ABSTRACT
We show in this study that toxin production in Clostridium difficile is altered in cells which can no longer form flagellar filaments. The impact of inactivation of fliC, CD0240, fliF, fliG, fliM, and flhB-fliR flagellar genes upon toxin levels in culture supernatants was assessed using cell-based cytotoxicity assay, proteomics, immunoassay, and immunoblotting approaches. Each of these showed that toxin levels in supernatants were significantly increased in a fliC mutant compared to that in the C. difficile 630 parent strain. In contrast, the toxin levels in supernatants secreted from other flagellar mutants were significantly reduced compared with that in the parental C. difficile 630 strain. Transcriptional analysis of the pathogenicity locus genes (tcdR, tcdB, tcdE, and tcdA) revealed a significant increase of all four genes in the fliC mutant strain, while transcription of all four genes was significantly reduced in fliM, fliF, fliG, and flhB-fliR mutants. These results demonstrate that toxin transcription in C. difficile is modulated by the flagellar regulon. More significantly, mutant strains showed a corresponding change in virulence compared to the 630 parent strain when tested in a hamster model of C. difficile infection. This is the first demonstration of differential flagellum-related transcriptional regulation of toxin production in C. difficile and provides evidence for elaborate regulatory networks for virulence genes in C. difficile.
INTRODUCTION
Clostridium difficile is a Gram-positive spore-forming bacillus which is recognized to be the major cause of nosocomial diarrhea associated with antibiotic therapy (35). The incidence of C. difficile infection has been rapidly increasing in both Europe and North America, and this increase in infections has been associated with a significantly high mortality rate (2, 35). The broad spectrum of diseases caused by C. difficile, which range from antibiotic-associated diarrhea to the potentially lethal, pseudomembranous colitis, has been shown to depend on the level of toxin produced (1), and this production is recognized as a critical determinant of pathogenicity. Following antibiotic therapy when the microbiota of the gastrointestinal tract is disrupted, infection by C. difficile is mediated by spores which germinate in the gut, followed by vegetative cell proliferation and the subsequent secretion of the two major virulence factors, the Rho glucosylating toxins TcdA and TcdB.
The toxin-encoding genes (tcdA and tcdB) are localized to a 19.6-kb pathogenicity locus (PaLoc) which includes three other accessory genes, tcdR, tcdE and tcdC (43). TcdR is an alternative sigma factor required for transcription of the two toxin genes; TcdE has been described to be a putative holin-like protein involved in toxin secretion, although this role has recently been a source of debate (17, 32); and TcdC is an anti-sigma factor that negatively regulates tcdR-dependent transcription (28, 29). In addition, four other regulators of toxin synthesis have recently been identified: CepA (3), CodY (10), SpoOA (42), and SigH (36). The role of both toxins in virulence was recently reviewed (6).
Recent studies have demonstrated that the C. difficile flagellin structural protein FliC is glycosylated and glycosylation is required for assembly of functional flagella (41). In addition, comparison of C. difficile 630 and hypervirulent genome sequences revealed differences in the gene content of the flagellum locus, especially with respect to flagellar glycan biosynthetic genes (43).
Flagellar motility is critical for the colonization of respective hosts by many bacterial pathogens, most notably, the gastrointestinal pathogens Campylobacter jejuni (47), Vibrio cholerae (34, 37), and Helicobacter pylori (13). Additionally, in the case of V. cholerae, while motility is essential for initial infection, recent work has suggested that for the organism to persist and cause disease, the loss of motility is a critical factor (25, 39, 40). Studies of V. cholerae pathogenesis have revealed that virulence gene expression which includes toxin production is modulated by the regulatory network which governs flagellar assembly (25, 37, 39). Loss of flagella through breakage of the filament upon penetration of the mucosal layer results in loss of anti-sigma factor FlgM and subsequent activation of the alternative sigma factor, fliA. This sigma factor represses the quorum-sensing controlled transcriptional regulator HapR and allows increased expression of virulence factors (including cholera toxin), revealing an interesting interplay between motility and virulence gene expression (40). In C. jejuni, while regulation of virulence factors may not be controlled by the flagellar regulon, secretion of virulence proteins has been shown to be dependent on a functional flagellar export apparatus (23).
To date, little is known about the process of spore germination and C. difficile colonization in the gastrointestinal tract. Following germination, the organism would be required to penetrate the mucous layer and adhere to the underlying colonic epithelial cells, whereupon toxins would be released for uptake by host epithelial cell receptors. While both S layer and flagella have been reported to be involved in colonization (5), the process whereby the organism senses this new environment and regulates virulence gene expression, including production and secretion of toxins, is poorly understood. In the current study, we provide the first evidence that toxin production can be regulated by the flagellar regulon.
