Metabolic capacities and toxigenic potential as key drivers of Bacillus cereus ubiquity and adaptation

Abstract

Bacillus cereus is ubiquitous and is commonly found in a wide range of environments, including food. In this study, we analyzed 114 foodborne B. cereus strains isolated mainly from starchy and dairy products in order to investigate their phenotypic diversity (API system), antimicrobial resistance and toxigenic profiles (hblA, nheA, hlyII, cereolysin O, cytK2, cytK1 and EM1 genes). All isolates were confirmed as B. cereus using their 16–23S ribosomal DNA intergenic transcribed spacer (ITS) signature, and were shown to be Gram-positive, catalase and caseinase positive, hemolytic (97 %), and positive for lecithin hydrolysis and motility (97 and 87 %, respectively). PCR detection of B. cereus-specific toxin genes revealed occurrence rates of 100 % for cereolysin O, 98 % for nheA, 74 % for cytk2, 52 % for hblA, 28 % for hlyII, and the absence of cytK1. Only two strains (2 %), isolated from intestine of boar and pheasant, carried the emetic toxin genetic determinants (ces). The antimicrobial susceptibility of isolates was tested towards 15 different antimicrobial agents. We detected susceptibility of all strains to most antibiotics, intermediate resistance to clindamycin, and resistance to β-lactam antibiotics with 83 % of the resistant isolates producing β-lactamase enzyme. This large phenotypic diversity, combined with the toxigenic traits and antibiotic resistance, emphasize the high potential risk of food poisoning of B. cereus isolates. Additionally, a clear correlation between the metabolic features and the origin of isolation was shown. Most starchy isolates were able to hydrolyze starch while dairy strains were not able to produce amylases. Overall, our results reveal that metabolic flexibility and toxigenic potential represent the main drivers for B. cereus ubiquity and adaptation in a given ecological niche.

Introduction

Bacillus cereus is a constant problem due to its ubiquitous occurrence, nutritional versatility, ability to form endospores and to grow over a broad temperature range (Setlow and Setlow 1994). It is commonly found in a wide range of environments including foods (Bartoszewicz et al. 2008). Bacillus cereus causes two kinds of food poisoning: diarrheal and emetic syndromes (Drobniewski 1993). The emetic syndrome is characterized by nausea and vomiting, starting 1–5 h after ingestion and is associated mostly with farinaceous foods such as cooked rice (Valéro et al. 2002). This emetic type of gastrointestinal disease is caused by the cereulide, a dodecadepsipeptide structurally related to valinomycin (Finlay et al. 2002). It has been shown that the genetic determinants of this toxin are located on a 270 kb megaplasmid (Hoton et al. 2005). This toxin results from non-ribosomal peptide synthesis (NRPS) performed by the cereulide synthetase (ces) gene cluster (Ehling-Schulz et al. 2006).

The diarrhoeal syndrome is characterized by abdominal pain, profuse watery diarrhoea and rectal tenesmus, which occur between 8 and 16 h after ingestion of B. cereus-contaminated food, generally of a proteinaceous nature (Notermans and Batt 1998). At least three different enterotoxins seem to be involved in this syndrome: hemolytic toxin (Hbl), non-hemolytic toxin (Nhe) (McKillip 2000) and cytotoxin K (CytK) (Lund et al. 2000). These enterotoxins are encoded chromosomally and are produced during vegetative growth of B. cereus in the small intestine. Expression of these three enterotoxins is regulated positively by the pleiotropic PlcR/PapR quorum-sensing system (Agaisse et al. 1999; Slamti and Lereclus 2002; Gohar et al. 2002). Both Nhe and Hbl are three-component toxins while CytK is a single-component toxin that belongs to the family of β-barrel pore-forming toxins. Two forms of CytK have been characterized, namely CytK1 and CytK2. The amino acids sequence of CytK1 (the original and particularly cytotoxic CytK variant) displays 89 % identity with CytK2. The strain NVH391/98, where CytK was first identified (CytK1), was shown to be rather peculiar since it represents a cluster of thermophilic strains among the B. cereus group and was suggested to be recognized as a novel species “B. cytotoxicus” (Auger et al. 2008; Guinebretière et al. 2008, 2013; Lapidus et al. 2008). Another single-component toxin of B. cereus, namely HlyII (hemolysin II), which belongs to the family of oligometric β-barrel channel-forming toxins, is hemolytic and cytotoxic towards human cell lines (Andreva et al. 2007).

