Glucose-Dependent Activation of Bacillus anthracis Toxin Gene Expression and Virulence Requires the Carbon Catabolite Protein CcpA

Abstract

Sensing environmental conditions is an essential aspect of bacterial physiology and virulence. In Bacillus anthracis, the causative agent of anthrax, transcription of the two major virulence factors, toxin and capsule, is triggered by bicarbonate, a major compound in the mammalian body. Here it is shown that glucose is an additional signaling molecule recognized by B. anthracis for toxin synthesis. The presence of glucose increased the expression of the protective antigen toxin component-encoding gene (pagA) by stimulating induction of transcription of the AtxA virulence transcription factor. Induction of atxA transcription by glucose required the carbon catabolite protein CcpA via an indirect mechanism. CcpA did not bind specifically to any region of the extended atxA promoter. The virulence of a B. anthracis strain from which the ccpA gene was deleted was significantly attenuated in a mouse model of infection. The data demonstrated that glucose is an important host environment-derived signaling molecule and that CcpA is a molecular link between environmental sensing and B. anthracis pathogenesis.

The capacity for sensing and responding to their surroundings is an essential feature of all living organisms. Bacteria have developed an array of highly sophisticated sensing and adaptation systems, such as two-component systems, quorum-sensing systems, and transport systems (9, 15, 21, 42, 58). Investigations into bacterial pathogenesis have revealed an intuitive connection between basic metabolic processes of adaptation and virulence. For example, it has been reported that bacteria alter the transcription of carbohydrate utilization genes and virulence factor-encoding genes in response to the sugar availability encountered in the environment, in particular, in the infected host (44, 65, 67, 69, 79). Consequently, regulatory mechanisms that connect carbohydrate metabolism to virulence factor production must have evolved.

A key role in carbohydrate utilization in both Gram-negative and Gram-positive organisms is played by the complex sugar-transporting phosphotransferase system (PTS), but additional proteins and different mechanisms are involved in the two types of bacteria (4, 16, 53). One of these mechanisms is the carbon catabolite repression (CCR) system that is based on the Crp-cyclic AMP system in Gram-negative bacteria, while the HPr protein of the PTS system, together with the transcription factor CcpA, is the master regulator in Gram-positive bacteria (27, 59, 75).

Extensive work carried out mostly in Bacillus subtilis has demonstrated that HPr can be phosphorylated on the His15 residue (B. subtilis numeration) by the enzyme I (EI) of the PTS in response to the phosphoenolpyruvate-to-pyruvate ratio or on the Ser46 residue by the HPr kinase/phosphorylase (HPrK/P) enzyme in response to the concentrations of ATP, P_i, PP_i, and fructose-1,6-bisphosphate (FBP) (23, 30, 43, 56). HPr phosphorylated on Ser46 interacts and stimulates CcpA DNA binding activity, thus providing the link between carbon availability and transcriptional adaptation (23, 30, 61).

CcpA is a member of the LacI protein family of transcription factors, and it can be either a positive or a negative regulator (34, 39, 48, 49). CcpA binds to target promoters at catabolite response element (cre) sites, which are 14-bp cis-activating palindromic sequences (46).

The role of CcpA in carbohydrate metabolism has been extensively investigated in the model organism B. subtilis, but accumulating evidence has revealed a role for this protein in the virulence of several low-G+C-content Gram-positive pathogens. Several ex vivo studies have shown a role for CcpA in, for example, regulation of expression of virulence factors, biofilm formation, and antibiotic resistance in Staphylococcus aureus (63, 64, 66); enterotoxin expression and biofilm formation in Clostridium perfringens (73, 74); and modulation of pathogenic potential and biofilm formation in Streptococcus mutans (1, 78). Additionally, animal models of infection have shown that CcpA is required for nasopharyngeal colonization, lung infection, and bacteremia in the mouse by Streptococcus pneumoniae (24, 28). Attenuation of virulence in mouse models of infection has also been demonstrated for the ccpA mutants of Streptococcus pyogenes (group A streptococcus) (32, 67), although a conflicting result has also been reported (35). Recently, the ccpA mutant of S. aureus was also shown to be less virulent in a murine model of staphylococcal abscess formation (38).

Environmental sensing plays a major role in the growth and virulence of the Gram-positive spore-forming bacterium Bacillus anthracis, the causative agent of anthrax (47). The dual lifestyle of B. anthracis requires the ability to recognize the need to enter the ecological cycle, which results in the production of dormant spores, or to initiate the virulent cycle in an infected host, which results in the production of virulence factors. The pathways leading to sporulation or to virulence are mutually exclusive (52), suggesting the existence of sophisticated regulatory mechanisms that ensure the initiation of the correct pathway in any of the two environments in which the organism finds itself.

Virulence of B. anthracis is associated with the production of two virulence factors, the tripartite anthrax toxin comprised of protective antigen (PA; encoded by pagA), lethal factor (LF; encoded by lef) and edema factor (EF; encoded by cya) and the poly-d-glutamic acid capsule. Expression of both virulence determinants strictly depends on the AtxA transcription factor, whose full expression requires the activity of two distinct promoters (7, 14, 18, 19). The mechanisms regulating the activity of these promoters are largely unknown. Expression of the toxin- and capsule-encoding genes is induced by the presence in the environment of bicarbonate, a major compound in the mammalian body essential for acid-base homeostasis (5, 68).

Glucose is a critical element in human and animal cells used as a primary source of energy. In this work, we analyzed the mechanism of induction of B. anthracis toxin gene expression by glucose and determined that CcpA plays a positive role by indirectly regulating the transcription of the gene encoding AtxA. Furthermore, a ccpA mutant strain expressed a severely reduced level of toxin and its virulence was significantly attenuated in a mouse model of infection. The results indicated that, for B. anthracis, glucose is an additional signaling molecule linking environmental sensing to virulence factor production via the transcription regulator CcpA of Gram-positive organisms.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

All the Bacillus anthracis strains used in this study are derivatives of the Sterne strain 34F2 (pX01 positive, pX02 negative). Strains were routinely grown in LB medium with the appropriate antibiotics. Cultures for β-galactosidase assays were grown in R medium supplemented with 0.8% sodium bicarbonate and 0.25% glycerol (57), unless otherwise indicated. Assays were carried out as previously described (7). Glucose (when required) was added to a 0.25% final concentration immediately after samples for the second time point (T2) were taken. At the same time, an additional 0.25% glycerol was added to the control culture (referred to as “growth in the absence of glucose” throughout this work). Growth was carried out in a 5% CO₂ atmosphere, unless otherwise specified. Experiments for β-galactosidase analysis were carried out at least in duplicate to ensure reproducibility and that a representative result is reported here. Within each set of experiments, culture samples were taken and assayed in duplicate. The values shown are the averages of the two samples per each time point.

