Correspondence: Maximilian Moravek, Institut für Hygiene und Technologie der Lebensmittel tierischen Ursprungs, Tierärztliche Fakultät, Universität München, Schönleutnerstraße 8, 85764 Oberschleißheim, Germany. Tel.: +49 89 2180 78600; fax: +49 89 2180 78602; e-mail: email@example.com
Haemolysin BL (HBL) and nonhaemolytic enterotoxin (Nhe), each consisting of three components, represent the major enterotoxins produced by Bacillus cereus. To evaluate the expression of these toxins, a set of 100 B. cereus strains was examined. Molecular biological characterization showed that 42% of the strains harboured the genes for HBL and 99% for Nhe. The production of all Nhe and HBL components were analyzed using specific antibodies and, in culture supernatants, detectable levels of HBL and Nhe were found for 100% of hbl-positive and 96% of nhe-positive strains. The concentrations of the HBL–L2 and NheB component ranged from 0.02 to 5.6 μg mL−1 and from 0.03 to 14.2 μg mL−1, respectively. Comparison of the amount of NheB produced by food poisoning and food/environmental strains revealed that the median value for all food poisoning strains was significantly higher than for the food/environmental isolates. The data presented in this study provide evidence that specific and quantitative determination of the enterotoxins is necessary to evaluate the toxic potential of B. cereus. In particular, the level of Nhe seems to explain most of the cytotoxic activity of B. cereus isolates and may indicate a highly diarrheic potential.
Bacillus cereus strains, culture medium and culture conditions
The Bacillus cereus strains used in this study were collected and characterized within the EU-research project ‘Preventing Bacillus cereus food borne poisoning in Europe’ (QLK1-CT-2001-00854). The strain set includes clinical isolates and isolates from food remnants connected to food-borne outbreaks, as well as isolates from diverse foodstuffs and the environment (Table 1). For cytotoxicity testing and enzyme immunoassay (EIA) analyses, cells were grown under conditions optimized for toxin production (Glatz & Goepfert, 1976; Beecher & Wong, 1994). In detail strains were incubated in CGY medium supplemented with 1% glucose for 6 h at 32°C or overnight at 25°C (psychrotolerant strains only; n=4) with shaking. EDTA (1 mM) was added at the time of harvesting. Cell-free supernatants were obtained by centrifugation (10 000 g at 4°C for 20 min) and filtration through 0.2 μm Millipore filters and were used for purification of proteins and as coating antigens in the indirect EIA.
Table 1. Characteristics of Bacillus cereus isolates collected in Belgium, Denmark, Finland, France, Germany, India, Japan, Norway, Sweden, Thailand, United Kingdom and USA from 1972 to 2002
Purified HBL–L2 and NheB components which served as standards in the specific immunoassays were prepared by immunoaffinity chromatography as described previously (Dietrich et al., 1999, 2005). Briefly, monoclonal antibodies were attached to CNBr-activated Sepharose 4b (Amersham Biosciences, Freiburg, Germany) according to the manufacturer's instructions. The purification procedure comprised the following steps: (i) storage buffer [phosphate-buffered saline (PBS), containing 0.1% sodium azide] was replaced with PBS; (ii) sample (B. cereus supernatant diluted five times in PBS) was applied; (iii) column was washed with PBS; (iv) bound NheB was eluted with glycine/HCL buffer (pH 2.5); (v) column was washed with PBS and stored in storage buffer. During all steps, the flow rate was set to 1 mL min–1.
Indirect enzyme immunoassays (EIAs)
Using specific antibodies the absence or presence of HBL L1, HBL B, NheA and NheC in cell-free culture supernatants of the B. cereus isolates was determined by indirect EIAs as described recently (Dietrich et al., 1999, 2005).
Sandwich enzyme immunoassay
To obtain quantitative data on the amount of toxin produced by B. cereus isolates, novel Sandwich enzyme immunoassays were established within this study based on previously described monoclonal antibodies. In detail, microtitre plates were coated with 10 μg mL−1 monoclonal antibody 1A12 (directed against HBL–L2) or 5 μg mL−1 monoclonal antibody 2B11 (reactive with NheB) overnight at room temperature. Free protein-binding sites of the plates were blocked with PBS (pH 7.3) containing sodium caseinate (30 g L−1) for 45 min. Then the plates were washed with Tween 20 solution (0.25 mL l−1 of 0.15 mol L−1 of sodium chloride solution) and made semidry. Subsequently, serial dilutions (in PBS containing 0.5% of Tween 20) of cell-free, crude culture supernatants of B. cereus strains were added to each well (100 μL per well) and incubated for 1 h. After a washing step, monoclonal antibody 8B12 (L2) or monoclonal antibody 1E11 (NheB) labelled with horseradish peroxidase [1 : 1000 or 1 : 4000 in PBS containing sodium caseinate (10 g L−1)] was added and again incubated for 1 h at room temperature. Then the plate was washed, and 100 μL of substrate–chromogen solution [1 mmol of 3,3′,5,5′-tetramethylbenzidine, 3 mmol of H2O2 per litre of potassium citrate buffer (pH 3.9)] per well was added. After 20 min, the colour development was stopped with 1 mol of H2SO4 (100 μL per well), and the absorbance was measured at 450 nm. For quantification calibration curves using purified toxin components (L2 and NheB) were prepared and toxin concentration was calculated by linear interpolation (Fig. 1).
