|Home | About | Journals | Submit | Contact Us | Français|
Clostridium difficile is responsible for significant mortality and morbidity in the hospitalized elderly. C. difficile spores are infectious and are a major factor contributing to nosocomial transmission. The Spo0A response regulator is the master regulator for sporulation initiation and can influence many other cellular processes. Using the ClosTron gene knockout system, we inactivated genes encoding Spo0A and a putative sporulation-associated sensor histidine kinase in C. difficile. Inactivation of spo0A resulted in an asporogeneous phenotype, whereas inactivation of the kinase reduced C. difficle sporulation capacity by 3.5-fold, suggesting that this kinase also has a role in sporulation initiation. Furthermore, inactivation of either spo0A or the kinase resulted in a marked defect in C. difficile toxin production. Therefore, Spo0A and the signaling pathway that modulates its activity appear to be involved in regulation of toxin synthesis in C. difficile. In addition, Spo0A was directly phosphorylated by a putative sporulation-associated kinase, supporting the hypothesis that sporulation initiation in C. difficile is controlled by a two-component signal transduction system rather than a multicomponent phosphorelay. The implications of these findings for C. difficile sporulation, virulence, and transmission are discussed.
Clostridium difficile is a major nosocomial enteropathogen and is the primary cause of infectious hospital acquired diarrhea, mainly occurring in patients with a prior history of antibiotic therapy (9). The virulence of C. difficile is generally attributed to the production of toxins, which cause colonic damage and inflammation, leading to colitis and diarrhea. Transmission of C. difficile in health care institutions is mediated primarily by spores (31). Spores are infective and highly resistant and facilitate C. difficile persistence in aerobic environments outside of the host. Spores may also contribute to survival of C. difficile inside the host during antimicrobial therapy and subsequent recrudescence following the cessation of therapy (42). Interestingly, epidemic strains of C. difficile, such as ribotypes 001 and 027 (NAP1/BI), have an increased inherent sporulation capacity, indicating that this may be a factor in their spread (1, 10).
Generally, sporulation only occurs when efforts to maintain vegetative growth have failed. The signal transduction pathway that controls sporulation initiation has been most extensively studied in the genus Bacillus and is composed of an expanded variant of a two-component signal transduction system (TCS) known as the sporulation phosphorelay (6) (see Fig. S1 in the supplemental material). Environmental and cellular signals that indicate that vegetative growth is no longer possible are sensed by sporulation-associated sensor histidine kinases, resulting in autophosphorylation of a specific histidine in the catalytic domain. The phosphoryl group is transferred to an aspartate on the Spo0F response regulator and is subsequently relayed to the Spo0A response regulator transcription factor by the Spo0B phosphotransferase, which is transiently phosphorylated on a histidine. Phosphorylation of the active site aspartate of Spo0A promotes binding to a specific target sequence (the “0A box”) in or near the promoters of genes under Spo0A control, resulting in gene activation or repression (34).
Sensor kinases are responsible for activation of TCSs and phosphorelays via the input of phosphoryl groups. Sensor kinases have a modular domain organization with an N-terminal signal input domain and a catalytic C-terminal kinase domain containing the dimerization and histidine phosphotransfer (DHpt) subdomain and an ATP binding subdomain. In the sporulation phosphorelay, multiple sensor kinases phosphorylate the same response regulator substrate, Spo0F (35). For example, B. subtilis has five sensor kinases, KinA to KinE, which have highly conserved active sites (Fig. (Fig.1)1) and are capable of influencing sporulation (17). Generally, one sensor kinase is responsive to one specific signal ligand; thus, having several kinases increases signal sensing diversity and allows multiple signals to influence Spo0A activation.
Spo0A is highly conserved in Bacillus and Clostridium species, and 0A boxes can be identified upstream of genes likely to be controlled by Spo0A in clostridia (such as spo0A itself), suggesting that the mechanism of Spo0A-mediated gene regulation is also the same. Furthermore, the key residues of Spo0A known to mediate the interaction with the bases of the 0A box are highly conserved in Bacillus and Clostridium species (47). Despite these facts, true Spo0F and Spo0B homologues do not appear to be encoded within the genomes of sequenced Clostridium species, at least on the basis of protein sequence homology (35). The generally accepted hypothesis is that in Clostridium species the sporulation initiation pathway has not evolved into a multicomponent phosphorelay but has remained more like a TCS in which signals are sensed by sporulation-associated sensor kinases that phosphorylate Spo0A directly (39, 45).