B. anthracis markerless mutant strains were generated essentially by using the protocol of Janes and Stibitz (29).

Escherichia coli strain DH5α or TG1 was used for plasmid construction and propagation. E. coli strain C600 was used to generate unmethylated plasmid DNA for electroporation into B. anthracis. Transformation of B. anthracis by electroporation was carried out as described previously (36).

Plasmid construction.

All PCR amplifications of B. anthracis DNA fragments were carried out using genomic DNA of strain 34F2 as the template. The oligonucleotide primers used in this study are listed in Table 1. Oligonucleotide primers were designed on the basis of the genome sequence of the B. anthracis Sterne strain (GenBank accession number NC_005945).

TABLE 1.

Oligonucleotide primers used in this study

Primer name	Sequencea
BAccpA5′Nhe	5′-TGAGGCTAGCATGAACGTAACAATATATG-3′
BAccpA3′Bam	5′-ATACGGATCCTTACTTCGTTGAATCTC-3′
BAptsH5′Nde	5′-ATAAATTCATATGGAAAAAATCTTTAAG-3′
BAptsH3′Xho	5′-AATGTCTCGAGTCATTATTCTCCTAATCC-3′
BAccpAL5′Kpn	5′-AAGGGTACCATAGGACAGAGCGCTG-3′
BAccpAL3′Mlu	5′-TGTTACGCGTATCTCATCGCACACTC-3′
BAccpAR5′Mlu2	5′-ACCTACGCGTATCCAATTTAGAGATTC-3′
BAccpAR3′Pst	5′-TTTCTGCAGTCTCGTCCCAATACAG-3′
BAccpBL5′Kpn	5′-GACGGTACCATTAAATGAGAAGTTTGAG-3′
BAccpBL3′Mlu	5′-TATTACGCGTACACTCACCTCTAC-3′
BAccpBR5′BSSHII	5′-ACAAGCGCGCTAAACAAGAAAAGGTGG-3′
BAccpBR3′Pst	5′-ACACTGCAGCACTATCTCCTAGAGG-3′
AtxA3′Xho	5′AAGGGCTCGAGTATATCTTTTTGATTTG-3′
AtxA5′promEco	5′-TATAAGAATTCTATGTTAATATGCT-3′
AtxA5′upEcoRI	5′-CTATAGGATCCAAAAATTTCAAGGTG-3′
AtxAseq1	5′-GTAGGAGCTTTATACCC-3′
AtxAH199A	5′-CACACAATTTTGCTTTTGAATAG-3′
AtxA3′EcoRV	5′-CGATGGATATCGGTGTTAGCATGTCT ATAA-3′
2lacZ5′	5′-GGGATGTGCTGCAAGGCG-3′
AtxApromBam4	5′-ATAGGATCCTTATAATTTTTATTTATTA TTCC-3′
BAS44885′	5′-ACTTGTTGTTTTGAGCGTGAGG-3′
BAS44883′	5′-TTCTAAACCTCGAATAACAGTC-3′
BAS44885′Eco	5′-TTTTTGAATTCTTTCTGTTCACAAAAG TCAG-3′
BAS44883′BamI	5′-GCTACGGATCCTTCTAAACCTCGAATAAC-3′
BAS38935′EcoRI	5′-ATTAAGAATTCAAGGGTATTAAGCGCT ACCG-3′
BAS3893′Bam	5′-TAAACGGATCCCTATTTGAACGTGTGC-3′
Spac-upPst	5′-TATCCCTGCAGCACAAGAGCGGAAAGATG-3′
AtxA5′Eco2	5′-ATTTAGAATTCAGGAAAGGAGAATCAATT ATAG-3′
AtxA3′SacI	5′-CAAAGAGCTCAGGGCATTTATATTATC-3′

^aRestriction sites are underlined.

The construction of plasmids pAtxA10 and pAtxA12 was described by Bongiorni et al. (7) (Fig. 1). Plasmid pAtxA26 was generated by cloning an EcoRI-BamHI-digested fragment, amplified by PCR using oligonucleotides AtxA5′upEcoRI and AtxApromBam4, into a similarly digested transcriptional fusion vector, pTCV-lac (54).

FIG. 1.

Schematic representation of the region in plasmid pXO1 containing the atxA gene. The arrows indicate the positions of the open reading frames, and the lines delineate the extent of the fragments cloned in the indicated plasmids. The plasmid carrying the lacZ fusion constructs was generated in vector pTCV-lac (54), while the plasmid carrying the atxA coding sequence under the control of the spac promoter is a derivative of pTCV-spac (see Materials and Methods). The extent of the fragment used as nonspecific (NS) DNA in competition EMSA is also shown.

Plasmids pTCV-4488 and pTCV-3893, carrying the citZ-lacZ and BAS3893-lacZ fusion constructs, respectively, were generated by amplifying the citZ (BAS4488) and the BAS3893 promoter regions, respectively, with oligonucleotide primer set BAS44885′Eco-BAS44883′BamI or BAS38935′EcoRI-BAS38933′Bam, respectively, and cloning the resulting fragments in pTCV-lac after digestion with EcoRI and BamHI.

Plasmid pTCV-spac, used for the construction of plasmid pAtxA28, was generated as follows: a fragment carrying the spac promoter was obtained by PCR amplification using oligonucleotide primers Spac-upPst and 2lacZ5′ with plasmid pMutin 4 as the template (71). The fragment, digested with BamHI, was cloned in the SmaI-BamHI sites of vector pTCV-lac (54). The coding sequence and ribosome binding site of atxA were amplified using oligonucleotide primers AtxA5′Eco2 and AtxA3′SacI. The fragment was digested with SacI and cloned in pTCV-spac, which had been digested with SmaI and SacI.

