Bacillus cereus comprises a highly versatile group of bacteria, which are of particular interest because of their capacity to cause disease. Emetic food poisoning is caused by the toxin cereulide produced during the growth of emetic B. cereus in food, while diarrhoeal food poisoning is the result of enterotoxin production by viable vegetative B. cereus cells in the small intestine, probably in the mucus layer and/or attached to the host's intestinal epithelium. The numbers of B. cereus causing disease are highly variable, depending on diverse factors linked to the host (age, diet, physiology and immunology), bacteria (cellular form, toxin genes and expression) and food (nutritional composition and meal characteristics). Bacillus cereus group strains show impressive ecological diversity, ranging from their saprophytic life cycle in soil to symbiotic (commensal and mutualistic) lifestyles near plant roots and in guts of insects and mammals to various pathogenic ones in diverse insect and mammalian hosts. During all these different ecological lifestyles, their toxins play important roles ranging from providing competitive advantages within microbial communities to inhibition of specific pathogenic organisms for their host and accomplishment of infections by damaging their host's tissues.
The Bacillus cereus group comprises the following seven closely related species: B. cereus s. s., B. thuringiensis, B. anthracis, B. weihenstephanensis, B. mycoides, B. pseudomycoides and B. cytotoxicus, which can be divided into seven phylogenetic groups (Table 1; Guinebretiere et al., 2010, 2013). The taxonomy of the B. cereus group is complex and equivocal. Strains clustering in group I (B. pseudomycoides), group VI (B. weihenstephanensis and B. mycoides) and group VII (B. cytotoxicus) appear to be different species, but strains of B. cereus, B. thuringiensis and B. anthracis are spread over groups II, III, IV and V without formation of clusters according to their ‘species’. Comparison of housekeeping genes with multilocus sequence typing (MLST) and multilocus enzyme electrophoresis (MEE) indicated that B. cereus, B. thuringiensis and B. anthracis are one species (Helgason et al., 2000). Moreover, very high (> 99%) sequence similarity of the 16S rRNA gene sequences of B. anthracis, B. cereus and B. thuringiensis was demonstrated (Sacchi et al., 2002). Therefore, these B. cereus group strains appear to belong to a single species. However, DNA–DNA hybridization was observed between strains of the different groups just below the threshold of 70%, which indicates that they are different species (Nakamura & Jackson, 1995; Guinebretiere et al., 2013). To further complicate this matter, the most discriminatory characteristics of the different species are often plasmid-encoded toxins (Rasko et al., 2005a, b), which can be lost naturally in the environment (te Giffel et al., 1997) or in the laboratory (Bizzarri et al., 2008) and which have a high potential for horizontal transfer (Van der Auwera et al., 2007; Yuan et al., 2007; Modrie et al., 2010). Finally, the taxonomical discussion has been influenced by the practical usefulness of discriminating between strains based on phenotypical similarly such as specific pathogenicity (B. anthracis and B. thuringiensis), psychrotolerance (B. weihenstephanensis) and colony morphology (B. mycoides and B. pseudomycoides) rather than on genetical relatedness (Guinebretiere et al., 2008). Bacillus cereus, B. thuringiensis and B. anthracis share a large core set of conserved genes (75–80%) and a highly similar and syntenic chromosomes, suggesting that these strains are at least originating from a common ancestor and/or have been under multidirectional influence by intensive horizontal gene transfer (Ivanova et al., 2003; Rasko et al., 2005a). Phylogenetic groups can be distinguished, but strains of the three ‘species’ occur intermingled in these phylogenetic clusters (Table 1; Guinebretiere et al., 2010). Phenotypic distinction mainly originates from plasmid-borne virulence properties (Rasko et al., 2005a, b), such as those with insecticidal crystal toxins (cry genes) in B. thuringiensis (Schnepf et al., 1998), the anthrax toxins (pXO1 and pXO2 plasmids) in B. anthracis (Mock & Fouet, 2001) and the emetic toxin (ces genes) in B. cereus (Ehling-Schulz et al., 2006a). Most enterotoxins and extracellular virulence factors are encoded on the chromosome and are thus shared by all B. cereus group strains, but B. anthracis deviates from the other strains due to a nonsense mutation in the important virulence regulator PlcR, preventing expression of plcR-regulated genes (Slamti et al., 2004; Gohar et al., 2008). The reader is referred to previously published reviews for a detailed review of the taxonomy of B. cereus (Maughan & Van der Auwera, 2011), their clinical significance (Bottone, 2010), their toxins (Stenfors-Arnesen et al., 2008) and the regulation of toxin production (Ceuppens et al., 2011). In this review, the emetic food poisoning and diarrhoeal food poisoning by B. cereus are elaborately explained, followed by an overview of the diversity of B. cereus group strains in pathogenicity and ecological lifestyles. The focus of this review is on the significance and role of B. cereus toxins in different environments.
Table 1. Phylogenetic groups and species of the Bacillus cereus group and their cytotoxic and growth characteristics (adapted from Guinebretiere et al., 2010)
Growth temperature range (°C)
B. cereus, B. thuringiensis
B. cereus, B. thuringiensis, emetic B. cereus in groups III-2 and III-3, B. anthracis in group III-4
B. cereus, B. thuringiensis
B. cereus, B. thuringiensis
B. weihenstephanensis, B. mycoides
Emetic and diarrhoeal food poisoning caused by B. cereus
According to the European Food Safety Authority (EFSA) reports on food-borne outbreaks in the EU, B. cereus was responsible for only 1.88% of the reported food-borne outbreaks in 2010 (EFSA, 2012). However, B. cereus was identified as the causative agent in 56% of the food-borne outbreaks reported in 2011 to the Laboratoire Central des Services vétérinaires (situated at Anses, Maisons-Alfort) in France for which the causative agent was identified, which was the case for 60 of the 197 outbreaks in total (Anses, Laboratoire de sécurité des aliments de Maisons-Alfort, pers. commun.). Bacillus cereus is underestimated as a food-borne pathogen due to a number of causes. Firstly, underreporting exists due to the mild and transient symptoms of the illness. Secondly, B. cereus food poisoning is not classified by EFSA as a zoonosis, a disease transmitted through animals, and therefore has received less attention in reporting and surveillance compared with zoonotic agents. Thirdly, atypical and emetic strains, which often display no or low lecithinase and haemolytic activity, are not detected with the standard enumeration of B. cereus by plating on MYP according to ISO7932:2004 (Ehling-Schulz et al., 2005; Fricker et al., 2008). Fourthly, food products implicated in food poisoning outbreaks are likely to contain multiple B. cereus strains (Pirhonen et al., 2005). To identify the strain responsible for the illness, multiple B. cereus colonies of different morphology should be selected for further investigation and, ideally, linked to identical isolates from the stool and/or vomit of patients.