Once activated, Spo0A controls many post-exponential-phase phenomena. Spore-forming bacterial pathogens have evolved to utilize Spo0A to modulate the virulence and survival responses that are essential for their pathogenic lifestyle (39). For example, Spo0A is involved in both sporulation and toxin production in Clostridium perfringens, presumably by binding to the putative 0A box upstream from the cpe cytotoxin gene (15). In C. difficile there is conflicting evidence for the existence of firm regulatory links between sporulation and toxin production. Several reports describing positive correlations between sporulation, stationary-phase events, and toxin production (8, 18, 20) are contradicted by those showing negative correlations or no link at all (2, 19). This confusing picture has been compounded by differences in the strains and the growth media used in the studies and by the paucity of C. difficile gene inactivation systems to construct defined mutants in key genes that would allow the dissection of the pathways involved. In the present study we have used the recently developed ClosTron gene knockout system (14) to inactivate the genes encoding Spo0A and a putative sporulation-associated sensor kinase to investigate their involvement in sporulation and toxin production in C. difficile.
To allow selection of ClosTron mutants the erythromycin-sensitive C. difficile strain 630 Δerm (16) was used. This strain is a spontaneously cured derivative of the genome sequencing strain 630, which is a virulent and multidrug-resistant isolate from an outbreak in Switzerland (32). Escherichia coli DH5α was used as the host for cloning steps and E. coli Rosetta (DE3; Novagen) was used as the host for overexpression and protein purification. Conjugative transfer of plasmids to C. difficile was carried out with E. coli CA434 (27) as the donor strain. The ClosTron delivery vector pMTL007 (14) was used as the parental plasmid for the construction of pMTL007::Cdi-spo0A-178a and pMTL007::Cdi-CD2492-254a, which retarget the group II intron of pMTL007 to spo0A and CD2492 of C. difficile. pGEX-6PI (GE Healthcare Life Sciences) was used as the vector for the expression and purification of proteins as GST fusions. All DNA manipulations were carried out according to standard molecular biological protocols (30). The sequences of the primers used in the present study are shown in Table S1 in the supplemental material.
Liquid cultures of C. difficile were grown in brain heart infusion (BHI; Oxoid) broth or Schaedlers anaerobic broth (SAB; Oxoid). Solid C. difficile cultures were grown on fresh blood agar (FBA) or BHI agar. In all cases, C. difficile was grown at 37°C in an anaerobic environment (80% N2, 10% CO2, and 10% H2 [vol/vol]). E. coli was grown aerobically in Luria-Bertani (LB) broth or on LB agar at 37°C. For conjugative transfer to C. difficile, E. coli was grown anaerobically on FBA. Unless stated otherwise, the following antibiotics were included in liquid and solid media as required; cycloserine (250 μg/ml), cefoxitin (8 μg/ml), chloramphenicol (12.5 μg/ml), thiamphenicol (15 μg/ml), and erythromycin (5 μg/ml).
The ClosTron system was used as described previously (14), in conjunction with the commercially available TargeTron gene knockout system kit (Sigma-Aldrich). Briefly, the TargeTron design software was used to design primers to retarget the group II intron on pMTL007 to spo0A (primers Cd-spo0A-178a-IBS, Cd-spo0A-178a-EBS1d, and Cd-spo0A-178a-EBS2; see Table S1 in the supplemental material) and CD2492 (primers IBS_CD1A, EBS1d_CD1A, and EBS2_CD1A; see Table S1 in the supplemental material). These primer sets were used with the EBS universal primer (see Table S1 in the supplemental material) and the intron template DNA provided with the TargeTron kit to generate a retargeted intron DNA fragment for each gene by overlap PCR according to the manufacturer's instructions. The two resultant DNA fragments (353 bp each) were cloned into the HindIII and BsrGI restriction sites of pMTL007 using E. coli DH5α as the host to produce pMTL007::Cdi-spo0A-178a and pMTL007::Cdi-CD2492-254a. The fidelity of the cloned inserts was verified by DNA sequencing.
Retargeted plasmids were transformed into E. coli CA434 using chloramphenicol as the selective marker and subsequently mated with C. difficile 630 Δerm on FBA. C. difficile transconjugant colonies were selected by subculturing on BHI agar containing cycloserine, cefoxitin, and thiamphenicol. To make the gene knockouts, C. difficile transconjugant colonies were grown in SAB, and the integration of the group II intron RNA into the spo0A and CD2492 genes was induced by the addition of 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside), followed by plating onto BHI agar containing erythromycin. Erythromycin-resistant (and thiamphenicol-sensitive) C. difficile colonies are produced, following plasmid loss and after insertion of the group II intron into the chromosome, which is accompanied by splicing out of the td group I intron from the ermB retrotransposition-activated marker (RAM).
Cultures of C. difficile were grown overnight in BHI broth containing antibiotics as required. The cells from 1.5 ml of culture were harvested by microcentrifugation, and the cell pellets were resuspended in 500 μl of solution I (50 mM Tris-HCl [pH 7.5], 20 mM NaCl, 2 mg of lysozyme/ml) and incubated at 37°C for 30 min. Then, 20 μl of proteinase K solution (20 mg/ml) and 350 μl of 2% (wt/vol) sodium dodecyl sulfate (SDS) were added to each sample, followed by incubation at 60°C for 2 h. The DNA was extracted using phenol-chloroform, precipitated with ethanol, and washed with 70% (vol/vol) ethanol. Precipitated DNA pellets were dissolved in 50 μl of distilled water, and contaminating RNA was removed by the addition of 1.25 μl of RNase (10 mg/ml), followed by incubation at 37°C for 30 min.