Bacillus cereus is ubiquitous in highly diverse ecological niches and often encountered in the food matrix. Indeed, B. cereus is able to multiply from biotic or abiotic soil fractions (Vilain et al. 2006) to the human gastro-intestinal tract, passing through food environments. This ecological flexibility is made possible by its metabolic versatility and is, in general, due to its genotypic and phenotypic diversity (Chaves et al. 2011; Chon et al. 2012). The species is known as panmictic, displaying a wide diversity of virulence factors (Helgason et al. 2004), growth behavior and survival characteristics (Choma et al. 2000). To face changing and sometimes hostile environments, B. cereus has to develop adaptive strategies by elaborating different physiological responses depending on the stress and habitat type encountered (Carlin et al. 2009).

The present study was undertaken to assess the occurrence of B. cereus in different food matrices (e.g., dairy and starchy products) and to elucidate their phenotypic, metabolic and toxicological profiles in correlation with their ecological features. The study was based on the investigation of phenotypic traits using the API system, biochemical tests and antibiotic resistance, and the use of PCR analysis to detect virulence-related genes and the presence of emetic strains and/or high virulent strains of B. cereus (B. cytotoxicus).

Materials and methods

Isolation of B. cereus strains

A total of 114 B. cereus strains was isolated from dairy, starchy and other food products according to the AFNOR method (AFNOR 1996) (Table 1). In brief, 10 g food was homogenized in a Stomacher with 90 mL sterile peptone water. Next 0.1 mL of the suspension was spread on the selective polymixin, mannitol, egg yolk, phenol red agar medium (MYP), containing 100 mg L⁻¹ polymixin B sulphate and 100 mL L⁻¹ egg yolk (Mossel et al. 1967). The plates were incubated at 30 °C for 24 h and colonies displaying a pink or purple color with an irregular edge surrounded by a white area were considered as positive. The isolates were confirmed subsequently as B. cereus by Gram staining, shape and biochemical tests. Cultures were then streaked in tryptone soybean agar (TSA) slants and stored refrigerated as pure cultures. Strain suspensions were maintained at −80 °C in culture medium supplemented with 20 % glycerol.

Table 1 Origin of Bacillus cereus isolates (n = 114)

Phenotypic characterization: biochemical and enzymatic tests and psychrotrophy

Biochemical tests based on the API 50CHB gallery were used to investigate the fermentation of 49 sugars, and the API 20E gallery was used for 12 additional tests (ONPG, ADH, LDC, ODC, CIT, H2S, URE, TDA, IND, VP, GEL and NIT) (Biomérieux, Marcy-l’Etoile, France) according to the manufacturer’s instructions. All strains were cultured on TSA at 30 °C for 24 h and suspended in each medium used as inoculates for the biochemical tests. Biochemical kits inoculated with suspended medium were incubated at 30 °C for 24 h. The biochemical profiles obtained from the API 50CHB and API 20E were analyzed using APILAB Plus software. Catalase and oxydase activities were also determined, the hemolytic activity in blood agar medium was determined as described by (Pruss et al. 1999), and lecithinase, caseinase and motility were evaluated. The overall results (18 characters) of the phenotypic characterization were statistically analyzed using the Jaccard coefficient and unweighted pair-group method with arithmetic mean (UPGMA) aggregation method, by multivariate statistical package (MVSP) software to construct a dendrogram. To determine the growth of B. cereus at a psychrotrophic temperature, all strains were inoculated onto nutrient agar followed by incubation at 7 °C up to 21 days. Isolates giving visual colonies at 7 °C within this time frame were considered to be psychrotrophic.

