Proteomic Evidences for Rex Regulation of Metabolism in Toxin-Producing Bacillus cereus ATCC 14579

The facultative anaerobe, Bacillus cereus, causes diarrheal diseases in humans. Its ability to deal with oxygen availability is recognized to be critical for pathogenesis. The B. cereus genome comprises a gene encoding a protein with high similarities to the redox regulator, Rex, which is a central regulator of anaerobic metabolism in Bacillus subtilis and other Gram-positive bacteria. Here, we showed that B. cereus rex is monocistronic and down-regulated in the absence of oxygen. The protein encoded by rex is an authentic Rex transcriptional factor since its DNA binding activity depends on the NADH/NAD+ ratio. Rex deletion compromised the ability of B. cereus to cope with external oxidative stress under anaerobiosis while increasing B. cereus resistance against such stress under aerobiosis. The deletion of rex affects anaerobic fermentative and aerobic respiratory metabolism of B. cereus by decreasing and increasing, respectively, the carbon flux through the NADH-recycling lactate pathway. We compared both the cellular proteome and exoproteome of the wild-type and Δrex cells using a high throughput shotgun label-free quantitation approach and identified proteins that are under control of Rex-mediated regulation. Proteomics data have been deposited to the ProteomeXchange with identifier PXD000886. The data suggest that Rex regulates both the cross-talk between metabolic pathways that produce NADH and NADPH and toxinogenesis, especially in oxic conditions.


Introduction
Bacillus cereus is a Gram-positive, facultative-anaerobe, rodshaped endospore-forming human pathogen. Most of the reported illnesses involving B. cereus are food-borne intoxications, classified as emetic and diarrheal syndromes [1,2]. Diarrheal disease is due to vegetative outgrowth and secretion of various extracellular factors, including enterotoxins [3]. The most extensively studied diarrheal enterotoxins are hemolysin BL (Hbl), nonhemolytic enterotoxin (Nhe), and cytotoxin K (CytK) [4,5]. These enterotoxins are secreted via the Sec translocation pathway [6]. Although Hbl, Nhe and CytK are currently considered as the etiologic agents of diarrheal syndrome, other toxins, such as EntA, EntB and EntC, may also contribute to the pathogenicity of B. cereus [7,8]. To grow and produce virulence factors in the human intestine, B. cereus must adapt its metabolism, and regulates its proteome [9] in response to changes in oxygen availability. Indeed, B. cereus encounters oxic conditions in zones adjacent to the mucosal surface [10] and anoxic condition in the intestinal lumen [11]. Changes in oxygen availability can influence the relative levels of the dinucleotide, NAD + and NADH, in the cell, and such changes are sensed by the transcriptional regulator, Rex, in B. subtilis [12] as in other Gram-positive bacteria [13,14,15,16,17,18,19]. Depending on the cellular NAD + /NADH ratio, Rex regulators modulate the expression of genes involved in fermentative metabolism, biofilm formation, and oxidative stress [18,20,21]. Structural studies of Rex proteins have identified dinucleotide-binding pockets in the C-terminal domain of the protein. NADH binding in this region leads to a conformational change in the Rex homodimer, triggering a displacement of Rex from its recognition sites on DNA, and thus leading to derepression of the downstream genes [15,19]. A Rex homologue has been detected in the cellular proteome of B. cereus [9]. Furthermore, we found a canonical Rex binding motif [15] overlapping ResD and Fnr binding motifs [22,23,24] in the ldhA promoter region [25]. In B. cereus, ResD, Fnr and LdhA regulate both catabolism and enterotoxin production under both aerobic and anaerobic growth conditions, probably through a regulatory complex [26]. Rex could be, thus, a regulator of both catabolism and toxinogenesis in B. cereus. A role of Rex in toxinogenesis has not yet been reported in B. cereus or in any other organism.
In this study, we show that Rex from B. cereus, like its orthologues, is a transcriptional factor capable of interacting with DNA in an NADH/NAD + -responsive manner. We demonstrate that B. cereus Rex is a key regulator of anaerobic fermentation, aerobic respiration, resistance against external reactive oxygen species, and toxinogenesis, by modulating the cellular and extracellular proteome in an oxygen-dependent manner. This study provides the most comprehensive experimental information on proteins whose synthesis was changed in the presence of Rex. All together, our results offer new information about the metabolic events that maximize B. cereus growth in environments with varying oxygen conditions.

Materials and Methods
Bacterial strains, media, and growth conditions Escherichia coli TOP 10 (Invitrogen) was used as the host for cloning experiments, and E. coli SCS110 (Stratagene, La Jolla, CA) was used to prepare DNA for B. cereus transformation. B. cereus ATCC 14579 [27] was used as the parent strain for the construction of the rex deletion mutant. E. coli strains were grown at 37uC, with agitation, in Luria broth (LB). B. cereus strains were cultured in batches (three independent cultivations per strain) at two oxygen availabilities, i.e. pO 2 = 0% and pO 2 = 100% [8,9]. Each batch culture was inoculated with a subculture grown overnight at an initial optical density at 560 nm (OD 560 ) equal to 0.02. For B. cereus cultivation, the minimal MOD medium was supplemented with 30 mM glucose as the carbon source [28]. Anaerobic and aerobic batch cultures were performed at 37uC in a 2 L bioreactor (BioFlo/CelliGen 115, New Brunswick), and the working volume was maintained at 1.8 L. The pH was kept at a controlled value of 7.2 by automatic addition of 5 M KOH. B. cereus growth was monitored spectrophotometrically at 560 nm and calibrated with cell dry-weight measurements as previously described [25]. B. cereus cells were harvested by centrifugation when they reached their maximal growth rate (m = m max ) and immediately frozen until proteomic analysis. Supernatants were kept at 220uC for glucose and glucose-by-product assays and exoproteomic analysis.

