Thomsen Bioscience A/S, Strømmen 6, DK-9400 Nørresundby, Denmark.
Occurrence and significance of Bacillus cereus and Bacillus thuringiensis in ready-to-eat food
Article first published online: 9 JAN 2006
FEMS Microbiology Letters
Volume 250, Issue 1, pages 129–136, September 2005
How to Cite
Rosenquist, H., Smidt, L., Andersen, S. R., Jensen, G. B. and Wilcks, A. (2005), Occurrence and significance of Bacillus cereus and Bacillus thuringiensis in ready-to-eat food. FEMS Microbiology Letters, 250: 129–136. doi: 10.1016/j.femsle.2005.06.054
Edited by W. Kneifel
- Issue published online: 9 JAN 2006
- Article first published online: 9 JAN 2006
- Received 1 April 2005, Revised 27 May 2005, Accepted 28 June 2005
- Bacillus cereus;
- Bacillus thuringiensis;
- Ready-to-eat food;
- Emetic toxin;
- Insecticidal toxins
Among 48,901 samples of ready-to-eat food products at the Danish retail market, 0.5% had counts of Bacillus cereus-like bacteria above 104 cfu g−1. The high counts were most frequently found in starchy, cooked products, but also in fresh cucumbers and tomatoes. Forty randomly selected strains had at least one gene or component involved in human diarrhoeal disease, while emetic toxin was related to only one B. cereus strain. A new observation was that 31 out of the 40 randomly selected B. cereus-like strains could be classified as Bacillus thuringiensis due to crystal production and/or content of cry genes. Thus, a large proportion of the B. cereus-like organisms present in food may belong to B. thuringiensis.
Bacillus cereus-like organisms can readily be isolated from various food products. The organisms are common in nature, and due to their resistant endospores they may survive different stresses during food production, e.g. drying and heat treatment. The group of B. cereus-like organisms comprises, besides B. cereus sensu stricto, the insect pathogen Bacillus thuringiensis, the human pathogen Bacillus anthracis, and the rhizoid Bacillus mycoides. Recently, Bacillus pseudomycoides and the psychrotolerant Bacillus weihenstephanensis have also been included into this group. B. cereus and B. thuringiensis are closely related and genomic studies have proposed that they should be merged into a single species. However, the name B. thuringiensis is retained for those strains that produce crystalline parasporal inclusions.
B. cereus is a well-known food borne pathogen causing two types of illness: the emetic and the diarrhoeal syndrome. The former is due to a small-molecular weight cyclic toxin, cereulide, while the diarrhoeal syndrome results from the production of enterotoxins. Cereulide is produced in the food, whereas the enterotoxins are believed to be produced in the intestine after ingestion of B. cereus-like organisms. The two most well-characterised enterotoxins are haemolysin BL (HBL) and the non-haemolytic enterotoxin (NHE). Both are three-component toxins requiring expression of all three components for virulence [8,9]. These two enterotoxins typically give relatively mild and short-lived diarrhoeal syndrome. A third enterotoxin, the single-component cytotoxin K (CytK) has so far only been reported to be involved in a single case of severe food poisoning outbreak including the death of three persons.
B. thuringiensis is used worldwide as a biological insecticide due to the production of crystal (cry) proteins (δ-endotoxins) with highly specific activity against certain insects. For example, strains of B. thuringiensis subspecies kurstaki producing the crystal protein Cry 1, are toxic for insect species within Lepidoptera, strains of B. thuringiensis subsp. israelensis producing Cry 11 are toxic for species within Diptera, and strains of B. thuringiensis subsp. tenebrionis producing Cry 3 are toxic for species within Coleoptera. More than 100 different δ-endotoxins have been characterised, the majority with insecticidal activity. Besides the Cry proteins other insecticidal proteins unrelated to Cry are also produced by some strains of B. thuringiensis, e.g. the cytolytic (Cyt) proteins. These toxins are thought to act in combination with Cry proteins to cause cytolysis in insects, however, Cyt proteins are in contrast to the δ-endotoxins not specific against insects, but are also haemolytic and cytolytic against many mammalian cell lines.