DNA extraction and PCR analysis

Total DNA was extracted using a modification of the Kalman’s method (Kalman et al. 1993; Cherif et al. 2003). DNA samples were stored at −20 °C until further use. In order to confirm the phenotypic identification of the isolates, the 16–23S rRNA internal transcribed spacers (ITS) were amplified by PCR using the primers S-D-Bact-1494-a-S-20 and S-D-Bact-0035-a-A-15 (Jensen et al. 1993) as described previously (Cherif et al. 2003).

Virulence-related genes of B. cereus were amplified using the primers listed in Table 2. This includes nheA, hblA, cytk2, cereolysin O and hlyII as described (Guinebretière et al. 2002; Yang et al. 2005), Other high virulence-related genes were also detected: the EM1 gene specific for the emetic toxin gene (Ehling-Schulz et al. 2004) and cytk1 gene for the detection of high virulent food poisoning strain (Guinebretière et al. 2006).

Table 2 Nucleotide sequences, locations and references of the B. cereus virulence-related gene PCR primers

The PCR reactions were performed using a thermocycler (Applied Biosystems, Foster City, CA) in 25 μL reaction volumes using 1 μL DNA extract, 1x Taq polymerase buffer, 2 mM final concentration of MgCl₂, 0.2 mM for each dNTP, 1 μM for each primer and 1 U Taq DNA polymerase. For the cereulide genetic determinants the PCR protocol started with a denaturation step for 15 min at 95 °C, followed by 30 cycles of 30 s at 95 °C, 30 s at 60 °C and 60 s at 72 °C, with a final elongation step at 72 °C for 5 min (Ehling-Schulz et al. 2004). PCR reaction condition for cytk1 and cytk2 genes consisted of an initial denaturation step at 94 °C for 5 min followed by 30 cycles of 95 °C for 15 s, 57 °C for 30 s and 72 °C for 30 s ended with a final extension at 72 °C for 7 min (Guinebretière et al. 2006). PCR amplification of nheA, cereolysin O, hblA and hlyII, and the 16–23S rRNA ITS, was performed with an initial denaturation step at 94 °C for 5 min followed by 30 cycles of 94 °C for 15 s, 55 °C for 45 s and 72 °C for 2 min completed with a final extension at 72 °C for 10 min (Hansen and Hendriksen 2001). The PCR products were then analyzed on a 1 or 2 % agarose gel containing ethidium bromide (0.5 μg mL^–1).

Antibiotic resistance

The Kirby-Bauer disk diffusion method (Drew et al. 1972) was used to evaluate the antibiotic susceptibility patterns of B. cereus isolates. All strains were grown in Mueller Hinton agar (Biokar Diagnostics, Beauvais, France) in the presence of various antimicrobial impregnated filter paper disks and incubated for 18 h at 37 °C. The used disks (Liofilchem, Roseto degli Abruzzi, Italy) and antibiotic concentration were as follows: amoxicillin-clavulanic acid (30 μg), ampicillin (10 μg), penicillin (10 U), cefotaxim (30 μg), oxacillin (1 μg), cefoxithin (30 μg), vancomycin (30 μg), ciprofloxacin (5 μg), gentamicin (10 μg), rifampicin (5 μg), amikacin (30 μg), tetracycline (30 μg), clindamicin (2 μg), erythromycin (15 μg) and streptomycin (10 μg). All isolates were classified as susceptible, intermediate susceptible, or resistant according to standard CLSI guidelines (Clinical and Laboratory Standards Institute 2010).