Construction of the B. cereus Drex mutant strain
The deletion mutant, ATCC 14579 Drex, was constructed as follows. Two DNA fragments encompassing the 59untranslated region (UTR) and 39UTR of rex (BC 0291) were generated by PCR using primer pairs, rex1F (59-GCCATGTTAATGTTTC-GATGTCT-39) and rex2R (59-CCCGGGATCTTTTAG-CAGTGGCTTGTGG-39, SmaI restriction site is underlined), and rex3F (59-CCCGGGGTTTACTTTTTGAAAAACTATC-CACAA-39, SmaI restriction site is underlined) and rex4R , respectively. The resulting 840 bp 39SmaI and 827 bp 59SmaI DNA fragments were cloned into the TA cloning vector, pCR4-TOPO (Invitrogen, La Jolla, CA), generating plasmids, pCR4mutrex1 and pCR4mutrex2, respectively. The 840 bp DNA fragment encompassing the 59UTR region of rex was isolated from pCR4mutrex1 with PstI and SmaI, and subcloned into pCR4mutrex2 to generate pCR4mutrex3. A 1.5 kbp SmaI fragment containing the entire spectinomycin gene, spc [29], was purified from pDIA [25]. This purified fragment was ligated into SmaI-digested pCR4mutrex3. The resulting plasmid, pCR4mutrex4, was digested with EcoRI and the resulting 3067 bp 59rexUTR-spc-39rexUTR was subsequently inserted into the EcoRI site of pMAD [30]. The resulting plasmid was introduced into B. cereus cells by electroporation. The rex ORF was deleted and replaced with spc via a doublecrossover event [30]. Chromosomal allele exchanges were confirmed by PCR with oligonucleotide primers located upstream and downstream of the DNA regions used for allelic exchange. To complement the rex gene in trans, a DNA fragment of 932 bp encompassing the rex ORF (630 bp) and its promoter region (219 bp) was first PCR amplified using the primer pairs, rex-compF (59-GGATCCCGTTCGAAAGCGCGTTTACTTG-39; the BamHI restriction site is underlined) and rexcompR  the SacI restriction site is underlined), and then cloned into the pCRXL-TOPO plasmid (Invitrogen). The PCR fragment was then cut with BamHI and SacI and ligated into pHT304 [31], digested with the same restriction enzymes. The integrity of the recombinant vector (pHT304rex) insert was verified by sequencing.
Phenotypic characterization of B. cereus Drex using the API-50CHB testsystem The carbohydrate metabolism of the wild-type strain (WT) transformed or not with pHT304, Drex mutant transformed or not with pHT304 and Drex mutant transformed with pHT304rex was examined using API 50 CHB strips (BioMérieux SA, France). The results showed that WT and WT(pHT304) did not ferment turanose. In contrast, Drex and Drex (pHT304) showed a typical positive reaction in turanose test fermentation. Transformation of Drex mutant with pHT304rex inhibited the capacity of Drex to ferment turanose.

Measurement of glucose and by-product concentrations
Enzymatic test kits from Diffchamb (Lyon, France), R-Biopharm (Saint-Didier au Mont-d'Or, France), and Roche (Meylan, France) were used to analyze the glucose, lactate, ethanol, formate, acetate, and succinate concentrations in the supernatants of 4 mL cell cultures obtained after centrifugation at 10,0006g for 5 min (4uC). The specific glucose consumption rate, defined as the differential change in glucose concentration with time, was calculated from the equation, q glucose = m/Y x , where m is the specific growth rate (h 21 ) and Y x is the biomass yield (g.mol carbon substrate 21 ).

Gene expression analysis by RT-PCR
RT-PCR was performed using SYBR Green technology on a Lightcycler instrument (Roche applied Science) as described previously [32]. The primers used in this study have been described previously [8].