MATERIALS AND METHODS
Bacterial strains and growth conditions.Strains were routinely grown on brain heart infusion (BHI) agar medium supplemented with 5 g/liter yeast extract, 1.2 g/liter NaCl, 0.5 g/liter cysteine HCl, 5 mg/liter hemin, 1 mg/liter vitamin K, and 1 mg/liter resazurin. Erythromycin was added (2.5 μg/ml) for flagellar mutant strains. For toxin expression studies and assessment of toxin production, strains were inoculated into preequilibrated TY broth (3% tryptone, 2% yeast extract) and the flasks were incubated in an anaerobic chamber without shaking for 4, 7, and 24 h.
Construction of mutants.Target sites were identified for each gene using a Targetron gene-knockout system (Sigma-Aldrich), and mutants were generated according to the method of Heap et al. (18, 19) using C. difficile strain 630Δerm. Transconjugants were streaked onto BHI plates supplemented with 2.5 μg/ml erythromycin to select for integrants and 250 μg/ml cycloserine and 8 μg/ml cefoxitin to select against the Escherichia coli donor. A minimum of two clones for each mutant strain were checked by specific PCR using flanking primers to each gene to verify gene disruption by the erm cassette and for further phenotypic studies. For complementation studies of the 630Δerm fliC::erm mutant, a 1,141-bp fragment encompassing the fliC structural gene and 277 bp of 5′ noncoding region was amplified by PCR and cloned into plasmid pMTL84151 to generate plasmid pMTL-pfliC. For complementation studies of the 630Δerm fliM::erm mutant, the heterologous promoter and ribosome binding site of the thiolase gene (thl) were cloned upstream of the fliM structural gene in pMTL84151, generating pMTL-PthlfliM, as described by Heap et al. (19), and the sequence was confirmed by sequence analysis of the plasmid. Both pMTL-pfliC and pMTL-PthlfliM were transferred into the respective C. difficile 630Δerm mutant strains by conjugation, and these transformants were tested for restoration of motility. Empty pMTL84151 vector was used as a control for each strain.
Motility assay.Motility assays were performed using motility agar tubes containing BHI medium containing 0.175% agar. These tubes were stab inoculated, and the contents were allowed to grow anaerobically at 37°C for 16 h (41).
Cell monolayer rounding assay.Human lung fibroblast (HLF) cells were used to assess the presence of TcdA and TcdB in C. difficile culture supernatants. HLF cells were maintained and seeded in 96-well microtiter plates as described previously (21). Before they were added to HLF cells, C. difficile culture supernatants (from 4, 7, and 24 h of growth) were centrifuged, sterile filtered, and diluted in TY broth. Thirty microliters of supernatants (undiluted and 1/10, 1/100, and 1/1,000 dilutions) was added to 200 μl of confluent HLF monolayers (time zero), and the extent of cell rounding was observed at 24 h posttreatment using the cytopathic effect (CPE) scoring system (4). Three independent assays were performed with duplicate experimental wells in each assay. Statistical analysis was performed in GraphPad Prism software (version 4.02; La Jolla, CA) using one-way analysis of variance (ANOVA), with Bonferroni's test comparing the level of cell rounding from mutant strains to that of the 630Δerm parent strain.
