Small Bacillus cereus ATCC 14579 subpopulations are responsible for cytotoxin K production

Authors

  • S. Ceuppens,

    1. Laboratory of Food Microbiology and Food Preservation (LFMFP), Faculty of Bioscience Engineering, Ghent University, Gent, Belgium
    2. Laboratory of Microbial Ecology and Technology (LabMET), Faculty of Bioscience Engineering, Ghent University, Gent, Belgium
    Search for more papers by this author
  • S. Timmery,

    1. Laboratory of Food and Environmental Microbiology, Faculty of Bioscience Engineering, Université catholique de Louvain, Louvain-la-Neuve, Belgium
    Search for more papers by this author
  • J. Mahillon,

    1. Laboratory of Food and Environmental Microbiology, Faculty of Bioscience Engineering, Université catholique de Louvain, Louvain-la-Neuve, Belgium
    Search for more papers by this author
  • M. Uyttendaele,

    1. Laboratory of Food Microbiology and Food Preservation (LFMFP), Faculty of Bioscience Engineering, Ghent University, Gent, Belgium
    Search for more papers by this author
  • N. Boon

    Corresponding author
    1. Laboratory of Microbial Ecology and Technology (LabMET), Faculty of Bioscience Engineering, Ghent University, Gent, Belgium
    • Laboratory of Food Microbiology and Food Preservation (LFMFP), Faculty of Bioscience Engineering, Ghent University, Gent, Belgium
    Search for more papers by this author

Correspondence

Nico Boon, Laboratory of Microbial Ecology and Technology (LabMET), Faculty of Bioscience Engineering, Ghent University, Coupure Links 653, B-9000 Gent, Belgium. E-mail: nico.boon@ugent.be

Abstract

Aims

Bacillus cereus diarrhoeal food poisoning can be caused by several potential enterotoxins, including the nonhaemolytic enterotoxin (Nhe), haemolysin BL (Hbl) and cytotoxin K (CytK). To get more insights into the CytK expression, a fluorescent reporter strain was created for CytK expression.

Methods

Bacillus cereus ATCC 14579 was used as the reporter strain that contained the cyan fluorescent protein (CFPopt) gene under control of the cytK promoter. Transcription of enterotoxin genes nheB, hblC and cytK was assessed by messenger RNA analysis (RT-qPCR), and their full expression was assessed by immunological protein detection in the case of Nhe and Hbl and fluorescence microscopy in the case of CytK, using the reporter gene CFPopt.

Results

Transcription of enterotoxins Nhe, Hbl and CytK showed similar kinetics with a peak during the late exponential growth phase. Toxin expression of the reporter strain was unaltered in comparison with the wild type. However, fluorescence, and thus CytK expression, only occurred in a small (1–2%) portion of the cell population.

Conclusions

These results suggest that a small subpopulation of B. cereus ATCC 14579 is responsible for CytK production in a homogeneous monoculture.

Significance and Impact of the Study

Future research is warranted to determine whether genetically homogeneous B. cereus populations utilize differential gene expression for other toxins and virulence genes than CytK and whether this also applies to other B. cereus strains. If so, differential expression of toxin genes could be used by these bacteria to increase the fitness and survival chances of their population by diversification and specialization into different subpopulations.

Introduction

In 1998, 44 elderly patients in a French nursing home suffered from severe diarrhoeal food poisoning caused by Bacillus cereus NVH 391/98 (alias B. cereus ssp. cytotoxicus) in vegetable purée (Lund et al. 2000). This lethal food poisoning outbreak (6·8% mortality) led to the discovery of a new enterotoxin: the 34-kDa single-component β-channel-forming cytotoxin K (CytK). Further research showed that CytK exists in two distinct variants, of which the firstly discovered CytK-1 displayed stronger haemolytic and cytotoxic activity than the second variant CytK-2 (Fagerlund et al. 2004). However, most recent research has shown that strains harbouring CytK-1 are actually another more rare species, namely B. cytotoxicus, distinct from B. cereus strains, which harbour the CytK-2 variants (Guinebretiere et al. 2006, 2012). Thus, it became apparent that CytK-1 and CytK-2 are actually orthologs of two different bacterial species rather than gene polymorphisms. Moreover, the cytotoxicity of B. cereus strains is not merely determined by the genomic presence of enterotoxin gene variants. It is the overall combination of enterotoxin genes and the actual expression level of each toxin gene that determines the food poisoning potential of a particular strain. For instance, B. cytotoxicus NVH 883/00 possesses the cytK toxin gene in its genome, but in contrast to the highly toxic strain B. cytotoxicus NVH 391/98, this strain is not cytotoxic due to the lack of CytK expression (Fagerlund et al. 2004).