Insecticides based on different B. thuringiensis strains producing different δ-endotoxins are commercially available as powdered or granulated formulations containing a mixture of endospores and crystal proteins. The products are sprayed onto crops such as cabbage, wine grapes, tomatoes, cucumbers, and peppers. Approximately 50% of the bio-pesticides on the Danish market are B. thuringiensis-based, and B. thuringiensis-containing products constitute 90% of microbial bio-insecticides on the world market.
After the commercialization of B. thuringiensis-based insecticides, studies have shown that B. thuringiensis (including commercial strains used for insect control) like B. cereus produces enterotoxins responsible for human diarrhoea. B. thuringiensis has, however, only in one case been described to be implicated in food borne disease. Despite the pathogenic characteristics of B. thuringiensis, the presence of this bacterial species in food and food borne disease is not well described, probably because methods for detection and enumeration of B. cereus-like organisms in food and clinical settings do not distinguish between B. cereus and B. thuringiensis. Given the taxonomic similarity of B. cereus and B. thuringiensis and the introduction of high numbers of B. thuringiensis spores onto vegetables treated with B. thuringiensis-based insecticides, we speculate that a proportion of B. cereus-like organisms present in ready-to-eat food are B. thuringiensis and consequently, that some of the food borne diseases diagnosed as B. cereus infections are actually caused by B. thuringiensis.
The aims of the present study were to determine the occurrence of B. cereus-like organisms in ready-to-eat food at retail level in Denmark, to estimate the relative occurrence of B. thuringiensis in randomly selected food samples, to evaluate the potential of B. cereus and B. thuringiensis in food borne disease based on their production of emetic and diarrhoeal toxins, and finally to examine the antibiotic resistance of B. cereus-like organisms in food in order to evaluate if antibiotic resistance could be used for species differentiation.
2Materials and methods
Samples of ready-to-eat food products were randomly collected nationwide from local retail establishments (greengrocers, butchers, supermarkets, restaurants, etc.) by the Regional Veterinary and Food Authorities according to standard procedures as a part of the authorities' routine control of the microbiological quality of food. The random sampling independent of food conditions such as shelf-life and storage temperature was carried out to estimate the variety in the microbial quality of food ready for consumption and thereby the variety in human exposure from these products. After sampling the food samples were kept at 0–5 °C until analysis within 24 h.
2.2Enumeration and confirmation of Bacillus cereus-like strains
The number of B. cereus-like organisms was estimated according to a Nordic standard procedure using plate spreading of known amounts of sample on blood agar and/or on B. cereus selective agar (Oxoid, Basingstoke, UK). This method detects the mesophilic species B. cereus sensu stricto and B. thuringiensis, but also the psychrotolerant B. weihenstephanensis. In this study, we have not included methods to differentiate between B. weihenstephanensis and B. cereus sensu stricto. In the present study, we only mention B. cereus, although we are aware of the fact that the strains may belong both to B. cereus sensu stricto and B. weihenstephanensis. A number of 40 strains from different food sources, containing both high and low concentrations of B. cereus-like organisms, were randomly selected for further characterization. It was confirmed that the strains belonged to the B. cereus group by carrying out biochemical tests comprising fermentation of nine different carbohydrates and cleavage of several enzyme substrates linked to fluorophores using the Sensititre system for identification of Gram-positive bacteria (Trek Diagnostic Systems Ltd., East Grinstead, UK),. Three commercially available B. thuringiensis-based insecticides, Dipel®, Vectobac® and Bactimos® were also included in the investigation.
2.3Detection of crystal proteins and selected insecticide toxin genes
The isolates were streaked onto Agar Starch Medium (BD Bioscience, Erembodegem, Belgium) and inspected by phase-contrast microscopy for intracellular crystals, which are characteristic for B. thuringiensis after growth for 2–3 days at 30 °C.