Toxin detection by ELISA.Two sandwich enzyme-linked immunosorbent assays (ELISAs) were performed to quantify the amount of TcdA or TcdB in cell culture supernatants. In the first assay for TcdA quantification, 2.5 μg/well of the anti-TcdA single-domain antibody A20.1 (21) was coated per well overnight in phosphate-buffered saline (PBS) at 4°C. At the same time, C. difficile culture supernatants (7 and 24 h) from the 630Δerm, 630Δerm fliC::erm, and 630Δerm fliM::erm strains were dialyzed against PBS overnight using 10-kDa-molecular-mass-cutoff dialysis tubing at 4°C. On the next day, ELISA wells were blocked with 3% (wt/wt) milk diluted in PBS at 37°C. Next, 100 μl of dialyzed culture supernatants was added to wells for 1 h at 37°C. Serial dilutions of purified TcdA in PBS were incubated for the same duration and used to generate a standard curve. Following 3 washes with PBS-Tween 20 (0.05%, vol/vol), the anti-TcdA monoclonal antibody PCG4 (1.1 mg/ml; Novus Biologicals, Littleton, CO) was added to wells (1:1,000 dilution in PBS) for 1 h at room temperature. After another set of washes, goat anti-mouse IgG conjugated to horseradish peroxidase (HRP; 1:5,000 dilution in PBS; Invitrogen, Carlsbad, CA) was added to wells for 1 h at room temperature. A final set of 3 washes preceded the addition of the HRP substrate tetramethylbenzidine (Mandel Scientific, Guelph, ON, Canada). The reaction was stopped with 1.5 M phosphoric acid, and the absorbance was measured using a plate reader at 450 nm. In the second assay for TcdB quantification, 2.5 μg/well of anti-TcdB fractionated llama serum (21) was coated overnight. Wells were blocked as described above and incubated with dialyzed culture supernatants or serial dilutions of purified TcdB to generate a standard curve. TcdB was probed with a TcdB-specific single-domain antibody (0.2 mg/ml; G. Hussack, unpublished data) diluted 1:1,000 in PBS for 1 h at room temperature. Single-domain antibody binding was detected with rabbit anti-His6 IgG conjugated to HRP (Cedarlane, Burlington, ON, Canada) diluted 1:5,000 in PBS. All other steps were identical to those in the ELISA described above. Standard curves of TcdA and TcdB were generated in GraphPad Prism software, and the concentrations of toxins in supernatants were calculated from sigmoidal 4-parameter logistic curve fitting.
RNA isolation.We extracted total RNA from strains grown without shaking in 40 ml TY medium under anaerobic conditions at 37°C. Cells were harvested at either 4 or 7 h by centrifugation for 5 min at 4°C. Cells was resuspended in 7 ml TRIzol and disrupted by passing through a 10-ml syringe with an 18-gauge needle (10 times). Following incubation of the homogenized samples at room temperature for 15 min, 1.4 ml chloroform was added, and samples were shaken for 2 min and then centrifuged at 12,000 × g for 10 min at 4°C. The supernatant was carefully removed, and an equal volume of cold isopropanol was added, followed by incubation at room temperature for 10 min. Samples were centrifuged at 12,000 × g for 10 min at 4°C, and the supernatant was decanted. Pellets were resuspended in 75% ethanol, centrifuged at 7,500 × g for 5 min at 4°C, and washed with 5 ml 100% cold ethanol. RNA was finally resuspended in 200 μl H2O and quantified using an ND-1000 spectrophotometer (Fisher Thermo Scientific, Nepean, ON, Canada). For removal of contaminating DNA, RNA (5 μg) was incubated with 5 U DNase I (Fermentas, Burlington, ON, Canada) for 30 min at 37°C. Following phenol-chloroform extraction, RNA was ethanol precipitated and resuspended in H2O.
Quantitative reverse transcription-PCR (qRT-PCR).cDNA was synthesized from 2.5 μg of total RNA by adding 1 μg N6 random primer, 10 mM deoxynucleoside triphosphates, 5× buffer, 100 mM dithiothreitol (DTT), 40 U RNase inhibitor (Life Technologies, Burlington, ON, Canada) in a total volume of 50 μl. Following incubation at 42°C for 2 min, 1 μl Superscript II (Life Technologies) was added and the reaction mixture was incubated for 1 h at 42°C. The primers used to examine PaLoc gene transcripts were either synthesized using previously published sequences (for tcdA and tcdB [46]) or designed using Beacon Designer (version 4) software (Premier Biosoft, Palo Alto, CA) (for tcdE, tcdR, tcdC, fliA) and synthesized (see Table S1 in the supplemental material). The primer sequences used for the housekeeping gene rpsJ, encoding 30S ribosomal protein S10, were as described by Metcalf et al. (30). Real-time quantitative PCR was performed in a 20-μl reaction mixture volume containing 50 ng cDNA, 10 μl SYBR mix (Perfecta SYBR PCR fast mix, low carboxy-X-rhodamine), and 1 μl each primer. Amplification and detection were performed as described above. Melting curves generated at the end of each quantitative PCR (qPCR) were also examined to ensure that single PCR products were made for each experiment. qPCRs were performed in duplicate for each of three independent biological replicates for each strain at 4 and 7 h of growth in TY broth. The relative quantitation method (26) was used to determine gene expression changes in the mutant relative to the control. The quantity of cDNA for each gene was normalized to that of the cDNA of the housekeeping gene rpsJ. The relative change in toxin-related gene expression of each of the tcd genes from each mutant strain compared to that of the genes from strain 630Δerm was then determined. Statistical analysis was performed using GraphPad Prism software (version 5.04) for Windows (GraphPad Software, La Jolla, CA) using a one-way ANOVA repeated-measures test with Dunnett's multiple-comparison test for analysis of the transcript level in the mutant strains relative to the level in the 630Δerm parent strain. For the FliA mutant analysis, a paired one-tailed t test was used to compare transcript levels in the mutant strain relative to those in the control strain.
