Bacillus cereus NVH 0500/00 Can Adhere to Mucin but Cannot Produce Enterotoxins during Gastrointestinal Simulation

Abstract

Adhesion to the intestinal epithelium could constitute an essential mechanism of Bacillus cereus pathogenesis. However, the enterocytes are protected by mucus, a secretion composed mainly of mucin glycoproteins. These may serve as nutrients and sites of adhesion for intestinal bacteria. In this study, the food poisoning bacterium B. cereus NVH 0500/00 was exposed in vitro to gastrointestinal hurdles prior to evaluation of its attachment to mucin microcosms and its ability to produce nonhemolytic enterotoxin (Nhe). The persistence of mucin-adherent B. cereus after simulated gut emptying was determined using a mucin adhesion assay. The stability of Nhe toward bile and pancreatin (intestinal components) in the presence of mucin agar was also investigated. B. cereus could grow and simultaneously adhere to mucin during in vitro ileal incubation, despite the adverse effect of prior exposure to a low pH or intestinal components. The final concentration of B. cereus in the simulated lumen at 8 h of incubation was 6.62 ± 0.87 log CFU ml⁻¹. At that point, the percentage of adhesion was approximately 6%. No enterotoxin was detected in the ileum, due to either insufficient bacterial concentrations or Nhe degradation. Nevertheless, mucin appears to retain B. cereus and to supply it to the small intestine after simulated gut emptying. Additionally, mucin may play a role in the protection of enterotoxins from degradation by intestinal components.

INTRODUCTION

The food-poisoning bacterium Bacillus cereus produces one or more enterotoxins, including the nonhemolytic enterotoxin (Nhe), hemolysin BL (Hbl), and cytotoxin K (CytK) (1). These toxins are pore-forming proteins (2, 3) that can cause diarrhea, presumably through colloid osmotic lysis of the epithelial cell membranes (3, 4). The expression of enterotoxins appears to be controlled by quorum sensing (5, 6), which suggests that B. cereus must outlive gastrointestinal passage and proliferate in the intestine in order to produce toxins (6, 7). Additionally, enterotoxins must be formed in sufficient amounts and maintain their functionality long enough to damage the intestinal epithelium and cause diarrhea. Both bacterial survival and enterotoxin production depend on the environmental determinants (history) and intrinsic properties of B. cereus, as well as the host status (e.g., health, age) and defenses (8–12).

In vitro data suggest that B. cereus vegetative cells are generally sensitive to the low pH encountered in the stomach (9, 10), and they may be inactivated by bile secreted in the duodenum (12–14). In contrast, spores resist both acidic conditions and the presence of bile (9, 10, 12, 13), but they do not germinate at bile salt concentrations between 5 and 10 g liter⁻¹ (12). Incubation of B. cereus in the presence of microbial competition inhibits the outgrowth of germinated spores (15), and the effect is more pronounced when the intestinal bacteria are more abundant than B. cereus (16). Other studies have shown that spore germination and growth are possible during gastrointestinal simulations; however, enterotoxin production was not assessed (17). Enterotoxins, when produced, are unstable (6) and can be degraded rapidly in the presence of digestive enzymes (18, 19). Evidently, B. cereus has to withstand and adapt to the environmental hurdles posed by the host (20) and requires a subset of specific conditions to express its virulence. The factors regulating enterotoxin gene expression in B. cereus have been comprehensively reviewed by Ceuppens et al. (1).

Adhesion of B. cereus to the intestine has been proposed as a mechanism of pathogenesis (19, 21, 22). B. cereus adhesion to the gut surfaces may increase bacterial transit time and result in enterotoxin secretion in the proximity of the epithelium. This would facilitate the rapid transfer of toxins to the target cells and protect them from extensive dilution or degradation in the intestinal lumen (15, 19, 22). Most relevant studies focus on the adhesion of B. cereus to Caco-2 cells, which may express a small-intestinal phenotype after differentiation, or other cell lines (22–25). Nevertheless, enterocytes are covered by mucus consisting mainly of mucin glycoproteins that serve as sources of nutrients and offer a plethora of adhesion sites for both commensal and pathogenic bacteria (26, 27). Yet to date, only a few studies have dealt with the attachment of B. cereus to mucin substrata. We have demonstrated recently that food-poisoning strains of the B. cereus group could adhere to mucin preparations to different extents during short term incubation under various environmental conditions (28). Sánchez et al. (29) demonstrated that the attachment of Bacillus isolates from probiotics to intestinal models was dependent on the bacterial strain, the physiology (spores or cells), and the adhesion substratum (mucin, Matrigel, or Caco-2 cells). The latter finding suggests that bacterial affinity for Caco-2 cells does not guarantee efficient attachment to other biological surfaces and that the specific interactions of B. cereus with mucin need to be further investigated.