Furthermore, PCR analyses were carried out to detect various groups of insecticide toxin genes from B. thuringiensis. Five general primer sets were employed. The primer sets for detecting cry1, cry3, cry11 and cyta were previously described. An additional primer set was designed to detect genes for the insecticidal toxin Cry4 (Cry4a: 5′-CAGGTACCGGTGGAATGAATTATA-3′ and Cry4b: 5′-GCTCTAGAGACTTCTACTTTAGTA) generating a fragment of 1932 bp. Total genomic DNA was isolated by the method of Boe and co-workers. All PCR amplifications were performed in a Programmable Thermal Controller PTC-100 (MJ Research, Bio-Rad, Waltham, MA, USA). One PCR comprised 24 μl PCR SUPERMIX (Gibco BRL, Invitrogen, Taastrup, Denmark), 20 μM primer, and 1 μg genomic DNA. The PCR conditions were the following: an initial denaturation step for 5 min at 94 °C, followed by 30 cycles of 1 min at 94 °C, 1 min at 52 °C, and 3 min at 72 °C, and a final extension at 72 °C for 5 min.
2.4Detection of genes and components for enterotoxins and emetic toxin
The gene for the B component of HBL (hblA) was amplified using the HblA primers described by Wencheng. The recently published CK primers and EM1 primers were used for detection of genes coding for CytK and the emetic toxin, respectively. DNA was extracted by boiling a bacterial colony for 10 min in TE-buffer. After centrifugation, the supernatant was used in a PCR consisting of one Ready-To-Go PCR bead (Amersham Pharmacia Biotech, Buckinghamshire, UK), 10 pmol of each primer and 5 μl of DNA. PCR was performed in a Peltier Thermal Cycler PTC-225 (MJ Research, Bio-Rad, Waltham, MA, USA) with the following conditions: an initial denaturation step at 94 °C for 10 min, followed by 30 cycles at 92 °C for 40 s, 55 °C for 40 s, 72 °C for 90 s, and a final extension at 72 °C for 7 min.
Enterotoxin production was tested using the B. cereus Enterotoxin reverse Passive Agglutination test from Oxoid (BCET-RPLA) (Oxoid, Basingstoke, UK) and the B. cereus Diarrhoeal Enterotoxin Visual Immuno Assay (TECRA BDE-VIA) (Tecra Diagnostics, Reading, UK). The BCET-RPLA kit detects the L2 component of HBL whereas the TECRA BDE-VIA kit detects the 45 kDa protein of the NHE complex. Overnight cultures originating from single colonies were diluted 1:20 in fresh BHI broth medium (Oxoid, Basingstoke, UK) and incubated with aeration at 30 °C for further 12–14 h. The supernatant was isolated by centrifugation (2000g, 20 min, 4 °C) and analysed in accordance with the manufacturers' instructions.
2.5Susceptibility to antimicrobials
The antimicrobial susceptibility was determined by broth dilution testing from Sensititre (Trek Diagnostic Systems Ltd., East Grinstead, UK). The microtitre wells were inoculated according to NCCLS guidelines and incubated aerobically at 37 °C for 18–24 h. The MIC-value was defined as the lowest concentration of antimicrobial that produced no visible growth. Isolates were tested with the following antibiotics and testing ranges: chloramphenicol (1–64 μg ml−1), ciprofloxacin (0.03–8 μg ml−1), gentamicin (0.5–32 μg ml−1), streptomycin (2–128 μg ml−1), penicillin (2–128 μg ml−1), erythromycin (1–32 μg ml−1), vancomycin (2–32 μg ml−1) and tetracycline (0.5–32 μg ml−1). For interpretation of the MIC results the following breakpoints were chosen: ciprofloxacin > = 4 μg ml−1, erythomycin > = 8 μg ml−1, gentamicin, vancomycin and tetracycline > = 16 μg ml−1, chloramphenicol, streptomycin, and penicillin > = 32 μg ml−1.