Purification of Rex
The rex ORF was amplified by PCR from B. cereus ATCC 14579 using the oligonucleotides, pET101rexF (59-ACCATG-GATCAGCAAAAGATTCCA-39) and pET101rexR . The amplicon was introduced as a blunt-end PCR product into pET101/D-TOPO (Invitrogen). The integrity of the inserted sequence was confirmed by DNA sequencing. The resulting construct was transformed into E. coli BL21-CodonPlus(DE3)-RIL strain (Stratagene) for protein production. BL21-CodonPlus(DE3)-RIL cells carrying the pET101-rex expression plasmid were grown in 1 L LB medium containing 100 mg.mL 21 ampicillin, at 30uC with agitation (200 rpm) until the cell density reached an OD 600 of about 0.5. After this, 0.5 mM isopropyl-b-D-thiogalactopyranoside (IPTG) was added to the culture, and growth was continued for four hours. The cells were harvested by centrifugation (6500 rpm for 10 min at 4uC), washed, and resuspended in 20 mL ice-cold extraction buffer consisting of 50 mM TRIS buffered at pH 8.0, 0.3 M NaCl, 10% glycerol, and a protease inhibitor cocktail (one tablet of Complete Mini, EDTA free, Roche). Cells were then incubated with 0.5 mg/mL lysozyme for 45 min under gentle agitation at 4uC and disrupted by sonication for 5 min at 4uC using a Vibra Cell ultrasonicator (Fisher Bioblock Scientific). The lysates were clarified by centrifugation at 9000 rpm for 30 min at 4uC and loaded onto a 2 mL Co 2+ -immobilized TALON metal affinity chromatography column (Clontech) equilibrated with the extraction buffer. The column was washed with 30 mL of extraction buffer containing 20 mM imidazole and Rex was eluted with 3 mL extraction buffer containing 200 mM imidazole. The eluted fraction was desalted by dialysis, concentrated using Nanosep 30 kDa-molecular-mass cutoff devices (Omega disc membrane; Pall filtron), and stored at 280uC until analysis. The purity of the protein was estimated to be above 90% by Coomassie blue-stained SDS-PAGE. The protein concentration was determined using a Bradford assay (Interchim) with bovine serum albumin as the reference.

Electrophoretic mobility shift assays (EMSAs)
Nucleic acid fragments containing the promoter regions of ldhA and rex were PCR-amplified using biotinylated forward primers, LdhAF (59-ACCTGCTAATCCGATGATTG-39) and RexF , and nonbiotinylated reverse primers, LdhAR  and RexR . The DNA used as negative control was a fragment of the ssuRNA BC0007 sequence (NC_004722), which was amplified with the biotinylated ssubioF (59-GGTAGTCCACGCCGTAAACG-39) and ssuR (5-GACAACCATGCACCACCTG-39) primer pair. The 59-labeled amplicons were purified using the High Pure PCR Product Purification Kit (Roche). Binding reactions were performed for 30 min at 37uC by incubating biotin-labeled DNA fragments (2 nM per reaction) with different amounts of Rex in 10 mM Tris-HCl buffered at pH 7.5, and containing 50 mM KCl, 2.5% glycerol, 5 mM MgCl 2 and 5 mg/L poly(dI2dC). The samples were resolved by electrophoresis on a 6% nondenaturing polyacrylamide gel and electrotransferred onto Hybond N+ Nylon membranes (Amersham). Biotin-labeled DNA was detected using the LightShift Chemiluminescent EMSA Kit (Pierce).

Proteomic sample preparation and nanoLC-MS/MS analysis of tryptic peptides
Three independent biological replicates were harvested for each of the two conditions (aerobiosis and anaerobiosis) and two strains (Drex mutant and its parent strain, ATCC 14579). The extracellular proteins of the 12 cultures were extracted by trichloroacetic acid precipitation [8]. The cellular proteins from the 12 samples were obtained as previously described [9]. The 24 resulting samples were subjected to SDS-PAGE, and then identified after trypsin proteolysis by nanoLC-MS/MS tandem mass spectrometry with an LTQ-Orbitrap XL mass spectrometer as previously described [8,9]. A total of 131 nanoLC-MS/MS runs were carried out to acquire the whole dataset. The MS/MS spectra were assigned to tryptic peptide sequences with the Mascot Daemon software (version 2.3.2; Matrix Science) with mass tolerances of 5 ppm on the parent ion and 0.5 Da on the MS/ MS, fixed modification for carbamidomethylated cysteine, and variable modification for methionine oxidation. Mascot results were parsed with a p-value threshold below 0.05 for peptide identification and proteins were validated when at least two peptides were detected. The number of MS/MS spectra per protein recorded by nanoLC-MS/MS was extracted for each sample. In each condition, proteins were further considered for comparison only if peptides were seen in at least two of the three replicates. The resulting datasets were normalized taking into account the total protein concentration of the corresponding pellet and supernatant. Protein concentrations in B. cereus were determined using the Reducing agent Compatible Detergent Compatible (RCDC) protein assay (Bio-Rad) following the supplier's instructions. MS/MS spectral counts were compared with the TFold method using the PatternLab software program 2.0.0.13 [33] using a p-value cut-off set at 0.05. Log 2 (fold-change) were then calculated for comparisons and only the proteins for which the p-value was lower than the 0.05 cut-off were considered. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (http://www. proteomexchange.org) via the PRIDE partner repository [34] with the dataset identifier PXD000856 and DOI 10.6019/ PXD000856.

Exposure of bacteria to H 2 O 2 and viability assays
Oxidative stress resistance of the B. cereus Drex mutant and the parent strain, ATCC 14579, was assessed by exposing aerobically grown (OD,0.4) and anaerobically grown (OD,0.2) cells to 20 and 5 mM H 2 O 2 , respectively. Samples were taken prior to oxidative stress (time zero) and after 20 min. Aliquots (100 ml) of the samples were diluted in H 2 O, appropriate dilutions of the culture were plated onto LB agar, and after overnight incubation at 37uC the colony forming units (CFUs) were counted. All the experiments were performed at least in triplicate, and at least two technical replicates from each dilution step were carried out to determine the number of CFUs.

