Assessment of CcpA-mediated catabolite control of gene expression in Bacillus cereus ATCC 14579

Background The catabolite control protein CcpA is a transcriptional regulator conserved in many Gram-positives, controlling the efficiency of glucose metabolism. Here we studied the role of Bacillus cereus ATCC 14579 CcpA in regulation of metabolic pathways and expression of enterotoxin genes by comparative transcriptome analysis of the wild-type and a ccpA-deletion strain. Results Comparative analysis revealed the growth performance and glucose consumption rates to be lower in the B. cereus ATCC 14579 ccpA deletion strain than in the wild-type. In exponentially grown cells, the expression of glycolytic genes, including a non-phosphorylating glyceraldehyde-3-phosphate dehydrogenase that mediates conversion of D-glyceraldehyde 3-phosphate to 3-phospho-D-glycerate in one single step, was down-regulated and expression of gluconeogenic genes and genes encoding the citric acid cycle was up-regulated in the B. cereus ccpA deletion strain. Furthermore, putative CRE-sites, that act as binding sites for CcpA, were identified to be present for these genes. These results indicate CcpA to be involved in the regulation of glucose metabolism, thereby optimizing the efficiency of glucose catabolism. Other genes of which the expression was affected by ccpA deletion and for which putative CRE-sites could be identified, included genes with an annotated function in the catabolism of ribose, histidine and possibly fucose/arabinose and aspartate. Notably, expression of the operons encoding non-hemolytic enterotoxin (Nhe) and hemolytic enterotoxin (Hbl) was affected by ccpA deletion, and putative CRE-sites were identified, which suggests catabolite repression of the enterotoxin operons to be CcpA-dependent. Conclusion The catabolite control protein CcpA in B. cereus ATCC 14579 is involved in optimizing the catabolism of glucose with concomitant repression of gluconeogenesis and alternative metabolic pathways. Furthermore, the results point to metabolic control of enterotoxin gene expression and suggest that CcpA-mediated glucose sensing provides an additional mode of control in moderating the expression of the nhe and hbl operons in B. cereus ATCC 14579.


Background
Bacillus cereus is an important Gram-positive, spore-forming food-borne pathogen. Many strains cause either an emetic or a diarrhoeal type of disease. The production of emetic toxin in foods, also referred to as cereulide, may cause nausea and vomiting. The diarrhoeal type of disease is associated with the production of enterotoxins in the intestines and may involve Nhe, Hbl and CytK [1][2][3]. Food-borne disease caused by B. cereus is generally characterized by mild symptoms. However, recently more severe cases with a lethal outcome have been described [4,5]. B. cereus can also be the causative agent of other diseases, such as periodontitis, fulminant endophthalmitis, and meningitis in immuno-compromised patients [1,[6][7][8]. B. cereus is ubiquitously found in the environment, including in soil. Therefore, the transfer to food is not surprising and causes many problems [1]. In nutrient-rich environments, such as food, B. cereus shows low generation times putatively gaining advantage from its capacity to use various carbohydrates and proteinaceous substrates [9]. The regulation of gene expression plays an important role in the efficient selection of the preferred carbon and energy source for growth. Annotation of the genome of B. cereus ATCC 14579 predicted the regulation of gene expression to be highly complex involving over two hundred transcriptional regulators managing its 5370 open reading frames (ORFs) [9,10]. One of these putative regulators is the catabolite control protein CcpA, which is a member of the LacI-family of transcriptional regulators. CcpA and the regulatory mechanism of the catabolite repression are highly conserved in low-GC Gram-positives [11]. B. cereus ATCC 14579 CcpA shows 77% identity with B. subtilis CcpA. Furthermore, CcpA in B. subtilis has been shown to have a role in optimizing glucose metabolism and the underlying regulatory mechanisms have recently been reviewed [12][13][14]. Regulation of gene expression by CcpA is mediated by its binding to DNA at a specific cis-binding sequence, the Catabolite Responsive Element (CRE) [14][15][16].