Proteomics comparisons of secreted proteins.C. difficile strains were grown at 37°C in TY medium to stationary phase (24 h). Cells were removed by centrifugation at 8,000 × g for 10 min. The supernatants were retained and sterilized by filtration through a 0.22-μm-pore-size Millex GP filter (Millipore, Billerica, MA). Next, 20 ml of culture supernatants was precipitated by the addition of 100% (wt/vol) trichloroacetic acid (TCA) at −20°C to a final TCA concentration of 10%. Solutions were incubated on ice for 45 min before centrifugation at 4,500 × g for 45 min at 4°C. The supernatants were discarded, and 3 ml of ice-cold acetone was added to the pellet. The sample was then held on ice for 15 min before centrifugation at 4,500 × g for 45 min. The supernatant was discarded, and the pellet was air dried before resuspension in 130 μl of rehydration buffer (7 M urea, 2 M thiourea, 4% CHAPS {3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate}, 1% DTT, 0.5% ASB-14 detergent in Milli-Q water). Protein concentrations were determined using a Bradford protein assay (Bio-Rad, Mississauga, ON, Canada). A standard curve was generated using a bovine serum albumin (BSA) standard kit (Bio-Rad, Mississauga, ON, Canada). The absorbance was read and analyzed using a BioTek spectrophotometer plate reader at 595 nm with Gen5 software (BioTek Instruments, Winooski, VT). One-dimensional gel liquid chromatography (LC) was performed using a 1-mm 12% polyacrylamide gel. Equal volumes of each sample (approximately 10 μg protein) were combined in a 1:1 ratio with Laemmli buffer and heated at 90°C for 10 min before the mixture was loaded onto the gel. Gels were run at 150 V for 1 h and then silver stained and imaged as previously described (16). Approximately 30 bands of 2 mm by 7 mm from each lane were excised from the gel, reduced, and alkylated and digested with trypsin (14). Protein digests were analyzed by nano-liquid chromatography tandem mass spectrometry (MS/MS) using an LTQ XL linear trap MS (Thermo Fisher Scientific) coupled with a nanoAcquity ultra-high-pressure liquid chromatography system (Waters, Milford, MA), as described in our earlier work (41). Protein identification was carried out using Mascot (version 2.3.02) software (Matrix Science Inc., Boston, MA), searching the C. difficile 630 genome sequence (NCBI reference sequence with accession number NC_009089). The searches were limited to consider only tryptic cleavages (one miss allowed), carbamidomethylation of cysteine, and oxidation of methionine. Ion scores of at least 30 were required for positive peptide identification. Default tolerances were used for the LTQ XL linear trap MS (1.5 peptides and 1.2 MS/MS units). Using an in-house Java program and Mascot Parser (version 2.3) software (Matrix Science Inc.), the 514 search results were consolidated into a presence/absence report based on the sample source (wild type/mutants) using a peptide ion score cutoff of at least 30 and a minimum protein probability of 0.05. The relative peptide abundance value was assigned by the Mascot software using the exponentially modified protein abundance index (emPAI) (22) and also incorporated into the report. The estimated false-positive rate for peptide identification was less than 1%. The protein subcellular localization for all proteins in the genome was predicted using both the PSORTb (version 3.0) and LocateP (http://www.cmbi.ru.nl/locatep-db) databases (48).
Immunodetection of TcdA and TcdB.Precipitated proteins from C. difficile supernatants and purified TcdA and TcdB were run on three 3 to 8% NuPage Novex Tris acetate minigels following the manufacturer's instructions (Invitrogen). The protein volumes for each culture supernatant remained the same as those determined for the one-dimensional gel LC. High-molecular-mass protein standards (31 to 500 kDa) were used as markers, and for each of the toxin standards, 1 μg was loaded. The first of the three gels was silver stained as a reference gel. The other two gels were transferred to polyvinylidene difluoride membranes using a semidry transblotter cell (Bio-Rad). Posttransfer, membranes were blocked and washed as described in our earlier work (14). One membrane was probed with anti-TcdA primary antibody, while the other was probed with anti-TcdB primary antibody (Novus Biologicals). Both primary antibodies and goat anti-mouse HRP-labeled IgG secondary antibody (Perkin Elmer Life and Analytical Sciences, Shelton, CT) were diluted to 1:5,000 in 5% (wt/vol) milk in PBS containing 0.1% Tween (PBST). Blots were imaged using an ECL Western blotting detection kit (GE Healthcare, Baie D'Urfe, QC, Canada) as per the manufacturer's instructions, followed by exposure to X-ray film.