The food-borne pathogen B. cereus is widespread in the environment. As a result, diverse raw food ingredients such as vegetables, potatoes, milk, herbs and spices are often contaminated with B. cereus spores. For example, the cultivation soil of courgettes contained 4.0 × 104B. cereus spores per gram soil, which constituted an important source of B. cereus contamination in the finished food product, courgette purée (Guinebretiere & Nguyen-The, 2003). Thus, a wide variety of finished food products frequently contains B. cereus, such as cooked chilled foods [Refrigerated Processed Foods of Extended Durability (REPFEDs)], vegetable purées, sauces, rice, cereals, fish and seafood, milk and dairy products, although usually at low levels (< 103 CFU g−1; Choma et al., 2000; Wijnands et al., 2006a; Rahmati & Labbe, 2008; Park et al., 2009; Samapundo et al., 2011). Germination and outgrowth of spores in the finished food products during storage are possible, depending on the food properties, storage temperature and minimal growth temperature of the B. cereus strains. In this way, both spores and vegetative cells can be present in food products. Few studies regarding the prevalence of B. cereus in food distinguished between the cellular and spore forms. Therefore, it is difficult to estimate the relative prevalence of B. cereus vegetative cells and spores in food products at the time of consumption. Bacillus cereus spores are definitely expected to be present in food, but their prevalence may vary among different food types. For example, 37% of the B. cereus present in Belgian retail lasagne were spores, in comparison with only 6% in raw rice (Samapundo et al., 2011). Similarly, retail raw rice from the United States often (53%) contained B. cereus spores, albeit in very low numbers (33 spores g−1 on average; Ankolekar et al., 2009).
It is overall assumed that only consumption of food containing between 105 and 108B. cereus cells and/or spores will cause disease (Granum & Lund, 1997; EFSA, 2005). Emetic food poisoning is usually caused by preformed cereulide in food, because this emetic toxin is not inactivated during food processing or gastrointestinal passage due to its high resistance against heat treatments, extreme pH values and protease activities (Shinagawa et al. 1996; Agata et al., 2002; Rajkovic et al., 2008). As a consequence, ingestion of living B. cereus is not necessarily required to experience this type of illness. In contrast, diarrhoeal food poisoning is not caused by preformed enterotoxins in food, but by viable vegetative B. cereus cells producing enterotoxins in the small intestine, because spores produce no enterotoxins, and these proteins are rapidly (≤ 0.5 h) degraded under gastrointestinal conditions by proteases present in the host's digestive secretions, even at their basic fasted levels (Turnbull et al., 1979; Granum et al., 1993; Wijnands et al., 2005; Ceuppens et al., 2012a). The current hypothesized course of B. cereus diarrhoeal food poisoning is presented in Fig. 1. The host ingests food contaminated with B. cereus spores, vegetative cells and/or enterotoxins already produced by B. cereus during growth in the food. During gastric passage, preformed enterotoxins and the majority of vegetative cells are inactivated by gastric acid, while the spores remain viable but inactive (Turnbull et al., 1979; Granum et al., 1993; Clavel et al., 2004, 2007; Wijnands et al., 2005, 2006b, 2009; Ceuppens et al., 2012a-e). After gastric passage, the remaining vegetative cells and enterotoxins are rapidly inactivated by the digestive secretions in the small intestine (Ceuppens et al., 2012a, d). It is hypothesized that only cells that instantly reach the mucus layer upon arrival in the small intestine can survive, adhere to host cells, produce enterotoxins and cause diarrhoeal disease (Ceuppens et al., 2012d). In contrast, all spores survive and germinate in the distal small intestine (ileum), where the bile concentration is reduced due to re-absorption by the host (Clavel et al., 2004, 2007; Wijnands et al., 2006b; Ceuppens et al., 2012b, c). Bacillus cereus outgrowth and enterotoxin production in the intestinal lumen are highly unlikely. Firstly, the normal indigenous intestinal microbial community inhibits the vegetative outgrowth of germinated B. cereus spores in the intestinal lumen (Ceuppens et al., 2012e). Secondly, enterotoxins produced in the intestinal environment are rapidly degraded by digestive enzymes (Turnbull et al., 1979; Granum et al., 1993; Wijnands et al., 2005; Ceuppens et al., 2012a). Therefore, B. cereus spores probably first adhere to the mucus layer near the intestinal epithelium of the host, followed by germination, vegetative outgrowth and enterotoxin production. These toxins then damage the nearby epithelial cells by pore formation, resulting in microvilli damage and osmotic lysis of the host's epithelial cells and eventually diarrhoea (Beecher et al., 1995; Hardy et al., 2001; Minnaard et al., 2001; Lindbäck et al., 2004; Ramarao & Lereclus, 2006; Fagerlund et al., 2008).
Pathogenic B. cereus strains
As an indication of the pathogenic potential of B. cereus strains, the presence of toxin genes can be determined by PCR.