In this study, the food-poisoning bacterium B. cereus NVH 0500/00 was exposed to various gastrointestinal hurdles in vitro prior to evaluation of its attachment to mucin microcosms. The ability of this strain to produce Nhe toxin under these conditions was studied by using a commercial immunoassay. Subsequently, we investigated the role of mucin in the persistence of B. cereus NVH 0500/00, by using the mucin adhesion assay (30), and the stability of enterotoxins in vitro in the presence of intestinal components. The aim of this work was to gain insight into the adhesion potential of B. cereus after exposure to a simulated host environment, to elucidate its involvement in pathogenesis, and to uncover potentially protective effects of mucin for B. cereus or its enterotoxins.

MATERIALS AND METHODS

Definitions

Some key terms used in this article are defined as follows: (mucin-) adherent bacteria, bacteria attached to mucin agar or microcosms; luminal bacteria, nonadherent bacteria; supernatant, a bacterium-free solution obtained after the centrifugation of a bacterial culture; (simulated) lumen, a solution with intestinal components and bacteria (in contact with mucin agar or microcosms); intestinal water, a solution with intestinal components but no bacteria (in contact with mucin agar).

Gastrointestinal simulation. (i) Bacterial strains and preparation of precultures.

B. cereus NVH 0500/00 was kindly provided by M. H. Guinebretière of the National Institute for Agricultural Research (INRA; France). This strain has been involved in food poisoning (31) and produces Nhe (32) but does not contain hbl or cytK (33, 34). B. cereus NVH 0500/00 adheres well to mucin agar (28, 30).

A loopful of B. cereus NVH 0500/00 was transferred aseptically from a −80°C stock culture to test tubes containing 10 ml of brain heart infusion (BHI; Oxoid) broth. The preculture was incubated overnight at 37°C (atmospheric headspace) without shaking. The overnight culture was 10-fold diluted in fresh BHI broth and was incubated at 37°C (atmospheric headspace) on a rotary shaker (110 rpm) for approximately 2.5 h. Subsequently, the culture was centrifuged at 5,000 × g for 20 min at 4°C, and the pellet was washed twice with 0.1 M potassium phosphate-buffered saline (PPBS) at pH 7.0. The cells were resuspended in the same buffer to an optical density at 610 nm (OD₆₁₀) of approximately 1.5 (±0.15).

Five independent biological replicates were performed on different days. In total, 13 technical replicates were executed (2 technical replicates for each of 2 biological replicates, and 3 technical replicates for each of the other 3 biological replicates).

(ii) Stomach simulation.

The suspension with an OD₆₁₀ of ca. 1.5 was centrifuged at 5,000 × g for 20 min at 4°C. The pellet was resuspended in 280 ml of simulated food solution and was placed in an incubation vessel. The initial bacterial concentration was ca. 7.5 to 8.0 log CFU ml⁻¹ and was determined as described under “Sampling” below. The simulated food solution contained (per liter) 1g arabinogalactan (Sigma), 2 g pectin from apple (Sigma), 1 g xylan from beechwood (Sigma), 3 g starch from potato (Anco), 0.4 g d-glucose (Sigma), 3 g yeast extract (Oxoid), 1 g proteose peptone (Oxoid), 0.5 g l-cysteine (Sigma), and 4 g type II mucin from porcine stomach (Sigma). During the 2-h gastric incubation, a stepwise decrease of 1 pH unit per 30 min, from pH 5.0 to pH 2.0 (±0.1), was accomplished by automated addition of 0.5 N HCl (VWR) using multiparameter controllers (Consort R305) equipped with pH electrodes.

(iii) Upper small intestine (duodenum/jejunum) simulation.

The simulated gastric suspension was transferred to a new vessel with the simultaneous addition of 120 ml of a fresh intestinal solution using peristaltic pumps with a mixing ratio of 7:3 (the pumping of the two solutions started and ended simultaneously). The intestinal solution contained (per liter) 12.5 g NaHCO₃ (Sigma-Aldrich), 1 g pancreatin from porcine pancreas (Sigma-Aldrich), and 6 g oxgall bile (Difco). The final concentrations of bile and pancreatin (intestinal components) were 1.8 g liter⁻¹ and 0.3 g liter⁻¹, respectively. The pH of the final mixture was adjusted to 7.0 ± 0.1 by automated addition of 0.5 N NaOH (VWR) using pH controllers and was kept constant until the end of incubation. The retention time was 2 h.