C. difficile sporulation was measured by direct microscopic counting of vegetative cells and spores as described previously (10). Overnight cultures of C. difficile were grown in SAB and used to inoculate BHI broth to an optical density at 550 nm (OD550) of 0.1. The cultures were incubated for 72 h, after which time samples were removed, and the cells or spores in each sample were pelleted by microcentrifugation. Cell or spore pellets were resuspended in phosphate-buffered saline (PBS), spread evenly over microscope slides, and air dried. The slides were stained with malachite green and counterstained with carbol fuchsin over boiling water to discriminate between spores and vegetative cells, respectively. C. difficile sporulation capacity was determined by counting at least five fields of 100 cells or spores per slide using a light microscope at ×1,000 magnification, and the number of spores was expressed as a percentage of the total number of cells.
Proteins in C. difficile culture supernatants or cell extracts were separated by SDS-polyacrylamide gel electrophoresis (PAGE) on a Mini-Protean II electrophoresis cell (Bio-Rad) using 7% (wt/vol) acrylamide gels. To obtain the samples for SDS-PAGE, 1-ml volumes of C. difficile cultures were removed, and the cells and supernatants were collected after microcentrifugation. Supernatant samples were loaded without further manipulation. Cell extracts representing total intracellular proteins were prepared by resuspending the cell pellets in 100 μl of lysis buffer (50 mM Tris-HCl [pH 7.5], 2 mg of lysozyme/ml, 2 mg of DNase I/ml), followed by incubation at 37°C for 30 to 45 min. The samples were boiled for 3 min, and then 900 μl of 50 mM Tris-HCl (pH 7.5) was added. Where required, proteins in C. difficile culture supernatants (1 ml) were precipitated by the addition of 500 μl of 50% (wt/vol) trichloroacetic acid (TCA; Sigma), followed by incubation on ice for 60 min. Precipitated proteins were pelleted by microcentrifugation, washed twice with acetone, and resuspended in 100 μl of 20 mM Tris-HCl (pH 7.5). For all samples, prior to gel loading, 5× SDS-PAGE sample buffer (100 mM Tris-HCl [pH 6.8], 5% [vol/vol] β-mercaptoethanol, 4% [wt/vol] SDS, 20% [vol/vol] glycerol, 0.2% bromophenol blue) was added, and the samples were boiled for 3 min. Where required, the volumes loaded onto the gels were corrected to take into account minor differences in culture ODs. Separated proteins in gels were stained with Coomassie blue as required.
To detect C. difficile toxin A in culture supernatants and cell extracts, proteins were separated by SDS-PAGE and transferred to nitrocellulose membranes by using a Mini-Transblot blotting cell (Bio-Rad). Membranes were blocked in Western blotting buffer (PBS, 0.1% Tween 20) containing 5% (wt/vol) skim milk and then incubated overnight at room temperature with the primary rabbit anti-toxin A polyclonal antibody (Fitzgerald Industries). After extensive washing in blotting buffer, membranes were incubated for 2 h at room temperature with a secondary goat anti-rabbit peroxidase-conjugated antibody (Jackson Immunoresearch). The secondary antibody was removed by extensive washing in blotting buffer, and the immobilized toxin A on the membranes was detected by using the SuperSignal West Pico chemiluminescent substrate (Pierce) according to the manufacturer's instructions, followed by exposure to X-ray film, scanning, and qualitative image analysis.
C. difficile cytotoxin titers (i.e., total toxins A and B) in culture supernatants were determined by using a Vero cell cytotoxicity assay as described previously (5, 12). Briefly, replicates of C. difficile culture supernatants were 10-fold serially diluted in sterile PBS and applied to Vero cell monolayers in 96-well microtiter plates. Plates were incubated at 37°C in a 5% (vol/vol) CO2 environment for 48 h and examined by using an inverted microscope. Positive toxin reactions were indicated by the characteristic rounding of Vero cells accompanied by parallel neutralization of cytotoxicity with Clostridium sordellii antitoxin (Prolab Diagnostics). Total cytotoxin titers in C. difficile culture supernatants were expressed as log10 relative units (RU), where 1 RU represents ~80% cell rounding by the undiluted sample, 2 RU represents ~80% cell rounding by the 10−1 dilution of the sample, and so on.
DNA fragments encoding full-length products of the spo0A and CD1579 genes were generated by PCR from C. difficile 630 genomic DNA using primer pairs CDSpo0A_For/CDSpo0A_Rev and CD3B_For/CD3B_Rev, respectively (see Table S1 in the supplemental material). The resulting DNA fragments were ligated into the BamHI and XhoI restriction sites of pGEX-6P1 to produce genes encoding fusion proteins with glutathione S-transferase (GST) at their N termini. The fidelity of the process was confirmed by DNA sequencing. Recombinant proteins were overexpressed in E. coli Rosetta (DE3) cells grown at 37°C in LB broth supplemented with ampicillin (270 μM) and chloramphenicol (105 μM). Expression was induced by the addition of 1 mM IPTG at an OD550 of 0.5. Incubation was continued for 4 h, and the cells were harvested by centrifugation at 4,500 × g for 20 min.