The plasmid used for the generation of a B. anthracis ccpA deletion strain was constructed as follows: oligonucleotide primers BAccpAL5′Kpn and BAccpAL3′Mlu were used to amplify a 500-bp fragment upstream of the start codon of ccpA (BAS4929), while oligonucleotide primers BAccpAR5′Mlu2 and BAccpAR3′Pst were used to amplify a 630-bp fragment downstream of the ccpA stop codon. Each PCR fragment was cloned in the TOPO vector (Invitrogen), sequenced, and then recovered by digesting the plasmids with KpnI-MluI and MluI-PstI, respectively. The fragments were ligated to the pORISce vector (8), which had been digested with KpnI and PstI, generating plasmid pORISce-ΔccpA. A similar strategy was used to generate the ccpB (BAS0208) deletion plasmid pORISce-ΔccpB, for which the upstream and downstream PCR fragments were obtained using oligonucleotide primer sets BAccpBL5′Kpn-BAccpBL3′Mlu and BAccpBR5′BSSHII-BAccpBR3′Pst, respectively. In this case, the downstream fragment was digested with BssHII and PstI.

For the expression of N-terminal His-tagged CcpA, the ccpA gene was amplified by PCR using the oligonucleotide primers BAccpA5′Nhe and BAccpA3′Bam. The fragment was digested with NheI and BamHI and ligated to similarly digested pET28a (Novagen), thus generating a fusion to 10 His codons at the 5′ end of the ccpA gene (plasmid pET28-CcpA).

In order to generate the N-terminal His-tagged HPr protein, the ptsH gene (BAS3959) was amplified by PCR using oligonucleotide primers BAptsH5′Nde and BAptsH3′Xho. The fragment was digested with NdeI and XhoI and cloned in similarly digested pET28a, thus generating plasmid pET28-HPr. The construct generated a fusion to 10 histidine codons at the 5′ end of the gene.

Plasmid pORI-lacPA was constructed by transferring an approximately 4-kb fragment from plasmid pJM115-pagA (70) to the temperature-sensitive plasmid pORI-Cm (10). The fragment, carrying the pagA promoter fused to the lacZ gene, was recovered by EcoRI-NsiI digestion and cloned in the vector digested with EcoRI-PstI. The resulting plasmid, upon transformation into B. anthracis at 30°C, integrated by single crossover in the pagA region of pXO1 when cells were shifted to the nonpermissive temperature of 37°C.

A ccpA complementation plasmid was constructed by amplifying a fragment by PCR using oligonucleotide primers BAccpAL5′Kpn and BAccpA3′Bam. The fragment was digested with KpnI and BamHI and cloned in similarly digested vector pHT315, which is presumably present at 15 copies per cell (3).

The fidelity of the PCR for all resulting plasmids was confirmed by DNA sequencing.

Protein expression and purification.

The pET28a derivatives carrying the ccpA or ptsH gene were transformed in the E. coli BL21(DE3) strain (Novagen). Cells were grown in LB medium supplemented with kanamycin, and protein expression was obtained by induction with 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) when cells reached an optical density (OD) at 600 nm (OD₆₀₀) of approximately 0.7. After induction, growth was continued for 3 h at 37°C before the cells were harvested by centrifugation.

The CcpA protein was purified by first resuspending the cells in a lysis buffer containing 100 mM Tris-HCl, pH 8.0, 150 mM KCl, 1 mM β-mercaptoethanol, and 5 mM imidazole. Cells were lysed by sonication, and the lysate was recovered after 30 min of centrifugation at 31,000 × g. The lysate was loaded onto a nickel-nitrilotriacetic (Ni-NTA) agarose column (Qiagen). The CcpA protein was eluted with the lysis buffer containing 50 mM imidazole. The eluate was dialyzed in buffer containing 100 mM Tris-HCl, pH 8.0, 150 mM KCl, 1 mM EDTA, and 0.1 mM dithiothreitol. The protein was then concentrated using an Amicon Centricon concentrator (Millipore) and stored at −20°C in the presence of 40% glycerol.

The HPr protein was purified from cells resuspended in a lysis buffer containing 50 mM Tris-HCl, pH 8.0, 500 mM NaCl, 10 mg/ml lysozyme, 2% Triton X-100, and 2 mM phenylmethylsulfonyl fluoride (PMSF). Cells were lysed using a French press, and the clear lysate was recovered after centrifugation at 31,000 × g for 30 min. The lysate was loaded onto an Ni-NTA agarose column equilibrated with 50 mM Tris-HCl, pH 8.0, and 500 mM NaCl. The column was washed with the same buffer containing 30 mM imidazole, and the HPr protein was eluted using 300 mM imidazole. The eluate was dialyzed in buffer containing 50 mM Tris-HCl, pH 8.0, and 20 mM NaCl. The protein was then concentrated and stored at −20°C with 30% glycerol.

Electrophoretic mobility shift assays (EMSAs).

Labeled double-stranded probes were generated by PCR amplification. One oligonucleotide per amplification reaction was end labeled using T₄ polynucleotide kinase (PNK; New England Biolabs) in the presence of [γ-³²P]ATP (6,000 Ci/mmol; Perkin Elmer) according to the enzyme manufacturer's recommendation. The PCR amplification was carried out with Accuzyme polymerase (Bioline). The probes were purified from agarose gels using a PureLink gel purification kit (Invitrogen). One microliter was used to determine the radioactive counts (cpm).

The oligonucleotide primer sets used to generate the probes were the following: for atxA, AtxA5′promEco-AtxA3′EcoRV (228 bp; this fragment is equivalent to the one carried by plasmid pAtxA10); for citZ, BAS44885′-BAS44883′ (236 bp); and for BAS3893, BAS38935′EcoRI-BAS38933′Bam (290 bp).

EMSAs were performed using a constant amount (approximately 1 ng, equivalent to 4,000 cpm) of labeled probe. The concentration of proteins used is indicated in Fig. 6 and 7 and the legends to those figures.