Cytotoxicity of the B. cereus culture supernatants was determined using Vero cells as described recently (Dietrich et al., 1999). Briefly, serial dilutions of the supernatants were placed into microtiter plates (0.1 mL per well) and Vero cell suspensions (0.1 mL; 104 cells per well) were added immediately afterwards. The growth medium and diluent consisted of Eagle minimum essential medium (Biochrom KG, Berlin, Germany) with Earle salts supplemented with 1% foetal calf serum and 2 mmol L−1 glutamine. The test was incubated for 24 h at 37°C in a 5% CO2 atmosphere and then the mitochondrial activity of viable cells was determined at 450 nm by using the tetrazolium salt WST-1. The resulting dose–response curve was used to calculate the 50% inhibitory concentration (expressed as the reciprocal dilution that resulted in 50% loss of mitochondrial activity) by linear interpolation.
As several recent studies (Hansen & Hendriksen, 2001; Veld et al., 2001; Guinebretiere et al., 2002) indicate that nearly all strains of Bacillus cereus possess the genes of at least one of the diarrhoeal enterotoxins, the present study was initiated to provide data about the expression of the HBL and Nhe components under in vitro conditions. To cover the biodiversity of B. cereus, the strains used in this study were selected on differences in strain properties, geographical origin and source of isolation. To address the pathogenicity of this species about half of the strains were selected from documented food poisoning outbreaks (Table 1).
First, the strains were analysed by PCR for the presence of the hbl and nhe genes. As a high degree of sequence polymorphism was found for hbl in food associated strains of B. cereus (Guinebretiere et al., 2002), two different sets of primers were used for this purpose. In the first assay, performed according to Guinebretiere et al., (2002), 53 isolates out of the 100 strains tested reacted positive, 42 could be confirmed with a primer pair specific for hblC (Moravek et al., 2004). A similar percentage of PCR-positive strains has been reported by (Prüßet al., 1999), whereas in other studies one or more hbl genes were detected in 59–73% of the strains analysed (Hansen & Hendriksen, 2001; Veld et al., 2001). By using specific immunoassays, expression of the HBL components could be verified for 38 out of the 42 hbl-positive strains enriched at 32°C in CGY broth. After modifying the enrichment conditions, i.e. incubation at 25°C overnight, HBL components were also detected in the supernatants of the remaining four strains. This unusual productivity behaviour was ascribed to the psychrotrophic growth characteristics of these strains (details not shown). Furthermore, 40 (95%) out of the 42 hbl-positive strains produced complete HBL consisting of the three components; the two remaining hbl-positive strains were negative for the B component but produced traces of the L2 and L1 component. By using the newly developed sandwich EIA (Fig. 1), enabling the quantification of the L2 component, 0.02–5.6 μg mL−1 (Table 1) of this protein could be found in the culture supernatants of the analysed strains. As the genes encoding for the components of HBL are transcribed from the same operon in one mRNA (Ryan et al., 1997) and maximum biological activity is obtained at an approximately equimolar ratio of the single compounds (Beecher et al., 1995; Dietrich et al., 1999) it may be assumed that HBL B and L1 are produced to a similar level as the L2 component. All 42 hbl-positive strains carried the nhe operon and produced detectable levels of Nhe.