Results

Isolation and phenotypic characterization of foodborne Bacillus cereus

A total of 114 presumptive B. cereus were isolated on MYP (mannitol egg yolk polymixine) selective medium mainly from starchy and dairy products with the following percentages: starchy products (rice, pasta, potato salad, n = 49; 433 %), dairy products (ricotta, milk, butter, and leben, n = 37; 32 %) and other products (spices, animal intestine, honey, n = 28; 24 %; Table 1). All isolates were confirmed as Gram-positive, catalase positive, caseinase positive and oxydase negative, non psychrotrophic (unable to grow at 7 °C after 21 days) and 97 % of the isolates were shown to be hemolytic on blood agar with 74 % β-hemolytic and 26 % α-hemolytic. From the 114 B. cereus isolates, 111 (97 %) and 99 (87 %) strains were positive for lecithinase and motility tests, respectively. These isolates were further characterized and identified using the API 50CH/20E phenotypic system and APILAB Plus software. Most of the isolates belonged to the profile B. cereus 1 (65 %) and 33 strains (29 %) presented the API profile B. cereus 2. Five strains (4 %: RC11, RC33, S1, F1, BC922) showed a profile intermediate between B. anthracis and B. cereus 2, and two strains (RC35 and R35) were assigned to an intermediate profile between B. megaterium and B. cereus 2. The result of API 50CHB and API 20E characterization combined with other classic biochemical tests (motility, hemolysis and lecithinase) confirmed the identification of 100 % of the isolates as B. cereus. To further confirm this identification, all isolates were subjected to ITS-PCR targeting the 16–23S rRNA internal transcribed spacer (ITS). PCR results revealed identical patterns for all isolates, comprising two bands of 260 and 500 bp, which corresponds to the standard profile observed for the B. cereus group species (Wunschel et al. 1994).

The dendrogram generated from phenotypic patterns represented in Fig. 1 showed four groups (A, B, C and D). Groups A and B included ten isolates, originating mainly from ricotta (90 %) and characterized by their negative response to starch hydrolysis (90 %) and their incapacity to ferment either salicin (100 %) or esculin (60 %). Group C was composed of 73 isolates, dominated largely by strains from starchy products and spices. These isolates showed positive starch hydrolysis activity (99 %) and were able to ferment salicin (73 %). Group D was formed by 31 strains isolated mostly from dairy products. Isolates of this cluster were shown to be unable to degrade starch (100 %) and positive for salicin fermentation (64 %). Interestingly, from the eight isolates of the whole collection able to ferment lactose, six belong to group D, indicating that lactose degradation may be associated with dairy isolates.

Fig. 1

The unweighted pair-group method with arithmetic mean (UPGMA) dendrogram of foodborne Bacillus cereus isolates based on phenotypic traits analysis (API and biochemical tests). Toxin Toxic profile (from I to XI); Hemolysis hemolytic trait (β, α or non hemolytic); Origin/Biochem. origin of isolates (dairy or starchy) and biochemical tests salicin, starch hydrolysis and lactose fermentation. A RC35, RC34, RC20, RC14, R35, RC12; B RC30, RC32, RC17, RC9; C1 PMD, PMB, PMA; C2 IM1, LB3, R16; C3a BC847, MK1, R19, R25, R28, R27, R18, CO1, ML3, C3, CER5, PMC, ML2, COR2, COR1, R30, R39, R24, R20, R17, PA1; C3b C2, R13, R3, A1, R8,R34, R9, R7, BC848, BC87, P3, SRG1, CUM1, R6, P1, RC31, R41, BC117, BC134, R33, C4, R15, R10, CER3, CER6, R26, PA2, R23, R22, R21, MY1, BR2, R12, R11, PA3, R38, R36, R31, RC19, C1, R29, R14, R5, R4, RC27, RC7; D1 RC4; D2a RC4, RC33, R42, BC922, F1, S1, RC15, RC11, P2, RC10, RC22; D2b RC6, L5, L4, L3, ML1, R37, RC18, LB2, L2, R32, R2, LB1, RC26, RC25, RC3, L1, R40, A2, BR1, RC2