Cells were resuspended in 20 ml of buffer A (20 mM Tris-HCl [pH 7.5], 200 mM NaCl) and disrupted by sonication. GST-tagged proteins were purified from cleared lysates by using GST-Sepharose columns (GE Healthcare Life Sciences). Briefly, cell lysates were applied to columns pre-equilibrated with three column volumes of buffer A. Columns were washed with five column volumes of buffer A, and proteins were eluted with the same buffer containing 50 mM reduced glutathione. Fractions containing the desired proteins were identified by using SDS-PAGE, pooled, and then treated with 80 μg of C3 protease at 4°C for 14 h to cleave the GST tag. The proteins were concentrated by using Centricon ultrafiltration devices (Millipore) of appropriate molecular weight cutoff and separated by size exclusion chromatography using Superdex S75 or S200 (GE Healthcare Life Sciences) for Spo0A or CD1579, respectively. Proteins were concentrated to ~2 mg/ml as before and stored at −80°C until required.
The autophosphorylation activity of sensor kinases and phosphotransfer to Spo0A were assayed as described previously (21, 40). Reactions (20 μl) were carried out in 50 mM EPPS buffer (Sigma) containing 20 mM MgCl2, 0.1 mM EDTA, and 5% (vol/vol) glycerol. Autophosphorylation reactions contained 1 μM kinase and, where required, Spo0A was included at a final concentration of 10 μM to measure the phosphotransfer. In all cases, the reactions were initiated by the addition of 100 μM ATP containing 50 μCi of [γ-32P]ATP (Perkin-Elmer). At various time intervals, reactions were terminated by the addition of 5× SDS-PAGE sample buffer on dry ice. Labeled protein samples were separated by SDS-PAGE on a 10% (wt/vol) acrylamide gel and visualized by phosphorimaging using a Fuji BAS1000 phosphorimager.
In Bacillus species sporulation-associated phosphorelay sensor kinases have certain characteristics that distinguish them from those of TCSs (39). Unlike most TCSs, the genes encoding sporulation-associated kinases are chromosomally located as orphans, i.e., they are not adjacent to a gene encoding their cognate response regulator. Furthermore, since all of the sporulation kinases have the same substrate, Spo0F, they share a high degree of sequence conservation around the active site histidine and in the other regions of the DHpt subdomain that interact with Spo0F and/or determine specificity (Fig. (Fig.1).1). Using these features, it is possible to identify potential sporulation-associated sensor kinases within the genome sequences of spore-forming bacteria (35).
The genome of C. difficile 630 (32) encodes 50 sensor kinases and 51 response regulators that are organized into 45 TCSs, with five orphan kinases and six orphan response regulators scattered around the chromosome. The orphan kinases are encoded by the CD1352, CD1492, CD1579, CD1949, and CD2492 genes. From here on, sensor histidine kinase proteins encoded by the relevant genes will be denoted by the prefix HK followed by the locus tag.
Sporulation-associated sensor kinases of B. subtilis have highly variable signal input domains but are generally well conserved (39 to 50% amino acid identity) in their catalytic domains. Similarly, HK CD1492, HK CD1579, and HK CD2492 share 39 to 44% amino acid identity in their catalytic domains, but the signal input domains are variable in sequence. In addition, in a phylogenetic analysis of all C. difficile 630 sensor kinases HK CD1492, HK CD1579, and HK CD2492 group together in a single cluster, whereas the other orphan kinases fall into separate and unrelated clusters (Y.-F. Lin and K. Stephenson, unpublished data). The DHpt subdomains, including the active sites of the orphan kinases, were identified by using the Pfam database (http://pfam.sanger.ac.uk/) (Fig. (Fig.1).1). The active sites of HK CD1492, HK CD1579, and HK CD2492 share a high degree of sequence identity, which is at a level comparable to that shared by B. subtilis KinA to KinE (Fig. (Fig.1).1). The active sites of HK CD1352 and HK CD1949 do not exhibit good homology to each other or to the other C. difficile orphan kinases. Furthermore, a number of the amino acids that define the interaction surface with the response regulator and/or are involved in specificity of the interaction (33, 44, 46) are strictly conserved in HK CD1492, HK CD1579, and HK CD2492 but are different in HK CD1352 and HK CD1949 (Fig. (Fig.1).1). Taken together, this suggests that HK CD1492, HK CD1579, and HK CD2492 may be able to phosphorylatie the same response regulator substrate, Spo0A, and are therefore potentially involved in C. difficile sporulation initiation.