The food-borne pathogen B. cereus can cause two types of food poisoning, namely the diarrhoeal and emetic syndrome (Ceuppens et al. 2011). Emetic food poisoning is characterized by vomiting, and it is caused by one peptide toxin, cereulide. In contrast, diarrhoeal food poisoning can be caused by multiple toxins, such as cytotoxin K (CytK) and the tripartite nonhaemolytic enterotoxin (Nhe) and haemolysin BL (Hbl). All major B. cereus potential enterotoxins, that is Nhe, Hbl and CytK, are regulated by the pleiotropic phospholipase C regulator (PlcR), which belongs to the PlcR/PapR quorum-sensing system (Slamti and Lereclus 2002; Gohar et al. 2008). As a consequence, high bacterial numbers (>6 log CFU ml−1) are the prerequisite, but no guarantee for enterotoxin production, as other bacterial and environmental signals further regulate the toxin expression, for example, growth rate, oxygen, pH and carbohydrates (Ceuppens et al. 2011). The regulation of toxin gene expression is very complex and remains incompletely understood, but it remains of interest in the quest to discriminate between pathogenic and harmless B. cereus strains and to identify toxin production stimuli.

The expression and regulation of enterotoxin genes encoding Nhe, Hbl and CytK were investigated in numerous studies on the population level by mRNA analysis with quantitative reverse-transcription real-time PCR (RT-qPCR) or microarrays, and by studies with mutant and reporter strains. Commercial immunological protein detection methods are also available for enterotoxin Nhe and Hbl components, but not for CytK. As a consequence of this lack of easily implemented toxin detection test and the rather recent discovery of CytK, this toxin is scarcely studied in comparison with the other enterotoxins Nhe and Hbl. The current knowledge regarding specific regulation of CytK production is insufficient to estimate the role and importance of CytK in B. cereus diarrhoeal food poisoning syndromes.

In this study, CytK expression by B. cereus was investigated by developing a fluorescent B. cereus reporter strain. The type strain B. cereus ATCC 14579 was isolated in 1916 from the air in a cowshed in the USA. This strain was chosen as the reporter strain and also as the donor of the CytK promoter sequence, because its genome has been completely sequenced (GenBank AE016877, AE016878; Ivanova et al. 2003) and the expression of the Nhe, Hbl and CytK enterotoxins was established on the protein level (Gohar et al. 2002), which proved the functionality and activity of the nhe, hbl and cytK promoters in this strain.

Materials and methods

Construction of reporter strain Bacillus cereus (pHT304-CytK-CFPopt)

Unless stated otherwise, the B. cereus and E. coli strains were cultivated in Luria broth (LB, 10 g l−1 tryptone (Oxoid, Clintonpark ‘Keppekouter’, Erembodegem, Belgium), 5 g l−1 yeast extract (Oxoid), 5 g l−1 NaCl (Sigma-Aldrich, Bornem, Belgium) containing the appropriate antibiotics at 30 or 37°C with shaking at 120 rev min−1 (Table 1). PCR was conducted with GoTaq polymerase and GoTaq Flexi buffer (Promega Benelux, Leiden, the Netherlands), 200 μmol l−1 of each dNTP, 1 μmol l−1 of each primer and 1·5 μmol l−1 MgCl2 during 30 cycles of 30 s at 95°C, 30 s at 55°C and 60 s at 72°C (C1000 Thermal Cycler; Bio-Rad). Purification of PCR products or plasmid DNA was carried out with the GenElute PCR clean-up kit (Sigma-Aldrich) or with the QIAquick Gel Extraction Kit (Qiagen Benelux, Venlo, the Netherlands). Plasmids were extracted with the GenElute Plasmid Miniprep Kit (Sigma-Aldrich) or with the JetStar Plasmid Purification Midi Kit (Genomed, ImTec Diagnostics, Antwerpen, Belgium). T4 DNA ligase (Promega) was used for DNA ligation and shrimp alkaline phosphatase (SAP; Fermentas GmbH, St Leon-Rot, Germany) for DNA dephosphorylation. Restriction of plasmid DNA was performed overnight at 37°C with PstI, SalI and HindIII (Fermentas) in the recommended buffers (Fermentas). Cloning of PCR products in pCR®4-TOPO® vectors was performed with the TOPO TA® Cloning Kit (Invitrogen, Life Technologies Europe, Gent, Belgium) according to the manufacturer's instructions. Electroporation was performed in 2-mm cuvettes with the GenePulserXcell (Bio-Rad, Eke, Belgium) at 2500 V, 25 μF and 200 Ω for E. coli and at 1400 V, 25 μF and 400 Ω for B. cereus. Inserted constructs were amplified by PCR with the M13 or pHT primers on the plasmids TOPO or pHT304-18Z, respectively (Table 2), and sequenced (GATC Biotech, Germany).