In recent years the regulon of B. subtilis CcpA has been studied extensively by transcriptome analyses, revealing genes and operons under direct and indirect control of CcpA [17][18][19][20]. Furthermore, Moreno et al. [21] showed a clear correlation between the glucose-repressed genes and the presence of predicted CRE-sites. Moreover, they showed CcpA-mediated glucose-independent regulation of expression [21]. Other organisms for which the role of CcpA in carbon metabolism was established are Lactobacillus acidophilus [22] and Lactococcus lactis [23]. Recently, a role for CcpA in the control of virulence of Staphylococcus aureus [24], Streptococcus pneumoniae [25], and Clostridium perfringens was reported [26] and reviewed [27].
Notably, comparative genomics of the different species of the B. cereus group revealed reduced capacity to metabolize carbohydrates and increased potential for protein metabolism as compared to B. subtilis [28,29]. Here we report on the role of CcpA in regulation of metabolism and virulence in B. cereus ATCC 14579.

Results and Discussion
Growth and glucose utilization of the ccpA deletion strain compared to the wild-type Growth of the wild-type and the ccpA deletion strain was assessed under aerobic conditions in BHI containing 2 g/ L D-glucose, and revealed specific growth rates (μ) of 0.024 and 0.022 (h -1 ), respectively (Fig. 1). Statistical analysis showed the growth rates of the wild-type and deletion strain to be significantly different. Assessment of glucose concentrations at different time points during growth revealed a reduction in the glucose consumption rate for the ccpA deletion strain compared to that of the wild-type (Fig. 1). The glucose concentration at earlyexponential growth for both the wild-type and the ccpA deletion strain was around 1.2 g/L, and glucose was still available in the mid-exponential growth phase. Notably, glucose was depleted in the wild-type culture upon entry into the transition phase, whereas glucose depletion in the ccpA deletion strain culture was only observed in the stationary phase of growth.

Overview of the transcriptome data
Analyses of the transcriptome data of the ccpA deletion strain compared to the wild-type in samples taken at the four time points indicated in figure 1, showed expression of a large number of ORFs to be affected upon ccpA deletion. Remarkably, the number of genes differentially expressed increased from 147 at early exponential phase to over 700 genes in the stationary phase (Fig. 2). The large differences in gene expression in transition and stationary phase cells of the ccpA deletion strain compared to that of the wild-type are conceivably affected by the respective presence and absence of glucose. Furthermore, these differences may point to the initiation of secondary effects of the ccpA deletion in these growth phases on gene expression and consequently cellular performance. Therefore, transcriptome analysis was focused on the early-and mid-exponential phase samples where glucose is still present at high levels in the cultures of both the ccpA deletion strain and the wild-type. In early-exponential phase, 103 genes showed higher expression and 44 genes showed a lower expression in the ccpA deletion strain compared to the wild-type. For mid-exponential phase cells these numbers were 127 and 54, respectively. When corrected for overlap between regulated genes in the earlyand mid-exponential phase, a total of 173 genes expres-sion in the exponential phase was shown to be higher in the ccpA deletion strain compared to the wild-type and for 80 genes expression in the ccpA deletion strain was observed to be lower than in the wild-type (Fig. 2). Consequently, genes that show a higher expression in the ccpA deletion strain are possibly repressed by CcpA, whereas genes that show a lower expression in the ccpA deletion strain may be activated by CcpA in the wild-type. The number of genes putatively regulated are similar to those described for CcpA-regulated genes in B. subtilis [17][18][19]21].
Expression ratios of six randomly chosen genes with significantly altered expression in the ccpA deletion strain (cggR, acoR, gapB, ymfC, fruR and odhA) were quantified using qPCR. Expression ratios obtained by microarray analysis were compared to ratios obtained by qPCR. The tested genes showed the same trend in expression, with differential expression only slightly more pronounced in qPCR experiments, a feature observed before in such comparisons [30]. This indicates the microarray platform to be suited for gene expression analysis (for qPCR data see Additional file 1).