Results
Expression analysis of rex in B. cereus ATCC 14579 at two different pO 2 We identified BC_0291 in the B. cereus ATCC 14579 genome [27] as the homologue to the Bacillus subtilis transcriptional repressor, Rex [12,35,36], with 90% sequence identities. We established by 59RACE PCR (Fig. S1) that a transcriptional start site (G) was located 41 bp from the translational start site (ATG) of the BC_0291 open reading frame. Upstream of this transcriptional start site, we identified a potential housekeeping sA-type promoter, TATACAN(17)TAAACT. The stop codon (TAA) overlapped an inverted repeat (AAAACGCAGAGG(N6)C-CTCTGCGTTTT; DG = 223.2 kcal/mol) that may be a transcriptional terminator, suggesting that rex was monocistronically transcribed. Using real-time RT-PCR, we investigated the expression of B. cereus rex under aerobiosis and anaerobiosis in early-growing ATCC 14579 cells. The results indicated that rex expression was significantly higher under aerobiosis than under anaerobiosis (log 2 fold-change = 1.1, p-value,0.05).

Rex regulates glucose catabolism in B. cereus
Given that Rex orthologues function as NAD + /NADHresponsive regulators of metabolism in several Gram-positive bacteria (such as B. subtilis, S. aureus and S. coelicolor [12,13,14,15,36,37]), we sought to determine if B. cereus rex contributed to anaerobic fermentative and aerobic respiratory growth. As shown in Table 1, rex deletion slightly increased the B. cereus growth rate under both anaerobiosis and aerobiosis without a significant change in final biomass. Under anaerobiosis, the increase of growth rate was not related to glucose consumption, suggesting perturbation of the fermentative pathways. In accordance with this hypothesis, the spectra of fermentation end products were significantly modified by the rex deletion (Table 1). We observed that formate, acetate and ethanol production were promoted at the expense of lactate production. In addition, the ethanol-to-acetate ratio increased in Drex mutant, while NADH recovery levels remained unchanged and ATP yields tended to increase. Under aerobiosis, respiratory Drex cells secreted ,8-fold higher amounts of lactate compared with the wild-type cells, while producing lower levels of acetate. We conclude that rex deletion affects the carbon flux through the NADH recycling lactate pathway under both anaerobiosis and aerobiosis (Fig. 1). Complementation for growth of the non-polar rex mutant in B. cereus ATCC 14579 was not possible using a pHT304-based plasmid. This is probably due to the presence of multiple copies of the rex promoter region and/or overexpression of rex [38]. Therefore, to validate the role of Rex in aerobic respiration and anaerobic fermentation in B. cereus, we deleted the rex gene of B cereus F4430/73 [39] and cultured the mutant strain under both aerobiosis and anaerobiosis. The results showed that the growth phenotype of Drex was the same in strains ATCC 14579 and F4430/73 (Table S1).

Rex regulates hydrogen peroxide resistance of B. cereus cells
To evaluate the impact of Rex deficiency on B. cereus resistance to oxidative stress, aerobically and anaerobically grown ATCC 14579 cells were exposed to hydrogen peroxide (20 and 5 mM H 2 O 2 , respectively). Figure 2 shows that aerobically grown Drex cells were less susceptible to H 2 O 2 harmful effects than WT cells, while the anaerobically grown Drex cells were more susceptible. These data indicate that Rex restricts the resistance to oxidative stress of aerobically grown B. cereus cells, while sustaining a high resistance to oxidative stress of anaerobically growing cells. To validate the role of Rex in the B. cereus resistance to external H 2 O 2, we repeated the experiments using the F4430/73 Drex strain (Fig. S2). The results showed that Rex deficiency similarly impacts F4430/73 and ATCC 14579 cells while the survival of F4430/73 anaerobically grown cells was strongly higher than the survival of ATCC 14579 cells.

DNA binding activity of B. cereus Rex
Previous studies have shown that proteins belonging to the Rex family function as dimers that bind to promoter regions containing DNA motifs with the 59-TTGTGAAnnnnTTCACAA-39 consensus sequence [21]. This typical binding motif is missing in the rex regulatory region of B. cereus, as reported for S. aureus and B. subtilis [12,15]. However, a putative binding motif with two mismatches compared with the known consensus motif was found upstream of the transcription start site of ldhA [25] in B. cereus, as in S. aureus and B. subtilis. To test whether B. cereus Rex binds to the ldhA promoter region, we overexpressed B. cereus rex in E. coli, purified the dimeric His6-tagged recombinant protein (Fig.  S3), and performed electrophoretic mobility shift assays (EMSA). As a negative control, we assessed the binding of the Rex protein to the ssu DNA fragment (Fig. 3A). The results showed that a concentration of 0.6 mM Rex led to a complete shift of the ldhA fragment (2 nM). As expected, no shift was observed with the rex DNA fragment. The specificity of the binding was demonstrated by the absence of DNA shift with the negative control. The second EMSA experiment (Fig. 3B) showed that 10 mM NADH impaired Rex binding to the ldhA promoter region while 100 mM NAD + did not interfere with Rex-DNA complex formation. An NAD + / NADH ratio of five led to an incomplete shift of the ldhA fragment. We conclude that in B. cereus, as in B. subtilis and S. aureus [15], Rex binding activity depends on the NADH/NAD + ratio.