Competition assays were carried out in the presence of 300 ng of unlabeled fragments. The specific DNA fragments were generated with the same sets of oligonucleotide primers used to generate the labeled probes (see above). The unlabeled nonspecific DNA fragment was obtained by PCR amplification using oligonucleotide primer set AtxAseq1-AtxAH199A (260 bp; Fig. 1).

The reaction mixtures were incubated for 15 min at room temperature, mixed with loading dye, and separated by electrophoresis on a 5% acrylamide gel run at room temperature using Tris-acetate-EDTA buffer. The gels were dried and exposed to a PhosphorImager screen (Molecular Dynamics). Screens were scanned using a Storm 840 scanner (Molecular Dynamics) and analyzed with ImageQuant software.

DNase I footprinting.

The labeled probes (20,000 cpm) were incubated in the presence of HPr (5 μM) and CcpA at the concentrations indicated in Fig. 6 and 7 and their legends in a 15-μl reaction mixture containing 1× DNase I buffer (New England Biolabs). The reaction mixtures were incubated for 4 min at room temperature before the addition of DNase I and further incubation for 1 min. Stop solution (6 μl; 0.1 M sodium EDTA, 0.5% SDS, 0.4 mg/ml sheared salmon sperm DNA) was added, and DNA precipitation from the reaction mixtures was accomplished with 500 μl of ethanol in an ethanol-dry ice bath. The DNA was collected by centrifugation, dried, and resuspended in 6 μl of loading dye (0.1% bromophenol blue, 0.1% xylene cyanol, 10 mM sodium EDTA, 95% deionized formamide). Four microliters of each sample was loaded on the gel. DNA sequence ladders were generated by the dideoxy chain termination method (60) using a USB Sequenase (version 2) sequencing kit. Plasmids pAtxA10 and pTCV-3893 were used as templates, and the same 5′ and 3′ oligonucleotide sets used to amplify the fragments were used as sequencing primers. The sequencing reactions were run simultaneously with the footprinting products on a 5% polyacrylamide-6 M urea gel.

Western blot analysis.

Strain 34F2 was grown in the presence and absence of 0.25% glucose in R medium, as described above for the β-galactosidase analysis. Samples were taken at T7, and the OD at 600 nm was recorded. Cells (the equivalent of 2.6 OD units for each strain) were harvested by centrifugation and resuspended in 250 μl of H₂O with 5 mg/ml lysozyme and 50 U/ml of mutanolysin. Cells were incubated at room temperature for 30 min. An equal volume of 0.2 M EDTA and protease inhibitor (Roche) was added, and the resuspensions were sonicated three times for 10 s each time. Lysates were mixed with SDS-loading dye, boiled for 10 min, and directly applied to 12% SDS-polyacrylamide gels. Western blotting was carried out using an anti-AtxA polyclonal antibody (70), and signals were detected with an ECL Plus Western blotting detection system (Amersham-GE Healthcare) using a Storm PhosphorImager. The level of AtxA protein was quantitated with ImageQuant software.

Animal infection model.

Six-week-old female A/J mice (Jackson Laboratory) were injected subcutaneously with ≈1 × 10⁶ spores of strain 34F2 or 34F2ΔccpA in a 100-μl volume (77). Colony counts were verified by plating for viable colonies and hemocytometer counting of individual cells. Spores were purified on Renografin gradients (50). Mice were housed and maintained at The Scripps Research Institute animal facility, and the protocol was carried out under the approval of the Institutional Animal Care and Use Committee. Survival data were analyzed by Kaplan-Meier survival analysis and were tested for significance by the log-rank test using Prism software (GraphPad Software, Inc.).

RESULTS

pagA toxin gene expression is induced by glucose.

We previously reported that the AtxA transcription factor, which is required for toxin gene expression in B. anthracis, contains PTS regulation domains (PRDs) domains generally known as PTS regulatory targets for phosphorylation/dephosphorylation in response to the availability of carbon sources (70). This observation raised the question of whether sugars would affect pagA gene expression. A report by Cataldi et al. (11) indeed indicated that induction of pagA expression occurred upon addition of 0.25% glucose to a culture otherwise grown in R medium supplemented with 0.25% glycerol as the carbon source and 0.8% sodium bicarbonate in a CO₂ atmosphere. Using essentially the same growth conditions, we also observed, by means of a lacZ reporter fusion construct in a replicative plasmid, that the transcription of pagA remained at a low constitutive level in the culture grown with glycerol as the sole carbon source but that it was induced 12-fold upon addition of glucose to 0.25% final concentration after T2 (Fig. 2 A). No induction was observed when cells were grown in air in the same medium in the presence of 0.25% glucose but without CO₂-sodium bicarbonate (Fig. 2A). Also, no induction was observed when cells were grown in the presence of glycerol and either fructose, mannitol, cellobiose, or the non-PTS-dependent sugar melibiose was added as described for glucose (data not shown).

FIG. 2.

Glucose induces expression of the pagA gene. Time courses of β-galactosidase activity were taken for strains 34F2, 34F2ΔccpA, and 34F2ΔccpB carrying a pagA-lacZ fusion construct on the replicative plasmid pTCV-lac (54). Cells were grown in R medium with 0.8% sodium bicarbonate and 0.25% glycerol in a 5% CO₂ atmosphere. Glucose or glycerol (without glucose) (0.25%) was added at T2. (A) β-Galactosidase activity expressed in Miller units (45); (B) growth curves. Symbols: open symbols, cultures grown in the presence of glucose; closed symbols: cultures grown in the absence of glucose; circles, parental strain; squares, ΔccpA; triangles, ΔccpB; open inverted triangles, level of β-galactosidase activity obtained with strain 34F2 carrying the pagA-lacZ fusion construct when it was grown in R medium with glucose in air without CO₂-bicarbonate.

These results confirmed a specific role for glucose in induction of toxin gene expression in an environment thought to mimic the infected host.

Glucose induction of pagA expression is mediated through atxA expression.

In order to understand the mechanism of glucose induction of pagA gene expression, we first asked the question whether the effect was direct on the pagA promoter or whether it was mediated through AtxA activity or atxA gene expression. Analysis of pagA expression in an atxA mutant strain showed that no increase occurred upon addition of 0.25% glucose at T2 (Fig. 3 A), indicating that the induction effect required AtxA.

FIG. 3.