(iv) Ileum simulation.

Subsequently, vessels were operated either for four more hours without further manipulation (control) in order to harbor only luminal B. cereus (2 biological replicates with 2 technical replicates each; n = 4) or after the incorporation of a mucosal environment (3 biological replicates with 3 technical replicates each; n = 9). Approximately 230 mucin-covered microcosms (AnoxKaldnes K1 carrier; AnoxKaldnes AB) per liter of (food and intestinal) solution were prepared by submerging the microcosms in freshly prepared mucin agar (35). The mucin agar (pH 6.8) consisted of 5% (wt/vol) porcine mucin type II (Sigma) and 1% (wt/vol) bacteriological agar (Oxoid). Mucin-covered microcosms were prepared the day before use, placed in groups of five, and stored at 4°C above a physiological peptone solution (PPS) containing (per liter) 1 g neutralized bacteriological peptone (Oxoid) and 8.5 g NaCl (Merck). The percentage of adhesion at the end of the ileal incubation was calculated as [(adherent bacteria_{8 h})/(adherent bacteria_{8 h} + luminal bacteria_{8 h})] × 100, where bacteria are measured in CFU per milliliter.

Table 1 gives an overview of the experimental setup during gastrointestinal simulations.

TABLE 1.

Experimental setup used during simulated gastrointestinal passage of B. cereus NVH 0500/00

Time (h)	pH	Simulation	Action	Effect studied
0–2	5.0–2.0 (stepwise decrease of 1 pH unit per 30 min)	Stomach (luminal bacteria)	280 ml simulated food inoculated with B. cereus at 0 h	pH
2–4	7.0	Upper intestine (luminal bacteria)	120 ml intestinal solution at 2 h	Bile/pancreatin
4–8	7.0	Ileum (luminal/adherent bacteria)	230 mucin-covered microcosms liter−1 at 4 h or not	Adhesion

(v) Sampling.

Samples were collected with a sterile syringe during incubation in order to determine the concentration of luminal bacteria in all the simulated gastrointestinal compartments. Adherent bacteria were extracted from mucin-covered microcosms collected during ileal incubations (5 microcosms per sample). First, the microcosms were immersed in beakers containing sterile 0.1 M PPBS (pH 7.0) to remove loosely adherent bacteria. Then the mucin agar was removed from the microcosms with a sterile needle, diluted 1:10 (wt/wt) in PPS, and homogenized for 5 min in a stomacher (Stomacher Lab-Blender 400; Seward). Luminal and adherent bacteria were plated on mannitol-egg yolk-polymyxin (MYP; Oxoid) agar after 10-fold serial dilutions in PPS. MYP was prepared according to the manufacturer's instructions. Plates were incubated overnight at 32°C.

The production of toxin (Nhe) from B. cereus NVH 0500/00 was determined in a 150-μl undiluted ileal sample using the Duopath Cereus Enterotoxins kit (Merck).

Persistence of mucin-adherent B. cereus after simulated lumen emptying as determined using the mucin adhesion assay. (i) Bacterial strains and preparation of cultures.

Cultures of B. cereus NVH 0500/00 were prepared as described in the preceding section. Bacteria with an OD₆₁₀ of 1.5 (±0.15) in 0.1 M PPBS (pH 7.0) were 1,000 times diluted in the same buffer before use in the mucin adhesion assay.

(ii) Mucin adhesion assay.

The mucin adhesion assay was performed as described by Tsilia et al. (30). One milliliter of the diluted bacterial suspension was added to 12-well plates (VWR) containing 1.2 ml mucin agar (of the same composition as that used to cover the microcosms). This system was used to simulate lumen and mucus. Multiwell plates were incubated at 37°C for 90 min on a rotary shaker (30 to 50 rpm) using a MART jar (Mart Microbiology) with AnaeroGen (Oxoid) sachets (phase I).

After incubation, the mucin agar surface was rinsed twice with 0.1 M PPBS (pH 7.0) to remove loosely adherent cells. Subsequently, the mucin agar was covered with 1 ml of sterile BHI broth or 0.1 M PPBS (pH 7.0). Plates were incubated for 2 h under the conditions described for phase I (phase II).