Spore inoculum.To obtain spores for animal studies, each strain was grown on BHI agar plates for 7 days. Cells were harvested, washed 3 times (by centrifugation at 8,000 × g), and resuspended into PBS. Samples were heat shocked at 56°C for 10 min to kill vegetative cells. Spores were centrifuged, resuspended in Dulbecco modified Eagle medium, and frozen at −80°C. Spore numbers were determined before use by plating serial dilutions on BHI containing 0.1% taurocholate (33). Spores were diluted in saline for orogastric inoculation into hamsters.
Hamster model of infection.Female Golden Syrian hamsters aged 6 to 7 weeks were purchased from Charles River Laboratories (St. Constant, QC, Canada). The animals were maintained and used in accordance with the recommendations of the Canadian Council on Animal Care Guide to the Care and Use of Experimental Animals (5a), and experimental procedures were approved by the institutional animal care committee at the National Research Council of Canada. Clindamycin (2 mg/hamster, 200-μl volume) was given orally 4 h prior to oral infection with approximately 100 C. difficile spores in 200 μl saline. Hamsters were monitored twice daily for clinical signs of disease (lethargy, hunched posture, and wet tail/diarrhea), and all moribund animals were immediately euthanized. Due to the sensitivity of 630Δerm to clindamycin, we compared the effects of insertional inactivation of genes from distinct regions of the flagellar regulon to the parent C. difficile 630 strain. This is the strain from which 630Δerm was derived and which we demonstrated had identical sensitivity to clindamycin as the mutant strains (500 μg/ml, as measured by broth MIC assay).
RESULTS
Generation of flagellar mutant strains.To examine the effect of the flagellar regulon on toxin production and secretion, we generated a number of insertionally inactivated flagellar mutants of C. difficile 630Δerm using the ClosTron targeted mutagenesis approach (19). The selected genes were chosen on the basis of their roles in assembly and export of flagellar components, in addition to their location within the C. difficile 630 flagellar operon (8, 27). The major flagellar operon is located between bp 293002 and bp 333302 within the C. difficile 630 chromosome, and the organization of genes within this locus is presented in Fig. 1. The first region (F1; bp 293002 to 304049) of the flagellar regulon (CD0226 to CD0240) encodes late-stage flagellar regulon genes and includes the flagellar filament gene (fliC) and flagellar capping protein gene (fliD), which are typically regulated in Gram-positive organisms by the alternate transcriptional regulator fliA, a homolog of the Gram-negative flagellar transcriptional regulator σ28 (31). In addition, the CD0240 gene from this locus, which encodes a putative glycosyltransferase, has been shown to be involved in glycosylation of the flagellar filament structural protein FliC (41). Promoter sequence analysis of the 630 genome using promoter mapping software identified three fliA promoter sequences upstream of CD0226, flgM, and fliS2 within this region (Fig. 1). No fliA promoter sequences were found upstream of either the fliC or CD0240 gene (Fig. 1, gray box).
Schematic of flagellar locus from C. difficile 630 (drawn to scale). Black arrows, open reading frames with gene annotations above or below the respective genes. The three regions of the flagellar regulon are indicated by a solid bracket above the gene locus (F1, late-stage flagellar genes; F2, flagellar glycosylation genes; F3, early-stage flagellar genes). Broken arrows below the flagellar locus, location of putative fliA promoter sites; white triangles, genes selected for mutagenesis using the ClosTron system; gray shading, noncoding regions of DNA. Targetron insertion sites and the orientation of the Targetron within each gene are as follows: fliC, base pair (bp) 260, antisense; CD0240, bp 864, sense; fliF, bp 240, antisense; fliG, bp 669, sense; flhB-fliR, bp 266, antisense; fliM, bp 342, sense; fliA, bp 228, sense. Numbers indicating the base pair position (start/end) within the genome are indicated for the F1, F2, and F3 regions.
