µL template DNA (colony material). Initial denaturation at 95 • C for five minutes was followed by 30 reaction cycles: denaturation at 95 • C for 15-60 s, annealing for 15-60 s at the appropriate temperature (see Table S1) and extension for 20-60 s at 72 • C. The final extension was performed at 72 • C for seven minutes. PCR fragments were separated on 1% agarose gels by electrophoresis and made visible by UV illumination. Primers used for all PCR analyses are listed in Table S1. First, the panC (pantothenate synthetase) gene was amplified for sequence analysis (see Section 2.4). Subsequently, all strains were screened for cry toxin genes and allocated to five groups according to Ben-Dov et al. [24]. Furthermore, the strain set was tested for the presence of the (entero)toxin genes hblC, nheA, ces and cytK1 via multiplex PCR according to Wehrle et al. [25], as well as cytK2 according to Ehling-Schulz et al. [26].

PanC Sequence Typing
To allocate the B. thuringiensis isolates tested in this study to phylogenetic groups, sequencing of a 485 bp fragment of the pantothenate synthetase gene (panC; [27]) was applied (Eurofins Genomics Germany GmbH, Ebersberg, Bavaria, Germany). Sequences were aligned using CLUSTAL Ω [28]. Further cluster analysis was carried out using SplitsTree software (version 4, Eberhard Karls University Tübingen, Tübingen, Baden-Württemberg, Germany ) [29]. The sequences of the following reference strains were added to the analysis:  4 and 1 M maltose in bacto nutrient broth (Oxoid, Wesel, Northrhine-Westphalia, Germany); pH 7.6) were inoculated with 0.5% overnight culture (see Section 2.1) and incubated for three days at 32 • C and 200 rpm. Sporulation was monitored via bright-field microscopy. When >80% spore content was reached, the cultures were centrifuged at 3500 rpm and 4 • C for 10 min, washed three times in pre-cooled spore washing buffer (1 M KH 2 PO 4 , 1 M K 2 HPO 4 , pH 7.4) and stored at 4 • C.

Germination
For assessing germination of B. thuringiensis spores, media were inoculated to OD 600 = 1. Two biological with three technical replicates for each strain were incubated in 96-well plates at 37 • C for one hour. As described before [14], OD 620 was measured every three minutes in a Tecan photometer using Ridawin software. Germination is shown as the decrease of OD 620 (in %) per time. Germination Foods 2020, 9, 1484 5 of 19 was compared in CGY and cRPMI medium, as well as with and without prior heat activation for 10 min at 80 • C.

Motility Assay
To evaluate swimming behaviour of the B. thuringiensis strains, three replicates each were tested. For this, 1 µL CGY overnight culture was injected into the centre of 53 mm diameter CGY agar plates containing 0.25% agar and incubated at 37 • C. Motility diameters were assessed after 24 h.

Enzyme Immunoassays (EIAs)
Sandwich enzyme immunoassays for the detection of the enterotoxin components NheB and Hbl L2 were performed as described previously [33][34][35]. The following antibodies were used for detection: 5 µg/mL mAb 2B11/1E11-HRP 1:2000 (NheB) and 10 µg/mL mAb 1A12/1H9-HRP 1:1000 (Hbl L2). Titres are defined as the reciprocal of the highest dilutions resulting in an absorbance value of ≥1.0. Results were compared according to those titres, which were calculated from at least three replicates each.

WST-1 Bioassay
WST-1 bioassays were performed on CaCo-2 cells as described earlier [15]. Briefly, serial dilutions of the culture supernatants were applied to 96-well plates in RPMI 1640 medium (100 µL/well) before 2 × 10 4 three days old CaCo-2 cells/well (100 µL) were added. After 24 h incubation at 37 • C and 7% CO 2 , cell viability was determined by addition of the cell proliferation regent WST-1 (water soluble tetrazolium, Roche diagnostics, Penzberg, Bavaria, Germany). Increasing amounts of formazan were detected in a Tecan infinite F50 photometer at 450 nm. Dose-response curves and thus, 50% lethal concentrations of at least two replicates were calculated with Magellan software and are shown as reciprocal titres.