3.1Occurrence of B. cereus-like organisms in ready-to-eat food
In the routine surveillance of ready-to-eat foods for sale at the retail market in Denmark, a total of 48,901 randomly collected samples with unknown preparation and storage conditions were tested for the occurrence of B. cereus-like organisms at the point of sampling in the period 2000–2003 (Table 1). The products analysed were fresh fruits and vegetables, heat-treated products such as ready-prepared dishes, sauces, meat, pasta, and rice, and products with both fresh and heat-treated ingredients, e.g. sandwiches, pasta salad, vegetable/meat/fish mayonnaise, and desserts including ice cream and cream-cakes. The desserts analysed contained typically heat-treated starchy ingredients and/or milk and milk products. Enumeration of B. cereus-like organisms showed that 98.7% of the products had counts below 103 cfu g−1, 0.7% were in the range 103–104 cfu g−1, and 0.5% of the samples had counts above 104 cfu g−1 (Table 1). The high counts were most frequently found in fresh cucumbers and tomatoes, heat-treated rice, cake custard, and in desserts with milk and rice. The latter is a Danish speciality called ‘ris a la mande’, which is made from rice boiled in milk added almonds and whipped cream.
|Food category||Number of samples investigated||Percentage of samples containing B. cereus-like organisms within the range:|
|N||<1000 cfu g−1a||1000–10,000 cfu g−1a||>10,000 cfu g−1a|
|Pâté, liver paste, etc.||529||99.8||0.2||0|
|Meat for open sandwiches||4,215||99.7||0.3||0.1|
|A combination of raw and heat treated food|
|Desserts with milk and flour||2350||97.5||1.5||1.0|
|Desserts with milk and rice||223||95.1||1.8||3.1|
|Ice-cream with milk products||4751||99.3||0.5||0.2|
3.2Detection of crystals and crystal protein genes
The ability of 40 randomly selected strains to produce intracellular crystals, together with their content of selected cry genes, is presented in Table 2. Twenty-eight of the strains produced visible crystals and were therefore classified as B. thuringiensis. Of these 28 strains, 10 were positive for cry1 and two positive for cry11. Furthermore, one strain was positive for cry11 and two positive for cyta without a visible crystal protein giving a total of 31 B. thuringiensis strains. None were found to harbour genes for production of Cry3 and Cry4. Nine strains contained no visible crystal protein and were negative in PCR for the cry genes tested.
|Strain||Source||Insecticidal proteins||Enterotoxins||Emetic toxin|
3.3Distribution of genes and components involved in human disease
The production of selected protein components, and the presence of selected genes coding for proteins involved in the diarrhoeal and emetic syndrome, is listed in Table 2. The 45-kDa protein of the NHE complex was produced by all 40 strains, 36 strains also produced the L2 component of the HBL complex or had genes coding for the B component of HBL, and 27 strains had genes coding for CytK. These results show that all strains examined had genes or components for toxins involved in human diarrhoeal disease. Genes for the emetic toxin were only found in one strain identified as B. cereus.
3.4Geno- and phenotypic profiles
Grouping of the 40 randomly selected strains based on their geno- and phenotypic characteristics as regards crystal proteins, enterotoxins and emetic toxin (Table 2) showed that these characteristics were independent on the food source of the various strains. By comparing the profiles of the food isolates with the profiles of the commercial biopesticides it was observed that five strains isolated from sausage, pasta, red pepper (×2), and cauliflower stowage had profiles similar to the commercial Dipel® strain.
