Correspondence: Monika Ehling-Schulz, Microbial Ecology Group, Department of Biosciences, WZW, Technische Universität München, D-85354 Freising, Germany. Tel.: +49 8161 713851; fax: +49 8161 714492; e-mail: firstname.lastname@example.org
Very different toxins are responsible for the two types of gastrointestinal diseases caused by Bacillus cereus: the diarrhoeal syndrome is linked to nonhemolytic enterotoxin NHE, hemolytic enterotoxin HBL, and cytotoxin K, whereas emesis is caused by the action of the depsipeptide toxin cereulide. The recently identified cereulide synthetase genes permitted development of a molecular assay that targets all toxins known to be involved in food poisoning in a single reaction, using only four different sets of primers. The enterotoxin genes of 49 strains, belonging to different phylogenetic branches of the B. cereus group, were partially sequenced to encompass the molecular diversity of these genes. The sequence alignments illustrated the high molecular polymorphism of B. cereus enterotoxin genes, which is necessary to consider when establishing PCR systems. Primers directed towards the enterotoxin complex genes were located in different CDSs of the corresponding operons to target two toxin genes with one single set of primers. The specificity of the assay was assessed using a panel of B. cereus strains with known toxin profiles and was successfully applied to characterize strains from food and clinical diagnostic labs as well as for the toxin gene profiling of B. cereus isolated from silo tank populations.
Toxin producing Bacillus cereus plays an important role as the causative agent of two types of food poisoning: diarrhea and emesis. The emetic syndrome is mainly characterized by vomiting a few hours after ingestion of the contaminated food. In the diarrhoeal syndrome, symptoms appear 8–16 h after ingestion, and include abdominal pain and diarrhea. In general, both types of food borne illness are relatively mild and self-limiting. Nevertheless, more severe cases have occasionally been reported involving hospitalization or even deaths (Mahler et al., 1997; Lund et al., 2000; Dierick et al., 2005).
The two types of gastrointestinal disease caused by B. cereus are associated with very different types of toxins (Granum, 2001; Ehling-Schulz et al., 2004a). The emetic syndrome is caused by a single heat stable peptide toxin called cereulide (Agata et al., 1995), which is preformed in food. Cereulide has been shown to be toxic to mitochondria by acting as a potassium ionophore and has been reported to inhibit human natural killer cells (Paananen et al., 2002). Recently, the peptide synthetase genes responsible for the nonribosomal production of cereulide (ces genes) have been identified and characterized, and the first molecular assays for the detection of emetic toxin producers have been described (Ehling-Schulz et al., 2004b, 2005b, 2006). The diarrhoeal poisoning is caused by heat-labile enterotoxins produced during vegetative growth of B. cereus in the small intestine (Granum, 1994). At present, three different enterotoxins involved in food poisoning outbreaks are known: two protein complexes, hemolysin BL (HBL; Beecher et al., 1995) and nonhemolytic enterotoxin (NHE; Granum et al., 1999), and the single protein cytotoxin CytK (Lund et al., 2000). HBL is a three-component hemolysin that consists of two lytic components (L2 and L1, encoded by hblD and hblC) and a binding protein B (encoded by hblA). NHE is also a three-component, but nonhemolytic, toxin that is encoded by three genes nheA, nheB and nheC. Both toxin complexes are organized in operons and the corresponding genes of the enterotoxin complex NHE have been shown to be transcribed together (Lindback et al., 2004). Immunological assays are commercially available for the detection of NHE and HBL and monoclonal antibodies targeting these enterotoxin complexes have been generated (Dietrich et al., 1999, 2005) but no such tools are yet available for CytK or cereulide. Molecular assays for the detection of the different enterotoxin genes revealed a high degree of molecular diversity among the enterotoxin genes, which could lead to false negative results in PCR (Mäntynen & Lindström, 1998; Prüßet al., 1999; Hansen & Hendriksen, 2001; Guinebretiere et al., 2002).