Global comparative proteomics of B. cereus Drex and wild-type cells
To investigate the regulatory role of Rex in aerobic respiratory and anaerobic fermentative growth of B. cereus and decipher its role in toxinogenesis, we carried out a comparative proteome analysis from cells harvested in exponential growth phase (m = m max ). We compared the cellular and extracellular proteomes of aerobically and anaerobically grown Drex with those of its parental strain, with biological triplicates for each condition. A total of 1,620 proteins were identified in the whole study (Tables  S2, S3, S4 and S5), with 1,450 proteins specifically detected in the whole-cell shotgun analysis and 170 further proteins observed in the exoproteome study. These proteins were quantified by spectral count as previously described [8,9]. The Rex-dependent changes in protein abundance, expressed as log 2 of the ratio of a protein's abundance at a given condition (aerobiosis or anaerobiosis) relative to the wild-type, are presented in Tables S6 and S7, where the proteins with significant abundance changes based on the PatternLab Tfold student t-test (p-value below 0.05) are highlighted while proteins with non-significant changes are in grey color.
(i) Anaerobiosis. Among the whole cellular proteome (Table  S6), 145 proteins showed significant abundance level changes in Drex cells relative to WT (p-values,0.05): 89 proteins were significantly up-regulated and 56 proteins were significantly downregulated in Drex cells. In the exoproteome (Table S7), 41 proteins showed significant abundance level changes in Drex cells relative  Table S6. The form filling indicates fold-change values that satisfied the Student's t-test statistical criteria (p-value,0.05) in anaerobiosis (blue) and aerobiosis (yellow). Red and green BC numbers indicated significant increase and decrease, respectively, of abundance level of the proteins in Drex mutant compared with wild-type. The nicotinamide nucleotides are indicated in red (NADP + /NADPH) or blue (NAD + /NADH). doi:10.1371/journal.pone.0107354.g001 to WT: 17 proteins were significantly up-regulated and 24 proteins were significantly down-regulated in Drex cells.
(ii) Aerobiosis. Table S6 indicates that 132 proteins showed significant abundance level changes in Drex cellular proteome (pvalues,0.05): 82 proteins were significantly up-regulated and 50 proteins were significantly down-regulated in Drex cells. Table S7 indicates that 70 proteins showed significant abundance level changes in Drex cellular exoproteome (p-values,0.05): 35 were significantly up-regulated and 35 were significantly down-regulated in Drex cells.
(iii) Global analysis ( Fig S4). Only 8 cellular proteins showed similar behavior under both aerobiosis and anaerobiosis (4 were up-regulated and 4 were down-regulated in Drex cells). Only 3 extracellular proteins were down-regulated under both anaerobiosis and aerobiosis. However, all the extracellular proteins that were up-regulated under anaerobiosis were also up-regulated under aerobiosis. Taken together, the data indicate that Rex (i) differently modulates the cellular proteome of B. cereus anaerobically and aerobically grown cells in terms of the identity of proteins that show significant abundance level changes, (ii) acts mainly as a repressor at the cellular proteome level under both   aerobiosis and anaerobiosis and, has a stronger impact on the exoproteome of aerobically grown than anaerobically grown cells.