3.5Susceptibility to antimicrobials
Apart from intrinsic penicillin resistance found in 36 strains – caused by the production of β-lactamase by B. cereus-like organisms, the strains were sensitive to all the antimicrobials tested (Table 3). Two strains classified as B. cereus (10570, 11128) and two strains classified as B. thuringiensis (10584, 11294) were sensitive to penicillin as well.
|Antimicrobial agent||MIC (μg/ml)||% Resistant|
|Chloramphenicol||< = 1–4||2||2||0|
|Ciprofloxacin||< = 0.03–0.25||0.12||0.12||0|
|Gentamicin||< = 0.5–2||< = 0.5||< = 0.5||0|
|Streptomycin||< = 2–8||2||4||0|
|Tetracycline||< = 0.5–8||< = 0.5||4||0|
|Erythromycin||< = 1–4||1||2||0|
|Vancomycin||< = 2–2||2||2||0|
This study has shown that B. cereus-like organisms can be isolated from many different ready-to-eat food products for sale at retail. In most of the food samples studied, the counts of these organisms were below 103 cfu g−1 (satisfactory according to the Danish guidelines). In 0.5% of the samples, though, the counts were above 104 cfu g−1 (unacceptably high according to the Danish guidelines). As expected, the high counts were mainly found in heat-treated, starchy products, where growth of the organisms might have occurred as a consequence of improper cooling of the products after heat-treatment. Similar high counts of B. cereus in pasta and rice products have been reported in a Dutch study and in a study of ready-to-eat food in Taiwan. The present finding is in good agreement with the fact that the majority of food borne outbreaks of B. cereus-originated illness are caused by cooked, starchy products. Due to high toxin production in such products, the emetic syndrome is also related to these food types.
Valero and co-workers observed that cucumbers among several vegetables analysed presented high counts of B. cereus. Likewise, the present study found high counts of B. cereus-like organisms in cucumbers and tomatoes (>104 cfu g−1). These counts are most likely not due to growth of the organisms, but may be natural contaminants or residues of B. thuringiensis insecticides. Of the forty isolates further characterized, five isolates from different food categories grouped together with the commercial Dipel® strain according to content of cry genes and enterotoxin genes and proteins. However, further studies are needed to clarify the genetic relationship of the isolated strains to commercial B. thuringiensis strains.
Other food categories, which have occasionally been reported to present high numbers of B. cereus-like organisms (above 104 cfu g−1), are spices and milk [30,31]. These products were not included in our study but are subjects of investigations presently being conducted in our institute.
All forty strains tested had at least one gene or component of HBL and NHE. A high occurrence of protein components and/or genes involved in diarrhoeal disease has previously been described for B. cereus from food [29,32,33] and B. thuringiensis[34,35]. In addition, genes for the cytotoxin CytK were abundantly found among the isolates, as also reported in another study. The importance of this frequent occurrence of CytK in B. cereus-like organisms is unknown though, as the role of CytK in food borne disease is not yet fully understood. Recent research has identified a new variant of cytK, designated as cytK-2, with the original cytK being cytK-1. This new variant has only 89% identity at amino acid sequence level to the original CytK, and was shown to have a lower toxicity than CytK-1 against mammalian cell lines. Furthermore, it has recently been shown that in the clinical strain responsible for the death of three persons, cytK was more strongly transcribed. The high CytK expression may account of the high virulence of this strain. However, in our study we used primers that do not differentiate between these two types of CytK. Genes for the emetic toxin were only found in one strain of B. cereus isolated from pasta. The production of emetic toxin is believed to be restricted to a certain group of B. cereus[38,39], and in agreement with the findings in an earlier study, the mentioned strain did not possess genes or protein products for HBL.
Within a given strain, the presence of toxic components or genes encoding them does not necessarily lead to food borne disease following ingestion. Therefore, the exact influence of B. cereus and B. thuringiensis on human disease cannot be estimated from the present results. Similarly, such estimates are not possible from Danish surveillance data on human gastrointestinal diseases, since individual cases caused by B. cereus-like organisms are not registered. Nonetheless, outbreaks have been described in Denmark as well as in other countries. In the Netherlands and in England and Wales, B. cereus has been reported to be the causative organism of approximately 2% of the outbreaks of known origin. In France, the reported frequency of B. cereus outbreaks was 4–5%, and in the United States, 1–2% of the outbreaks have been attributed to B. cereus.