Induction of pagA expression by glucose is abolished by the deletion of atxA or by expressing AtxA from the spac promoter. (A) Time courses of β-galactosidase activity in strain 34F2ΔatxA carrying the pagA-lacZ fusion construct in vector pTCV-lac; (B) time courses of β-galactosidase activity in strain 34F2ΔatxA harboring the pagA-lacZ fusion construct in plasmid pORI-lacZ integrated at the pagA locus and the replicative plasmid pAtxA28 expressing the AtxA protein from the constitutive spac promoter. Symbols: ▪, cells grown with glucose; •, cells grown without glucose.

The results suggested that the glucose effect could be either on AtxA activity or on atxA gene expression. To test this we generated a plasmid that carried the atxA gene under the control of the constitutive spac promoter (plasmid pAtxA28) (Fig. 1). β-Galactosidase assay of the pagA-lacZ reporter construct integrated at the pagA locus in pXO1 (plasmid pORI-lacPA) showed that the addition of 0.25% glucose did not result in induction of pagA expression when AtxA was expressed from the spac promoter on the replicative vector pAtxA28 in an otherwise atxA mutant strain (Fig. 3B). These results indicated that glucose did not affect the activity of AtxA.

The atxA gene is transcribed from two promoters which are approximately 620 bp apart (Fig. 1) (7, 14). In order to determine whether an inducing effect by glucose, if any, was exerted on promoter P1, promoter P2, or both promoters, three atxA promoter fusions to the E. coli lacZ gene were analyzed in parental strain 34F2. The results, shown in Fig. 4, indicated that addition of 0.25% glucose to the culture medium at T2 significantly induced transcription from the P1 promoter in plasmid pAtxA10 (Fig. 4A) but did not induce transcription from the P2 promoter in plasmid pAtxA26 (Fig. 4C). Moreover, transcription from the fusion construct in plasmid pAtxA12, which contains P1 and P2, was also induced by the addition of glucose. Similar induction levels were obtained with glucose at a 0.1% final concentration (data not shown).

FIG. 4.

Glucose-dependent induction of atxA transcription from the P1 promoter requires CcpA. Time courses of β-galactosidase activity were carried out on B. anthracis cultures grown in the presence (open symbols) or absence (closed symbols) of glucose. Glucose or glycerol (0.25%) was added to the R medium at T2 as described in Materials and Methods. Strain 34F2 harbored plasmid pAtxA10 (A), plasmid pAtxA12 (B), or plasmid pAtxA26 (C). (D) Growth curves of the strains whose β-galactosidase activities are shown in panel A as representative growth curves for each set of experiments. Symbols: circles, parental strain 34F2; squares, 34F2ΔccpA mutant strain.

In order to ensure that induction of atxA expression by glucose was associated with an increased level of AtxA protein, a Western blot analysis was carried out on cell lysates of strain 34F2 grown in the presence and absence of 0.25% glucose. The results, shown in Fig. 5, confirmed that glucose addition to the culture medium increased the level of AtxA protein approximately 4- to 6-fold compared to the level observed when glycerol was the only carbon source. These results indicated that the induction of pagA expression by glucose correlated with an induction in gene expression of the AtxA transcription factor that is essential for toxin expression.

FIG. 5.

Western blot analysis of AtxA induction by glucose. (A) Coomassie-stained 12% SDS-polyacrylamide gel. Strain 34F2 was grown in R medium with CO₂-bicarbonate in the absence (lane 1) or presence (lane 2) of 0.25% glucose starting at T2. Strain 34F2ΔatxA (lane 3) was grown under the same conditions in the presence of glucose. Samples were taken at T7. Lanes 1, 2, and 3 contained 10 μl of cell lysate normalized to a cell OD₆₀₀ equivalent to 0.05. (B) Western blotting carried out with polyclonal antibodies raised against AtxA (70). Lanes 1, 2, and 4 correspond to lanes 1, 2, and 3 of panel A, respectively. Lane 3 contained 0.006 μg of purified AtxA protein modified to carry a 6× His tag at the N-terminal end, which accounts for the slightly higher molecular weight. The band corresponding to AtxA is indicated by the arrow. The molecular weight standards (lanes MW) are Page Ruler (Fermentas) in panel A and the Magic Mark XP Western standard (Invitrogen) in panel B.

CcpA is required for glucose induction of atxA expression.

Transcription regulation in response to carbon source availability in Gram-positive organisms is mainly carried out by the CcpA transcription factor (75). Therefore, we asked whether CcpA was involved in the glucose-dependent induction of atxA transcription. A markerless deletion of the ccpA gene was generated in strain 34F2, and atxA-lacZ fusion reporter plasmids pAtxA10, pAtxA12, and pAtxA26 were introduced in the resulting strain via electroporation.

Time course studies of β-galactosidase activity were carried out on cells grown in R medium with CO₂-bicarbonate in the presence or absence of 0.25% glucose starting at T2. The results shown in Fig. 4 indicated that the deletion of ccpA almost completely abolished the glucose induction of atxA transcription from promoter P1 (Fig. 4A). Correspondingly, transcription of pagA in the ccpA mutant was severely reduced in the presence of glucose compared to that in the parental strain (∼4.5-fold reduction) (Fig. 2A). However, some pagA induction compared to the rate of transcription in cells grown in the presence of glycerol only was still observed, consistent with a reproducible, slightly higher level of transcription of atxA in the ccpA mutant strain grown with glucose than in the strain grown without glucose (Fig. 4B and data not shown). Notably, the growth rate of the ccpA mutant strain is essentially identical to the growth rate of the parental strain when cells were grown in glycerol only, but the rate differed when the cultures were grown in the presence of glucose, which resulted in reduced growth of the mutant strain (Fig. 2B and 4D). This is consistent with the severe growth defect described for the ccpA mutant in B. subtilis grown on glucose as the sole carbon source (41). Notably, the R medium contains methionine (73 mg/liter), glutamate (612 mg/liter), and the branched-chain amino acids isoleucine (170 mg/liter) and leucine (230 mg/liter), which were reported to be necessary in B. subtilis to restore growth of the ccpA mutant to the wild-type level (40), but they appear not to be sufficient to overcome the growth defect in B. anthracis.