All experiments were performed in quadruplicate.

(iii) Sampling.

Samples representing the simulated lumen were collected (postincubation) after mixing five times by pipetting to determine the levels of luminal bacteria (phases I and II). The mucin agar surface was washed as described under “Mucin adhesion assay” above, and the whole mucin layer was transferred with a UV-sterilized spatula to a stomacher bag with 10 ml PPS. The mixture was homogenized for 5 to 10 min in a stomacher to extract adherent bacteria (phases I and II). Luminal as well as adherent B. cereus concentrations were determined by plating on BHI agar after 10-fold serial dilutions in PPS. Plates were incubated overnight at 32°C.

Degradation of Nhe enterotoxin by intestinal components. (i) Bacterial strains and preparation of cultures (step 1).

An overnight culture of B. cereus NVH 0500/00 was diluted 10-fold in fresh BHI broth buffered with 0.1 M PPBS (pH 7.00) and was incubated on a rotary shaker (110 rpm) at 37°C for 3 h (atmospheric headspace). Production of the Nhe enterotoxin was confirmed with the Duopath kit (Fig. 1, step 1). The pH of the culture was then corrected to 7.00 ± 0.05 using 1 N NaOH.

FIG 1.

Flow chart of simulation experiments used to investigate the stability of Nhe in the presence of intestinal components. Step 1, production of Nhe enterotoxin after growth of B. cereus NVH 0500/00 in buffered BHI; step 2, transfer of enterotoxin to the mucin agar layer during incubation with culture supernatant positive for Nhe; step 3, stability of Nhe in the presence of mucin agar after the addition of intestinal components (experiments 5 and 6). Culture (CUL) and supernatant (SUP) were also incubated without mucin agar in the absence (experiments 1 and 3, respectively) and presence (experiments 2 and 4, respectively) of intestinal components (controls). During each step, the presence of Nhe in the mucin agar extract (mucin) or other samples (culture, supernatant, intestinal water [IW]) was evaluated using Duopath. NA, not applicable; Nhe+, Nhe positive; Nhe−, Nhe negative; *, Nhe negative for 0.1 M PPBS. Exp., experiment.

(ii) Transfer of Nhe to mucin agar (step 2).

A 1.2-ml volume of the Nhe-positive culture (Fig. 1, step 2, experiments 1 and 2) or its supernatant obtained after centrifugation at 5,000 × g for 15 min at 4°C (Fig. 1, step 2, experiments 3 and 4) was placed in empty wells of a 12-well plate (controls). Wells containing 1.2 ml mucin agar also received 1.2-ml aliquots of the Nhe-positive supernatant (Fig. 1, step 2, experiments 5 and 6). Plates were incubated at 37°C for 4 h at 110 rpm. The supernatant incubated with the mucin agar was then collected, pooled, and evaluated for the presence of Nhe. The mucin surface was then rinsed twice with 0.1 M PPBS at pH 7.0 to remove any remaining toxin. Subsequently, the mucin agar was transferred with a UV-sterilized spatula to a stomacher bag containing 3 ml 0.1 M PPBS (pH 7.0) and was homogenized for 5 min in the stomacher. The sample was centrifuged briefly to remove large mucin agar particles and was tested for the presence of Nhe using Duopath.

(iii) Stability of Nhe in the presence of intestinal components (step 3).

The supernatant collected after a 4-h incubation with mucin agar in step 2, as well as 0.1 M PPBS (pH 7.00), was mixed with intestinal components to a final concentration of 2.5 g oxgall and 0.815 g pancreatin per liter solution (Fig. 1, step 3, experiments 5 and 6, respectively). Aliquots (1.2 ml) of each solution were placed in wells with Nhe-positive mucin agar (from step 2). Both the supernatant and PPBS supplemented with intestinal components and placed on mucin agar are referred to as intestinal water (IW). Plates were incubated at 37°C on a rotary shaker at 110 rpm. The culture and culture supernatant (controls) were also incubated either without (experiments 1 and 3, respectively) or with (experiments 2 and 4, respectively) intestinal components. During incubation, the culture, the culture supernatant, the intestinal water, and the mucin layer were tested for the presence of toxin.

Statistical analysis.