Haemolysis Assays
CGY culture supernatants (10 µL) were filled into stamp holes with five mm diameter on sheep blood agar plates. After five hours incubation at 32 • C, haemolysis patterns were assessed.

Antimicrobial Broth Dilution Assays
EOs with antibacterial effects in the disk diffusion tests (see Section 2.11) were applied in antimicrobial broth dilution assays to determine minimal inhibitory as well as maximal tolerated concentrations according to Dussault et al. [36]. The EOs were set to a concentration of 2.5% (dissolved in sterile H 2 O with 5% Tween 80). The first column of a 96-well microtitre plates was filled with 167.7 µL CGY medium and 33.3 µL of the prepared EO solution. Columns two to 12 were filled with 100 µL CGY and a two-fold dilution series was prepared. EO concentrations thus ranged from 0.42 to 0.00041%. Column 12 was not filled with EOs and served as positive control. The B. thuringiensis CGY overnight cultures were set to OD 600 = 0.5 (approx. 10 7 CFU/mL) and further diluted to approx. 10 4 CFU/mL. Rows A and H served as negative control, the remaining wells were filled with 10 µL bacterial suspension. For the determination of the minimal inhibitory concentration (MIC) a total of 10 6 CFU/mL were used, 10 3 CFU/mL to establish the maximal tolerated concentration (MTC). After 24 h incubation at 37 • C, 10 µL iodonitrotetrazolium chloride (Sigma-Aldrich/Merck KGaA, Darmstadt, Hesse, Germany) were added to each well. After further 60 min incubation at 37 • C, bacterial growth was measured at OD 540 in a Tecan infinite F50 photometer. The MIC was calculated from the mean values of the lowest EO concentrations that prevented bacterial growth. The MTC was determined by the mean values of the maximal concentrations that allowed bacterial growth. The experiment was done twice with three technical replicates per strain.

Statistical Analyses
Data were statistically validated using the column statistics program of GraphPad Prism version 5.00 for Windows (GraphPad Software, San Diego, CA USA). Unpaired t-tests were applied with two-tailed p-values and 95% confidence interval. Significant differences were marked with * (p-value ≤ 0.05). Furthermore, Spearman correlation tests were performed with two-tailed p-values and 95% confidence interval to determine correlations between data sets. Significant correlation was marked with * (p-value ≤ 0.05).