Table 1. Transformant E. coli and Bacillus cereus strains constructed in this study
Host strainPlasmidConstruct and plasmidBCCM-LMBP no.
  1. BCCM-LMBP, Belgian Coordinated Collections of Micro-organisms – Laboratory of Molecular Biology of the Ghent University, Ghent, Belgium.

E. coli TOP 10 (Invitrogen)pCR4-TOPO-cytKCytotoxin K (cytK) promoter of B. cereus ATCC 14579 with the start codon (ATG) and 27 bp of the upstream sequence (176 bp) in pCR®4-TOPO® (Invitrogen)LMBP 7638
E. coli TOP 10 (Invitrogen)pCR4-TOPO-CFPoptCyan Fluorescent Protein (CFPopt, 20) gene without the first two codons and with 83 bp of the downstream sequence (797 bp) in pCR®4-TOPO® (Invitrogen)LMBP 7639
E. coli TOP 10 (Invitrogen)pCR4-TOPO-cytK-CFPoptFusion of cytK promoter region before the CFPopt gene (973 bp) in pCR®4-TOPO® (Invitrogen)LMBP 7640
E. coli ER (Invitrogen)pHT304-18Z-cytK-CFPoptThe cytK promoter with the CFPopt gene in plasmid pHT304-18Z (Agaisse and Lereclus 1994)LMBP 7641
B. cereus ATCC 14579pHT304-18Z-cytK-CFPoptThe cytK promoter with the CFPopt gene in plasmid pHT304-18Z (Agaisse and Lereclus 1994)LMBP 7642
B. cereus ATCC 14579pSW4-CFPopt (20)The PA promoter with the CFPopt gene in plasmid pSW4 (Sastalla et al. 2009)LMBP 7643
Table 2. Primers used in this study for cloning and gene transcription analysis with reverse-transcriptase real-time PCR (RT-qPCR)
NameSequence (5′ to 3′)aPurposeReference
  1. a

    Restriction sites are indicated in the primer sequences in bold and underlined, and restriction enzyme is indicated between parentheses.

CytK-prom_FAAGCTTATGGACATTTTCGAGTCTTAAG (HindIII)Amplification of the cytK promoter regionThis study
CytK-prom_RGTCGACCATACAATCACTTCCTTTTATC (SalI)
CFP_FGTCGACTCAAAAGGAGAAGAATTATTTACAGG (SalI)Amplification of the CFPopt gene regionThis study
CFP_RAAGCTTTAGTTATTGCTCAGCGGTG (HindIII)
M13_FPTGTAAAACGACGGCCAGTSequencing of the insert in the pCR®4-TOPO® plasmidInvitrogen
M13_RPCAGGAAACAGCTATGACC
pHT_OVGCGTAATCTTACGTCAGTAACTSequencing of the insert in the pHT304-18Z plasmidAgaisse and Lereclus (1994)
pHT_F0GGGCCTCTTCGCTATTACGC
CS-F GCGCTGATAAACAGATTGCCGTQuantification of cytK cDNAThis study
CS-R GAAGCTTTAACAGAGCCACCA
SG-F3GCACTTATGGCAGTATTTGCAGCQuantification of nheB cDNAYang et al. (2007)
SG-R3GCATCTTTTAAGCCTTCTGGTC
SH-FGGCTGCTATTTCACCTTGTTTGCQuantification of hblC cDNAThis study
SH-RAGTGCAAGAAGCAACAGAGC
SFP-FAGAAGGAGATGCAACATACGGAQuantification of CFPopt cDNAThis study
SFP-RACGTATCCTTCAGGCATTGCT
gatB_Yqey-FAGCTGGTCGTGAAGACCTTGQuantification of gatB_Yqey cDNAReiter et al. (2011)
gatB_Yqey-RCGGCATAACAGCAGTCATCA
udp-FACTAGAGAAACTTGGAAATGATCGQuantification of udp cDNAReiter et al. (2011)
udp-RGACGCTTAATTGCACGGAAC
rpsU-FAAGATCGGTTTCTAAAACTGGTACAQuantification of rpsU cDNAReiter et al. (2011)
rpsU-RTTTCTTGCCGCTTCAGATTT
ssu-FGCAAGGCTGAAACTCAAAGGQuantification of 16S rRNA cDNASuzuki and Yamasato (1994)
ssu-RTGTAGCCCAGGTCATAAGGG