A replicative vector containing the ccpA coding sequence and promoter region was constructed in plasmid pHT315 (see Materials and Methods) for complementation studies. Unfortunately, we were never able to obtain viable transformants of strain 34F2ΔccpA carrying this plasmid. It is possible that not only deletion of ccpA is deleterious to the cells but also overexpression of CcpA may deregulate essential functions.

A second catabolite control protein, CcpB, is known to contribute to carbon catabolite repression in B. subtilis (12). The possibility that the B. anthracis CcpB protein also contributed to glucose regulation of pagA and/or atxA expression was analyzed. A markerless deletion of ccpB was generated in strain 34F2, and the β-galactosidase activity of the pagA reporter plasmids was measured in cells grown in R medium with CO₂-bicarbonate in the presence and absence of 0.25% glucose. As shown in Fig. 2A, the ccpB deletion did not affect the induction of pagA transcription by glucose. Similarly, the transcription of atxA from the pAtxA10 and pAtxA12 reporter constructs was not affected by the ccpB deletion (data not shown).

These data indicated that CcpA is mainly responsible for the induction of atxA gene transcription from the P1 promoter resulting from the addition of glucose to the culture medium when cells were grown in a CO₂ environment in the presence of sodium bicarbonate.

CcpA does not bind specifically to the atxA promoter.

CcpA is a DNA binding protein that recognizes target promoters containing the catabolite response element (cre) site, which is a 14-bp palindromic sequence (22). Visual analysis of the atxA promoter region did not identify any sequence with strong similarity to the CcpA consensus sequence of B. subtilis, S. pyogenes, or Bacillus cereus (2, 72, 76). Nevertheless, we set forward to determine whether CcpA was controlling atxA transcription by directly binding to the atxA promoter region.

After purification of the B. anthracis CcpA protein from an overexpressing E. coli strain, a series of EMSAs were carried out using radioactively labeled fragments containing the atxA promoter region. The results obtained with the same fragment present in plasmid pAtxA10 (Fig. 1 and 6 A) showed that CcpA slightly lowered the mobility of the probe but only when it was present at relatively high concentrations (1 to 1.5 μM). Notably, the B. subtilis CcpA protein has been shown repeatedly to affect the mobility of target DNA fragments at nM concentrations (6, 34, 55).

FIG. 6.

Electrophoretic mobility shift assay to determine conditions of CcpA binding to atxA, citZ, and BAS3893 promoter regions. Fragments were generated by PCR amplification and end labeled with [γ-³²P]ATP via previous phosphorylation with PNK of one oligonucleotide primer per each set of amplification reactions. A constant amount of probes (1 ng) was incubated at room temperature with the indicated concentrations of CcpA (without HPr [−HPr]) or with CcpA and HPr at a 10 μM final concentration (+HPr). Samples were run on 5% polyacrylamide gels.

The addition of purified, unphosphorylated B. anthracis HPr stimulated CcpA binding to DNA, as mobility of the pAtxA10 fragment was detected at a CcpA concentration of 0.5 μM (Fig. 6A). Whether HPr in the phosphorylated form could further stimulate CcpA binding to DNA was tested by carrying out the EMSA reaction in the presence of HPr, HPrK/P, and ATP. Under these conditions we did not observe any additional stimulation of CcpA activity, but under our assay conditions we could not obtain more than 30% of HPr in the phosphorylated form (data not shown).

In order to determine whether the CcpA binding to the fragment corresponding to plasmid pAtxA10 was specific or not, the EMSA was carried out in the presence of HPr with the addition of a 300-fold excess of the same unlabeled PCR-amplified fragment present in pAtxA10 (specific fragment) or an unlabeled PCR-amplified fragment containing a 260-bp fragment internal to the atxA coding sequence (nonspecific fragment). As shown in Fig. 7 A, competition of CcpA binding to the labeled fragment was observed upon addition of the specific fragment, as expected. However, the nonspecific fragment also competed away CcpA from the labeled probe with the same efficiency as the specific fragment.

FIG. 7.

CcpA does not bind specifically to the atxA promoter. Competition EMSAs were carried out on the atxA, citZ, and BAS3893 promoter fragments (1 ng) in the presence of CcpA at the indicated concentrations and HPr at 10 μM. Specific DNA (+ S-DNA) or nonspecific DNA (+ NS-DNA) was added to 300-fold excess. The specific DNAs were the atxA (A), citZ (B), or BAS3893 (C) unlabeled fragments, while the nonspecific DNA in all three panels was the atxA fragment in the coding region generated with oligonucleotide primer set AtxAseq1-AtxAH199A.

These results raised the possibility that the DNA retardation observed was the result of nonspecific binding of CcpA due to its intrinsic DNA binding nature. The lack of a recognizable cre site in the probe being tested was consistent with this conclusion. Nevertheless, to further support the hypothesis that CcpA would not specifically bind to a DNA region containing the atxA P1 promoter, additional fragments spanning the atxA P1 and P2 promoter region were tested in the mobility shift assay, but essentially the same mobility profiles were obtained (data not shown).

These results also raised the possibility that the condition used for the mobility shift assay could affect CcpA binding specificity. Therefore, we searched for a positive-control promoter in order to test our assay conditions. We first used the promoter of the citZ gene (BAS4488) because citZ of B. subtilis is repressed by CcpA (34), the expression of the citZ gene of B. cereus is 2- to 3-fold higher in the ccpA mutant than in the parental strain, according to a comparative transcriptome study (72), and a palindromic sequence resembling the B. subtilis cre site was present approximately 70 bp upstream of the translational start codon. Thus, we labeled a 236-bp fragment containing the B. anthracis citZ promoter and used it in a mobility shift assay carried out under the same conditions used for the AtxA10 fragment. The results shown in Fig. 6B are essentially identical to the ones obtained with the atxA promoter fragment (Fig. 6A); also essentially identical were the results of the competition EMSA experiment (Fig. 7B), in which both the specific and nonspecific unlabeled probes competed for CcpA-dependent retardation of the labeled citZ probe. This suggested that either the EMSA conditions were not appropriate or neither fragment was specifically recognized by AtxA, raising the need to find another promoter that would act as a positive control.