Except from penicillin, all 40 isolates were susceptible to the antibiotics tested. Hence, potential infections caused by the B. cereus-like organisms deriving from the present study may be treated with antimicrobial agents. Antimicrobial resistance among B. cereus has previously been reported in milk and dairy products. Phenotypic characterisation of the isolates based on antibiotic resistance profiles was not possible in this investigation, though other studies have shown that the resistance patterns of different Bacillus spp., in part, are species related [46–48].
The majority of the strains isolated from the food samples belonged to B. thuringiensis due to the presence of intracellular crystals and/or genes for selected cry genes. We believe this to be the first larger study with the specific aim to identify B. thuringiensis in food, since earlier studies did not distinguish between B. thuringiensis and B. cereus sensu stricto. Previously, B. thuringiensis has been isolated from grapes [49,50], farm bulk tank and creamery silo milk, and in a small sample of pasta, bread and milk. These observations indicate that B. thuringiensis could actually be responsible for many of the food borne outbreaks previously attributed to B. cereus sensu stricto.
We thank the Regional Veterinary and Food Authorities for collecting and analysing the food samples, Bodil Madsen and Rikke Kubert, the Danish Institute for Food and Veterinary Research, for technical assistance regarding characterisation of the isolates, and Stephen On for correcting the English language. The Danish Veterinary and Food Administration and the Danish Environmental Protection Agency partly funded this study.
- Kramer, J.M., Gilbert, R.J. (1989) Bacillus cereus and other Bacillus species. In: Foodborne Bacterial Pathogens (Doyle, M.P., Ed.), pp.21–70 Marcel Dekker Inc., New York and Basel.
- Granum, P.E. Bacillus cereus. Doyle, M.P., Ed. Food Microbiology: Fundamentals and Frontiers, 2nd Edn. (2001) ASM Press, Herndon, VA. 373–381
- Hansen, B.M., Damgaard, P.H., Eilenberg, J. and Pedersen, J.C. (1996) Bacillus thuringiensis. Ecology and environmental effects of its use for microbial pest control. Report no. 316, Danish Environmental Protection Agency.
- Anonymous (1997) Bacillus cereus. Determination in Foods (NMKL 67). Nordic Committee on Food Analysis 67 (4th Edn.).
- 1998) Study of the bceT and hblA genes and the hemolysin BL of Bacillus thuringiensis group. Chin. J. Microbiol. Immunol. 18, 428–433.(
- National Committee for Clinical Laboratory Standards (1997) Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically. Approved Standard M7-A4, National Committee for Clinical Laboratory Standards, Wayne, PA.
- Anonymous. Guidelines for the Assessment of Pathogenic Microorganisms in Food. (1999) Danish Veterinary and Food Administration, Copenhagen. 80 pp
- 1996) Incidence of Bacillus cereus and Bacillus subtilis in foods in the Netherlands. Food Microbiol. 13, 53–58., , , (
- The Danish Veterinary and Food Administration. http://www.foedevarestyrelsen.dk/Foedevare/Foedevarekontrol/Smitteopsporing/Forside.htm.
- WHO (2000) Surveillance Programme for Control of Foodborne Infection and Intoxications in Europe, 8th report 1993–1998 and 1999–2000.
- 2002) Les toxi-infections alimentaires collectives en France en 2001. Bull. Epidémiol. Hebdom. 50, 249–253., , , , , (
- Granum, P.E., Baird-Parker, T.C. (2000) Bacillus species. In: The Microbiological Safety and Quality of Food (Lund, B.M., Baird-Parker, T.C., Gould, G.W., Eds.), pp.1029–1056 Aspen Publishers, New York.
- 1987) A Bacillus thuringiensis isolate found on grapes imported from California. J. Food Prot. 50, 857–858., , (