HK CD1492 and HK CD2492 are putative integral membrane proteins, whereas HK CD1579 is likely to be a cytosolic kinase (see Fig. S2 in the supplemental material). Domains with recognized signaling functions were not detected in HK CD1492 and HK CD2492 using Pfam. HK CD1579 was predicted to contain a PAS domain in its signal input domain (see Fig. S2 in the supplemental material). A PAS domain at the N terminus of B. subtilis KinA has a critical and indispensable role in the activity of the kinase (38, 43).
We used the ClosTron system to construct a knockout mutant in which the gene encoding HK CD2492 was inactivated by insertion of the group II intron from pMTL007. In parallel, we also constructed a mutant in which the spo0A gene had been inactivated as described previously (14). The general gene inactivation process is represented schematically in Fig. Fig.22.
pMTL007::Cdi-spo0A-178a and pMTL007::Cdi-CD2492-254a retarget the group II intron to insert into the spo0A and CD2492 genes in the antisense orientation immediately after the 178th or 254th nucleotide in the coding sequence, respectively. The plasmids were transferred to C. difficile 630 Δerm by conjugation, and intron insertion was induced by IPTG. Genomic DNA was isolated from erythromycin-resistant transconjugants. This DNA was subjected to PCR using (i) the RAM-F/RAM-R primer pair, (ii) the Target_R_Spo0A/EBS universal primer pair, and (iii) the Target_R_CD1A/EBS universal primer pair. As a control, unmodified pMTL007 plasmid DNA was also subjected to PCR with the same primer pairs.
In the pMTL007 control only the RAM-F/RAM-R primer pair produced a PCR product (Fig. (Fig.2).2). This DNA fragment (1,300 bp) represents the erm RAM prior to splicing out of the td group I intron. As expected, PCR products were not amplified from the pMTL007 template using the Target_R_Spo0A/EBS primer pair or the Target_R_CD1A/EBS primer pair since this requires integration of the group II intron into the respective target genes (Fig. (Fig.2).2). When genomic DNA isolated from C. difficile 630 Δerm carrying pMTL007::Cdi-spo0A-178a or pMTL007::Cdi-CD2492-254a was used as the template for PCR with the RAM-F/RAM-R primer pair, a product of 900 bp was amplified in both cases (Fig. (Fig.2B).2B). These products are 400 bp smaller than the one amplified from the unmodified pMTL007 template with the same primer pair and represent the erm RAM after splicing out of the td group I intron. Since splicing out of the group I intron accompanies insertion of the group II intron into the chromosome, the presence of these smaller RAM-F/RAM-R products confirms the insertion event. Furthermore, the failure to amplify a 1,300-bp product using the RAM-F/RAM-R primers confirms the absence of extrachromosomally replicating plasmid, as would be expected due to the known instability of this plasmid replicon in C. difficile (14, 27).
To verify that the group II intron had inserted into the correct target gene, PCR was carried out by using both the Target_R_Spo0A/EBS and the Target_R_CD1A/EBS primer pairs. PCR products could only be amplified from the pMTL007::Cdi-spo0A-178a and pMTL007::Cdi-CD2492-254a DNA templates by using the respective Target_R_Spo0A/EBS and Target_R_CD1A/EBS primer pairs (Fig. (Fig.2B).2B). These spo0A and CD2492 PCR products were 425 and 246 bp in size reflecting differences in the relative site of annealing of the gene specific primers within their target genes. Overall, this set of PCRs confirms that the erm RAM inserted into the C. difficile chromosome at the correct site and in the correct orientation, resulting in the inactivation of the spo0A and CD2492 genes. For unknown reasons, despite multiple attempts, it was not possible to generate knockout mutants of CD1492 or CD1579 using the ClosTron system.
Batch cultures of C. difficile 630 Δerm, C. difficile 630 Δerm/pMTL007::Cdi-spo0A-178a, and C. difficile 630 Δerm/pMTL007::Cdi-CD2492-254a were cultivated in BHI broth, and growth was monitored by measuring the OD550. All strains grew well in BHI, and vegetative cells of the parental strain and knockout mutants were indistinguishable using light microscopy (data not shown). The spo0A and CD2492 mutants grew with the same exponential growth rate as the parental strain and the strains had a mean generation time of 66 min (see Fig. S3 in the supplemental material). All cultures reached the same OD550 at the end of the exponential phase and exhibited the same pattern of stationary phase growth with a gradual reduction in OD over time (see Fig. S3 in the supplemental material). Therefore, neither inactivation of the genes encoding Spo0A or HK CD2492 or the presence of the integrated group II intron in the genome had any significant effects on the growth or viability of vegetative C. difficile when cultured in batch in a rich growth medium.