Emetic B. cereus comprise a subgroup of strains, which possess the plasmid-encoded cereulide synthase genes (Ehling-Schulz et al., 2006a). This nonribosomal peptide synthase gene cluster cesHPTABCD, located on a pXO1-like megaplasmid of 200–270 kb, is responsible for the synthesis of the emetic toxin cereulide (Ehling-Schulz et al., 2006a; Dommel et al., 2011). Moreover, the transcription level of ces genes corresponds well with the levels of cereulide produced (Dommel et al., 2011). As a consequence, several PCR assays targeting the ces genes have been described to screen for the emetic food poisoning genotype (Ehling-Schulz et al., 2004, 2005; Horwood et al., 2004; Nakano et al., 2004; Fricker et al., 2007; Seong et al., 2008). As always, careful primer design and thorough evaluation of the primer pairs are required to include all different polymorphisms in such PCR screening for toxin genes, as was underlined by the discovery of B. weihenstephanensis strains, which produced cereulide, but were PCR negative (Hoton et al., 2009). Emetic strains are widely distributed in food products and the natural environment. For example, 4.7% of the B. cereus isolates in ice cream were emetic strains, 1.6% in fish products, 11.0% in ready-to-eat foods, 3.9% in other food samples not implicated in food poisoning, 1.7% in soil, 1.5% in cow milk, 1.2% in cow bedding and 3.9% in farm rinsing water (Yang et al., 2005; Altayar & Sutherland, 2006; Svensson et al., 2006; Wijnands et al., 2006a; Rahmati & Labbe, 2008; Messelhausser et al., 2010). Although the overall prevalence of emetic strains among B. cereus isolates is low, c. 1.5%, it is much higher in nonrandom food and clinical samples, namely 32.8%, indicating the enhanced pathogenic potential of these strains (Hoton et al., 2009). Emetic strains often display certain specific characteristics compared with other B. cereus strains, such as no or weak haemolysis, the inability to hydrolyse starch and salicin, an elevated lower temperature limit for growth of minimum 10 °C and an increased heat resistance of their spores (Shinagawa, 1993; Andersson et al., 2004; Carlin et al., 2006). As a consequence, emetic strains were initially considered to be a clonal B. cereus lineage, but several independent emetic B. cereus lineages have been discovered (Apetroaie et al., 2005; Vassileva et al., 2007). Food isolates were often indistinguishable regardless of their origin, but environmental and intestinal isolates demonstrated more phenotypic variability. The majority of the cereulide producing strains are mesophilic, but other variants of the ces operon were discovered on a larger unrelated plasmid or even in the chromosome of two emetic B. weihenstephanensis and three emetic B. cereus strains, which were able to grow at 8 °C (Thorsen et al., 2006; Hoton et al., 2009). Enterotoxin nhe genes have been detected in the majority of the emetic strains (Yang et al., 2005; Rahmati & Labbe, 2008; Kim et al., 2010), so it seems plausible that emetic strains can also cause diarrhoeal food poisoning.
Similarly, various (q)PCR assays have been designed to detect diarrhoeal toxins and virulence factors, such as nonhaemolytic enterotoxin (Nhe) genes, nheA, nheB and nheC (Granum et al., 1999; Guinebretiere et al., 2002, 2010; Yang et al., 2005, 2007; Ehling-Schulz et al., 2006b; Ngamwongsatit et al., 2008; Wehrle et al., 2010); haemolysin BL (Hbl) genes, hblC, hblD, hblA and hblB (Ryan et al., 1997; Hansen & Hendriksen, 2001; Guinebretiere et al., 2002; Thaenthanee et al., 2005; Yang et al., 2005; Ehling-Schulz et al., 2006b; Ngamwongsatit et al., 2008; Wehrle et al., 2010); cytotoxin K genes, cytK-1 and/or cytK-2 (Fagerlund et al., 2004; Yang et al., 2005; Ehling-Schulz et al., 2006b; Guinebretiere et al., 2006; Ngamwongsatit et al., 2008; Wehrle et al., 2010); enterotoxin FM gene entFM (Kalman et al., 1993; Yang et al., 2005; Ngamwongsatit et al., 2008); enterotoxin T gene bceT (Agata et al., 1995; Yang et al., 2005), haemolysin II gene hlyII (Fagerlund et al., 2004; Cadot et al., 2010); the phospholipase C gene PC-PLC (Schraft & Griffiths, 1995; Martinez-Blanch et al., 2009); cereolysin AB genes cerAB (Schraft & Griffiths, 1995); and cereolysin O gene clo (Wang et al., 1997). Many of the primer pairs do not detect all toxin gene variants, which may lead to underestimation of the enterotoxin gene prevalence. Guinebretiere et al. (2002) designed one of the most frequently used primers to determine the enterotoxin genotype of B. cereus strains. However, the authors had already demonstrated with Southern blotting in the original publication that not all toxin variants were detected with these primers. They recently presented new primer pairs, which detect all Nhe variants, including the most divergent polymorphisms of groups I and VII (Guinebretiere et al., 2010). Studies using the most recent and broad primer pairs show that enterotoxin genes are widely spread among B. cereus strains, also among strains originating from environmental, food and faecal samples in a nonclinical setting. For example, all B. cereus strains seem to possess the nhe genes, even though highly divergent variations exist (Ngamwongsatit et al., 2008; Guinebretiere et al., 2010).