The next region of the flagellar regulon, F2 (CD0241 to CD0244, bp 304766 to 308251), encodes additional flagellar glycan biosynthetic genes. The predicted start codon of CD0241 is located 717 bp downstream from CD0240, and the locations of genes CD0242, CD0243, CD0244 suggest that they are likely to be cotranscribed with CD0241.
The third region of this flagellar operon, F3 (flgB [CD0272], bp 309272 to 333302) resembles the large 27-kb fla-che operon of Bacillus subtilis (9). In B. subtilis, the genes are cotranscribed from a common promoter under the control of RNA polymerase and the vegetative σ70 homolog σA. The genetic organization of the F3 locus of C. difficile suggests that a similar cotranscription of the F3 region genes from a common promoter may be occurring (45). Included in this group of early genes are fliF (MS ring), fliG, and fliM (C ring motor switch), flhB and fliR (export apparatus membrane proteins), and the fliA (σ28 homolog) transcriptional regulator of the F1 locus. Each of these genes was targeted for insertional inactivation using the ClosTron system. In the C. difficile 630 genome, flhB and fliR genes are found as a single open reading frame (flhB-fliR fusion). Construction of fliC and CD0240 mutant strains with mutations in the F1 region was completed in an earlier study (41). The location of each of these flagellar genes within the flagellar locus is highlighted in Fig. 1, and the site and orientation of the Targetron insertion for each gene are described. Insertion of the Targetron erythromycin (erm) resistance marker into each gene was confirmed by PCR using both primers flanking the gene of interest and primers specific to the erm resistance marker. Growth curves in broth culture for each mutant strain were assessed and shown to be comparable to those for the C. difficile 630 and 630Δerm strains (data not shown). Loss of motility for each mutant strain was confirmed using motility agar tubes as previously described (Fig. 2). A minimum of two independent colonies were examined for each mutant generated to confirm the observed phenotype. While we were able to successfully complement the 630Δerm fliC::erm mutant strain and restore a motile phenotype using the fliC gene with its native promoter (Fig. 2, lane 11), we were unsuccessful in restoring motility in the 630Δerm fliM::erm strain with a mutation in the early-stage F3 locus (data not shown). It should be noted that the F3 locus early-stage genes are all part of a long 30-gene operon regulated by a single promoter, and so the construct used for complementation in this case utilized a nonflagellar clostridial promoter, thl, for fliM expression (19). As bacteria are known to use a complex hierarchy of gene regulation to control the synthesis and secretion of flagellar structural components, it is possible that the level of expression of FliM was not optimal in this construct, and so FliM was unable to restore a motile phenotype. Although the lack of complementation of 630Δerm fliM::erm is a possible shortcoming of the current study, it should be noted that multiple gene knockouts from distinct regions of this long operon all produced the same phenotype with respect to motility.
Motility assays of flagellar mutant strains. The motility of each strain was determined by assessing growth patterns in agar stabs. A motile phenotype produced diffuse growth spreading away from the inoculum stab. Lane 1, 630Δerm; lane 2 630Δerm fliC::erm; lane 3, 630Δerm CD0240:erm; lane 4, 630Δerm fliF::erm; lane 5, 630Δerm fliG::erm; lane 6, 630Δerm flhB-fliR::erm; lane 7, 630Δerm fliM::erm; lane 8, 630Δerm fliA::erm; lane 9, 630Δerm CD3350::erm; lane 10, 630Δerm fliC::erm pMTL84151; lane 11, 630Δerm fliC::erm/pMTL-pfliC.
While toxin production has been shown to increase during the stationary phase of growth (12, 20, 32), genes involved in motility are expressed at higher levels during the exponential phase of growth (36). Correspondingly, we chose to examine toxin production in flagellar mutants during the logarithmic phase of growth, where flagellar gene expression and assembly of the flagellar apparatus are optimal.
Toxin activity.HLF cells are sensitive to both TcdA and TcdB, and the HLF cell rounding assay is routinely used for analyzing C. difficile toxins (4, 21). The level of toxin secreted by cells growing in broth culture for each flagellar mutant was compared to that for the 630Δerm parent strain. Broth supernatants were collected at 4, 7, and 24 h of growth, filter sterilized, and then applied to confluent monolayers of HLF cells (undiluted, diluted 1/10, diluted 1/100, or diluted 1/1000), and the HLF monolayers were then examined at 24 h posttreatment. Supernatants from 4, 7, and 24 h of growth in TY broth were selected because they represent mid-exponential (4 h), late exponential (7 h), and stationary (24 h) phases of growth, according to our growth curves. Cell rounding was assessed using the CPE scoring system, and the results are presented in Fig. 3. As expected, the toxin levels in broth culture supernatants of strain 630Δerm were clearly increased in cultures grown for 24 h compared to those in cultures grown for only 7 h, as evidenced by the degree of cell rounding at higher dilutions in these samples. Correspondingly, supernatants from 630Δerm cultures grown for 7 h showed higher levels of toxin than supernatants from cultures grown for 4 h. This correlated well with previous studies indicating that toxin production is increased when cells reach the end of exponential phase/beginning of stationary phase.