Genetic Prerequisites of the Tested Strains
In this study, 24 B. thuringiensis isolates of different origin (see Table 1) were characterized. For a rough overview on their phylogenetic relationship, the isolates were subjected to panC sequence typing, allocating them to panC type IV ( Figure 1A). Distinct clusters within this group appeared ( Figure 1B). In cluster a, MHI 2882 (B. thuringiensis ssp. israelensis isolated from larva) was found, as well as MHI 3241 and the older, barely documented isolate MHI 2878 (both B. thuringiensis ssp. tenebrionis), which might be identical. Cluster b comprised MHI 271, MHI 2873 (ATCC type strain) and MHI 3370. The reference strain B. cereus ATCC 14579 and MHI 2870 (B. thuringiensis ssp. entomocidus) belonged to cluster c. All further isolates clustered closer together. Cluster d comprised four biopesticidal strains and four food isolates, while in cluster e mainly soil and salad isolates appeared, with the exception of biopesticidal isolate MHI 3240. Interestingly, MHI 2872 and MHI 3189 (from ready-to-eat salad) in cluster d as well as MHI 2874 (B. thuringiensis ssp. kurstaki, isolated from soil) and MHI 3164 (also from ready-to-eat salad) in cluster e seemed to be identical.
Furthermore, all isolates were tested for the presence of cry genes via PCR with universal primers according to Ben-Dov et al. [24]. Thus, they could be verified as B. thuringiensis and additionally grouped in cry-type gene profiles (see Table 1). Twenty strains (83%) were classified as cry1, while 13 of those (54%) were also positive for cry2. Two strains (MHI 2878 and MHI 3241; B. thuringiensis ssp. tenebrionis) were allocated to the cry3-type and one strain (MHI 2882; Culicidae larva isolate) to cry4. MHI 3370, the ancrystalliferous derivate Bt 407 Cry − [37,38], was negative for all cry-types. Equally, all isolates were screened for the presence of (entero)toxin genes. All isolates harboured nheA as well as hblC, but were negative for ces as well as cytK1 (see Table 1). Twenty two strains (92%) were tested positive for cytK2 and could thus be assigned to toxin profile A (nhe+, hbl+, cytK2+), while only two isolates (MHI 2878 and MHI 3241; B. thuringiensis ssp. tenebrionis) were cytK2 negative and thus, allocated to toxin profile C (nhe+, hbl+) [39]. Interestingly, these were the two isolates assigned to the cry3-type. However, investigating a larger set of strains would be necessary to contextualize these observations. Phylogenetic relationship between the 24 tested B. thuringiensis strains, based on panC sequence analysis. Sequences were aligned with Clustal Ω [28] and further analysed using SplitsTree4 [29]. (A) Overview. (B) Enlargement of the B. thuringiensis isolates, which were all assigned to panC type IV. Within these, five distinct clusters (a-e) were identified. Blue: animal isolate, green: food, red: biopesticide, brown: soil, black: unknown origin.
Within the B. thuringiensis strain set a high variability was detected, from maximal germination of 45% in CGY medium with heat activation to no germination. Furthermore, different germination patterns emerged (see Figure 2 and Table 2). Pattern A is characterized by minimal or no germination (MHI 2870 and MHI 2873). Strong germination in nutrient-rich medium with heat activation, but little germination in cRPMI medium was classified as pattern B (six food isolates, three biopesticidal strains and MHI 2882 isolated from larva). Pattern C germination seemed to be heat-as well as nutrient-dependent. This was detected for spores of biopesticidal strains as well as isolates of unknown origin. Pattern D comprised mostly soil isolates, which showed heat-and nutrientindependent germination. Spores with heat-dependent, but mostly nutrient-independent germination were summarized in pattern E.
Beyond germination, the swimming ability of the isolates was investigated using CGY plates with 0.25% agar. With the exception of insect isolates MHI 2873 and MHI 2882, and biopesticide isolate MHI 3241, all strains were able to actively move after 24 h incubation at 37 °C, while MHI 2870, as well as the salad and biopesticide isolates MHI 3186, MHI 3187, MHI 3188, MHI 3189 and Phylogenetic relationship between the 24 tested B. thuringiensis strains, based on panC sequence analysis. Sequences were aligned with Clustal Ω [28] and further analysed using SplitsTree4 [29]. (A) Overview. (B) Enlargement of the B. thuringiensis isolates, which were all assigned to panC type IV. Within these, five distinct clusters (a-e) were identified. Blue: animal isolate, green: food, red: biopesticide, brown: soil, black: unknown origin.
Within the B. thuringiensis strain set a high variability was detected, from maximal germination of 45% in CGY medium with heat activation to no germination. Furthermore, different germination patterns emerged (see Figure 2 and Table 2). Pattern A is characterized by minimal or no germination (MHI 2870 and MHI 2873). Strong germination in nutrient-rich medium with heat activation, but little germination in cRPMI medium was classified as pattern B (six food isolates, three biopesticidal strains and MHI 2882 isolated from larva). Pattern C germination seemed to be heat-as well as nutrient-dependent. This was detected for spores of biopesticidal strains as well as isolates of unknown origin. Pattern D comprised mostly soil isolates, which showed heat-and nutrient-independent germination. Spores with heat-dependent, but mostly nutrient-independent germination were summarized in pattern E.  MHI 3190 showed maximal diameters ( Figure 3). Generally, various strains-including isolates from biopesticides-were able to germinate in low-nutrient cRMPI medium and also to actively move; hence, further prerequisites for the onset of the disease are given.   Beyond germination, the swimming ability of the isolates was investigated using CGY plates with 0.25% agar. With the exception of insect isolates MHI 2873 and MHI 2882, and biopesticide isolate MHI 3241, all strains were able to actively move after 24 h incubation at 37 • C, while MHI 2870, as well as the salad and biopesticide isolates MHI 3186, MHI 3187, MHI 3188, MHI 3189 and MHI 3190 showed maximal diameters (Figure 3). Generally, various strains-including isolates from biopesticides-were able to germinate in low-nutrient cRMPI medium and also to actively move; hence, further prerequisites for the onset of the disease are given.