In contrast to the clear association between emetic food poisoning cases and emetic B. cereus strains possessing the ces genes, no specific genotype could be linked to diarrhoeal food poisoning cases. No link has been found between diarrhoeal food poisoning strains and the presence of certain diarrhoeal toxin genes, virulence factors and not even their regulatory genes (Fagerlund et al., 2007). A preliminary risk assessment of B. cereus isolates can be made by determining its phylogenetic group (Table 1; Guinebretiere et al., 2010). Each group is associated with a certain cytotoxicity class and food poisoning risk. However, this classification gives only an estimation of the most probable level of toxicity, because several classes contain exceptional strains, which deviate considerably from the class average. Rather than presence or absence of enterotoxin genes, their expression level seems to correlate with the potential of the strains for causing food poisoning. Food poisoning strains displayed higher prevalence of enterotoxin genes, less variation in the toxin gene sequences and higher toxin gene expression than random B. cereus food isolates (Guinebretiere et al., 2002; Moravek et al., 2006). The same phenomenon was observed for the 41 B. cereus strains analysed in this study (Table 2). All strains showed the genomic presence of the nheB gene, but only 71% expressed this gene. All of the nonproducing strains were found in the category of psychrotolerant strains isolated from randomly sampled food and human faeces, which were thus not implicated in diarrhoeal food poisoning cases. The enterotoxin Nhe and Hbl production by psychrotolerant strains can be underestimated due to their reduced growth rate at high temperature. Nevertheless, if these psychrotolerant strains fail to grow sufficiently to produce the enterotoxins after limited incubation time (8–16 h) at body temperature, they will probably not be able to do so in vivo and are unlikely to cause diarrhoeal illness. Few studies have investigated the specific contributions and (synergistic) interactions of the enterotoxins to cytotoxicity and diarrhoeal food poisoning. The cytotoxicity assay with Vero cells indicated the predominant contribution of Nhe to cytotoxicity rather than Hbl (Moravek et al., 2006), but it remains uncertain whether this accurately reflects the in vivo situation. Synergistic haemolysis has been reported for enterotoxin Hbl, sphingomyelinase (SM-PLC, SMase) and phosphatidylcholine phospholipase C (PC-PLC, lecithinase) during in vitro tests with erythrocytes (Beecher & Wong, 2000). Synergism occurred under specific conditions, determined by many factors such as the erythrocytes' membrane composition (variable among mammalian species), the concentrations of the enzymes and the presence of cations. Nevertheless, relevant interactions between enterotoxins and virulence factors may also exist during gastrointestinal illness and infections. In conclusion, no specific B. cereus genotype(s) has been correlated with diarrhoeal food poisoning, because the enterotoxin genes are highly prevalent among B. cereus strains and their expression levels vary greatly between different strains with the same genes.
Table 2. Comparison of the genomic presence of nheB gene and the production of enterotoxins Nhe and Hbl of 41 Bacillus strains
The food product strongly influences the production of both emetic and diarrhoeal toxins through its specific nutritional composition, other intrinsic food properties and storage (Ceuppens et al., 2011). Emetic food poisoning usually involves food products with high starch contents, such as pasta, rice, mashed potatoes, bread and pastries, which stimulate the production and accumulation of the stable toxin cereulide (Jaaskelainen et al., 2003; Rajkovic et al., 2006b), although the majority (> 90%) of emetic B. cereus strains are unable to hydrolyse starch (Kim et al., 2010). It is also important to note that under anaerobic conditions with < 1 to 2% O2, cereulide production no longer occurs (Jaaskelainen et al., 2004; Rajkovic et al., 2006a). Emetic B. cereus strains in the exponential growth phase can produce 0.004–0.130 μg of cereulide per 106 cells, independent of sporulation (Haggblom et al., 2002; Jaaskelainen et al., 2004). This means that populations in the range of 105 to 108 can produce sufficient amounts of toxin to cause gastrointestinal illness, because the estimated required dose of cereulide is between 0.02 and 1.83 μg per kg body weight (Ceuppens et al., 2011).
In contrast, diarrhoeal food poisoning cases occur with a wide variety of food commodities. The influence of food type is less pronounced in these cases, because preformed enterotoxins in food are rapidly degraded during gastrointestinal passage. The food matrix significantly influences the gastrointestinal survival of B. cereus vegetative cells by modifying the acid resistance and bile resistance at ≤ 3.0 g L−1 (Clavel et al., 2004, 2007), while this influence is no longer observed for continuously decreasing gastric pH values and higher bile concentrations of ≥ 5.0 g L−1 (Ceuppens et al., 2012d).
In general, cooked chilled foods (REPFEDs), milk and dairy products, dried herbs and spices and to-be-reconstituted formulates are risk food products for B. cereus food poisoning, because these products commonly contain low numbers of B. cereus, while the competitive microbiota is inactivated by drying or heat treatment (pasteurization). In Belgium, 70% of the retail lasagne in Belgium contained B. cereus, as well as 77% of the bolognese sauce samples, 81% of the béchamel sauce, 20% of the cooked pasta, 100% of the raw rice, 15% of the fresh minced beef, 40% of the carrots, 35% of the celery, 30% of the Chinese cabbage and 5% of the bell pepper (Samapundo et al., 2011). To investigate the level and source of B. cereus contamination of REPFEDs, several Belgian food companies were sampled before, during and after production of various composite meals with meat or fish, vegetables and potatoes and various pasta dishes with bolognese or béchamel sauce (J. Daelman, LFMFP-UGent, unpublished data). Screening of the food ingredients showed that B. cereus only occurred in herbs and spices, except for two positive samples of raw meat (102.5 CFU g−1) and potato flakes (103.5 CFU g−1). Although the majority (72%) of herbs and spices were free of B. cereus (< 102 CFU g−1) and the contamination was usually at tolerable levels (97% ≤ 104 CFU g−1), high levels of B. cereus in herbs and spices were sporadically detected (3% > 105 CFU g−1). However, none of the finished products contained B. cereus at > 103 CFU g−1 after production or after storage at the time of consumption. In the Netherlands, REPFEDs and a wide variety of other retail food products were also contaminated with B. cereus, but the majority (> 99%) of the contamination levels were situated below the tolerance level of 105 CFU g−1 at the time of consumption (Wijnands et al., 2006a). In the USA, 18% of retail seafood samples contained B. cereus, although usually (62%) at low levels (< 102 CFU g−1; Rahmati & Labbe, 2008). Also in Korea, B. cereus was frequently isolated from glutinous rice (37%), Job's tears (27%), barley (21%), brown rice (18%) and soybean sprouts (71%), although again at low levels (≤ 102 CFU g−1; Kim et al., 2004; Park et al., 2009). The overall prevalence of B. cereus in Australian retail food samples was surprisingly low, being absent (< 102 CFU g−1) in 98% of the cases and most frequently found in chilled raw chicken (6% containing on average 2.0 × 104 CFU g−1), frozen cooked meat pies (5% with 1.6 × 102 CFU g−1) and pizzas (5% with 2.5 × 103 CFU g−1; Eglezos et al., 2010). Specific food products are inherently associated with different numbers and different types of B. cereus strains. For example, milk and fish products were associated with psychrotolerant strains (Wijnands et al., 2006a), and fried rice dishes in restaurant are associated with high B. cereus levels of 104 to 105 CFU g−1 (Chang et al., 2011). In addition, the storage temperature of the final food product is the major factor influencing the number and type of B. cereus present at the end of shelf life. For example, storage of zucchini purée at 4 °C for 21 days did not reveal any B. cereus present (< 5.0 × 101 CFU g−1), while storage at 10 °C for 21 days resulted in 4.0 × 104 CFU g−1 of mesophilic and psychrotolerant soil strains and storage at room temperature for 5 days revealed 2.5 × 106 CFU g−1 mesophilic strains from soil and milk proteins (Guinebretiere & Nguyen-The, 2003; Guinebretiere et al., 2003). In conclusion, B. cereus is a ubiquitous environmental and food contaminant, which is present in a wide variety of retail food products worldwide, although seldom in high (> 105 CFU g−1) concentrations.