In contrast, the sporulation capacities of C. difficile 630 Δerm/pMTL007::Cdi-spo0A-178a and C. difficile 630 Δerm/pMTL007::Cdi-CD2492-254a were markedly affected compared to the parental strain. After 72 h of incubation in BHI, C. difficile 630 Δerm produced 14% spores (Fig. (Fig.3).3). Although low compared to other spore-forming bacteria such as B. subtilis (38), this inherent sporulation capacity of C. difficile 630 Δerm is reproducible and comparable to batch cultures of other virulent C. difficile isolates cultivated for the equivalent time period (10). Consistent with previous observations (14), inactivation of spo0A in C. difficile resulted in a complete inability to sporulate (Fig. (Fig.3).3). Furthermore, inactivation of the gene encoding HK CD2492 in C. difficile also resulted in a reduction in the number of spores produced but did not completely abolish sporulation. The sporulation capacity of C. difficile 630 Δerm/pMTL007::Cdi-CD2492-254a was reduced to 4% after 72 h of cultivation (Fig. (Fig.33).
For several other spore-forming pathogens, Spo0A is known to influence the production of key virulence factors such as toxins (13, 15, 22, 29). To determine whether a similar phenomenon exists in C. difficile, we investigated toxin production in batch-grown cultures of C. difficile 630 Δerm, C. difficile 630 Δerm/pMTL007::Cdi-spo0A-178a, and C. difficile 630 Δerm/pMTL007::Cdi-CD2492-254a.
Cultures were grown in BHI broth and, after 48 h of incubation, the extracellular and intracellular levels of toxin A were detected by using Western blotting. Toxin A could be detected intracellularly and extracellularly in all strains. In the parental strain high levels of toxin A were detected intracellularly, and this was accompanied by high levels of toxin A released into the culture supernatant (Fig. (Fig.4,4, lane A). In stark contrast, the amount of toxin A detected in C. difficile 630 Δerm/pMTL007::Cdi-spo0A-178a was reduced significantly in both the intracellular fraction and the culture supernatant (Fig. (Fig.4,4, lane B). Using ImageJ image analysis software (http://rsb.info.nih.gov/ij/), we estimated the level of toxin A in intracellular and extracellular fractions of the spo0A mutant to be <10% of that of the parental strain. Furthermore, the amount of toxin A detected in C. difficile 630 Δerm/pMTL007::Cdi-CD2492-254a was also reduced but not to the same extent as the spo0A mutant (Fig. (Fig.4,4, lane C). In this case, using ImageJ, we estimated the level of intracellular and extracellular toxin A to be 50 to 60% of that of the parental strain.
Although highly informative, Western blotting did not take into account the activity of toxins or any effects of the gene inactivation on the production of toxin B. Therefore, these data were confirmed by using the Vero cell cytotoxicity assay, which measures the combined activity of toxin A and toxin B. Using this assay, we measured the toxin titers in the culture supernatants of C. difficile 630 Δerm, C. difficile 630 Δerm/pMTL007::Cdi-spo0A-178a, and C. difficile 630 Δerm/pMTL007::Cdi-CD2492-254a after 48 h of incubation in BHI to be 4, 1, and 3 log10 RU, respectively. All cultures were at the equivalent point in stationary phase and had the same OD550 on sample removal. Since each toxin titer unit represents a log10 difference in the total cytotoxic activity in the supernatant, inactivation of spo0A resulted in a 1,000-fold reduction in activity compared to the parental strain, whereas inactivation of CD2492 resulted in a 10-fold reduction. These data are in agreement with the qualitative Western blotting data.
To rule out the possibility that the lower levels of extracellular toxin in the spo0A and CD2492 mutants were a consequence of a general defect in protein export, the total proteins in supernatants of stationary phase cultures were analyzed by SDS-PAGE. As shown in Fig. Fig.5,5, many proteins are exported by C. difficile, but there were no large-scale reductions in the number or amounts of total proteins released by the spo0A and CD2492 mutants that could account for the observed reductions in extracellular toxin. In addition, comparable reductions in the amount of a high-molecular-mass protein, most likely toxin A/B, are clearly visible on the stained gel of the proteins isolated from the supernatants of C. difficile 630 Δerm/pMTL007::Cdi-spo0A-178a and C. difficile 630 Δerm/pMTL007::Cdi-CD2492-254a (Fig. (Fig.5,5, asterisk).
The genetic evidence discussed above suggests that HK CD2492 is involved in sporulation initiation in C. difficile. The absence of Spo0F or Spo0B in C. difficile raises the possibility that HK CD1492, HK CD1579, and HK CD2492 phosphorylate Spo0A directly. HK CD1492 and HK CD2492 are large (>100 kDa) integral membrane proteins (see Fig. S2 in the supplemental material). The integral membrane nature of many sensor kinases limits investigation of their biochemical properties due to problems with heterologous overexpression, solubility, and kinase activity. There are only a few examples of overexpression and purification of intact and active full-length integral membrane sensor kinases (4, 11, 26).
In the present study, it was not possible to overexpress and purify the intact full-length HK CD2492 to investigate its biochemical properties. However, due to the high degree of overall conservation in the active site and DHpt subdomain (Fig. (Fig.1),1), we reasoned that it is probable that both HK CD2492 and HK CD1579 recognize and phosphorylate the same response regulator in C. difficile. Like the main B. subtilis sporulation-associated sensor kinase KinA (23), HK CD1579 is likely to be cytosolic, rendering the full-length protein amenable to heterologous expression and purification. The genes encoding HK CD1579 and Spo0A from C. difficile were cloned into pGEX-6PI and purified as described above. On the basis of SDS-PAGE, the HK CD1579 and Spo0A proteins were estimated to be 80 and 95% pure, respectively (data not shown).