Statistical analysis was performed using R, version 3.2.0 (http://www.R-project.org). The normality of log-transformed data was assessed with the Shapiro-Wilks test, as well as Q-Q plots. For a low number of observations (n, 3 to 5), no distributional assumptions were made. Homoscedasticity for pairwise comparison was performed using a bootstrapped classical Levene's test (with a correction factor) or a modified Levene's test (with a modified structural zero removal method and correction factor) for normally or nonnormally distributed data, respectively. In the two-sample case, nonnormally distributed data were compared with the Wilcoxon rank sum test, while a two-sample t test was used for normally distributed data (a Welch modified t test was used if the homoscedasticity assumption was not met). For multiple comparisons, analysis of variance (ANOVA) was carried out if the required assumptions were withheld. When admissible, post hoc testing with the Tukey honestly significant difference (HSD) correction was used. If normality was violated but homoscedasticity was withheld, a Kruskal-Wallis rank sum test was executed, followed by a pairwise Wilcoxon rank sum test with the Holm correction (36). When the homoscedasticity assumption was rejected for normally distributed data, a weighted least-squares analysis was performed (with the square rooted residuals as weights); however, if weighting did not stabilize the homoscedasticity, a Welch-corrected ANOVA was carried out. Nonparametric multiple-contrast tests with simultaneous confidence intervals (37, 38) were carried out for the nonnormally distributed data. With these tests, Tukey HSD contrasts were used with a Fisher transformation function. The significance level was set at 5%. The results are expressed as the average log CFU per milliliter with standard deviations.

RESULTS

Gastrointestinal simulation.

The effect of pH on the survival of B. cereus was determined during gastric simulations. Figure 2 (stomach) shows the concentrations of total B. cereus NVH 0500/00 bacteria detected during a stepwise reduction in pH (1 unit per 30 min) from 5.0 to 2.0. A small but significant (P, 0.026 for 0 h versus 1 h) inactivation of B. cereus was observed during the first hour of incubation, corresponding to pH values higher than 4.0. Additionally, a 30-min incubation at pH 3.0 resulted in a large and significant decrease (P < 0.0001) in the total microbial population (Δlog = 2.48 ± 0.56 log CFU ml⁻¹). Further acidification of the medium to pH 2.0 had a limited but statistically significant (P < 0.01) effect on bacteria compared to pH 3.0. After the simulated gastric passage, the final concentration of B. cereus was 4.71 ± 0.34 log CFU ml⁻¹. Two hours after inoculation, the gastric contents were mixed with an intestinal solution in order to simulate the effect of host intestinal secretions (bile/pancreatin) in the upper small intestine. B. cereus NVH 0500/00 could survive without extensive inactivation (P > 0.32) during a subsequent 2-h incubation at pH 7.0 in the presence of 1.8 g bile and 0.3 g pancreatin per liter of solution (Δlog = 0.29 ± 0.75 log CFU ml⁻¹) (Fig. 2, duodenum/jejunum).

FIG 2.

Total luminal (bars) and mucin-adherent (circles) Bacillus cereus NVH 0500/00 bacteria and pH profile (lines) imposed during gastrointestinal simulation. The results are expressed as average log CFU per milliliter with standard deviations. Statistical comparisons were performed for each gastrointestinal compartment separately (e.g., stomach, duodenum/jejunum, or ileum). Capital and lowercase letters refer to luminal and adherent bacteria, respectively. Values designated with the same letter (in a given gastrointestinal compartment) do not differ significantly from each other (P > 0.05).

The ileal phase started with the addition of mucin microcosms in order to study the ability of B. cereus to adhere to relevant intestinal surfaces. The concentration of luminal bacteria in the presence of mucin microcosms did not differ (P > 0.05) from that in the control experiments without mucin microcosms (results not shown); therefore, the luminal bacterial concentration at each time point is given as the average value for both systems (Fig. 2, ileum). After the gastric and duodenal/jejunal challenge, B. cereus could recover during the ileal simulation. This resulted in a significant increase (P < 0.0001) of 2.34 log CFU ml⁻¹ in the luminal bacterial concentration within 4 h of incubation. The concentration of luminal bacteria at the end of the gastrointestinal simulation (8 h) was 6.62 ± 0.87 log CFU ml⁻¹. Nhe was not detected at any point of the ileal incubation. An hour after the addition of mucin microcosms (5 h), 2.8 ± 1.1 log CFU ml⁻¹ was adherent, and the bacterial concentration continued to increase, reaching a significantly higher level, 5.47 ± 0.93 log CFU ml⁻¹ (P = 0.030), at the end of the experiment (Fig. 2, ileum). At this point, 6.02% ± 9.07% of the B. cereus population was adhering to mucin microcosms.