Growth under Simulated Intestinal Conditions and Production of Enterotoxins
The crucial step for B. cereus-associated food infections is growth and enterotoxin production of the bacteria in the human intestine. Representatively, reciprocal titres of toxin components NheB and Hbl L2 were determined in culture supernatants after six hours of growth. All B. thuringiensis isolates tested in this study were able to grow under simulated intestinal conditions (cRPMI medium, 37 °C, 7% CO2) and to secrete significant amounts of NheB ( Figure 4A, B). Applying a classification scheme developed for B. cereus (high NheB production: titres >2000; medium: 1000-2000; low: <1000; [15]), a uniform distribution of high, medium and low toxin producing strains was observed (see Table 1 and Figure 4B). Under these terms, eight strains (mainly biopesticide and salad isolates) showed NheB titres below 1000 ( Figure 4B), but conclusions should be drawn carefully, as this is still sufficient to cause significant damage to CaCo-2 cells (see Section 3.4). Additionally, NheB titres were determined after six hours growth in CGY medium, as this is applied in routine diagnostics of B. cereus [39]. Due to accelerated growth, titres were enhanced in CGY medium compared to cRPMI. Furthermore, all isolates secreted significant amounts of Hbl L2 under both conditions, similar to B. cereus [15,39], with the exception of Culicidae larva isolate MHI 2882 ( Figure 4C). Motility of the B. thuringiensis isolates. The 1 µL overnight culture was injected into the centre of CGY plates with 0.25% agar. Swimming diameters were measured after 24 h incubation at 37 • C. *: Significant difference compared to full-plate expansion (p-value ≤ 0.05; unpaired t-test, 95% confidence interval). Blue: animal isolate, green: food, red: biopesticide, brown: soil, black: unknown origin.

Growth under Simulated Intestinal Conditions and Production of Enterotoxins
The crucial step for B. cereus-associated food infections is growth and enterotoxin production of the bacteria in the human intestine. Representatively, reciprocal titres of toxin components NheB and Hbl L2 were determined in culture supernatants after six hours of growth. All B. thuringiensis isolates tested in this study were able to grow under simulated intestinal conditions (cRPMI medium, 37 • C, 7% CO 2 ) and to secrete significant amounts of NheB ( Figure 4A,B). Applying a classification scheme developed for B. cereus (high NheB production: titres >2000; medium: 1000-2000; low: <1000; [15]), a uniform distribution of high, medium and low toxin producing strains was observed (see Table 1 and Figure 4B). Under these terms, eight strains (mainly biopesticide and salad isolates) showed NheB titres below 1000 ( Figure 4B), but conclusions should be drawn carefully, as this is still sufficient to cause significant damage to CaCo-2 cells (see Section 3.4). Additionally, NheB titres were determined after six hours growth in CGY medium, as this is applied in routine diagnostics of B. cereus [39]. Due to accelerated growth, titres were enhanced in CGY medium compared to cRPMI. Furthermore, all isolates secreted significant amounts of Hbl L2 under both conditions, similar to B. cereus [15,39], with the exception of Culicidae larva isolate MHI 2882 ( Figure 4C).   Reciprocal NheB titres after six hours growth in CGY medium (nutrient-rich, aerobic conditions) are shown for comparison (orange). (C) Hbl L2 production after six hours growth under simulated intestinal conditions (cRPMI, dark blue). Reciprocal Hbl L2 titres after six hours growth in CGY medium (nutrient-rich, aerobic conditions) are shown for comparison (light blue). *: Significant difference compared to the isolate with highest toxin production (p-value ≤ 0.05; unpaired t-test, 95% confidence interval). Blue: animal isolate, green: food, red: biopesticide, brown: soil, black: unknown origin.