The ccpC gene of B. subtilis is also regulated by CcpA via a cre site located approximately 90 bp upstream of the translational start site. Binding of CcpA to this cre site was shown to repress readthrough transcription from the upstream gene ykuL in response to the presence of glucose (33). Analysis of the nucleotide sequence surrounding the ccpC gene of B. anthracis (BAS3892) did not reveal the presence of any cre site in the putative promoter region. However, a cre site was proposed to exist in the putative promoter region of the upstream gene BAS3893, which encodes a protein that shares 55% identical residues with ykuL of B. subtilis (72). Thus, we amplified a 289-bp fragment encompassing this putative cre site and analyzed by EMSA whether CcpA could bind to it. As shown in Fig. 6C, using the same protein concentrations used for the AtxA10 and citZ fragments, mobility of the BAS3893 fragment was already detected with 0.25 μM CcpA. Mobility of this fragment was even obtained with concentrations as low as 50 nM CcpA (Fig. 6D). Furthermore, in an EMSA-competition assay, the labeled BAS3893 fragment was competed from CcpA binding by the specific DNA fragment (unlabeled BAS3893) but not by the nonspecific fragment (unlabeled atxA internal fragment) (Fig. 7C). The competition by the specific fragment was also higher than the competition by the nonspecific fragment at the same concentrations of CcpA used for the atxA and citZ promoter fragments (data not shown).

In order to determine whether CcpA had any role in regulating citZ or BAS3893 transcription in vivo, we generated lacZ fusion constructs in vector pTCV-lac and measured β-galactosidase activities in the parental strain and its ccpA mutant derivatives. As shown in Fig. 8, transcription of citZ was not affected by the ccpA mutation (Fig. 8A), while the transcription of BAS3893 was derepressed in the mutant strain (Fig. 8B).

FIG. 8.

Effect of a ccpA mutation on the expression of citZ and BAS3893. Time courses of β-galactosidase activity were carried out with B. anthracis strains harboring the pTCV-citZ (A) or the pTCV-BAS3893 (B) promoter-lacZ fusion constructs. Strains 34F2 (•) and 34F2ΔccpA (▪) were grown in R medium with 0.25% glucose and CO₂-bicarbonate.

To further understand the role of CcpA on the regulation of atxA, citZ, and BAS3893 and distinguish between direct or indirect effects, DNase I protection reactions were carried out on the same fragments used in the mobility assays. The results, shown in Fig. 9, indicated that while CcpA interacted with the atxA promoter throughout its length, as suggested by dispersed sites of protection and enhancement, no clearly defined protected region was obtained in multiple independent experiments (Fig. 9A and B). Similar patterns were observed with the citZ promoter fragment (data not shown). On the contrary, a clearly defined region of protection was identified on both strands of the BAS3893 promoter, in addition to a DNase I-hypersensitive site (Fig. 9C and D). The region of protection corresponded to the cre site predicted by the comparative transcriptome analysis carried out on B. cereus by van der Voort et al. (72) (Fig. 9E). This is consistent with CcpA being a direct regulator of BAS3893 transcription.

FIG. 9.

DNase I footprinting analysis of CcpA binding to the atxA (A and B) and BAS3893 (C and D) promoters. Fragments labeled with γ-³²P at the 5′ end (coding strand) (A and C) or at the 3′ end (noncoding strand) (B and D) were incubated with CcpA at the following concentrations: 0 μM (lanes 1 and 2), 0.4 μM (lane 3), 1 μM (lane 4), or 2 μM (lane 5). Protected regions in panels C and D are identified by thick lines. Hypersensitive sites are indicated by arrows. The known −10 and −35 promoter regions of atxA are labeled (13). (E) Nucleotide sequence of the BAS3893 promoter region identified by the thin line to the right of panel C. The regions protected in the 5′-labeled fragment are identified by lines above the sequence, while the region protected in the 3′-labeled fragment are shown by the line below the sequence. The region in gray highlights the cre site proposed by van der Voort et al. (72). The cre consensus sequences for B. cereus (38) and B. subtilus (22) are also shown.

Overall, this comparative analysis supports the conclusion that regulation of atxA transcription by glucose occurs through an indirect effect exerted by CcpA. Incidentally, we also show that the transcription of citZ in B. anthracis is not regulated by CcpA, while the BAS3893 promoter is now experimentally demonstrated to be directly regulated by CcpA.

CcpA is required for full virulence of B. anthracis.

In light of our observation that CcpA was required for full activation of atxA and pagA expression by glucose and with the assumption that glucose could be a signaling molecule of the mammalian host, we next tested whether a ccpA mutant strain would be less virulent than its isogenic parental strain in a mouse model of infection. Purified spores of strain 34F2 and strain 34F2ΔccpA were used to subcutaneously inoculate A/J mice, which are highly susceptible to the unencapsulated Sterne strain (77). Groups of 5 mice per strain were infected with ≈1 × 10⁶ spores. After 72 h from the time of inoculation, all the mice inoculated with the parental strain had succumbed to the infection. In contrast, significant attenuation (P = 0.0116) of virulence was shown in the group of mice infected with the ΔccpA strain, with the mice surviving up to 8 to 9 days after inoculation before succumbing to the infection (Fig. 10).

FIG. 10.

Inactivation of CcpA significantly decreased B. anthracis virulence. Groups of five A/J mice were inoculated subcutaneously with ≈1 × 10⁶ spores of strain 34F2 (•) or 34F2ΔccpA (▪). Percent survival was graphed with Kaplan-Meier survival analysis, and the P value by the log-rank test was 0.0116 (GraphPad Prism, version 5, software).

These results indicated that CcpA not only contributes ex vivo to full expression of toxin genes but also participates in vivo in B. anthracis virulence. Thus, sensing glucose is an important mechanism for triggering virulence in B. anthracis.

DISCUSSION

Recognition of the signals from the host environment is an essential bacterial function required for successful pathogenesis. Sensing of bicarbonate by B. anthracis is a well-recognized mechanism that induces virulence gene expression in response to this signaling molecule, an essential component of all mammalian bodies (5). In this study we have identified glucose as an additional signaling molecule that contributes to full expression of virulence factors in this organism. Because the systemic form of anthrax is characterized by extensive bacterial growth in the bloodstream, it seems reasonable that sensing glucose (normally present in the blood at a concentration of 0.1 to 0.15%) may provide further stimulation of toxin production.