For the detection of the nhe operon, gene primers, amplifying successively nheA, B and C, were used and like in other recent PCR studies (Hansen & Hendriksen, 2001; Guinebretiere et al., 2002; Ehling-Schulz et al., 2005) with the exception of strain 391/98 (producing cytotoxin K) all strains (99) were positive in this assay. However, by applying specific immunoassays it could be shown that only 93 strains were capable to produce NheA, NheB and NheC simultaneously. Mutations in the PlcR regulon described by (Gohar et al., 2002) and (Slamti et al., 2004) may explain that two strains produced only two components and five isolates were negative for all three compounds, including strain 391/98, which has an nhe operon quite different from other strains (A. Fagerlund and P. E. Granum, unpublished result). Overall, the levels of NheB found in the culture supernatants ranged from 0.03 to 14.2 μg mL−1 (Table 1), thus covering a broader and higher range of toxin concentrations than HBL L2. Moreover, nearly all strains expressing both enterotoxin complexes simultaneously produced higher amounts of Nhe. As also shown in the study of Dietrich et al., (2005) compared with NheA and B the EIA results for the NheC component were very low, which could be explained by a predicted stem–loop structure (in mRNA) between the nheB and nheC gene (Granum et al., 1999). Also the recent report of (Lindbäck et al., 2004) showing maximum cytotoxic activity of the Nhe enterotoxin complex at a molar ratio of NheA, B and C of 10 : 10 : 1, would suggest that NheC is probably produced at low levels.
The cytotoxicity titres of the strains producing all three Nhe components varied from 37 to 3030. As expected, negative results were obtained for the two strains producing only two components of Nhe and lacking the hbl gene. Cytotoxicity of HBL-positive strains ranged from 30 to 1700 but could not be attributed to HBL alone as all strains positive for HBL also produced Nhe. Therefore, comparison of the amount of L2 detectable in the culture supernatants with the cytotoxicity titres resulted in a very poor correlation (Fig. 2). On the other hand, a good correlation between the concentration of NheB and the toxic activity of culture filtrates of B. cereus on Vero cells was obtained (Fig. 2). This interesting finding clearly indicates that cytotoxicity on Vero cells is dominated by Nhe rather than HBL and shows that B. cereus strains producing two enterotoxin complexes are not more cytotoxic than sole Nhe producers. The reasons for this striking result need to be clarified, particularly with regard to earlier studies which suggest comparable toxic activities of the both enterotoxins, HBL and Nhe (Lund & Granum, 1997).
To get further insight into the respective role of the two complexes in enterotoxicity we compared the amount of NheB produced by food poisoning and food/environmental strains. Overall, the median value (5.10 μg mL−1, Fig. 3a) of all food poisoning strains was c 1 μg mL−1 higher than that of the food/environmental isolates (4.10 μg mL−1, Fig. 3b). Less pronounced results, 1.10 vs. 0.85 μg mL−1 (Figs 3c and d), were obtained for the HBL–L2 production of hbl-positive strains. Altogether the results demonstrate that food poisoning strains tend to produce higher amounts of enterotoxins, particularly Nhe. While interpreting these data it must be kept in mind that ‘food/environmental strains’ probably also include strains, which could cause food poisoning. On the other hand, the ‘food poisoning strains’ also could include isolates which lost some of their toxic potential and it cannot fully ruled out, that some strains may have been wrongly associated with illness. We also know far too little about expression of the toxins when B. cereus is growing in the gut. With the protein amounts found in this study for NheB and for HBL L2, both enterotoxin compounds are expressed at a high level, a finding which particularly for NheB is consistent with data reported by Gohar et al. (2002)).
On the other hand, the amount of enterotoxins produced by individual B. cereus strains showed an extremely broad variation. Consequently, the question arises if the detection of enterotoxin genes in B. cereus by PCR or an absence/presence test for toxin production is an appropriate method to estimate the virulence of B. cereus isolates. The data obtained in this study rather show that the quantitative determination of the enterotoxins allow a better assessment of the toxic potential. Particularly, the newly developed sandwich assays (Fig. 1) for L2 and NheB represent valuable tools enabling accurate measurements of the toxin productivity of B. cereus isolates.
In conclusion, this is the first report on qualitative and quantitative aspects of enterotoxin production by B. cereus strains. The most striking result was that the toxic activity of strains producing both enterotoxin complexes, HBL and Nhe, was not significantly different from that of sole Nhe producers. The cytotoxicity of the B. cereus strains was dominated by the amount of secreted Nhe. Even though cytotoxicity on Vero cells only partly reflects the mechanisms involved in the aetiology of diarrhoea, the results obtained suggest that Nhe is the most important toxin in food poisoning. To verify this statement further research on the pathogenic role of the two enterotoxin complexes produced by B. cereus is required.
This work has been carried out with the financial support from the Fifth European Community Framework Programme, under the Quality of Life and Management of Living Resources specific programme, Contract QLK1-2001-00854, Preventing Bacillus cereus foodborne poisoning in Europe – Detecting hazardous strains, tracing contamination routes and proposing criteria for foods. The authors thank Bouziane Moumen for his participation to the study.