Minimal numbers of B. cereus required for disease
Bacillus cereus food poisoning is not simply linked to the amount of ingested B. cereus. For example, consumption of 250–450 mL milk which was naturally contaminated with up to 108 CFU mL−1 psychrotolerant B. cereus provoked no gastrointestinal complaints (Langeveld et al., 1996), while three elderly people died after consumption of vegetable purée containing 3.2 × 105B. cereus per gram (Lund et al., 2000). The numbers of B. cereus required for disease are estimated between 105 and 108 living B. cereus based on epidemiological data from food poisoning cases (Granum & Lund, 1997; EFSA, 2005). Obtaining reliable data on dose-response are hampered by the time delays between the consumption, the development of food poisoning symptoms and the food sample analysis. Moreover, relevant and adequately stored food samples from food poisoning outbreaks are seldom available for microbiological analysis. Furthermore, the identification of an exact ‘infective dose’ for diarrhoeal food poisoning is complicated by the exceptionally large variability in B. cereus strains, host physiology and food commodity, which greatly influence the development and severity of disease (Fig. 2). Firstly, B. cereus can possess many toxin genes in different combinations and different expression levels (see 'Pathogenic B. cereus strains'). Germination, growth and enterotoxin production of psychrotolerant strains are generally decreased in comparison with mesophilic strains at high temperature (37 °C; Wijnands et al., 2007). Finally, the cellular type is of prominent importance, because vegetative cells lack the acid and bile resistance of spores (see 'Emetic and diarrhoeal food poisoning caused by B. cereus'). As a consequence, the capacity of vegetative cells to survive gastrointestinal passage is very low, in contrast to that of spores, presumably resulting in elevated required numbers for causing disease. Secondly, nearly all host aspects are highly variable. Each person has a specific intestinal microbiota, health status and medication history (Jernberg et al., 2007; Palmer et al., 2007). Moreover, people display large variability in eating habits, gastric emptying and secretion of gastric acid and bile (Dressman et al., 1990; Russell et al., 1993; Clarkston et al., 1997). These characteristics vary mostly in function of age and sex, but the immune system and the use of drugs and antibiotics also have a significant influence (Vakevainen et al., 2000; Lahner et al., 2009). Thirdly, the presence of food particles had no direct influence on the gastrointestinal survival and enterotoxin production of B. cereus during our standardized in vitro experiments (Ceuppens et al., 2012a, d). However, the food is expected to play several indirect roles. The food type determines the available nutrients and the composition of the competitive microbiota. Moreover, storage temperatures (and temperature abuse) determine whether B. cereus can multiply, and if so, whether this will be predominately psychrotrophic or mesophilic strains. Finally, the food type, meal size and time of consumption influence the kinetics of many digestive processes (Marciani et al., 2001), which in turn influences the bacterial survival during gastrointestinal passage.
Ecological diversity of B. cereus
The soil constitutes the primary environmental reservoir for B. cereus, where it can complete its saprophytic life cycle (Vilain et al., 2006; Fig. 3). Bacillus cereus spores germinate in soil when conditions are favourable, grow on decaying organic matter in the multicellular filamentous phenotype for translocation through the soil and finally resporulate when nutrients are depleted and conditions are no longer favourable. Similarly, growth of B. thuringiensis was also demonstrated in artificial soil microcosms (Ellis, 2004), while it is generally assumed that B. anthracis spores remain dormant in soil. During growth in a mixed microbial community, such as growth on decaying plant material in soil, toxin production can achieve a competition advantage and increased fitness (Tempelaars et al., 2011; see 'Ecological significance of toxins for B. cereus').
In the rhizosphere, some B. cereus strains positively interact with certain plants by stimulating their growth and offering protection against certain plant diseases. For example, B. cereus UW85 stimulated the growth of soy bean and prevented damping-off by alfalfa by zwittermicin A and kanosamine production (Handelsman et al., 1990; Bullied et al., 2002). These antimicrobial compounds synergistically inhibited the growth of various Gram-negative and Gram-positive bacteria, plant pathogenic fungi and oomycetes (Silo-Suh et al., 1998; see 'Ecological significance of toxins for B. cereus').