Toxin activity in broth supernatants measured by cell monolayer rounding assays. Each flagellar mutant strain was grown in TY broth for 4, 7, or 24 h, and toxin activity was compared to that of strain 630Δerm. Appropriate dilutions of sterile culture supernatants (30 μl) were added to 200 μl of HLF cell monolayers, and the extent of cell rounding was monitored at 24 h postincubation using the cytopathic effect scoring system. (A) Four-hour C. difficile supernatants diluted 1/10; (B) 7-h C. difficile supernatants diluted 1/10; (C) 24-h C. difficile supernatants diluted 1/100. (Inset in panel C) Cell rounding generated from the addition of 630Δerm and 630Δerm fliC::erm 24-h supernatants diluted 1/1,000. Columns and error bars represent the mean percent cell rounding ± SEM generated by supernatants from each strain (n = 6). The data were analyzed for significance by one-way ANOVA, followed by Bonferroni's multiple comparison test (**, P < 0.01; ***, P < 0.001; ns, not significant). CD0240, 630Δerm CD0240::erm; fliC, 630Δerm fliC::erm; fliF, 630Δerm fliF::erm; fliG, 630Δerm fliG::erm; fliM, 630Δerm fliM::erm; flhB/fliR, 630Δerm flhB-fliR::erm; CD3350, 630Δerm CD3350::erm.
As indicated in Fig. 1, the genes for early-stage flagellar basal body assembly appear to be coexpressed in a long 27-gene flagellar locus (F3) located downstream of the glycosylation locus, and complementation experiments with individual genes under alternative promoter systems were found to be challenging to construct and interpret. As such, we chose to examine the effect of inactivation of genes located at the start, middle, and end of this locus. A minimum of two independent colonies of each mutant strain generated were examined in toxin assays, and identical patterns of toxin secretion were observed for each of the independent colonies of each mutant strain examined.
In contrast to strain 630Δerm, the level of toxin in broth cell culture supernatants from flagellar mutants was clearly altered. Toxin levels in 630Δerm fliC::erm were significantly increased compared to those in the parent strain at 4, 7, and 24 h of growth. For example, the mean cell rounding generated by 1/10 dilutions of 4-h supernatants from 630Δerm fliC::erm was 91.7% ± 3.1%, in comparison to 25.0% ± 6.7% for the parent 630Δerm strain. A similar pattern of higher toxin levels in 7- and 24-h culture supernatants was also observed, including in 1/1,000 dilutions of 24-h supernatants, generating mean cell rounding of 81.67% ± 4.41% and 3.33% ± 1.67% for 630Δerm fliC::erm and 630Δerm, respectively (Fig. 3C, inset).
At 4, 7, and 24 h of growth, the levels of toxin in broth cell culture supernatants of the 630Δerm fliF::erm, 630Δerm fliG::erm, 630Δerm fliM::erm, and 630Δerm flhB-fliR::erm strains were significantly reduced compared to those for the 630Δerm parent strain. For example, the mean cell rounding generated by 1/10 dilutions of 7-h supernatants from 630Δerm fliF::erm, 630Δerm fliG::erm, 630Δerm fliM::erm, and 630Δerm flhB-fliR::erm ranged from 3.3% ± 1.1% (630Δerm fliG::erm) to 13.3% ± 4.2% (630Δerm fliM::erm), in comparison to was 57.5% ± 11.9% for 630Δerm. Furthermore, the mean cell rounding generated by 1/100 dilutions of 24-h supernatants from these same strains ranged from 0% (630Δerm fliF::erm and 630Δerm fliG::erm) to 18.33% ± 1.67% (630Δerm flhB-fliR::erm), in comparison to 76.67% ± 3.34% for 630Δerm.