Persistence of mucin-adherent B. cereus after simulated lumen emptying as determined using the mucin adhesion assay.

During phase I of the mucin adhesion assay, the total amounts of luminal and adherent B. cereus NVH 0500/00 bacteria were 5.43 ± 0.31 and 4.28 ± 0.17 log CFU ml⁻¹, respectively (Table 2, phase I). After the mucin agar was washed, sterile PPBS or BHI was added to wells with mucin agar that contained the adherent bacteria from phase I. Within 2 h of incubation, both media were populated with bacteria originating from the mucin layer. The total amount of luminal bacteria (approximately 6 log CFU ml⁻¹) did not differ (P = 0.97) between PPBS and BHI (Table 2, phase II). These values were significantly higher than the luminal B. cereus values from phase I (P, 0.006 for PPBS and 0.01 for BHI). The concentrations of mucin-adherent bacteria in phase II ranged between 4.7 and 5.0 log CFU ml⁻¹ and were also significantly enhanced over those in phase I (P, <0.0006 for PPBS and 0.03 for BHI).

TABLE 2.

Total counts of luminal and mucin-adherent B. cereus NVH 0500/00 bacteria during phase I and phase II incubations^a

Medium	Phase	Avg concn of bacteria (log CFU ml−1) ± SDb
		Luminal	Adherent
PPBS	I	5.43 ± 0.31 A	4.28 ± 0.17 A
PPBS	II	6.09 ± 0.34 B	4.98 ± 0.09 B
BHI	II	6.04 ± 0.18 B	4.65 ± 0.30 B

^aPhase I, adhesion of luminal bacteria to mucin; phase II, repopulation of bacterium-free PPBS or BHI by adherent bacteria.

^bDetermined after plating on BHI agar. Values followed by the same letter in the same column are not significantly different from each other (P > 0.05). Values followed by the same letter in the same row are not indicative of the presence or absence of any significance.

Degradation of Nhe enterotoxin by intestinal components.

Nhe was produced during the growth of B. cereus NVH 0500/00 in buffered BHI (Fig. 1, step 1). The culture supernatant positive for Nhe was incubated for 4 h with (Nhe-negative) mucin agar. After incubation, the enterotoxin was detectable in the mucin agar layer by use of the Duopath test (Fig. 1, step 2, experiments 5 and 6). The stability of Nhe toxin in mucin agar was then investigated in the presence of intestinal water (Fig. 1, step 3, experiments 5 and 6). Control experiments without mucin agar (CUL and SUP) were also performed (Fig. 1, experiments 1 to 4). All results are presented in Table 3.

TABLE 3.

Stability of Nhe toxin in mucin agar, culture, culture supernatant, and intestinal water during incubation in the presence or absence of a mixture of bile and pancreatin

Expta	Codeb	Nhe toxin signal intensityc at the following time (h):					Avg pH during incubation
		0	1	3	6	9
1	CUL−	****	****	****	****	****	6.96 ± 0.04
2	SUP−	****	****	****	****	****	6.98 ± 0.04
3	CUL+	****	***	***	**	*	6.95 ± 0.05
4	SUP+	****	***	***	**	*	6.99 ± 0.05
5a	IW SUP+	****	***	***	***	**	7.04 ± 0.07
5b	Mucin SUP+	****	***	***	*	*	6.80 ± 0.05d
6a	IW PPBS+	*	**	**	**	**	7.03 ± 0.06
6b	Mucin PPBS+	****	***	***	*	*	6.80 ± 0.05d

^aSamples for experiments 5a and 5b were obtained from the same well, as were samples for experiments 6a and 6b.

^bCode of each experiment as shown in Fig. 1. Mucin, mucin agar; CUL, culture; SUP, culture supernatant; IW, intestinal water; +, samples with intestinal components (bile and pancreatin mixture); −, samples without intestinal components.

^cAsterisks indicate the toxin signal intensity as interpreted from the results of the Duopath kit. ****, very strong response; ***, strong response; **, average response; *, weak response.

^dInitial pH.

No changes in the detection of Nhe were noticeable during incubation of the culture and culture supernatant in the absence of intestinal components (CUL− and SUP−). However, the addition of bile and pancreatin (CUL+ and SUP+) resulted in a weaker band density even 1 h after the initiation of the experiment. At 9 h, the signal representing the presence of Nhe was weak. The average pH remained neutral during the experiment.