Gene expression analysis

Growth conditions

Bacillus cereus (pHT304-CytK-CFPopt) and B. cereus ATCC 14579 were first cultivated from the −80°C culture stock in tryptone soya broth (TSB) for 24 h at 30°C. A subculture of 200 μl was inoculated in 100 ml brain–heart infusion broth (BHI; Oxoid) supplemented with 10 g l−1 glucose (Sigma) (BHIG) in 250-ml Erlenmeyers at 37°C with vigorous shaking [Yellow Line OS10 shaker (IKA) at 200 rev min−1]. The media for cultivation of B. cereus (pHT304-CytK-CFPopt) were always supplemented with 25 mg l−1 erythromycin as a selective pressure for conservation of the pHT304-CytK-CFPopt plasmid.

Sampling

Samples were taken every 2 h for several analyses. Firstly, the B. cereus concentration was determined using plate count enumeration on tryptone soya agar (TSA; Oxoid). Secondly, fluorescence of the culture medium (100 μl) was measured in a 96-well plate (flat bottom transparent; Greiner Bio-one, Wemmel, Belgium) with GENiosTM Plus (Tecan Benelux BVBA, Mechelen, Belgium) using filters for excitation at 430 nm and emission at 535 nm. Thirdly, culture samples of 1 mL were filter-sterilized (0·2-μm syringe filters; Whatman, Voor't Labo, Eeklo, Belgium) followed by detection of enterotoxins Nhe and Hbl production with the Duopath® Cereus Enterotoxins kit (Merck, VWR, Leuven, Belgium) and haemolytic activity on BBL™ trypticase soy agar with 5% sheep blood (Becton Dickinson and Company, Erembodegem, Belgium). Fourthly, samples were treated with RNAprotect Bacteria Reagent (Qiagen) according to the manufacturer's instructions and stored at −80°C for mRNA analysis (see RT-qPCR below).

RT-qPCR

Total RNA was extracted according to the instructions of the RNeasy Mini Kit (Qiagen) for the enzymatic lysis protocol (TE buffer with 30 mg ml−1 lysozyme from chicken egg white, Sigma-Aldrich) and elution in 30 μl of RNase-free water (Applied Biosystems). Remnants of genomic DNA were removed in 10 μl of samples by Dnase I treatment (Fermentas) in the provided MgCl2 buffer at 0·2 U μl−1 for 30 min at 37°C, followed by inactivation of the DNase at 65°C for 10 min. RNA samples of 3 μl were supplemented with 7·5 μl of RNase-free water and 1 μl of random hexamers (50 μmol l−1) and incubated at 95°C for 2 min followed by 2 min on ice. Then each sample was supplemented with 8·5 mastermix (Applied Biosystems) consisting of 4 μl MgCl2 (25 mmol l−1), 2 μL PCR buffer II (10×), 1 μl dNTP (20 mmol l−1), 1 μl RNase inhibitor (20 U μl−1) and 0·5 μl MultiScribeTM reverse transcriptase (50 U μl−1) and subjected to following temperature protocol: 10 min at 22°C, 15 min at 42°C, 5 min at 99°C and 5 min at 5°C. The resulting cDNA (2 μl samples) was quantified by real-time PCR in 20-μl reaction volumes containing 0·2 μmol l−1 of each primer (Table 2) and 10 μl Power SYBR® Green PCR Master Mix (Applied Biosystems). Each gene was amplified with separate primer pairs in separate wells with SYBR Green detection of the amplification products. The temperature protocol was identical for all primer pairs, that is, 2 min at 50°C, 10 min at 95°C, 45 cycles of 10 s at 95°C, 30 s at 60°C, 30 s at 72°C and finally melting curve analysis from 95 to 55 to 95°C. The mRNA copy number of each gene was calculated with the genomic DNA standard curves and expressed as the percentage of the average transcription of the gatB_Yqey, udp and rpsU household genes (Reiter et al. 2011) to normalize the results and express them independently of the cell number in the samples.