Identification of the B. cereus CRE-site consensus and in silico analysis
Direct regulation exerted by CcpA is indicated by the presence of a CRE-site. By a series of subsequent alignments searches, the B. cereus ATCC 14579 consensus CRE-site was identified and visualized in Figure 3 (for details see Methods). This showed the consensus sequence to consist of only 16-bp for B. cereus ATCC 14579, WWGWAARCG-WWWWCAW, whereas an 18-bp consensus sequence was reported for B. subtilis, WWTGNAARCGNWWWCAWW, [15]. Nevertheless, a high similarity between these consensus sequences exists. The B. cereus ATCC 14579 genome was scanned with the obtained CRE-site consensus ( Fig. 3), identifying the putative CRE-sites (see Additional file 2). A number of 76 putative CRE-sites were identified for 83 out of 173 genes that show a higher expression in the ccpA deletion strain compared to the   nr. of genes wild-type in exponential phase cells, and these 83 genes are part of 49 putative operons [9,31]. Furthermore, 21 putative CRE-sites were identified in the promoter region of 18 out of 80 genes with a lower expression in the ccpA deletion strain (see Additional file 2) in exponential phase cells, these 18 genes are part of 17 putative operons [9,31]. These results are in agreement with the general observation that activation by CcpA occurs less frequently than repression by CcpA [21]. For B. subtilis not all identified putative CRE-sites could be shown to be functional [15] and the predicted CRE-sites in B. cereus should therefore be regarded putative.

Analysis of genes differentially expressed in the ccpA deletion strain
The COG annotation reveals that the differentially expressed genes are mainly involved in metabolic processes ( Fig. 4), including glycolytic and gluconeogenic genes and genes encoding the citric acid cycle (Table 1). For approximately 35 differentially expressed genes a homologous gene was identified to be CcpA regulated in B. subtilis [32]. Of these genes a total number of 17 genes were annotated to have a function in the glucose metabolism ( Table 1). Genes that were first identified in this study to be putatively regulated by CcpA with an apparent and interesting function are discussed below.
One of the glycolytic genes found to be expressed lower in the ccpA deletion strain was yfmT. This gene encodes a non-phosphorylating glyceraldehyde-3-phosphate dehydrogenase (GAPN), for which 2 putative CRE-sites could be identified. We propose to rename yfmT as gapN, since this gene shows similarity with gapN of the other members of the B. cereus group, and with gapN of B. halodurans, Streptococci and Clostridia [33]. Notably, the gene is lacking in B. subtilis [29,33]. Next to the non-phosphorylating glyceraldehyde-3-phosphate dehydrogenase, two phosphorylating glyceraldehyde-3-phosphate dehydrogenases encoding genes (gapA and gapB) are present on the B. cereus ATCC 14579 genome. The function of GAPN in microbial metabolism is yet unclear. Recently, Asanuma and Hino (2006) showed a role for CcpA in expression control of gapN in Streptococcus bovis and proposed that NADPH is provided by GAPN activity for NADPHdependent biosynthetic reactions, thereby maintaining an optimal redox balance at the same time. However, in hyperthermophilic archaea GAPN has been shown to play a role in accelerating glycolysis as it can produce 3-phospho-D-glycerate from D-glyceraldehyde 3-phosphate in a one-step reaction, instead of two steps [34]. As a putative drawback, ATP is not produced in this one-step reaction. The exact role of GAPN in B. cereus metabolism remains to be elucidated.