Insights into the cellular proteome of the Drex mutant
The list of cellular proteins showing significant abundance changes (p-values,0.05) due to the absence of Rex includes proteins related to glucose metabolism, amino acid, nucleotide and lipid metabolism, protein folding, response to stress, virulence, cell wall/membrane assembly, transport, and a number of proteins of unknown function under both aerobiosis and anaerobiosis (see Table S6 for details on protein characteristics, the fold-changes and p-values in each comparison).
(i) Impact of Rex on glucose metabolism-related proteins under anaerobic fermentative conditions. Figure 1 shows an overview of the glucose catabolic pathways of B. cereus. As illustrated in Fig. 1, the abundance level of key fermentative enzymes, such L-lactate dehydrogenases (Ldh; BC_1924, BC_4870, BC_4996), pyruvate formate lyase (PflB; BC_0491), alcohol dehydrogenase (BC_4365, BC_2220, BC_0802) and acetate kinase (BC_4634) were not significantly different in the mutant compared with wild-type cells. However, the lack of Rex resulted in a decrease of the abundance of the IIA component of the glucose-specific phosphotransferase (PTS) system (BC_5320, log 2 = 20.4; p-value = 0.021). A significant decrease in abundance was also observed for the pyruvate-formate-lyase-activating enzyme, PflA (BC_0492; log 2 = 21; p-value = 0.048). PflA is the sole enzyme able to activate the oxygen-sensitive PflB (BC_0491), which converts pyruvate to acetyl-CoA and formate (Fig. 1) [40]. In contrast, anaerobically grown Drex cells sustain a higher level of phosphoglucomutase (Pgm; BC_4919) than the wild-type cells (log 2 = 0.6; p-value = 0.036) and thus, could sustain a higher carbon flux through gluconeogenesis than glycolysis (Fig. 1). As for Pgm, the malate dehydrogenase (Mdh; BC_4592) abundance level was increased in the absence of Rex under anaerobiosis (log 2 = 1.7; p-value = 0.017). Mdh is one of the enzymes involved in the NADH-dependent reduction of oxaloacetate to malate, which is the first reaction in the succinate pathway (Fig. 1). The increase of Mdh abundance in the Drex mutant compared with wild-type was associated with the increase of NADPH-producing malic enzyme (ME; BC_4604; log 2 = 0.5; p-value = 0.033), which converts malate into pyruvate. Taken together, our data suggest that Rex may modulate the abundance level of key metabolic enzymes to control carbon flow through fermentative pathways, possibly at the expense of the pathway that couples NADH-recycling Mdh with NADPH-producing ME.
(ii) Impact of Rex on glucose metabolism-related proteins under aerobic respiratory conditions. Figure 1 shows that ME was the sole enzyme showing an abundance level change in both aerobically and anaerobically grown mutant cells (log 2 = 0.4; p-value = 0.01 under aerobiosis). 6-Phosphogluconolactonase (Pgl; BC_3368) showed a higher increase than ME (log 2 = 1.3; pvalue = 0.047) in Drex aerobically grown cells. The increase of the Pgl abundance level may have an impact on the activity of the NADPH-generating pentose phosphate pathway (PPP) in the Drex mutant (Fig. 1). Two glycolytic proteins showed significant abundance decreases in aerobically grown mutant cells: the NADPH-producing glyceraldehyde-3-phosphate-dehydrogenase (NADPH-GAPDH; BC_0868; (log 2 = 20.5; p-value = 0.012) and pyruvate kinase (PK; BC_4599; (log 2 = 20.4; p-value = 0.032), which catalyzes the last steps of glycolysis. The decrease of the NADPH-GAPDH level could increase the carbon flow through the NADH-producing GAPDH (BC_4583, BC_5333) and thus stimulate flux through the upstream glycolysis pathways, especially through the NADH-recycling lactate pathway (Fig. 1). Finally, Rex may modulate glycolytic flux by controlling the abundance level of key enzymes of glycolysis and pentose phosphate pathway when cells are grown under aerobiosis in glucose-containing medium.  ( Table 2). A protein annotated as a putative prolipoprotein diacylglyceryl transferase (Lgt; BC_5163), which may be required for lipid modification of the cysteine residue present within the lipobox of prolipoproteins [41], showed a significant change of abundance level under both conditions: it was more highly detected in anaerobically grown cells and less detected in aerobically grown cells when Rex was absent. Transport-related proteins. The largest change in transportrelated protein abundance was observed in anaerobically grown Drex cells: 8 proteins showed significant abundance level changes (p,0.05). This suggests that the Rex-dependent control of the abundance pattern of transport related-proteins was stronger in anaerobic fermentative conditions than in aerobic respiratory conditions.
Transcriptional regulation-related proteins. Drex up-regulated the synthesis of 6 and 4 transcriptional regulators under anaerobiosis and aerobiosis, respectively. Among these, we found regulators of the Mer and Mar families, which control many cellular processes including oxidative stress response and virulence [43,44]. Interestingly, we noted that ResD was significantly upregulated in anaerobically grown Drex cells but remained unchanged in aerobically grown Drex cells. ResD is involved in both metabolism and toxinogenesis in B. cereus [22,32].
Stress-response related proteins. Among the stress responserelated proteins, the antioxidant protein, OhrA [7,9], was significantly up-regulated in aerobically grown Drex cells and its abundance remained constant under anaerobiosis ( Table 2). Like OhrA, the ribosome-associated heat-shock protein, Hsp15 (BC_0062), showed, under aerobiosis, a significant change in abundance level. This protein could be involved in the recycling of free 50 S ribosomal subunits [45]. Interestingly, we noted significant changes in the abundance patterns of 50 S ribosomal subunits in aerobically grown Drex cells (Table S6). Unlike aerobically grown cells, anaerobically grown Drex cells exhibited higher levels of catalase, KatE (BC_1155), lower levels of the BC1603 protein, which is functionally annotated as a cold-shock protein and no significant level change of alkyl hydroperoxide reductase (AhpC, BC_0377). Clearly, Rex controls the abundance pattern of stress-related proteins in an oxygen-dependent manner.
Phage-related proteins. Five phage-related proteins were strongly up-regulated in anaerobically grown Drex cells (log 2 . 1.5, p-value,0.05) and their abundance remained constant in aerobically grown Drex cells. The functions of these proteins are unknown but they are considered to be potentially beneficial to B. cereus in facing adverse environmental conditions [42].