The aim of this study was to develop a simple multiplex PCR system that allows the detection of all B. cereus toxins so far known to be involved in food poisoning, in a single reaction. Such an assay could improve diagnosis of gastrointestinal diseases caused by B. cereus and facilitate toxin gene profiling in population studies. Special emphasis was placed on the design of primers for enterotoxin genes since these toxins show great diversity at a molecular level while the cereulide synthetase genes are highly conserved in emetic B. cereus (Guinebretiere et al., 2002; Ehling-Schulz et al., 2005b).
Materials and methods
Forty-nine Bacillus cereus strains from diverse origins, belonging to different phylogenetic branches of the B. cereus group (Guinebretiere et al., 2002, 2003) were used to assess the molecular diversity of enterotoxin genes (Fig. 1). Clinical strains and isolates from food with well-characterized toxin profiles were chosen for the development of the multiplex PCR assay. The reference set of strains comprised strains that carry the genetic loci of all three enterotoxins, namely LMG 17615 (F289/78), F3371/93, and 98HMPL63, and one strain, RIVM-BC91, that encode only the two enterotoxin complexes HBL and NHE. In addition, the original CytK strain NVH0391-98, which does not possess the genes encoding the enterotoxin complexes, and, as reference for the emetic toxin, the cereulide producing strain F4810/72 were added to the reference set. Details of strain characteristics are provided elsewhere (Ehling-Schulz et al., 2005a).
Isolation of DNA
Total DNA of bacteria was isolated using either the AquaPure Genomic DNA Isolation Kit (Biorad, Germany) or the DNeasy Tissue kit (Qiagen, VWR International AB, Sweden) according to manufacturer's instructions. In addition, DNA was extracted using a simple boiling method. In brief, cells from one colony were suspended in sterile water, heated at 95°C for 3 min and then placed on ice. After centrifugation the supernatant was used as template for PCR. Although the latter technique worked well for strains from culture collections, DNA prepared by this method from B. cereus isolated from food and clinical environments was not suitable for multiplex PCR.
PCR amplification of enterotoxin genes and sequence analysis
Fragments of the enterotoxin genes nheA, nheB, hblD, and hblA were amplified from diverse B. cereus strains and sequenced as described previously (Guinebretiere et al., 2001, 2002). Primers used for amplification and sequencing are provided in Table 1. The resulting sequences and sequence data from B. cereus enterotoxin genes retrieved from databases were aligned using the software packages Clustal W (Thompson et al., 1997) and the Multalign version 5.4.1 (Corpet, 1988). Positions of the partial sequences were determined with reference to GenBank nucleic acid sequence data; accession number AJ277962 for cytK, AJ007794 for the hbl genes, and Y19005 for the nhe genes (Fig. 1).
Table 1. Characteristics of oligonucleotide primers used in this study
Basic 18–20 bp oligonucleotide primers were designed using reference enterotoxin gene data (mentioned in Table 1) available from GenBank/EMBL databases and the Primer Designer software (Becker et al., 1995). In a second step, sequence polymorphisms, previously identified by sequence alignments, were taken into account by substituting the bases at variable positions by inosine (Fig. 1). The resulting primers are presented in Table 1. Before setting up the final multiplex PCR assay, the primers were tested in singleplex PCR and in duplex PCRs using different combinations of primer pairs.
The PCR conditions were optimized for primer and MgCl2 concentrations. The final reaction mixture (50 μL) contained 0.2 mM of each dNTP, 3 mM MgCl2, 0.2 μM of the oligonucleotide primers CesF1 and CesR2, 1 μM of HD2F and HA4R; 0.3 μM of NA2F and NB1R; 0.4 μM of CKF2 and CKR5; 1 U of ThermoStart Taq DNA polymerase (ABgene, Epsom, UK), 5 μL 10 × polymerase buffer and 1 μL template DNA. The PCR protocol started with a denaturation step for 15 min at 95°C, followed by 30 cycles of 30 s at 95°C, 30 s at 49°C, and 1 min at 72°C each, and ended with a final elongation step at 72°C for 2 min. Sequences of all primers used are provided in Table 1.