Virulence-related genes and toxicological potential of the B. cereus isolates

Using standard primers specific to the virulence-related genes of B. cereus (Table 2), all the PCR reactions generated fragments of the expected sizes. Cereolysin O was the most common gene and was detected in all the isolates. The nheA gene encoding protein A of the non-hemolytic enterotoxin Nhe, was found in 112 strains (98 %). With regard to hemolytic activities, the hblA gene encoding binding component B, which is one of three proteins composing the Hbl complex, was detected in 59 strains (52 %). The hlyII gene encoding hemolysin II was detected in 32 strains (28 %). For the cyt K gene, 84 strains were positive (74 %) using the general PCR primers (Guinebretière et al. 2002) and all were assigned to the cytK2 variant by specific primers (Yang et al. 2005). Only two isolates, S1 and F1, were positive for the emetic toxin gene EM1. These strains were isolated from intestine of boar and pheasant, respectively.

Combining these overall results, 11 different toxigenic patterns were identified with 2 patterns being strain-specific (LB2 and CER6 showing the X and XI patterns, respectively). Pattern IX was specific to the two emetic isolates (S1 and F1, Table 3). Three patterns were shown to be predominant. Pattern I contained 34 isolates (30 %) positive for the nheA and the cytK2 genes, and pattern II (n = 34, 30 %) included strains also positive for the nheA and cytK2 genes in addition to the hblA gene (hemolytic enterotoxin). Eleven strains (10 %) were positive for all the toxin genes except EM1 and formed toxigenic pattern III (Table 3). Five other toxin gene patterns were shown by a limited number of isolates. This includes pattern IV shown by eight isolates positive for nheA and hlyII; pattern V with six isolates positive for nheA, cytK2 and hlyII; and patterns VI, VII and IIX with isolates positive for nheA and variable for cytK, hblA and hlyII genes. No clear correlation was observed between the predominant toxigenic patterns and isolation origin.

Table 3 Summary of the toxigenic potential of the B. cereus isolates

Antibiotic susceptibility

All strains were susceptible to gentamicin, ciprofloxacin, streptomycin, vancomycin, amikacin and rifampicin, whereas 110 isolates (97 %) were susceptible to tetracycline and 100 strains (88 %) to erythromycin. Intermediate resistance was recorded for clindamycin (100 %) and for the remaining isolates for erythromycine (12 %) and tetracycline (3 %) (Table 4). Resistance profiles were recorded for the overall strain collection towards amoxicillin-clavulanic acid, penicillin, cefotaxim, oxacillin, cefoxithin and ampicillin. For the latter, β-lactamase production was tested using the cefinase test, which showed that 83 % of the isolates were positive. Antimicrobial resistance patterns found were not associated with the origin of isolation or type of food.

Table 4 Antibiotic susceptibility of B. cereus isolates. VA Vancomycin, GM gentamicin, CIP ciprofloxacin, AK amikacin, S streptomycin, RD rifampicin, TET tetracycline, E erythromycin, CD clindamicin, CTX cefotaxim, OX oxacillin, P penicillin, AMP ampicillin, AMC amoxicillin-clavulanic acid, FOX cefoxithin