Insights into the extracellular proteome of the Drex mutant
(i) Focus on toxin-related proteins. Although the intracellular abundance level of HlyI was higher in the aerobically grown Drex mutant strain, the extracellular level of this protein was not significantly different in the mutant compared with wild-type cells (Table S7). This was also the case for EntB, a putative enterotoxin [9]. However, 12 other toxin-related proteins showed a significantly higher abundance level in the exoproteome of aerobically grown Drex cells than wild-type cells ( Table 3). Some of these proteins also showed a higher abundance level under anaerobiosis, but to a lesser extent, i.e. the three components of Nhe (NheA, NheB and NheC), EntA, EntC and the L1 component of Hbl.
From these data, we conclude that Rex regulates the toxinogenic profile of the B. cereus exoproteome in response to oxygen availability.
(ii) Other predicted extracellular proteins. Like the toxinrelated proteins, many degradative enzymes, flagella components and proteins that are released from the cell wall, are more abundant in the exoproteome of the Drex mutant strain than wildtype cells, especially under aerobiosis ( Table 4). All of these proteins are predicted to be extracellular proteins and, except for flagella components, all of them encompass signal peptides in their N-terminal primary structure and/or transmembrane helices. One metabolism-related protein, the subunit a of F0F1 ATP synthase (AtpA), and three proteins of unknown function (BC_1894, BC_5360 and BC_4062) also showed a higher abundance level in the exoproteome of aerobically grown Drex cells. Except for BC_1894, which was annotated as a prophage protein and thus could be secreted via holin(-like) pathway [47], these proteins contain a signal peptide, and thus are predicted to be secreted by classical export pathways.
(iii) Predicted cytoplasmic proteins. In contrast to proteins predicted to be classically secreted, many proteins predicted to be cytoplasmic proteins, including metabolic proteins, stress response-related proteins, translation-related proteins, and other proteins were significantly less abundant in the exoproteome of the Drex mutant than the parental strain (Table 4). We noted that most of the cytoplasmic proteins that showed a significant abundance level decrease under aerobiosis did not show significant abundance level changes under anaerobiosis, and vice-versa (Table 4 and Table S7). Finally, the percentage of predicted cytoplasmic proteins significantly decreased in the Drex exoproteome compared with the wild-type exoproteome in the presence of oxygen, while the percentage of typically secreted proteins (toxin-related proteins and degradative enzymes and adhesins) was significantly higher (Fig. 4).

The absence of Rex does not affect the mRNA level of toxin-related proteins
An explanation for the increased abundance levels of toxinrelated proteins in the Drex mutant exoproteome could be a change in their mRNA levels. We thus tested whether transcription of genes encoding these proteins was altered in the Drex mutant compared with wild-type. Table 3 indicates that rex deletion significantly down-regulated entC and entFM under aerobiosis, but did not significantly change transcription of all other toxin-related genes. These transcriptomic results are not concordant with the proteomic data. This suggests that toxinrelated protein abundance may be mainly modulated at the posttranscriptional level when Rex is lacking.