The toxin present in the mucin layer was adversely affected by the presence of bile and pancreatin in intestinal water and was barely detectable after 6 h (Mucin SUP+ and PPBS+). Degradation of Nhe in the mucin agar (Mucin SUP+) was faster than in the intestinal water (IW SUP+). Interestingly, Nhe signal was weaker in the intestinal water that was not placed back to the mucin agar during step 3 than in its presence (results not shown). When toxin-free PPBS supplemented with bile and pancreatin (IW PPBS+) was placed on enterotoxin-positive mucin agar, the Nhe response in the mucin agar extract also became weaker during the experiment (Mucin PPBS+). The Nhe degradation behavior/pattern was similar in mucin agar (Mucin SUP+ versus PPBS+) despite the type of intestinal water used (IW SUP+ or PPBS+). Toxin was detected in the originally toxin-free intestinal water based on PPBS (IW PPBS+) 1 h after incubation with Nhe-positive mucin agar.

DISCUSSION

Studies on the interaction of pathogenic B. cereus with the epithelium suggest that adhesion is a possible mechanism of pathogenesis (19, 21, 22). The enterocytes are covered by a protective mucus layer that prevents the direct adhesion of bacteria to the intestinal epithelium. Only a limited number of studies have investigated the importance of simulated mucus in the adhesion and nutrition of B. cereus (28, 29). In these studies, adhesion experiments were short term (a maximum of 1.5 h), and bacteria were not exposed to stresses encountered during gastrointestinal passage prior to adhesion. The role of mucin in enterotoxin stability in the presence of intestinal components (host digestive secretions) has thus far not been investigated. Our results showed that the adhesion of B. cereus NVH 0500/00 to simulated mucus increased during a 4-h incubation under conditions mimicking those in the ileum. Approximately 5.5 log CFU ml⁻¹ was bound to mucin at the end of incubation, corresponding to ca. 6% of the total population. This percentage of adhesion is comparable to those of other bacteria known for their preferences for mucin, such as Lactobacillus mucosae, Lactobacillus rhamnosus GG, and adherent-invasive Escherichia coli (approximately 8 to 11%) (30, 35). An increase in the total population of B. cereus in the gut due to adhesion may be critical for the onset of pathogenesis, because the enterotoxins associated with diarrheal disease are positively regulated by a PlcR quorum-sensing system (39). Such cell-to-cell communication mechanisms often promote adhesion and/or biofilm development, as in E. coli, Vibrio scophthalmi, and Pseudomonas aeruginosa (40–42). In contrast, Hsueh et al. (43) demonstrated that biofilm formation by B. cereus ATCC 14579 is repressed by PlcR, at least when nutrients are limited. To what extent the adherent bacterial population contributes to toxin production under intestinal conditions needs to be studied further.

In this study, Nhe was not detected in the lumen during ileal simulations. These results are consistent with those for different B. cereus strains in lasagna verde or tryptone soy broth medium with intestinal components (18). Ceuppens et al. (15) also did not detect enterotoxins (Nhe or Hbl) during in vitro ileal growth of B. cereus in the absence of competitive microbiota. On the other hand, B. cereus F4430/73 incubated for 6 h in an intestinal medium with peas, milk, or chicken produced Hbl; however, the toxin was not detected in the presence of more than 1.2 g bile per liter of solution (14). In this study, the concentration of bile was 1.8 g liter⁻¹, which could explain the lack of enterotoxin. It was demonstrated previously that CytK is produced by a small subpopulation of B. cereus (44). If this is the case for Nhe, the amount of toxin produced by a small fraction of the population may be too small for detection by the Duopath kit. Due to the involvement of a quorum-sensing system in enterotoxin expression (39), a relatively high B. cereus concentration is required for detection. It is not clear if the lack of toxin detection is associated with inefficient cell counts, the detection limit of Duopath (6 ng NheB per ml of solution [45]), or toxin degradation. The stability of the toxin was therefore investigated.

In the absence of intestinal components, Nhe appeared to be stable during incubation in both culture and supernatant (CUL− and SUP−), suggesting that bacterial hydrolases do not affect the enterotoxin (18; this study). In contrast, Gilois et al. (6) have shown that NheB present in the B. cereus supernatant was moderately stable at 30°C, with a half-life of ca. 1.2 h. It is possible that we did not observe Nhe degradation under the conditions tested because the antibodies of the Duopath kit were saturated by the antigen. Nevertheless, the intensity of the Nhe signal was clearly reduced during the incubation of the culture and supernatant with pancreatin and bile (CUL+ and SUP+). Rapid degradation of Nhe in intestinal media (<1 h) has been shown previously (18, 19). Our results demonstrated a degradation rate much lower than those in these studies, possibly due to differences in the setup (e.g., different concentrations and strengths of host proteases, differences in pH, and a different starting concentration of Nhe) or the different potency of strains to produce toxin.