Discussion

The present study phenotypically characterized 114 B. cereus strains, isolated mainly from starchy and dairy products, and evaluated their toxigenic potential and antimicrobial resistance. The assignment of the isolates to B. cereus species was based on the API system (50 CHB and 20E), biochemical tests and ITS-PCR analysis. With this latter approach, we detected a uniform ITS profile for all isolates, confirming their identification as belonging to the B. cereus group. The 16–23S rRNA spacer region remains a very valuable tool for PCR-based typing and bacterial identification. However, the method does not allow clear discrimination between B. cereus, Bacillus thuringiensis and Bacillus anthracis, which are still considered as very closely related species forming a unique taxonomic unit and distinguished only through their phenotypic and virulence specificities (Wunschel et al. 1994; Cherif et al. 2003). This taxonomic ambiguity is not resolved by the application of phenotypic and biochemical characterization. Indeed, the results obtained from the API characterization assigned five isolates, including the two emetic isolates, to intermediate profiles between B. cereus and B. anthracis. The overall results of the biochemical and phenotypic tests highlighted the huge diversity in the B. cereus collection and confirmed the inadequacy of the API system for identification purposes compared to molecular techniques (Ettoumi et al. 2009; Martínez-Blanch et al. 2011). Interestingly, this method, which is based on enzymatic potential was shown to be rather useful for correlating groups of isolates to their origin. The “B. cereus 1” API profile was recorded in strains isolated mainly from starchy products (86 %) whereas 76 % of the isolates originated from dairy products were affiliated to the “B. cereus 2” API profile. Furthermore, in the generated dendrogram (Fig. 1), the majority of dairy isolates (78 %), including six out of the eight lactose-positive strains, were located in groups A, B and D and corresponded to strains unable to hydrolyze starch. In contrast, the majority of starchy isolates (82 %) were located in group C corresponding to strains able to hydrolyze salicin and starch. These results indicate that the adaptation and the colonization of a given food matrix, or ecological niche, depends on the phenotypic and enzymatic traits that the bacterium displays. In our study, B. cereus expressing amylase genes clearly thrive in starchy products while those able to ferment lactose were encountered in dairy products.

The inability to hydrolyze starch and ferment salicin are considered as key traits for distinguishing emetic strains of B. cereus (Apetroaie et al. 2005; Ehling-Schulz et al. 2005). Indeed, we noted that the API biochemical profiles of the two emetic B. cereus isolates in our collection were characterized by their incapacity to hydrolyze starch or to ferment salicin. However, 36 % of the non-emetic strains showed these emetic-like features, suggesting that these traits could not discriminate emetic B. cereus strains. This is in agreement with the findings of Ankolekar et al. (2009) and Chon et al. (2012) who reported that starch hydrolysis phenotype alone is not useful for discrimination because 18 % (in both studies) of non-emetic strains were negative for the reaction and gave emetic-like phenotypes.

By using standard biochemical tests, we report the high occurrence of lecithinase activity (97 %) and motility (81 %), which is in agreement with the findings of Kim et al. (2010b), who showed that 100 % of their isolates were positive for these tests, and with the results of Chon et al. (2012), demonstrating that 100 % and 89 % of the strains were positive for motility and lecithinase, respectively.

Hemolysis was shown to be a widely spread characteristic, with 97 % of the strains being hemolytic. Several authors reported similar percentages, such as Kim et al. (2010b) with 93 % hemolytic, Chon et al. (2012) with 89 %, Chaves et al. (2011) with 92 % and Abriouel et al. (2007), who reported 100 % of their strains to be hemolytic. There is no coherence between the hemolytic phenotype and the hemolysis gene content of the isolates. The three non-hemolytic strains in our collection amplify with at least one of the hemolysin primers targeting hblA, hlyII or cereolysin O. This could be due to the non-expression of these genes or their non-functionality because of silencing mutations or deletion of transcriptional regulators, as suggested in previous studies (Van der Voort et al. 2008).