Discussion
B. cereus Rex is an authentic Rex transcriptional factor because (i) it is able to bind to promoter fragments in vitro, indicating that at least in some cases the effect of Rex is mediated by direct interaction with the promoters, (ii) it impacts the cellular metabolism and oxidative stress tolerance, and (iii) it acts mainly as a repressor at the proteomic level [12,13,14,15,20,48]. However, B. cereus Rex also plays an important role in fully aerobic respiratory metabolism and it modulates the toxinogenic profile of exoproteome.
When oxygen was supplied to pH-regulated batch cultures at pO 2 = 100%, B. cereus metabolizes glucose to carbon dioxide by oxidation of glycolytic pyruvate in the tricarboxylic acid (TCA) cycle. This reaction produces NADH, which then fuels oxidative Concerning mRNA levels, each log 2 (fold-change) represents the mean level of mRNA in the Drex mutant samples relative to the mean level in the wild-type sample. The mean values were obtained from three measurements done on triplicate independent cultures. Only significant log 2 ratios are indicated in bold. Concerning protein levels, the relative amount of each protein in Drex mutant compared to its parental strain was determined using PatternLab software. Numbers in bold indicate data satisfied statistical criteria (p-value,0.05). The p-values were indicated in brackets. Plus and minus indicate increased and decreased abundance levels, respectively. NS: no significant change was observed. For details, see Tables S6 and S7. doi:10.1371/journal.pone.0107354.t003 Table 4. Changes in protein abundances in Drex exoproteome compared to wild-type exoproteome under anaerobiosis and aerobiosis.     Only changes satisfying statistical criteria (p-value,0.05) at least in one growth condition are shown.
a Each log 2 fold-change value represents the mean protein level of the Drex sample in relation to the wild-type sample. The relative amount of each protein was determined using PatternLab software. Plus and minus symbols indicate up-and down regulation of the protein, respectively. Numbers in bold indicate data that satisfied statistical criteria (p-value,0.05). ND: not detected. For details, see Table S7.
doi:10.1371/journal.pone.0107354.t004 phosphorylation to produce ATP, with minimal production of lactate and elimination of carbon excess through excretion of acetate. Rex deficiency facilitates the entry of carbon flow into the NADH-recycling lactate pathway at the expense of pyruvate oxidation into acetyl-CoA (Fig. 1). This promotes lactate production at the expense of acetate without impairment of TCA capacity, since acetate was always produced as a waste [28]. In the absence of oxygen or other external electron acceptors, ATP synthesis occurs at the level of substrate phosphorylation and NADH is reoxidized in terminal step fermentative reactions from pyruvate (Fig. 1). When grown in pH-controlled anaerobic batch cultures (pO 2 = 0%), B. cereus produces lactate as the main glucose by-product to satisfy the demand for redox balance. Rex deficiency favored the production of more reduced metabolites, as evidenced by the increase in the ethanol to acetate ratio and the decrease of lactate production. This indicates that Rex regulates the carbon flow distribution at the pyruvate node by favoring carbon flow through the NADH-recycling lactate pathway at the expense of Pfl-dependent fermentative pathways under anaerobiosis. A key question is how Rex controls the carbon flow into the NADH-recycling lactate pathway in B. cereus cells grown in fully aerobic respiratory and anaerobic fermentative conditions? This control process could involve interplay of transcriptional and/or post-transcriptional regulation of key glycolytic enzymes (such as Pgm, BC_4919 under anaerobiosis and PK, BC_4599 under aerobiosis) and post-translational regulation of lactate dehydrogenase. Evidently, further intensive work is required to unravel the complexity of the regulatory network involving Rex. Finally, by controlling the entry of carbon flow into the lactate pathway, Rex controls the availability of glycolytic intermediates for macromolecular synthesis as well as supporting NADPH production under both aerobiosis and anaerobiosis. In contrast to anaerobic fermentation, aerobic respiration is a major source of reactive oxygen species (ROS) generation [49]. B. cereus, like other facultative anaerobes, uses scavenging systems to control the level of ROS [50]. Sequestration of ROS leads to oxidative deactivation of these scavenging systems, which are then reactivated directly or indirectly depending on NADPH level [51]. By modulating the distribution pattern of antioxidant enzymes (Tables 2 and 4), and possibly NADPH production, Rex may thus modulate the effectiveness of antioxidative defense systems. Specifically, Rex may maximize antioxidant system activity under anaerobiosis while restricting this activity under aerobiosis; this could explain why the resistance of B. cereus cells towards external H 2 O 2 , when Rex is absent, is lower under anaerobiosis and higher under aerobiosis.
Except for HlyI, the abundance levels of classical extracellular proteins (such as toxins) were enhanced in B. cereus exoproteome when Rex was absent, especially under aerobiosis. In B. cereus, as in other Gram-positive bacteria, most extracellular proteins are  (Table S5). For readability reasons, standard deviations (below 10%) are not shown. doi:10.1371/journal.pone.0107354.g004 synthesized as precursors, with typical N-terminal signal peptides, and exported from the cytoplasm by the Sec-dependent pathway [52]. This is the case for toxins [6]. However, unlike other toxins, HlyI contains a putative N-terminal twin arginine motif that could target the protein to the Tat secretion pathway instead of the Secdependent pathway [53,54]. Besides classical export routes, B. cereus possesses specialized protein export systems such the flagellar type III system [5,55], which translocates most flagellar components across the cytoplasmic membrane, and possibly the ESAT-6 secretion system, which mediates the secretion of virulence factors belonging to the WXG100 protein family, such as the protein BC_5360 [56,57]. A feature of these two specialized secretion systems is the lack of a classical signal peptide to direct the substrate protein for secretion. However, like the Sec-dependent pathway and unlike the TAT-dependent pathway, these protein export systems require chaperones to prevent premature folding, aggregation and degradation by cytoplasmic proteases [58,59]. Rex may thus influence protein-assisted export by modulating the distribution pattern of the components of export systems as observed in our proteomic data ( Table 2, [60]). In addition, it has been reported that chaperones may be targets for intracellular ROS [61]. Keeping the functional integrity of these chaperones, specifically in the context of secretion, may thus depend on Rexdependent NADPH availability. The role and contribution of NADPH in classical and specialized secretory mechanisms remains an open question that deserves further experiments.
Our results show that many cytoplasmic proteins predicted to be involved in active growth, either directly (such as metabolic enzymes, translational and post-translational-related proteins) or indirectly by preventing inappropriate gene expression (such as AbrB and CodY under aerobiosis [62]), were less abundant in B. cereus exoproteome when Rex was absent. We assume that cell lysis cannot account for changes in the abundance level of these cytoplasmic proteins since: (i) cell viability count did not change (data not shown), (ii) EF-Tu did not accumulate in extracellular medium (Table S7, [63]), and (iii) the abundance pattern of cytoplasmic proteins was dependent on growth conditions and did not always correlate with intracellular abundance changes ( Table 2 and 4). Among the cytoplasmic proteins identified in B. cereus exoproteome, several have already been identified in the exoproteome of other bacteria, and many have been described as extracellular moonlighting components playing a role in bacterial virulence [64,65]. The non classical mechanism of secretion that could explain their release remains to be firmly described [47,66]. One hypothesis is that signal-less predicted cytoplasmic proteins might be secreted within outer membrane vesicles [63]. Such nanovesicles have been observed in Bacillus anthracis [67]. Interestingly, ATP is the primary metabolic factor involved in secretory vesicle movements. ATP generation is lower in fermenting cells than in non-fermenting cells, according to metabolism. In addition, Rex promotes ATP generation by enhancing NADH-producing pathways at the expense of NADPH pathways. Therefore, if ATP availability is a triggering factor for export of cytoplasmic proteins, this could explain the higher export of cytoplasmic proteins in aerobically grown cells compared to anaerobically grown cells, as well as the substantial decrease of cytoplasmic protein export under aerobiosis when Rex was absent.
In conclusion, Rex in B. cereus plays a pivotal role in controlling the cross-talk between the metabolic networks that control growth, oxidative defense machinery and extracellular accumulation of toxins. In the context of B. cereus natural environment, Rex may thus help the pathogen to (i) maintain its host alive until transmission to the next host can be achieved, by controlling its growth and production of its toxins [68], and (ii) better survive stress conditions under anaerobiosis, albeit to the detriment of maximizing its survival when oxygen is present [10,69,70].