Evaluation of the multiplex PCR assay
A panel of B. cereus group strains was compiled to evaluate the specificity of the multiplex PCR. The test panel included clinical strains and food strains with known toxin profiles (Ehling-Schulz et al., 2005a). In addition, strains from other B. cereus group members: B. anthracis (ATCC6602, Cepanzo, CIP A2), B. thuringiensis (WS2734T, WS28024, WSBC28009), B. mycoides (WS2641T, WSBC10297, WSBC10293), B. weihenstephanensis (WSBC10204T, INRA I20, INRA 1) and strains from other Bacillus species and non-Bacillus species were added to the test panel to assess the specificity of the assay (see Table 2).
Table 2. Bacterial species used to test the specificity of PCR assay
Bacterial species (no. of species)
No. of strains tested
Including 40 clinical isolates and isolates from food remnants connected to food poisoning, 10 isolates from food and environment with known toxin profiles (Ehling-Schulz et al., 2005b).
Toxin gene profiling of B. cereus strains from diagnostic laboratories and environment
The evaluated multiplex PCR assay was used to type 60 clinical and food isolates provided by diagnostic laboratories (Technische Universität München, Freising; Landesanstalt für Verbraucherschutz, Halle; Technische Universität Dresden; private diagnostic labs; Institut f. Hygiene und Umwelt, Hamburg) and to determine the distribution and occurrence of toxin genes in B. cereus group populations from dairy silo tanks (Table 3). Details of the study on B. cereus populations in Swedish silo tanks are provided elsewhere (Svensson et al., 2004).
Table 3. Toxin gene profiles of Bacillus cereus isolates obtained from clinical and food environments and selected isolates belonging to the B. cereus group
Development and evaluation of the multiplex PCR assay
Gene sequences from de novo sequenced enterotoxin genes of 49 B. cereus strains were aligned to enterotoxin sequences available from databases in order to design specific oligonucleotide primers. Our sequencing approach revealed high sequence polymorphisms of enterotoxin genes, which were not yet covered by the enterotoxin gene sequences available in databases (Fig. 1). These sequence polymorphisms might explain the false negative results observed in previously described PCR assays for the enterotoxins NHE and HBL (Mäntynen & Lindström, 1998; Hansen & Hendriksen, 2001; Guinebretiere et al., 2002). The observed point mutations were taken into account when oligonucleotide primers were designed and inosine was inserted at variable positions (Table 1). The designed primers allowed the amplification of enterotoxin genes from strains, which were previously only detected by Southern blot analysis (Guinebretiere et al., 2002). The forward and reverse primers were each located in two different genes of the corresponding operons, targeting two toxin genes in a single reaction. The forward primer, designed for the detection of the nhe complex, was located in nheA while the reverse primer was located in nheB, and primers for hbl were located in hblD and hblA, respectively (Table 1). Oligonucleotide primers for cytK were directed at highly conserved regions of the toxin gene so as to detect both forms of cytK (cytK-1 and cytk-2), recently described (Fagerlund et al., 2004), in a single reaction (Fig. 1). The primers for detection of emetic toxin producers were directed against a part of the cereulide synthetase essential for cereulide production. Disruption of this part of the ces genes by insertion mutagenesis led to a cereulide deficient phenotype (Ehling-Schulz et al., 2005b).