The gluconeogenic genes expressed higher in the ccpA deletion strain included ywjI and gapB. The ywjI gene is part of the putative murA2-ywjI operon and is annotated as a fructose-1,6-bisphosphatase. The ywjI gene of B. subtilis, that shows similarity to ywjI of B. cereus, is annotated to be a fructose-1,6-bisphosphatase, as a member of the glpX-family. The other annotated fructose-1,6-bisphosphatase gene for B. subtilis, fbp, which is no member of the glpX-family, is not regulated by CcpA and deletion of this gene seems to have no effect on gluconeogenesis [35]. A corresponding fbp homologue appears to be absent in B. cereus. However, on the B. cereus genome two members of the glpX-family of fructose-1,6-bisphosphatases are present, of which ywjI expression was found to be affected by ccpA deletion, accordingly a putative CRE-site was identified in the promoter region of its operon. The gapB gene encodes a NAD(P)-dependent glyceraldehyde-3phosphate dehydrogenase. Three putative CRE-sites were identified for gapB in B. cereus (Table 1). In contrast, no CRE-sites were identified for gapB in B. subtilis, and its expression was shown to be only indirectly affected by ccpA deletion [36]. Furthermore, gluconeogenesis in B. subtilis was shown to be regulated by the transcriptional regulator CcpN [37,38]. CcpN is also present on the genome of B. cereus ATCC 14579 and putative CcpN binding sites can be found in front of the gluconeogenic genes encoding GapB and PckA [37]. Combined with our results this would point to dual control of the expression of gluconeogenic genes in B. cereus by CcpN and CcpA.
The expression of genes with functions in routing metabolism towards the citric acid cycle appeared to be higher in the ccpA deletion strain compared to the wild-type and included ~30 genes involved in amino acid catabolism. This included the bkd-operon (BC4163 to BC4157) that is common among the Bacilli, and that is involved in the catabolism of Valine and Leucine [39]. One putative CREsite in front of the bkd-operon and two within the operon were identified in B. cereus ATCC 14579. Notably, CcpA was previously shown to play no role in the regulation of expression of this operon in B. subtilis [40]. In contrast, the ilv-leu-operon of B. subtilis coding for biosynthesis of branched chain amino acids has been shown to be activated by CcpA [41], whereas no involvement of CcpA was found in regulation of expression of this operon in B. cereus under the conditions tested. The fact that no regulation of the ilv-leu-operon was observed, although its expression on the microarray was apparent (data not shown), could be due to a complex regulation of this operon under the tested circumstances involving CodY and TnrA [41,42]. Expression of several genes encoding enzymes involved in catabolism of nucleosides was higher in the B. cereus ccpA deletion strain, this included the putative yuf-operon (BC3791 to BC3788) encoding nucleoside transport under the apparent control of the  [18,19,21]. # Genes of which effects of ccpA deletion on expression were shown to be indirect in B. subtilis [18,19,21]. * pgk is represented on the microarray by two different amplicons, as Pgk is annotated in the ERGO database to be encoded by two genes.
¶ Numbers indicate the order of genes within one operon. † Ratios are presented as expression in the ccpA deletion strain compared to the wild-type (WT), expression in the wild type is set to 1. Early-and mid-exponential phase data are shown. ‡ In this column it is indicated whether a putative CRE-site was identified for the gene. Numbers indicate the number of putative CRE-sites identified, a + indicates that a gene is in the same operon as a gene for which a putative CRE-site was identified, and a -indicates no putative CRE-site could be identified.