Microscopy

Samples of the duplicate cultures of B. cereus (pHT304-CytK-CFPopt) in BHIG were also visually inspected under the microscope (Axioskop 2 plus; Zeiss, Carl Zeiss NV-SA, Zaventem, Belgium). Using a Bürker counting chamber (Marienfeld Superior), the total number of cells was enumerated in 20 squares with normal light microscopy at 400× magnification (Plan-Neofluar 40×; Zeiss). Thereafter, the number of fluorescent cells was counted in the same 20 squares with fluorescence microscopy (HBO 100 Microscope Illuminating System; Zeiss) with the FS 38 filter (Zeiss) with excitation at 470 ± 40 nm and emission at 525 ± 50 nm, visualizing the CFPopt-expressing cells as green fluorescent. Microscopy images were taken with the Oscar F320C camera (Allied Vision Technologies GmbH, Stadtroda, Germany) and the AVT SmartView 1.8 software.

Results

Construction of a CytK reporter strain Bacillus cereus

To investigate the variability in the cytK promoter sequence in the B. cereus group, NCBI-Blast and ClustalW multiple alignment were performed and showed 90% sequence identity among the 16 retrieved promoter sequences (data not shown). The cytK promoter region including the start codon (M) and 27 bp of the upstream sequence (176 bp) was amplified from B. cereus ATCC 14579 genomic DNA with CytK_prom primers (Table 2), cloned in the pCR®4-TOPO® vector and electroporated in E. coli TOP10, resulting in the E. coli TOPO-CytK strain (Table 1). Similarly, the cyan fluorescent protein (CFPopt) gene (Sastalla et al. 2009) without the first two codons (MV) and with 83 bp of the downstream sequence on the pSW4-CFPopt plasmid was cloned with the CFP primers (Tables 1 and 2). The cytK promoter sequence was excised by PstI and SalI restriction and ligated before the CFPopt gene in the TOPO vector and in the pHT304-18Z plasmid (Agaisse and Lereclus 1994) after HindIII restriction. As a consequence of the restriction of SalI site, two amino acids (VD) were inserted after the start codon (M). As the first two codons (MV) of the CFPopt gene were omitted, this resulted in the insertion of D as the third amino acid in comparison with the original CFPopt sequence (GenBank accession FJ169509).

Toxin expression by Bacillus cereus ATCC 14579 and B. cereus pHT304-CytK-CFPopt during growth in Brain–Heart Infusion with 1% glucose (BHIG) at 37°C

Both the wild-type B. cereus ATCC 14579 and the reporter B. cereus (pHT304-CytK-CFPopt) produced enterotoxins in BHIG at 37°C after 8 h, during their late exponential growth phase, at population densities of 8·6 log CFU ml−1 and 7·8 log CFU ml−1, respectively (Fig. 1a,c). From this time on, the culture supernatant of both strains showed haemolytic activity and the presence of enterotoxin components NheB and HblC. Moreover, the transcription of the enterotoxin genes nheB, hblC and cytK also peaked at 8 h and stabilized thereafter at a slightly lower level for hblC and cytK, while that of nheB continued to decrease. The CFPopt gene transcription was similar to that of cytK, with the exception of higher levels at the early exponential and late stationary growth phase. Although the CFPopt gene is also under control of the cytK promoter, it is present on the pHT304-18Z plasmid with approximately 10 copies per cell (determined using qPCR on tenfold dilutions of DNA extracted from a stationary culture in BHIG, data not shown). This could explain the slightly higher and earlier transcription during the early exponential phase due to the decreased detection limit. Moreover, the plasmid copy number varies in function of the growth phase, independent of the chromosomal DNA (Turgeon et al. 2008). In contrast to cytK, the CFPopt gene transcription was again increased to its peak level at 24 h.

Figure 1.