A set of reference strains, carrying different combination of toxin genes, was compiled and used for the development of the multiplex assay (see Materials and methods for strain details). After optimization of MgCl2 concentration and adjustment of primer concentrations, the PCR system was evaluated using a panel of 50 B. cereus strains with known toxin profiles. Closely related members of the B. cereus group, other Bacilli and known food pathogens were added to the test panel, to assess the specificity of the established assay (Table 2). The toxin gene profiles revealed by the novel multiplex assay were in accordance with the typing results of all strains obtained previously by singleplex PCR and/or Southern blotting (data not shown). Enterotoxin genes from strains, which were previously detected only by Southern blot analysis (Guinebretiere et al., 2002; Ehling-Schulz et al., 2005a) could now be identified by the novel multiplex assay (Fig. 2). None of the non-B. cereus group species isolates cross-reacted with the primer system (data not shown). In addition, the system turned out to be, in principle, suitable for the detection of enterotoxin genes in other members of the B. cereus group (Fig. 2, see also Table 3). The specificity and robustness of the assay was tested in two independent labs on two different cycler systems.
Toxin gene profiling of strains from diagnostic laboratories and from environment
A total of 60 B. cereus strains from different food and clinical diagnostic labs were typed by the established multiplex PCR system and the system was successfully applied to obtain toxin gene profiles of 80 B. cereus group strains, which had been collected during a population study from dairy silo tanks (Svensson et al., 2004). Specific toxin gene profiles turned out to be more common than others. Only five of the seven toxin gene profiles described previously (Ehling-Schulz et al., 2005a), were detected in our survey, which covered a total of 125 food isolates and 15 clinical isolates. The population of the silo tanks was dominated by strains with the toxin profile ‘C’ (nhe+, hbl+, cytK−, ces−) and ‘F’ (nhe+, hbl−, cytK−, ces−), whereas the prevalence of the toxin profile ‘C’ was much lower in isolates from diagnostic labs (Table 3). The most common toxin profile found among the latter isolates was toxin profile ‘F’. However, NHE could contribute substantially to the enterotoxic activity of a strain in vitro (Moravek et al., 2006). Further research will therefore be necessary to elucidate the exact role and importance of NHE in diarrhoeal food poisoning.
The incidence of emetic strains was generally low, and emetic strains carrying cytK seem to be quite rare, nevertheless emetic strains were found in all environments sampled in this study, in diverse foods, including baby foods and dry food products, as well as in clinical settings and in the silo tank environments. These findings are in accordance with recently published results on the occurrence of emetic strains in soil, dairy plants and farms (Yang et al., 2005; Altayar & Sutherland, 2006; Svensson et al., 2006). cytK was mainly found in combination with the two other enterotoxin genes, none of the tested isolates carried only cytK. From this study and our previous work (Guinebretiere et al., 2002; Ehling-Schulz et al., 2005a) one could assume that the occurrence of strains that possess only cytK is quite limited, nevertheless such strains could be highly toxic (Lund et al., 2000).
In conclusion, the assay we developed allows the detection of all genes, so far known, to be connected to gastrointestinal diseases caused by B. cereus in a one-step PCR. Improved primers, taking the discovered sequence polymorphism in enterotoxin genes into consideration, allowed the detection of enterotoxin genes previously missed by PCR. The described assay can facilitate diagnostics and could provide a powerful tool for toxin gene profiling of B. cereus in population studies. Such studies could provide new insights into the occurrence and distribution of toxin genes in different environments and could contribute to developing a better understanding of the epidemiology of toxic B. cereus. More detailed analysis will be necessary to examine if specific toxin gene pattern correlate with specific environments or genotypes as has been shown recently for emetic strains (Ehling-Schulz et al., 2005a).
Note added in proof
The Genbank accession numbers for the sequences of the internal enterotoxin gene fragments reported in this paper are AJ937140-AJ937208.
This work was supported by the European Commission (QLK1-CT-2001-00854). We thank Kathrin Buntin and Kerstin Ekelund for excellent technical assistance. We are grateful to Evi-Lang-Halter, Herbert Seiler, Herbert Schmidt, Dietrich Mäde, Jochen Bockemühl and Reinhold Gruss, who provided cultures from food and clinical environments.