transcriptional regulator encoded by the also higher expressed ymfC (BC3792), for both the regulator and the yuf-operon putative CRE-sites could be identified (see Additional file 2). Interestingly, expression of two putative operons with an apparent function in catabolism was higher in the ccpA deletion strain, and possible functions were assessed using in silico analysis. The first operon (BC0378-BC0380), which is not present in B. subtilis, consists of three genes and based on phylogenetic studies of the protein sequences it is suggested to annotate the genes as an L-fuculose phosphate aldolase [43] (BC0378), an Lfucose and/or arabinose isomerase (BC0379) and an Lfuculose phosphate aldolase (BC0380). This suggest a function for this operon in the catabolism of fucose and/ or arabinose. Consequently, names and functions that we suggest are fclI for the L-fucose isomerase (BC0379), fclK for the L-fuculokinase (BC0378) and fclA for the L-fuculose phosphate aldolase (BC0380) (Fig. 5). A putative CRE-site could be identified in front of this operon. The second operon consists of three genes and in silico analysis of this operon (BC1739-1741) suggested the transporter protein (BC1739) to function as a proton/sodium-aspartate symporter [44], as an alternative to gltT. Subsequently, aspartate can be converted into ammonia and fumarate by the aspartate ammonia lyase (BC1740), after which fumarate can be metabolized to malate by one of the two fumarate hydratases encoded on the genome of B. cereus. The gene BC1741 encodes a malic enzyme, MalS, which is responsible for converting malate into pyruvate. This suggests that the operon encodes enzymes involved in the catabolism of aspartate to pyruvate. This operon is unique for the species of the B. cereus group, as the genes are not found as an operon in genomes of other bacteria. A putative CRE-site was identified for the first gene of the operon (Fig. 5). The two-component system (BC1742-BC1743) next to this operon shows high similarity with the two-component system GlnK-GlnL of B. subtilis, which is involved in glutamine utilization [45]. Whether BC1742-BC1743 functions as a two-component system, involved in aspartate sensing and triggering expression of enzymes involved in its metabolism, remains to be elucidated.
Glutamate is a product of protein and amino acid catabolism and acts as link between nitrogen and carbon metabolism, mediated by glutamate synthase and glutamate dehydrogenase, as described for B. subtilis [46,47]. The B. subtilis genome contains a glutamate synthase (gltAB) and two glutamate dehydrogenases (rocG and gudB) of which the GudB dehydrogenase appears to be inactive [47]. Glutamate biosynthesis in B. subtilis has recently been shown to be tightly regulated by interaction of RocG and GltC [48], but these genes are not encoded on the B. cereus genome. The microbial glutamate synthases generally belong to the NADPH-GltS family of glutamate synthases and are encoded by two genes [49]. In the glutamate synthase reaction L-glutamate is produced from L-glutamine and 2-oxoglutarate, with the latter compound derived from the citric acid cycle [49]. The gltAB operon of B. subtilis is suggested to be regulated by CcpA [46,50]. On the cereus encoding a glutamate dehydrogenase, and its expression was observed to be clearly higher in the ccpA deletion strain (Table 1). Glutamate dehydrogenase is responsible for the reaction from L-glutamate to 2-oxoglutarate linking L-glutamate to the citric acid cycle [51]. Higher expression of gudB in the ccpA deletion strain compared to the wild-type, together with the fact that it is the only glutamate dehydrogenase annotated on the genome of B. cereus suggests that GudB is the active glutamate dehydrogenase for B. cereus. This is supported by the fact that the 9-bp sequence, encoding the 3 amino acids causing the inactivity of GudB in B. subtilis [47], are absent in the gudB sequence of B. cereus ATCC 14579. Two putative CRE-sites in front of and one within gudB were identified pointing to CcpA-controlled expression in B. cereus.
B. subtilis contains a large number of carbohydrate catabolic pathways [29], whereas the number of these pathways is limited in B. cereus. The observed deficiency of the B. subtilis ccpA deletion strain in growth with ammonium as the sole nitrogen-source, has been attributed to the regulation of the gltAB genes by CcpA [46,50] and the readthrough transcription of the rocG gene [47]. The lack of regulation of gltA in B. cereus by CcpA, together with a similarity to the B. subtilis gltA of only 46%, offers an explanation for the observation that the ccpA deletion strain of B. cereus is able to grow with ammonium as the sole nitrogen source (data not shown), conceivably with GudB as the active glutamate dehydrogenase.