The ability of HK CD1579 to autophosphorylate and transfer the phosphoryl group to C. difficile Spo0A was investigated by using in vitro phosphorylation assays (21, 40, 43). The assay measures the incorporation of 32P into sensor kinases from [γ-32P]ATP and subsequent transfer to a response regulator. Upon the addition of [γ-32P]ATP to HK CD1579 in the absence of Spo0A, a radiolabeled protein of high intensity was detected after 20 min of incubation (Fig. (Fig.6A,6A, control), which corresponds to the phosphorylated kinase (HK CD1579~P). This confirms that the purified protein was active as a kinase under these reaction conditions. When Spo0A was included in the reaction mix from t = 0, the amount of HK CD1579~P increased with time, but the level was much reduced compared to the control, and this was accompanied by a steady increase in the amount of phosphorylated Spo0A (Spo0A~P) with time (Fig. (Fig.6A).6A). This demonstrates that HK CD1579 is able to phosphorylate Spo0A in vitro. In the absence of HK CD1579, Spo0A does not become phosphorylated by [γ-32P]ATP (data not shown).
To exclude the possibility that the phosphorylation of Spo0A by HK CD1579 was a result of nonspecific phosphorylation, the assay was carried out using a heterologous sensor kinase from Propionibacterium acnes, HK PPA0945, which was purified by using the same vector and conditions as part of an unrelated project in our laboratory. After 20 min of incubation in the absence of Spo0A, a radiolabeled protein corresponding to phosphorylated HK PPA0945 (HK PPA0945~P) was visible on the gel (Fig. (Fig.6B,6B, control) at a level comparable to HK CD1579~P (Fig. (Fig.6A,6A, control). This confirms that HK PPA0945 is active under these assay conditions and that the autophosphorylation activity is similar to that of HK CD1579. However, when C. difficile Spo0A was included in the reaction the level of HK PPA0945~P increased steadily with time but no Spo0A~P could be detected after 20 min (Fig. (Fig.6B);6B); at this time, the level of HK PPA0945~P was comparable to that of the control lacking Spo0A. Therefore, in contrast to HK CD1579, HK PPA0945 did not phosphorylate Spo0A. This is strong evidence that the direct phosphorylation of Spo0A by HK CD1579 is specific. Coomassie blue staining of all gels confirmed the presence of proteins at the expected concentrations (data not shown).
Sporulation is a complex developmental process that is used by certain bacteria to survive conditions that cannot support vegetative growth and/or those that threaten integrity or viability of the cell. Despite the central importance of spores in transmission, persistence and pathogenesis of C. difficile, the signal transduction pathway controlling sporulation initiation in this bacterium has not been studied in detail.
Using bioinformatics, we identified three putative C. difficile sporulation-associated sensor kinases, HK CD1492, HK CD1579, and HK CD2492. Inactivation of the gene encoding HK CD2492 had no observable effects on C. difficile growth, viability, or gross cell morphology but reduced the sporulation capacity by 3.5-fold, providing strong genetic evidence for the involvement of HK CD2492 in sporulation initiation. The reduction in sporulation capacity of the C. difficile CD2492 mutant is comparable to the ~3-fold reduction in B. subtilis sporulation resulting from insertional inactivation of kinA, the gene encoding the main sporulation-associated phosphorelay sensor kinase of this bacterium (23).
Inactivation of CD2492 did not abolish sporulation completely, suggesting that other sensor kinases are likely to be involved in sporulation initiation. In B. subtilis, the involvement of alternative sensor kinases such as KinB in sporulation only becomes apparent after inactivation of kinA (41). HK CD1492 and HK CD1579 are the likely candidates for this role in C. difficile, and the residual sporulation observed in C. difficile 630 Δerm/pMTL007::Cdi-CD2492-254a is likely due to the activity of these alternative kinases. Despite repeated attempts to retarget the group II intron to the same and different insertion sites, we were unable to obtain knockout mutants in CD1492 or CD1579. Although highly unlikely due to the active site/DHpt conservation and the fact that HK CD1579 can phosphorylate Spo0A, the possibility that the kinases themselves or an additional non-Spo0A response regulator substrate for these kinases is essential for viability cannot be ruled out. Nevertheless, the effect of HK CD2492 on C. difficile sporulation is clear and is comparable to the best characterized of all spore-forming gram-positive bacteria, B. subtilis.