The stimulatory effect of glucose was found to require the carbon catabolite protein CcpA through an indirect mechanism of transcription induction of the atxA gene at the P1 promoter. Deletion of the ccpA gene essentially abolished the induction of atxA transcription resulting from the addition of glucose to the culture medium, but no specific binding of CcpA to the atxA promoter region was detected and no region was protected in DNase I protection experiments. By means of comparisons of CcpA binding efficiency to a promoter not regulated in vivo by CcpA (citZ) and to a promoter regulated in vivo by CcpA (BAS3893) (Fig. 6 and 7), we convincingly concluded that the requirement for CcpA for full activation of atxA transcription in response to glucose is through an indirect transcriptional mechanism (although a requirement for an additional factor that may mediate CcpA interaction with the atxA promoter cannot be ruled out). Consistent with this conclusion was the observation that the entire atxA promoter region did not contain any cre site with nucleotide sequence similarity to the experimentally defined B. subtilis cre consensus sequence or the in silico-identified cre consensus sequence of B. cereus (46, 72).

It was reported that CcpA is required in S. pyogenes for expression of the Mga virulence regulator (2). Mga shares functional and structural similarities with AtxA, and like the atxA gene, its encoding gene is transcribed from two distinct promoters (7, 51, 70). The distal P1 promoter of mga is preceded by a putative cre site which is necessary for full mga expression. Binding of CcpA to an oligonucleotide containing the mga P1 cre site was reported, but an unusual excess of CcpA versus the amount of DNA was used and no footprinting confirmation of protein-DNA binding was obtained (2, 32). A similar approach was used to infer specific binding of CcpA to the S. pyogenes sag promoter (35), but a recent report by Kietzman and Caparon did not observe specific binding of CcpA to the CcpA-controlled sag promoter of S. pyogenes by means of in vivo protein-DNA pulldown assays (32). Thus, regulation of mga by CcpA may require further investigations.

The CcpA protein of B. subtilis has been shown to interact with the serine-phosphorylated form of the HPr protein, which stimulates binding to cre sites in response to accumulation of FBP, the activator of HPrK/P (23, 30, 43, 56, 61). Here we showed that the DNA binding activity of the B. anthracis CcpA protein was stimulated by the presence of unphosphorylated HPr (17, 37). We cannot exclude the possibility that fully phosphorylated HPr would have activated CcpA activity even further. Unfortunately, we could not get any more than 30% of HPr phosphorylated by HPrK/P, and under these conditions we did not detect further stimulation of CcpA binding activity (data not shown). Additionally, it was also shown that the DNA binding activity of the CcpA-HPr complex is enhanced by the presence of glucose-6-phosphate (G6P) and FBP (25, 62), but we did not detect any stimulation of B. anthracis CcpA binding activity under our assay conditions (data not shown). Further investigation of these subjects was outside the scope of our study.

In B. subtilis, three proteins, in addition to CcpA, were shown to be involved in carbon catabolite repression: CcpB, which regulates catabolite repression of the gluconate and xylose operons (12); CcpC, which is required for glucose repression of the aconitase and citrate synthase genes (31); and CcpN, which binds to a small regulatory RNA and two gluconeogenesis genes, gapB and pckA (20). Because the deletion of ccpA did not fully eliminate the induction of pagA expression by glucose, it appears that an additional regulatory mechanism that responds to glucose/carbon sources exists. The deletion of the ccpB gene alone did not affect pagA induction by glucose (Fig. 2) and it did not affect the pattern of pagA transcription when it was used in combination with the deletion of ccpA (data not shown). Genes encoding putative orthologues of B. subtilis CcpC and CcpN also exist in B. anthracis (56% and 74% identities, respectively), which could account for the residual induction of pagA expression in the ccpA mutant; however, their role, if any, in regulation of pagA transcription has yet to be determined.

Our study incidentally showed that, unlike in B. subtilis but similar to the situation in Listeria monocytogenes, in B. anthracis the citZ gene is not regulated by CcpA. Consistent with the in vivo data, the in vitro data did not detect specific binding of CcpA to the citZ promoter region, and no sequence was protected from DNase I treatment (data not shown). Also, regulation of ccpC expression in B. anthracis must differ from the regulation described for B. subtilis: we found that a predicted cre site in the promoter region of BAS3893 (ykuL) is bound by CcpA specifically in EMSA reactions and is protected from DNase I treatment in a footprint analysis. Thus, this site must be responsible for a direct role of CcpA in repressing the transcription of this gene, as the in vivo analysis also indicated (Fig. 8B). In B. subtilis, a cre site was identified downstream of the ykuL gene and shown to be a target of CcpA for repression of readthrough transcription, therefore affecting the overall level of transcription of the downstream ccpC gene. In B. anthracis the cre site is in the BAS3893 (ykuL) promoter region, and a strong stem-loop structure (ΔG = −18.1 kcal/mol) is predicted to be present at the 3′ end of the coding gene so that readthrough may not occur in this organism.

Overall, this suggests that, in addition to using slightly different cre sites, regulation of gene transcription by CcpA may differ substantially between B. subtilis and B. anthracis. Further support of this consideration comes from our observation that the B. anthracis promoter of the ackA gene (which is directly and positively regulated by CcpA in B. subtilis [26]) and the B. anthracis promoter of the phoP gene (which is directly and negatively regulated by CcpA in B. subtilis [55]) are not specifically recognized by the B. anthracis CcpA protein, suggesting the existence of further regulatory differences between these two organisms.

The virulence of a B. anthracis strain carrying a deletion of ccpA was found to be significantly attenuated, but the strain was still virulent in the A/J mouse model of infection. Attenuation, rather than abolishment, of virulence is consistent with the low and delayed expression of pagA observed by β-galactosidase assays (Fig. 2), although the reduced growth rate of the ccpA mutant ex vivo may also be a contributing factor in vivo. Regardless, the ccpA mutation affects the virulence of B. anthracis, and there seems to be a correlation between the level of growth/toxin gene expression ex vivo and the level of virulence in vivo, supporting further the assumption that growth in R medium in the presence of CO₂-bicarbonate mimics the growth conditions in the host environment.