The role of CcpA in optimisation of glucose metabolism is apparent, since glycolytic enzymes were expressed lower and expression of genes encoding citric acid cycle enzymes was shown to be higher in the ccpA deletion strain compared to the wild-type. Up-regulation in the ccpA deletion strain was found for ~30 genes encoding enzymes involved in protein, peptide and amino acid metabolism (see Additional file 1). This is seemingly in contrast with the proposed preferred use of proteinaceous Putative CcpA-controlled metabolic pathways in B. cereus substrates for growth of B. cereus, a hypothesis put forward by Ivanova et al. [9]. This hypothesis was supported by the annotation of a large number of genes encoding proteolytic enzymes, a multiplicity of peptide and amino acid transporters, and a large variety of amino acid degradation pathways. However, under the conditions tested glucose is used as the carbon and energy source (Fig. 1) and genes encoding enzymes involved in protein, peptide and amino acid catabolism appear to be subject to catabolite repression, indicating that under the conditions tested, glycolysis rather than protein catabolism is the preferred energy generation route of B. cereus ATCC 14579.

CcpA-mediated catabolite control of enterotoxin gene expression in B. cereus
In early-and mid-exponential phase cells the expression of the nhe-operon (BC1809-BC1811), coding for the nonhemolytic toxin Nhe [52], was shown to be higher in the ccpA deletion strain compared to the wild-type. This higher expression was even more apparent in the stationary phase where an almost 20-fold stimulation was observed (Fig. 6). Furthermore, a putative CRE-site was identified for this operon. For the hbl-operon (BC3104-BC3102), coding for the HBL enterotoxin [53], an 8-fold increase in the ccpA deletion strain was seen in the stationary phase and a putative CRE-site could be identified (Fig.  6). Differential expression of cytK, encoding the enterotoxin Cytotoxin K [4], was not observed under the conditions tested, and no putative CRE-site could be identified. These results indicate involvement of CcpA in the catabolite control of the expression of at least two major enterotoxins in B. cereus ATCC 14579. Many bacteria use glucose as their preferred carbon and energy source as it can be obtained from plant-derived polysaccharides, which can be found in many environments including foods. We hypothesize that glucose can function as a signalling molecule enabling CRE site-mediated, CcpA-dependent repression of gene expression for microorganisms such as Lactobacillus plantarum [54] and B. cereus that occur in several environments, including soil, plant and the human gastro intestinal (GI) tract. In the GI tract, in the absence of glucose, CcpA-dependent repression is thus lifted, and this allows for induction of pathways involved in catabolism of other substrates present in the GI tract such as fucose, a major degradation product of mucus related fucosylated glycans [55]. Notably, results of this study suggest the putative fucose utilisation operon of B. cereus to be repressed by CcpA in the presence of glucose, and this pathway may thus be activated in the GI-tract under glucose limiting conditions. A similar mode of regulation seems applicable for the nhe and hbl enterotoxin operons, for which the combination of transcriptome data and the identification of putative CRE-sites, suggest these to be CcpA controlled as well. Our results offer an explanation for the observations recently made by Ouhib et al. [56], who showed that final enterotoxin levels reached, depended on the carbon and energy sources used for growth, with the lowest levels reached with cells grown on glucose. The anaerobic growth transcriptional regulators FNR and ResD have recently been found to play a role in enterotoxin production also, next to PlcR [57][58][59], and combined with our data, this points to an elaborate control of enterotoxin production in B. cereus at the transcriptional level based on a variety of input signals mediated by a range of transcriptional regulators. Notably, CcpA has been reported also to play a role in toxin formation in C. perfringens [26], and in S. aureus deletion of ccpA was shown to affect the expression of the virulence gene hla, encoding α-hemolysin [24].

Conclusion
B. cereus ATCC 14579 catabolite control protein CcpA is involved in optimizing the catabolism of glucose with concomitant repression of a range of metabolic pathways. Furthermore, CcpA-mediated glucose sensing is shown to provide an additional mode of control in moderating the expression of the nhe-and hbl-operon in this human pathogen.