Exhibition of Cytotoxic and Haemolytic Activity
Beyond toxin production, all tested B. thuringiensis isolates grown under simulated intestinal conditions showed cytotoxic activity towards the human colon adenocarcinoma cell line CaCo-2 ( Figure 5A). According to a prior classification of B. cereus strains [15], 14 of the tested B. thuringiensis isolates were classified as highly (titres > 100) and eight isolates as medium (titres from 50 to 100) toxic (see Table 1). Only two strains exhibited low cytotoxicity (titres < 50), which were MHI 2870 and biopesticide isolate MHI 3241. Similar to toxin production (see Section 3.3), cytotoxicity titres where accordingly enhanced when the strains were grown in CGY medium ( Figure 5A). Data from WST-1 bioassays were supported by haemolysis ( Figure 5B). Only strain MHI 3241 exhibited no haemolytic activity on sheep blood agar. Extension of the haemolytic zone varied strain-specifically, with isolates MHI 271, MHI 2873 and MHI 3370 showing the discontinuous haemolysis zone phenomenon [13,47,48]. Furthermore, data obtained from EIAs and WST-1 bioassays were statistically analysed. Spearman correlation tests revealed that NheB, Hbl L2 and cytotoxicity titres of all tested B. thuringiensis isolates generally tended to increase together, but weakly and in a non-significant manner (Table 3A). accordingly enhanced when the strains were grown in CGY medium ( Figure 5A). Data from WST-1 bioassays were supported by haemolysis ( Figure 5B). Only strain MHI 3241 exhibited no haemolytic activity on sheep blood agar. Extension of the haemolytic zone varied strain-specifically, with isolates MHI 271, MHI 2873 and MHI 3370 showing the discontinuous haemolysis zone phenomenon [13,47,48]. Furthermore, data obtained from EIAs and WST-1 bioassays were statistically analysed. Spearman correlation tests revealed that NheB, Hbl L2 and cytotoxicity titres of all tested B. thuringiensis isolates generally tended to increase together, but weakly and in a non-significant manner (Table 3A).

Growth Inhibition by Essential Oils (EOs)
Nearly all B. thuringiensis isolates tested in this study secreted sufficient amounts of enterotoxins under simulated intestinal conditions to harm human colon cells in a B. cereus-like magnitude. Thus, the antimicrobial activity of essential oils (EOs) was assessed as a preventive measure against putatively enteropathogenic B. thuringiensis. Initially, inhibition zones in disk diffusion assays were investigated. Except thyme, each EO was able to inhibit growth of all tested B. thuringiensis strains ( Figure 6A and Figure S1). Susceptibility of all isolates to cinnamon OR was similar (inhibition zone approx. 11 mm), with the exception of highly sensitive MHI 3241 (22 mm). For the other EOs, especially oregano, winter savory and citral, higher strain-specific variations were detected ( Figure 6A and Figure S1).  Cinnamon Chinese cassia, citral, oregano and winter savory were then chosen to evaluate MIC and MTC values by performing antimicrobial broth dilution assays ( Figure 6B,C, Figures S2 and S3). Average MICs were 0.067% for winter savory, 0.044% for oregano, 0.025% for citral and 0.022% for cinnamon Chinese cassia. Average MTCs were 0.0037% for winter savory, 0.003% for oregano, 0.0019% for citral and 0.0016% for cinnamon Chinese cassia. Thus, cinnamon Chinese cassia showed the highest antimicrobial activity towards the tested B. thuringiensis isolates, followed by citral, oregano and winter savory. Nonetheless, the previously observed highly strain-specific antimicrobial activity was further intensified (Figure 6B,C, Figures S2 and S3). MIC and MTC values of the strain set were also statistically analysed. Results of Spearman correlation tests are summarized in Table 3B. With the exception of oregano, MIC and MTC values did not significantly correlate. MIC values of citral and cinnamon Chinese cassia (not significantly), citral and winter savory (not significantly), and winter savory and oregano (significantly) tended to increase together. This was also observed for the MTC values of cinnamon Chinese cassia and oregano, and winter savory and oregano (not significant). Thus, despite some correlation, the great variability between the tested isolates became yet again obvious. Nevertheless, soil isolate MHI 2876, which showed highest resistance to all tested EOs, as well as biopesticide isolate MHI 3241 (B. thuringiensis ssp. tenebrionis), which was most susceptible to all tested EOs except cinnamon Chinese cassia, were particularly remarkable.