It was not possible to clone, overexpress, and purify HK CD2492 in an intact form from E. coli to study its biochemical activity and confirm the genetic data. Similar problems have been reported previously with Clostridium botulinum sensor kinases (45). As a tractable alternative, we investigated the biochemical activity of HK CD1579, a smaller-molecular-mass sensor kinase (~72 kDa) lacking membrane spanning segments which, unlike HK CD2492, was amenable to cloning, overexpression, and purification in its full-length intact form. HK CD1579 was active as a kinase and could phosphorylate Spo0A efficiently and at a level similar to that of B. subtilis KinA on Spo0F (40), whereas the heterologous HK PPA0945 could not. This suggests that the phosphorylation of Spo0A by HK CD1579 is specific. Under similar reaction conditions KinA does not readily phosphorylate B. subtilis Spo0A (7). The data presented here represent the first report of the direct phosphorylation of Spo0A by a sensor kinase in C. difficile or other clostridia and strongly support the hypothesis that Spo0A is directly phosphorylated by sporulation-associated sensor kinases in the anaerobic spore formers. Due to the conservation of key residues, HK CD1492 and HK CD2492 would also be expected to possess the same specificity and to phosphorylate Spo0A.
The amino acid sequences of the signal input domains of HK CD2492, HK CD1492, and HK CD1579 are highly variable, indicating that each is responsive to a different signal ligand. Furthermore, BLAST analysis of these kinases with microbial genome sequences in the NCBI database (http://blast.ncbi.nlm.nih.gov/blast.cgi) reveals that orthologues of these proteins are found in all C. difficile strains currently held in that repository but not in closely related Clostridium species or other genera of bacteria. The C. difficile strains harboring these sensor kinase orthologues include highly virulent strains of current clinical importance such as ribotype 027 (NAP1/BI). This suggests that the signals required to initiate sporulation in C. difficile are unique and specific to this species, as is known to be the case for members of the genus Bacillus (35). The precise nature of sporulation-inducing signal ligands has generally remained elusive. In C. perfringens, inorganic phosphate has been shown to be a signal capable of promoting spo0A expression, sporulation, and toxin synthesis (25), but any involvement of sporulation-associated sensor kinases in this process has yet to be demonstrated.
Inactivation of spo0A in C. difficile 630 Δerm resulted in a mutant that was completely asporogenous. This is in agreement with previous studies on Spo0A in clostridia (3, 14, 15) and highlights the central and essential role of Spo0A in sporulation initiation in C. difficile. Inactivation of spo0A also led to a drastic reduction in the level of toxin produced during stationary phase in batch cultures. We have observed similar effects in a chemostat-based model of C. difficile infection of the human gut (unpublished data). Inactivation of CD2492 resulted in less severe, but nevertheless significant, reductions in toxin production, as would be expected if HK CD2492 was one of multiple sensor kinases contributing to Spo0A activation. The reduction in extracellular toxin A is mirrored by the drop in intracellular levels of this protein in the CD2492 and spo0A mutants. The defect in toxin production thus most likely occurs at the level of toxin transcription and/or translation, as would be expected given the well established role of Spo0A in transcriptional control and cellular regulation. Based on these data there appears to be a direct correlation between the level of toxin produced and the relative level of sporulation, which are influenced by HK CD2492 and Spo0A. This confirms the existence of firm regulatory links between the sporulation initiation pathway and toxin production in C. difficile.
Although Spo0A is known to be involved in the virulence responses of other spore-forming human pathogens, this is, to our knowledge, the first report in C. difficile. Spo0A is a highly pleiotropic cellular regulator that has a dramatic influence on global gene expression. Therefore, it is not surprising that C. difficile has adapted this regulatory protein and the pathway that controls its activation status to modulate virulence responses. A pertinent question is how does Spo0A exert its effects on C. difficile toxin production? In the C. difficile genome >200 open reading frames have potential 0A boxes within 200 bp of their start codons, indicating direct regulation by Spo0A (28). However, tcdA and tcdB, encoding toxin A and toxin B, respectively, are not among them. Therefore, Spo0A is likely to have an indirect influence on toxin production, and many candidate regulators are encoded by the genes putatively under the direct control of Spo0A in C. difficile (28). In other pathogenic spore-forming bacteria the regulatory networks that influence sporulation initiation and virulence are intimately intertwined (24), and it appears that a similar phenomenon exists in C. difficile. What is clear is that the regulatory networks that contribute to both C. difficile sporulation initiation and toxin regulation are likely to be highly complex.
Histidine kinase-mediated signal transduction pathways, such as phosphorelays and TCSs, are known to be integral elements of the virulence responses of bacterial pathogens and consequently are recognized targets for the development of novel antimicrobial agents (36, 37). Therefore, selective inhibition of Spo0A or other elements of the signaling pathway that influences Spo0A activity has the potential to simultaneously reduce toxin production and inhibit sporulation of C. difficile. Ultimately, this could have important consequences for the therapy of C. difficile infections and the control of transmission and persistence of this highly significant nosocomial pathogen.
The study was supported in part by BBSRC, Leeds University, The Royal Society, and The Hospital Infection Society.
We thank Nigel P. Minton and John T. Heap for providing the ClosTron gene knockout system.
Published ahead of print on 25 September 2009.
†Supplemental material for this article may be found at http://jb.asm.org/.