Methods
Bacterial strains, culture media, growth conditions, and genetic methods B. cereus ATCC 14579 and its ccpA deletion strain FM1403 were cultured in brain heart infusion broth (BHI, Becton and Dickinson, The Netherlands) medium at 30°C, with shaking at 200 rpm. The growth of the culture was monitored by measurement of the optical density at OD 600 . Dglucose concentrations were measured by use of a D-glucose measuring kit (Boehringer). Growth experiments and glucose measurements were performed in three fold. Plasmid DNAs were purified from E. coli with a Qiaprep Spin Miniprep kit (Westburg, Leusden, The Netherlands). Pwo polymerase (Roche Diagnostics, Almere, The Netherlands) was used for PCR generated fragments that were used in cloning and Taq polymerase (Fermentas, Amersfoort, The Netherlands) was used in control PCRs. E. coli HB101/pRK24 [60] was used as the donor host in conjugation experiments. The antibiotics used were ampicillin (Sigma, Zwijndrecht, The Netherlands) at a concentration of 50 μg/ml, kanamycin (Sigma) at a concentration of 70 μg/ml, erythromycin (Sigma) at a concentration of 150 μg/ml (for E. coli) or 5 μg/ml (for B. cereus), spectinomycin (Sigma) at a concentration of 100 μg/ml, and polymyxin B (VWR, Amsterdam, The Netherlands) at a concentration of 50 μg/ml for counter-selection against E. coli upon conjugation.

Construction of ccpA deletion strain
To construct a double cross-over deletion strain of ccpA, an ~3.5-kb PCR product, comprising ccpA and 1-kb flank-ing regions was obtained by use of forward primer ccpA-KOsacIforw (TCgagctcAGATTACGTTGATGTTATTC) and reverse primer ccpAKOxbaIrev (TGtctagaAGAAGAA-GAAAAAGAGGAAGAAAT). This PCR product was cloned into pGemT-easy (Promega, Leiden, The Netherlands) according to the manufacturer's protocol resulting in pGemTccpA. Subsequently, an erythromycin-resistance cassette amplified from pUC18ERY [61] with forward primer ErycasFBsrGI (TCtgtacaGTCCGCAAAA-GAAAAACG) and reverse primer ErycasRClaI (TCatcgat-CATACCTAATAATTTATCTAC) was cloned into pGemTccpA after digesting both with Bsp1407I (BsrGI) and Bsu15I (ClaI) (Fermentas). The insert of the resulting plasmid, comprising the 1-kb flanking regions and the erythromycin-resistance cassette was cloned into the conjugal vector pATΔS28 [62] by digestion of the insert and the vector with XbaI and SacI (Fermentas). The resulting plasmid pATΔCcpAery was isolated from DH5α and transformed into E. coli HB101/pRK24. The resulting strain was used in a conjugation experiment with B. cereus ATCC 14579 following established procedures [63]. Transconjugants were obtained by selection for spectinomycin sensitivity and erythromycin resistance and one was analysed in comparison with the wild-type strain. PCR and Southern Blot analysis confirmed the deletion of ccpA by double homologous recombination (data not shown). The B. cereus ccpA deletion strain was designated B. cereus FM1403.

RNA isolation
RNA was extracted from both the ccpA deletion strain and the wild-type at four time points in the growth curve at OD 600 of 0.2, 0.8, 4 and 8 which corresponds to earlyexponential, mid-exponential, transition and stationary phase of growth from two independent cultures per phase by using RNAwiz (Ambion, Huntingdon, United Kingdom) according to the manufacturers protocol. Residual DNA from the RNA preparations was enzymatically removed by using TURBO DNA-free (Ambion). Extracted RNA samples were stored in 70% EtOH, 0.3 M sodium acetate buffer (pH 5.2) at -80°C.