Discussion
Over 400 B. thuringiensis-based formulations have been registered, mostly consisting of insecticidal toxins and viable spores [22]. In Germany, there are currently 16 authorized plant protection products containing B. thuringiensis, including biopesticides 1, 2 and 4 tested in the present study. The license for biopesticide 3 ended in 2019, its grace period ends in October 2020. B. thuringiensis concentrations of up to 540 g/kg are used in these products [49]. The present study aimed to investigate the enteropathogenic potential of B. thuringiensis as well as its susceptibility to essential oils, with special attention being paid to differences between the single isolates from soil, animals, food and commercially available biopesticides. It is already known that B. thuringiensis, as member of the B. cereus group, harbours genes encoding enterotoxins, which were also detected in the present study (see Table 1). Remarkably, all strains harboured hbl, which is typically found in 40-70% of B. cereus group isolates under investigation [26,50,51]. One older study also described a significantly higher presence of hblCDA in B. thuringiensis than in B. cereus [52].
Furthermore, all isolates investigated in the present study were assigned to panC type IV. The authors of a previous study with similar results concluded that biopesticide strains can thus indeed be present in foods [23]. Meanwhile, B. cereus group isolates of panC type IV cannot only be found in soil, water or insects, but also in conjunction with diarrhoeal outbreaks [23,27,39]. Additionally, potentially toxigenic pesticidal strains were isolated from tomatoes, cucumbers and peppers [53]. B. thuringiensis strains from tomatoes and bell pepper matched MLST (Multilocus Sequence Typing) types of the biopesticidal strains B. thuringiensis ssp. kurstaki ABTS-351 and B. thuringiensis ssp. aizawai ABTS-1857 [54]. In another study, isolates from food and beverage samples were indistinguishable from B. thuringiensis ssp. kurstaki from commercial biopesticides [55]. B. thuringiensis ssp. kurstaki HD1 was detected in cabbage products for human consumption [56], and a biopesticidal strain was even isolated from faecal samples of greenhouse workers who were exposed to B. thuringiensis-based insecticides [57].
In order to assess the potential health risk originating from (biopesticidal) B. thuringiensis isolates in more detail, we additionally investigated germination and motility behaviour of the strain set, which are further crucial prerequisites for the onset of the diarrhoeal disease [14,[40][41][42][43][44][45][46]. Although different germination patterns emerged, spores of most isolates were able to germinate, even in cRPMI medium with preceding heat treatment mimicking re-heating of stored foods and subsequent consumption. Although highly strain-specific, no significant differences between biopesticidal and other isolates were detected, which was also true for swimming motility (see Figures 2 and 3 and  Table 2). A connection between motility and enterotoxin secretion in B. thuringiensis has already been made [58,59].
Despite all important prerequisites, production of enterotoxins in the intestine is the most relevant factor for the manifestation of the diarrhoeal disease. Although enterotoxin production by B. thuringiensis isolated from different sources including biopesticides has been shown in previous studies, comparative cytotoxicity assays were rarely performed [23,[60][61][62][63]. Moreover, to our knowledge, the present study is the first one to compare enterotoxin production and cytotoxicity of a set of B. thuringiensis isolates grown under simulated intestinal conditions. Generally, enterotoxin production as well as cytotoxic activity is highly variable within enteropathogenic B. cereus strains. The reasons for this are still not completely explored [39,64]. A similar variability in NheB and Hbl L2 production as well as cytotoxicity towards CaCo-2 cells could be shown for the B. thuringiensis isolates investigated in the present study (see Figures 4 and 5). Nevertheless, applying a classification scheme established for enteropathogenic B. cereus grown under simulated intestinal conditions [15], the majority of our B. thuringiensis isolates had to be classified as medium or highly toxic, including biopesticide strains (see Table 1). Hence, despite high strain-specific variations, a considerable hazardous potential due to B. thuringiensis isolates from various origins is indisputable.