Bacillus cereus spores present in soil (or faeces) are ingested by arthropods. In the moist- and nutrient-rich intestines of healthy soil arthropods, germination occurs, followed by motile and filamentous growth, attached to the chitinous intestinal epithelium, and finally resporulation in the distal unattached end of the filaments (Margulis et al., 1998). Similarly, B. cereus s. s., B. thuringiensis and B. mycoides/B. pseudomycoides have been isolated from the digestive tracts of terrestrial arthropods (sow bugs; Swiecicka & Mahillon, 2006). Defecation or natural death of the host releases the cells and/or spores again into the soil. This represents the symbiotic lifestyle of B. cereus, which can be either commensal or mutualistic, if there are no effects or beneficial effects for the arthropod host, respectively. Although no proof of mutual beneficial relations has been provided, it is plausible to assume that intestinal B. cereus aid in the nutrient digestion of their hosts, since B. cereus group strains produce extracellular enzymes to degrade complex organic materials in soil (Luo et al., 2007; Gao et al., 2008). Moreover, B. cereus can produce various toxins, bacteriocins and antibiotics in the competition with other microbiota, which could protect the arthropod host against enteric pathogens (Bizani et al., 2005; Tempelaars et al., 2011). However, the pathogenic life cycle is also observed in insects depending on the specific strain and insect, for example B. thuringiensis subsp. kurstaki in lepidopteran and dipteran insects (Sevim et al., 2012). In this case, B. cereus strains kill the host by intestinal toxin production, followed by colonization and multiplication in the cadaver and eventually resporulation and return to the soil as a spore (Swiecicka et al., 2008). Consequently, B. thuringiensis toxins have been widely applied as biological insecticides, thereby reducing the amount of chemicals required to control the major pests (Naranjo & Ellsworth, 2010). The most widely known and successful application consists of transgenic corn, producing B. thuringiensis (Bt) toxins against the European corn borer (Ostrinia nubilalis).
Similarly to the diverse relations with insects host, B. cereus strains can also live both their pathogenic and the symbiotic life cycle (commensalism or mutualism) in mammals. For example, the most notorious B. cereus group strains are undoubtedly B. anthracis strains, the causative agents of anthrax disease in livestock (Koch, 1878). Anthrax is a zoonosis which is contracted cutaneously, pulmonary or gastrointestinally from contaminated cattle, goat, sheep or soil, or by injection of contaminated illicit drugs (Ringertz et al., 2000; Mock & Fouet, 2001). In contrast, Bacillus cereus var. toyoi is a beneficial member of the intestinal microbiota, which leads to better feed conversion (Williams et al., 2009). These B. cereus spores are sold as a probiotic feed additive by Rubinum S.A. under the name Toyocerin®, which is authorized in the EU as feed additive for poultry, swine, bovines and rabbits by EFSA. This probiotic B. cereus strain functions as a stabilizer of the intestinal microbiota and inhibitor of enteric bacteria such as Salmonella, Escherichia coli and clostridia to improve the feed conversion and performance of the animals (website Rubinum S.A.). Daily administration of 107 to 109 spores per kg body weight to various animals (rabbits, pigs, chickens, turkeys and cattle for 8–78 weeks) and even to human volunteers (five healthy males for 8 days) was found to be safe (Williams et al., 2009). As a result of its saprophytic soil life cycle, B. cereus is found in water, vegetables and many other food ingredients, resulting in the contamination of a wide variety of finished food products (see 'Risk food products'). Ingestion of B. cereus by humans can lead to emetic or diarrhoeal food poisoning (Fig. 2; see 'Emetic and diarrhoeal food poisoning caused by B. cereus'). Few diarrhoeal food poisoning cases have been attributed to B. thuringiensis strains, but this may be underreporting due to difficulties in species differentiation (Jackson et al., 1995; te Giffel et al., 1997). Interestingly, the presence of human intestinal cells specifically induces spore germination, indicating that the human gut is a specific favourable environment for B. cereus, activating its growth (Andersson et al., 1998; Wijnands et al., 2007). After multiplication and possibly sporulation in the host's intestinal tract, B. cereus cells and/or spores are excreted with the vomit and/or faeces, eventually return to the soil and thus end their pathogenic life cycle. Sometimes, the food poisoning cases can even have a lethal outcome due to dehydration or liver and brain damages after translocation of cereulide to those tissues (Posfay-Barbe et al., 2008; Shiota et al., 2010). Alternatively, consumption of B. cereus provokes no gastrointestinal illness, which exemplifies its symbiotic life cycle. After consumption of nonpathogenic strains or low quantities of pathogenic ones, B. cereus is also a common component of the normal transient intestinal microbiota, illustrated by its frequent (in 14–43% of the cases) isolation from healthy people's faeces (Ghosh, 1978; Turnbull & Kramer, 1985). Nevertheless, nonpathogenic B. cereus strains are unwelcome contaminants in food products, since they cause spoilage and economical losses, for example in milk (De Jonghe et al., 2010). Particularly B. weihenstephanensis strains are regarded food spoilage strains, because these psychrotolerant bacteria are capable of growth at refrigerator temperatures ≤ 7 °C (Lechner et al., 1998; Baron et al., 2007). After excessive vegetative multiplication and/or resporulation of B. cereus in food products, these are lost for human consumption and discarded, which completes their saprophytic life cycle during food spoilage. Besides food-borne illness, B. anthracis, B. cereus and B. thuringiensis strains have significant pathogenic potential to cause nongastrointestinal diseases. Bacillus cereus has been described in lethal nosocomial infections of immunosuppressed patients through contaminated ventilation equipment, catheters or linen and towels (Hernandez et al., 1998; Bottone, 2010). Bacillus cereus is also known for postsurgical and posttraumatic open-wound infections such as gunshot wounds, injection drug abuse, ground-contracted open-wound fractures and severe war wounds. Moreover, primary cutaneous infections by B. cereus are also described, resulting in skin lesions which resemble B. anthracis skin infections. Finally, inhaled B. cereus spores can cause serious life-threatening pulmonary infections. Therefore, vegetative B. cereus cells can colonize the human body in various ways, but always using toxin production to damage or invade the host tissues, for example the intestinal epithelium (food poisoning), eye (endophthalmitis), brain (meningitis, meningoencephalitis, subarachnoid haemorrhage and brain abscesses) and blood (bacteremia). Bacillus cereus produces multiple toxins (Nhe, Hbl and CytK) with haemolytic and cytotoxic activity due to pore formation in the cell membrane during vegetative growth (Beecher et al., 1995; Hardy et al., 2001; Lindbäck et al., 2004; Fagerlund et al., 2008). Enterotoxin EntFM, haemolysins, phospholipases C and other degradative enzymes are not directly cytotoxic, but they contribute to the cytotoxic and haemolytic activity of B. cereus and its adhesion to epithelial cells (Wazny et al., 1990; Asano et al., 1997; Firth et al., 1997; Beecher et al., 2000; Luxananil et al., 2003; Tran et al., 2010), resulting in detachment of the host's epithelial cells, microvilli damage, membrane damage, decreased mitochondrial activity and cell lysis (Minnaard et al., 2001; Ramarao & Lereclus, 2006).