Our study shows that the enterotoxin can be detected in mucin agar after incubation with an Nhe-positive supernatant (Fig. 1). It remains unclear whether the enterotoxin is merely binding to the mucin surface (a receptor-mediated process) or is actually trapped in the mucus layer (a diffusion-associated process). It appears that Nhe degradation in the intestinal water (IW SUP+) is slower than that in the supernatant (SUP+), where no mucin is present, as indicated at 6 h of incubation (Table 3). This is more evident considering that the toxin level in the intestinal water (IW SUP+) at the beginning of the incubation was lower than that in the supernatant (SUP+) due to the partial transfer or binding of Nhe to the mucin (Fig. 1, step 2). These observations suggest that the presence of mucin agar may protect the toxin from degradation by host proteases. Although the toxin signal was more intense at 6 h in the intestinal water than in the mucin agar itself (IW and Mucin SUP+), the proposal of a protective effect may still stand because (i) it is not clear if the toxin concentrations in the intestinal water and mucin agar at 0 h are equal and (ii) there is Nhe release or transfer (flux) from the mucin agar (Mucin PPBS+) to the intestinal water (IW PPBS+). This toxin flux could be the rate-limiting step in toxin degradation. We therefore infer that the mucus layer may play an indirect role in protecting Nhe by restricting the availability of toxin to the lumen. Such a process resembles the protective effect that some extracellular polymeric substances have on bacterial biofilms when they are exposed to antibiotics (46, 47).

Degradation is, of course, relevant only if toxin is produced, which would require the presence of a sufficient B. cereus population in the ileum. Bacterial concentrations between 5 × 10⁶ and 10⁷ CFU ml⁻¹ were required for the detection of Hbl and Nhe by an enzyme-linked immunosorbent assay (ELISA) (21). However, no enterotoxin was previously detected in the presence or absence of intestinal proteases when the B. cereus concentration was <7 log CFU ml⁻¹ (15, 18). In this study, 6.6 log CFU ml⁻¹ of luminal bacteria was present only at the end of the ileal simulation, a value close to the limit mentioned above. The extent of B. cereus outgrowth in the ileum is determined by its survival in the upper gastrointestinal tract. We have observed extensive inactivation at pH <4.0, similar to that shown during the incubation of B. cereus NVH 1230/88 in gastric media (10, 12). The bacteria remaining from the stomach simulation survived the bile and pancreatin treatment and resumed growth 2 h after exposure to these components. Although this behavior is consistent with that of other B. cereus strains (16, 17), studies have shown that factors including bile concentration, pH, and food type may affect the responses of B. cereus to bile (12, 14).

Rapid peristalsis in the small intestine results in a short transit time of the luminal contents (48), including luminal bacteria. On the other hand, adherent bacteria remain in the intestine until the turnover of mucin and epithelium (5 days) (49). Using a mucin adhesion assay, we have shown that adherent bacteria (Table 2, phase I) could “contaminate/recolonize” the simulated lumen (Table 2, phase II). After the recolonization step, the total amounts of both luminal and adherent cells increased. This suggests that the transit of B. cereus between mucus and the lumen could eventually result in a bacterial concentration sufficient to activate toxin production. A related concept representing the release of adherent B. cereus during the renewal of the epithelium has been incorporated into a mechanistic model for the gastrointestinal behavior of B. cereus (21).

To summarize, B. cereus outgrowth and adhesion to mucin occur simultaneously in the simulated ileum. However, this was not sufficient to result in detectable Nhe. It is not clear if this is due to an insufficient B. cereus concentration or to Nhe degradation. Nonetheless, a high infectious dose such as the one used in this study would cause food poisoning according to Pielaat et al. (21). The role of mucin with respect to B. cereus appears to be 4-fold: (i) it serves as a nutrient source (28); (ii) it is a site for adhesion (28, 30; this study); (iii) it retains/supplies bacteria to the small intestine (this study); and (iv) it may protect enterotoxins from degradation by host digestive secretions (this study). The latter possibility needs to be further investigated using quantitative approaches, such as liquid chromatography (LC)-mass spectrometry (MS), and after the introduction of epithelial cells to the model to ensure that toxins reach their intended target in the presence of mucin.