Besides the large diversity among the B. cereus isolates collection as revealed by phenotypic tests, several strains exhibited identical phenotypes but showed diverse virulence-related genes content, indicating their intraspecific genetic diversity; for instance BC847 and MK1, C4 and R15, BR2 and R12, and R14 and R5 isolates. In total, 11 distinct toxigenic profiles were detected based on seven toxin genes with no clear correlation with isolation origin except for emetic strains retrieved exclusively in animal intestines. The relative frequencies of these genes were in agreement with several reports in literature. A high occurrence was observed for cereolysin O (100 %), nheA (98 %) and cytK2 (74 %)—frequencies very comparable to the reported results of many authors with B. cereus collections from diverse origins (Hansen and Hendriksen 2001; Ankolekar et al. 2009; Swiecicka et al. 2006; Chon et al. 2012; Ouoba et al. 2008; Godic Torkar and Seme 2009). Conversely, the hblA and hlyII genes were shown to be less frequent (52 and 28 %, respectively). Hence, the genes of the HBL complex are generally less common than those of the NHE complex (Al-Khatib et al. 2007; Granum 2007; Moravek et al. 2006), while hlyII exhibits a relatively low frequency (Shadrin et al. 2007). The remaining two virulence-related genes screened in this study (cytK1 and EM1) are known to be present only rarely in B. cereus. The cytk1 gene was not detected in any of the 114 isolates. According to Guinebretière et al. (2010), strains harboring this gene belong to a distant phylogenetic group that may represent a novel subspecies designed as B. cytotoxicus. The absence or very low frequency of these cytK1 positive strains was further described by Carlin et al. (2009) and Lapidus et al. (2008). Although cytK2 is not as toxic as cytK1, its potential to cause disease cannot be underestimated (Fagerlund et al. 2010).

The emetic genetic determinant EM1 was detected in only the two strains S1and F1 isolated from animal intestines. This very low occurrence, or absence, of emetic phenotype was previously reported for B. cereus collections from diverse origins (Svensson et al. 2006; Altayar and Sutherland 2006; Chon et al. 2012) and particularly in food-related isolates (Ouoba et al. 2008; Ankolekar et al. 2009; Chaves et al. 2011). The rationale behind the selection of boar and pheasant for the isolation of B. cereus was that the digestive tract of small mammals can be a reservoir of cereulide-producing strains (Hoton et al. 2009), which is confirmed in the present study. As described in other works, these emetic isolates also harbored putative enterotoxin genes, which could reinforce their toxigenic potential (Kim et al. 2010c; Chon et al. 2012). The virulence similarities of these isolates confirm the clonal nature of the emetic phenotype. This B. cereus pathogen variant is in fact considered as a single evolutionary lineage of closely related strains (Pirttijärvi et al. 1999; Ehling-Schulz et al. 2005).

Together with this toxigenic potential, our isolates were shown to possess antibiotic resistance traits, particularly to β-lactam antibiotics. This natural resistance, based on the production of β-lactamase, is well known and described in other studies (Andrews and Wise 2002; Park et al. 2009). However, the cefinase test results revealed that 17 % of the β-lactam resistance may be mediated by mechanisms other than enzymatic inactivation by β-lactamases. The susceptibility of all isolates to gentamicin, ciprofloxacin, streptomycin, vancomycin, amikacin, rifampicin and the intermediate resistance of some isolates to tetracycline and erythromycin, is in agreement with previous studies (Jensen et al. 2001; Roy et al. 2007; Banerjee et al. 2011). There were no differences between resistance rates of isolates from different food matrix origins nor were there reported differences between food and clinical isolates (Godic Torkar and Seme 2009). Finally, we noted that 100 % of the isolates showed intermediate resistance for clindamicin, which represents a relatively high proportion compared to previous works such as that of Ankolekar et al. (2009), Kim et al. (2011) and Chon et al. (2012). This intermediate resistance to clindamicin, and for some isolates toward erythromycin (12 %) or tetracycline (3 %), indicates the possible emergence of new multi-drug resistance patterns in foodborne B. cereus strains, which may constitute a sanitary risk.

In this study, we investigated the toxin profiles, phenotypic and metabolic patterns, and antibiotic resistance profile of B. cereus isolates from different food matrices (mainly starchy and dairy products). Our overall results indicate the huge intraspecific diversity of the foodborne collection. Most B. cereus isolates harbored the genetic determinants of several potential enterotoxins. Furthermore, B. cereus isolates demonstrated natural antibiotic resistance, indicating potential food poisoning and sanitary risks. The observed phenotypic diversity was correlated clearly to isolation origin, particularly for starch hydrolysis. These findings reinforce the postulate that metabolic flexibility and toxigenic potential represent the main drivers for B. cereus ubiquity and adaptation in a given ecological niche, particularly in food.