(a) Growth of reporter strain B. cereus (pHT304-CytK-CFPopt) (+) and wild-type strain B. cereus ATCC 14579 (×) in brain heart infusion with 1% glucose (BHIG) at 37°C was determined by plating. Production of enterotoxins Nhe and Hbl (Duopath kit; Merck) and haemolytic activity (sheep blood agar) occurred simultaneously and at the same time for both strains (indicated by grey bars). Gene transcription of reporter strain B. cereus (pHT304-CytK-CFPopt) (+) and wild-type strain B. cereus ATCC 14579 (×) was also determined for the enterotoxin gene cytK (b), hblC (c) and nheB (d). Transcription of the CFPopt gene (○) in the reporter strain was also presented in (b). Gene transcription levels were calculated as the percentage of the average transcription of the gatB_Yqey, udp and rpsU household genes. The average values and standard deviation of triplicate experiments with different inocula are presented.

As expected, the CFPopt proteins remained inside the cells. No secretion of CFPopt was observed in the BHIG medium, as the fluorescence of the BHIG medium of B. cereus (pHT304-CytK-CFPopt) was similar to that of the original B. cereus ATCC 14579 throughout the 24-h culture period (results not shown).

CFPopt expression by Bacillus cereus (pHT304-CytK-CFPopt) assessed by fluorescence microscopy

Light and fluorescence microscopy images were made from 24-h cultures in BHIG at 37°C of the B. cereus (pHT304-CytK-CFPopt) transformants, the wild-type strain B. cereus ATCC 14579 and the positive control strain for fluorescence B. cereus (pSW4-CFPopt) (Fig. 2). The wild-type strain never showed fluorescent cells, and the positive control strain was always visible as a homogeneously fluorescent cell population. Surprisingly, the transformant culture contained only a small proportion of fluorescent cells. The bacteria were always cultured under vigorous shaking because of oxygen requirement for the CFPopt maturation and fluorescence. Static cultures of the positive control strain showed incomplete expression of fluorescence when grown under partially anaerobic culture conditions and no fluorescence in case of the reporter strain (results not shown).

Figure 2.

Light microscopy (a, c, e, g) and fluorescence microscopy (b, d, f, h) images of the Bacillus cereus (pHT304-CytK-CFPopt) strain (a, b, c, d), the positive control strain B. cereus (pSW4-CFPopt) (e, f) and the negative control strain B. cereus ATCC 14579 (g, h) magnified 1000× (a, b) and 400× (c, d, e, f, g, h), cultivated in brain–heart infusion broth supplemented with 1% glucose (BHIG) medium at 37 °C for 24 h.

The number of fluorescent cells was monitored microscopically during the growth of B. cereus (pHT304-CytK-CFPopt) in BHIG at 37°C for 24 h (Fig. 3). After 6 h, when the culture was in the (late) exponential growth phase at 8·8 log cells ml−1, fluorescent cells were observed for the first time. Thereafter, the percentage of fluorescent cells in the population increased to 2% at 10 h and remained between 1 and 2% during the stationary growth phase. The number of fluorescent cells in the B. cereus (pHT304-CytK-CFPopt) population could not be enhanced by repeated subculturing: after 10 passages, fluorescent cells were still observed with a frequency of approx. 1% (results not shown).

Figure 3.

(a) The growth and Nhe and Hbl toxin production of reporter Bacillus cereus (pHT304-CytK-CFPopt) in BHIG at 37°C were assessed by the total cell count with light microscopy (+) and the Duopath® kit (positive results are indicated with grey bars), respectively. (b) In parallel, the CFPopt expression of reporter B. cereus (pHT304-CytK-CFPopt) was determined by the fluorescent cell count with fluorescence microscopy (×) and expressed as the percentage of the total cell population (white bars). The average values of duplicate experiments with different inocula are presented, and error bars indicate the standard deviations.