In conclusion, B. cereus strains display a wide variety of different lifestyles, depending on the environmental context or the host and the specific characteristics of the strain (Fig. 3). This huge lifestyle diversity among B. cereus group strains is well reflected by the variety of contrasting encounters and relations we as humans have with these bacteria. On the one hand, some B. cereus strains and toxins are highly regarded if they can be applied as probiotics, biological plant protection agents and insecticides. On the other hand, B. cereus strains and toxins are combated if they can cause food poisoning, spoilage of food products, infections and anthrax disease.
Ecological significance of toxins for B. cereus
Diarrhoeal and emetic toxin genes are highly prevalent among B. cereus strains, even among the psychrotolerant strains that are not associated with food poisoning, such as B. weihenstephanensis (Thorsen et al., 2006; Baron et al., 2007). Despite the high prevalence of the toxin genes, the enterotoxins and the emetic toxin cereulide appear an exclusive characteristic of B. cereus (Ceuppens et al., 2011). Extensive regulation of toxin production exists in function of internal and environmental signals, but B. cereus only initiates toxin production after high population densities (> 106 CFU mL−1) have been obtained (Ceuppens et al., 2011). Toxin production usually starts under optimal growth conditions during the exponential growth phase, and it continues throughout the stationary phase under stressful or nutrient-limited growth conditions, independent of sporulation. However, toxin production has been frequently observed under suboptimal growth conditions as well, for example under anaerobic conditions and low oxido-reduction potentials (Zigha et al., 2007; Ouhib-Jacobs et al., 2009). Significant amounts of energy and resources are thus spent on toxin production. Between 80% and 92% of the proteins secreted by B. cereus were identified as potential virulence factors, for example toxins, degradative enzymes, adhesions and flagellins (Gohar et al., 2002; Clair et al., 2010). A closer inspection of these secreted proteins revealed that Hbl constituted 3–7% of the total protein, Nhe 3–11% and CytK up to 2%, which means that enterotoxins comprised 6–20% of the total secreted proteins (Gilois et al., 2007). Another study found that toxins and putative toxins constituted 32–39% of the secreted proteins during the early exponential growth phase under nutrient limitation (Clair et al., 2010). The exact amount of toxins produced significantly varies according to the strain, growth phase and growth conditions. However, toxin production clearly presents a considerable cost to the B. cereus cell, which would not be maintained unless it also yielded noteworthy benefits.
Bacillus cereus can change dramatically in its lifestyle, from soil bacterium to intestinal symbiont and enteric pathogen, in some cases while the same toxins are produced. Therefore, toxins probably play more than one role for B. cereus. Depending on the situation, toxins have a function in virulence and in inter-microbial competition. Firstly, toxins are of utmost importance in the pathogenesis, since B. cereus virulence is closely linked with toxin production (Stenfors-Arnesen et al., 2008; Bottone, 2010). Clinical strains tend to possess higher numbers of toxin genes, which are more frequently and stronger expressed in comparison with environmental and food isolates (Guinebretiere et al., 2002; Stenfors et al., 2002; Moravek et al., 2006; Hoton et al., 2009). For example, the overall prevalence of emetic B. cereus group strains is only 1.5%, but it is elevated to 32.8% among clinical and food isolates (Hoton et al., 2009). Similarly, the combination of enterotoxin genes nhe, hbl and cytK was found in 63% of the diarrhoeal food poisoning strains and in only 33% of the food strains (Guinebretiere et al., 2002). Bacillus cereus toxins clearly play a prominent role in pathogenesis. Secondly, toxins are used as weapons during the competition with other microorganisms for nutrients. Many bacterial, archaeal and fungal species are known to produce antibiotics and quorum-sensing inhibitors (O'Connor & Shand, 2002; Riley & Wertz, 2002; Rasmussen et al., 2005). By suppressing or killing competitors, toxins function as regulators of the microbial interactions within the microbial community or even within the own species population (Riley & Wertz, 2002). In this way, the toxin-producing bacteria can establish themselves in an existing microbial community; or to the contrary, they can protect their established place and niche in a microbial community against invasion and replacement by other strains or species. For example, colicin production increased the ability of E. coli strains to establish themselves in the intestinal microbial community of mice (Gillor et al., 2009). Furthermore, competitive exclusion of bacterial pathogens by bacteriocin-producing strains was demonstrated in food: enterocin AS-48–producing enterococci were used to control B. cereus and Staphylococcus aureus in milk and cheese (Munoz et al., 2004, 2007).
Cereulide, the stable emetic toxin from B. cereus, was shown to inhibit the growth of Gram-positive bacteria under alkaline conditions by depolarization of their cell membrane (Tempelaars et al., 2011). Moreover, cereulide production by emetic B. cereus was shown to increase their fitness in environments with low potassium availability by scavenging the cellular K+ from other bacteria (Ekman et al., 2012). As a result, emetic toxin production constitutes a competitive advantage of emetic B. cereus strains during growth in alkaline environments, such as decaying plant material in the soil and the insect gut. Furthermore, this experimental evidence suggests that the B. cereus emetic toxin acts as a bacteriocin in the sense that the growth of other B. cereus strains was inhibited, while other emetic strains were resistant (Tempelaars et al., 2011; Ekman et al., 2012). Cereulide shows structural and functional similarity to valinomycin, the ionophore produced by Streptomyces spp., which displays antimicrobial, antiviral, antifungal, insecticidal, nematicidal and cytotoxic activity (Park et al., 2008; Saris et al., 2009; Pimentel-Elardo et al., 2010). Interestingly, Streptomyces spp. also predominantly occur in soil and decaying vegetation. The cereulide synthase genes also share significant homology (30–40% sequence identity) with the genes for zwittermicin A synthesis, the broad-range antibiotic, which is produced by many B. cereus group strains isolated from soil under regulation by nutrients and plant exudates (Milner et al., 1995; Silo-Suh et al., 1998). In the rhizosphere, some B. cereus strains stimulate the growth of certain plants and offer protection against certain plant diseases by inhibiting plant pathogens. For example, B. cereus UW85 stimulated the growth of soybean and prevented damping-off by alfalfa, by zwittermicin A and kanosamine production (Handelsman et al., 1990; Bullied et al., 2002). In conclusion, cereulide nicely illustrates how the same toxin molecule confers vastly different effects in different contexts, namely illness of the human host after ingestion by induction of vomiting, increased fitness during growth in microbial soil populations and protection of the plant host by growth inhibition of competing microorganisms in its rhizosphere. Although B. cereus enterotoxins have not been described to possess antibiotic properties, but this could be due to the unstable nature and variable expression, which impairs the study of the ecological function of the enterotoxins. These toxins probably play additional environmental roles besides the degradation of host tissues during infection and food poisoning, since they are produced under diverse optimal and suboptimal growth conditions.