Due to these facts, the scientific opinion of the EFSA (European Foods Safety Authority) BIOHAZ Panel (2016) is entirely comprehensible. Clear discrimination between B. cereus and B. thuringiensis in routine food and clinical diagnostics is required as well as unequivocal identification of biopesticide strains by whole genome sequencing or additional, specific markers. For an improved risk assessment, further data on dose-response and characterization of B. thuringiensis are urgently needed, as well as field studies after application of the biopesticides to determine pre-harvest intervals [22]. On the other hand, the major advantage of B. thuringiensis-its high specificity towards single insect species-must not be neglected, which led to a significant reduction of the use of chemical insecticides [1,65]. Moreover, defenders of B. thuringiensis claim that the genotypes of biopesticidal strains have never been found in outbreaks, and that only environmental isolates, not biopesticidal strains, have been detected in high concentrations in foodstuffs [66]. However, particularly to prove this, more data on the biopesticide strains as well as reliable tools for their clear identification are necessary.
Until these data are available, the use of EOs can be an additional preventive strategy against putatively enteropathogenic B. thuringiensis. It has been shown that B. cereus is sensitive to EOs from nutmeg, mint, clove, oregano, cinnamon, sassafras, sage, thyme and rosemary [67][68][69][70][71][72], Satureja, Origanum and Thymus species [73,74], lemon, orange and bergamot [75], cumin, black pepper and black cumin [76,77], Daucus gracilis [78], as well as carvacrol [79][80][81][82] and many others. Nevertheless, most of these studies have been conducted with only one or few representative isolates, and a comparative study including B. cereus strains of different origin, toxin profile, cytotoxic activity etc. is missing. Studies on the sensitivity of B. thuringiensis are rarely found, and-due to the high variability inside the B. cereus group-conclusions cannot necessarily be drawn for B. thuringiensis. Only two earlier studies showed susceptibility of B. thuringiensis towards EOs of Thymus vulgaris [83] and Ferulago angulata [84]. In contrast to that, none of the tested strains in the present study was inhibited by thyme oil (Figure 6), which once more underlines the high variability of EO effects. Comparing average MIC values, we determined an antimicrobial activity of cinnamon Chinese cassia > citral > oregano > winter savory, which largely corresponds with earlier studies on B. cereus [72]. However, as in all tests applied in this study, strain-specific differences must be considered.

Conclusions
Altogether, the present study stresses the enteropathogenic potential of B. thuringiensis, even of biopesticidal strains, which were generally indistinguishable from other isolates in the applied tests. Application of these pesticides might thus indeed constitute a hazard to human health. On the other hand, the use of B. thuringiensis in agriculture significantly reduced the application of chemical insecticides. We conclude that banning of B. thuringiensis biopesticides is not implementable due to its various advantages. Nevertheless, the applied strains have to be extremely well characterized (for instance by whole genome sequencing) and labelled to provide gapless information about their safety and to eliminate any human health hazard. At this, routine discrimination between B. cereus and B. thuringiensis is as important as controlled application of the biopesticides. Additionally, clear labelling of foods treated with B. thuringiensis can benefit consumers.
Until then, the use of essential oils as antimicrobials might diminish the risks by reducing the level of contamination, especially in times of increased antibiotic resistance and enhanced consumers' demand for natural antimicrobial agents. It can also be concluded from the present study that, whenever predicting the hazardous potential or using antimicrobials, the high variety within B. cereus group strains, even B. thuringiensis, must be taken into account, whether it be their ability to produce enterotoxins, their cytotoxic activity, or their resilience towards antimicrobials.