Discussion

A fluorescent B. cereus reporter strain for CytK expression was constructed in this study. The remarkable result was that only a small proportion of the cells showed fluorescence. To determine whether toxin production was indeed originating from a minority of cells, or whether there was a problem with the reporter strain itself, several potential issues were systematically analysed. First of all, plasmid instability was excluded, as conservation of the plasmid with the reporter gene CFPopt under control of the cytK promoter in the transformant strain population was assured by antibiotic selection. Transcription of the cytK gene in the reporter strain was unaltered in comparison with that of the wild-type strain. It is therefore unlikely that the presence of multiple extra-chromosomal copies of the cytk promoter affected its functionality. The transcription of the reporter gene CFPopt was indeed simultaneous with that of cytK, indicating that the fluorescence expression is representative for CytK transcription. The translation of CFPopt mRNA is not expected to be problematic, as the codon usage in the CFPopt gene was optimized for Gram-positive bacteria and the positive control strain B. cereus (pSW4-CFPopt) showed consistent CFPopt expression under its strong promoter, which is very active during vegetative growth (Sastalla et al. 2009). However, the CFPopt expression was very low for B. cereus (pHT304-CytK-CFPopt) on the population level: approximately 1·5% of the cells were fluorescent. The nheB, hblC, cytK and CFPopt gene transcription peaked during the exponential growth phase when more than 7·5 log CFU ml−1 were present. This concurred with the detection of enterotoxin components NheA and HblC, haemolytic activity and the occurrence of fluorescent cells (0·2–2% of the population). These results suggest that CytK production originated from a small virulent subpopulation in a laboratory monoculture of B. cereus ATCC 14579. It would be interesting to examine whether the expression of the nhe, hbl, plcR and other PlcR-regulated genes is also restricted to the same small subpopulation of cells. Moreover, similar experiments should be conducted with different B. cereus strains to determine whether this phenomenon of differential toxin expression in a monoculture is strain specific or a general characteristic of B. cereus.

Bacterial cells in a homogeneous isogenic culture are known to exhibit phenotypic heterogeneity, as for the case of antibiotic-tolerant cells called ‘persisters’ in otherwise sensitive populations (Dhar and McKinney 2007). Moreover, E. coli showed a bimodal distribution of lactose operon expression levels, visualized by GFP as a reporter: the induced cells were apparent as a brightly fluorescing subpopulation (Ozbudak et al. 2004). The growth of B. subtilis in biofilms constitutes another example of differentiation into several distinct cell types, such as motile cells, competent cells, matrix-producing cells, exoenzyme-producing cells, toxin-producing cells and spores (Lopez et al. 2009). Recently, B. cereus monocultures were also shown to constitute physiologically heterogeneous and dynamic populations, which was observed as variability in lag time, cell membrane properties and reductase activity for individual bacterial cells of the same culture (Want et al. 2011). These examples all show the existence of nongenetic and thus noninherited variation in gene expression among individual cells in an axenic bacterial culture, as it was observed in the present study for cytotoxin K expression by B. cereus.

Various mechanisms can cause variable gene expression in bacterial monocultures. Firstly, growth itself is a source of intercellular variation, as different cells in the same culture can be in different cell cycle stages. Secondly, epigenetic regulation confers metastable inherited changes in gene expression without changes in the DNA sequence, and quite interestingly, the virulence of several pathogenic bacteria such as Salmonella spp., Haemophilus spp. and Yersinia pseudotuberculosis is associated with epigenetic regulation (Casadesus and Low 2006). Thirdly, stochastic fluctuations in gene expression can also generate different cell phenotypes in a clonal population and can cause single-cell behaviour to deviate from the population average (Avery 2006). Whatever the mechanism, differentiation of cells into multiple coexisting phenotypes with different functions benefits all genetically identical bacteria by increasing the diversity, fitness and survival chances of their population as a whole of in a changeable environment.

The cytk promoter of B. cereus ATCC 14579 enables gene transcription and subsequent mRNA translation, resulting in CytK protein production (Gohar et al. 2002). Although the cytK promoter sequence appears rather conserved, B. cereus strains show variation in their promoter sequences, which may lead to variability in regulatory sequences. Moreover, the upstream region and the genetic background differ among strains. This could differentially affect the cytK gene expression, leading to differences in CytK production among (groups of) strains. Furthermore, the question arises whether similarly low percentages of enterotoxin-producing B. cereus occur for the other enterotoxin genes, under other culture conditions and in mixed microbial communities in the soil, food products and the host's intestinal tract. Future research should thus further explore the frequency and the quantity of B. cereus enterotoxin expression on the individual cell level and on the whole population level to answer some of these intriguing questions.

Acknowledgements

The authors declare that they have no conflict of interest. This work was supported by the Special Research Funds of Ghent University as a part of the project ‘Growth kinetics, gene expression and toxin production by Bacillus cereus in the small intestine' B/09036/02 fund IV1 31/10/2008 – 31/10/2012, by the committee for scientific research (CWO UGent), by the Federal Public Service (FOD) Health, Food Chain Safety and Environment project RT09/2 BACEREUS and the National Fund for Scientific Research (FNRS) grant to Sophie Timmery. The fluorescence measurements were conducted with the GENios™ Plus reader (Tecan) of the FWO project G.0412.07N.

Ancillary