Perspectives for future research
Adhesion of B. cereus could be a third prerequisite for diarrhoeal food poisoning, after survival of gastrointestinal passage and intestinal toxin production. Future in vitro gastrointestinal simulation experiments could be slightly modified to study adhesion to mucus layers (Van den Abbeele et al., 2012). By inserting mucin-covered microcosms in the intestinal vessel, while keeping a duplicate intestinal vessel without mucin microcosms as luminal control, the potential of B. cereus to adhere to mucin and colonize the mucus layer under intestinal conditions can be investigated. It was shown that B. cereus was unable to establish itself within the indigenous community in the intestinal lumen (Ceuppens et al., 2012e) and that bile instantly inactivated the majority of vegetative B. cereus cells (Ceuppens et al., 2012d). Residence in the mucus layer could decrease the bile exposure, the competitive pressure exerted by luminal bacteria and the washout rate of B. cereus, which could ultimately contribute to enhanced intestinal survival. Since interaction between B. cereus and the host cells is suspected to play a prominent role during diarrhoeal food poisoning, the need arises to also include this aspect in the in vitro gastrointestinal simulation experiments. The experiments could be extended with the host–microbiota interaction (HMI) module, containing human cells in a separated compartment, connected to the intestinal vessel to investigate the reciprocal interactions between the gut microbiota and the host cells (Marzorati et al., 2011; M. Marzorati, pers. commun.). Sampling of the host cell medium would reveal whether diffusion of enterotoxin over the simulated gut wall occurs faster than the degradation of toxins by digestive secretions in the intestinal lumen. If sufficient toxins reach the host cells, adverse effects, such as membrane damage and lysis, should become apparent, confirming the hypothesis that local microorganism–host interactions cause diarrhoeal food poisoning.
The difference between pathogenic strains with high toxin expression and harmless low toxin-producing strains remains obscure, despite numerous worthy efforts to comprehend the B. cereus genome and its variability. Nevertheless, there is a great need for reliable and easy-to-use discriminatory tests for pathogenic B. cereus strains, enabling definition of microbial criteria for relevant B. cereus (food poisoning) strains in the food supply chain. Sequencing of clinical and environmental- or food-associated B. cereus strains and comparative genomic studies should be continued in future research to elucidate the source of the apparent phenotypic diversity in toxin expression and to identify marker genes or virulence mechanisms. Most likely, the expression level of enterotoxin genes discriminates virulent strains from harmless ones, but then, differences in toxin genes and/or their regulatory genes are expected to be found. Alternatively, it is also possible that additional toxins or virulence factors are expressed by food poisoning strains, which are currently unknown or unacknowledged. For example, new toxins EntA, EntB, and EntC, which were not regulated by the main regulator PlcR, were identified, and their putative role in gastrointestinal illness remains to be investigated (Clair et al., 2010). Finally, unknown indirect virulence factors could cause the difference in pathogenicity of B. cereus expressing identical toxins. For example, the expression of presently unknown adhesion-related proteins could be a condition sine qua non for well-known virulence factors such as enterotoxins Nhe, Hbl and CytK to exert their in vivo effect. As a result, it could be the overall combination of both direct and indirect virulence factors that results in full virulence and thus the identification of a discriminating genotype for diarrhoeal food poisoning. A similar approach could be used as in the study on the verocytotoxin-producing E. coli (VTEC), in which a statistical correlation between the presence of the eae, efa1 and paa genes and VTEC isolates from severe food poisoning cases was identified (Andersson et al., 2011). The complex regulation of toxin gene expression should remain the subject of future investigation to elucidate the source of the huge strain-dependent variability in toxin production under identical circumstances. The toxin gene expression should be further investigated under relevant conditions, namely during growth of B. cereus in biofilms, as a member of the complex intestinal community in the gastrointestinal environment and in the presence of human epithelial cells.
At first sight, huge differences exist between the beneficial and pathogenic B. cereus group strains, but a closer look reveals that many B. cereus isolates can produce various toxic compounds, of which many were probably originally intended for inter-microbial competition (see 'Ecological significance of toxins for B. cereus'). The difference lies in the outcome for the host: whether the host is protected by B. cereus due to suppression of other pathogens or the host itself becomes the target of B. cereus toxins. For some diseases, such as anthrax and emetic food poisoning, it is well established, which strains and toxins are responsible. However, for other illnesses, such as diarrhoeal food poisoning and (opportunistic) infections, a considerable knowledge gap remains and no responsible genotypes have been identified (see 'Pathogenic B. cereus strains'). In conclusion, B. cereus is a complex microorganism, which displays large variability in pathogenicity and ecological lifestyle (Figs 2 and 3). Therefore, investigating the virulence and the food poisoning capacity of these enigmatic bacteria requires a multidisciplinary approach, combining epidemiological, molecular, taxonomical, metabolical, microbiological, ecological and host–interaction studies.
The authors declare that they have no conflict of interest. This work was supported by the Special Research Funds of Ghent University (BOF) project B/09036/02 ‘Growth kinetics, gene expression and toxin production by B. cereus in the small intestine’.