Microarray construction and transcriptome analysis
Amplicon based DNA-microarrays were constructed for B. cereus ATCC 14579 as described for L. lactis IL1403 [64,65] with modifications as detailed below. Amplicons were designed on 5199 genes selected from the 5311 annotated genes (ORFs smaller than 80-bp were omitted) on the genome of B. cereus ATCC 14579 [9]. To reduce cross-hybridization between probe and target DNA sequences the amplicons had sizes of 70 -700-bp (depending on gene sizes) and comprised the most unique part of a gene. The amplicons were synthesized by EuroGentec (Seraing, Belgium) in two amplification steps. In the first amplification step, primers were used with a unique tag-sequence for B. cereus ATCC 14579 (forward primers were extended with the sequence: 5'-TCG-GGCAGCTGCTCC-3'; and reverse primers with the sequence: 5'-TGGCGCCCCTAGATG-3'). Two copies of each amplicon were present per array, resulting in microarrays comprising 10398 spots.
Normalized expression data (Feature Extract, Agilent) for each spot was used in a statistical analysis. The biological replicate experiments were merged with the web-supported VAMPIRE microarray suite, based on a Bayesian frame work. Furthermore, VAMPIRE calculated p-values for individual spots and subsequently used this p-value to identify statistically differentially expressed spots between compared growth conditions by use of a false discovery rate (FDR) of 0.05 as a threshold [66,67]. In addition, only ORFs of which both individual spots passed the FDR based threshold were considered to be putatively differentially regulated. Expression ratios per ORF were established by calculating the average of the log-values of individual spots. This value (R) was then used to calculate the average expression ratio (10 R ) per ORF. Finally, only ORFs that showed a change in expression of at least 2-fold (up/down) were considered to be differentially expressed. Microarray data are submitted to the GEO database with accession number: GSE7843.
To determine gene similarity, homology and gene context NCBI BLAST and the ERGO database were used [10], Putative CcpA-regulated enterotoxin operons Figure 6 Putative CcpA-regulated enterotoxin operons. Positioning of CRE-sites and PlcR binding sites in two major enterotoxin operons encoding Nhe and Hbl of B. cereus ATCC 14579. Ratios of gene regulation are given as found in the ccpA deletion strain compared to the wild-type, for the four different growth phases, early-and mid-exponential, transition and stationary phase, respectively. Closed arrows indicate the approximate position of the identified CRE-site, open arrows indicate binding sites for the pleiotropic regulator PlcR [59]. while KEGG [68] was used for assessment of metabolic functions and pathways. Whether succeeding genes were part of one operon was determined according to operon prediction as performed by [31].

Definition and identification of B. cereus CRE-site
The 350-bp upstream and 150-bp downstream sequences of the translation start of genes identified by the array experiments to be significantly higher expressed in the ccpA deletion strain compared to the wild-type strain for early-and mid-exponential growth were analyzed with AlignACE 3.1 [69], which searches the input sequences for stretches of nucleotides which align between the different input sequences. The 13-bp CRE-sites identified by Alig-nACE were aligned with MUSCLE 3.6 [70] and a Hidden Markov Model (HMM) was constructed with the HMMER package [71]. The HMM was used to search CRE-sites in the complete genome sequence of B. cereus ATCC 14579 [9], and not only the 350-bp upstream and 150-bp downstream sequences significantly higher expressed genes in the ccpA deletion strain. Subsequently, to analyse whether the B. cereus consensus was longer than 13-bp, the obtained sites after the HMM search were extended to 18bp, as has been the reported length for B. subtilis [15]. The resulting extended CRE-sites were again aligned using MUSCLE 3.6 [70]. This alignment was subsequently visualized with WebLogo [72] and a iteration HMM search was performed on the B. cereus ATCC 14579 sequence [9] to identify all putative CRE-sites confirming to the new alignment in the B. cereus ATCC 14579 genome.