Inactivation of CD0240, a glycosyltransferase gene which is involved in glycosylation of flagellin monomers, appeared to have a negligible effect on toxin secretion levels in broth culture supernatants at 4 and 7 h of growth. The extent of cell rounding from 630Δerm CD0240::erm culture supernatants was essentially identical to that for 630Δerm. However, a small but significant increase in toxin levels was seen in 630Δerm CD0240::erm culture supernatants grown for 24 h relative to the toxin levels seen in 630Δerm supernatants (1/100 dilution, 91.67% ± 1.67% versus 76.67% ± 3.34%). In addition, inactivation of a gene from a distinct locus to the flagellar regulon (a nonflagellin gene, CD3350) had no effect on toxin production relative to that by the parent strain in culture supernatants from 4, 7, and 24 h of bacterial cell growth.
Comparative analysis of PaLoc gene expression.While the level of toxin in culture supernatant was clearly affected in the respective flagellar mutants during logarithmic growth in TY broth, it is not clear if this effect was due to changes in toxin secretion level related to a defective flagellar apparatus or due to a change in transcriptional regulation of the toxin genes. We next determined the effect of the respective flagellar mutations on gene expression levels of the pathogenicity locus of strain 630Δerm. qRT-PCR analysis of each of the PaLoc genes tcdR (positive regulator/sigma factor), tcdB (toxin), tcdE (putative holin), tcdA (toxin), and tcdC (negative regulator of tcdR) from cells grown in TY broth for either 4 or 7 h was completed, and expression of each gene relative to that of the control strain, 630Δerm, is presented in Fig. 4. In a manner similar to that described above for toxin activity in broth supernatants, the transcriptional regulation of PaLoc genes was altered by the flagellar mutations.
Measurement of PaLoc gene transcript levels in flagellar mutant strains. Expression levels of tcd genes were determined by quantifying by qRT-PCR the respective mRNA from samples of bacterial culture grown for 4 or 7 h in TY medium. The housekeeping gene rpsJ was used for normalization of cDNA for each gene, and the relative change in toxin-related gene expression for each flagellar mutant was then determined by comparison to the expression for 630Δerm. (A, B) Growth in TY broth for 4 h (A) and 7 h (B). Data represent the mean fold change in expression of mRNA ± SEM (n = 3) compared to the expression for 630Δerm. Bars correspond to tcdR (gray shade), tcdB (gray), tcdE (horizontal lines), tcdA (white/gray outline), tcdC (black), fliA (white/black outline). The flagellar mutant strain is indicated on the x axis, and the fold change (log2) in transcript level relative to that for 630Δerm is indicated on the y axis. Transcript levels that were significantly different from that for the 630Δerm control are indicated with asterisks (one-way ANOVA with Dunnett's multiple comparison test; *, P < 0.05, **, P < 0.01, ***, P < 0.001). 240, 630Δerm CD0240::erm; FliC, 630Δerm fliC::erm; FliF, 630Δerm fliF::erm; FliG, 630Δerm fliG::erm; FliM, 630Δerm fliM::erm; FlhB, 630Δerm flhB-fliR::erm; 3350, 630Δerm CD3350::erm.
At 4 h of growth, inactivation of fliC led to an approximately 9- to 10-fold increase in levels of tcdR, tcdB, and tcdA expression and an approximately 3-fold increase in the level of tcdE expression compared to the levels found in strain 630Δerm. In contrast, at 4 h, the level of tcdC and fliA expression was not significantly affected in 630Δerm fliC::erm compared to that in 630Δerm. Conversely, at 4 h of growth, inactivation of the fliF, fliG, flhB-fliR, or fliM gene resulted in an approximately 5- to 25-fold decrease in expression of the tcdR and tcdA genes. No significant changes in the levels of expression of the tcdB, tcdE, and tcdC genes were observed in these mutants. For 630Δerm CD0240::erm, no significant changes in the levels of expression of any of the PaLoc genes were observed (Fig. 4A).
Growth of each mutant strain in TY medium for 7 h (Fig. 4B) revealed a pattern similar to that observed at 4 h. PaLoc gene transcript levels in 630Δerm fliC::erm were increased approximately 4- to 9-fold for all transcripts except TcdC, where a 2-fold decrease was observed. In contrast, inactivation of genes from the F3 region of the flagellar regulon resulted in an approximately 9- to 33-fold reduction in tcdR and tcdB gene transcription and up to an approximately 100-fold reduction in tcdA gene transcription. Note that due to the high variability in tcdB transcript levels, reduced expression was statistically significant only for the fliF and flhB mutants (P < 0.05). At 7 h, insertional inactivation of CD0240 (which encodes a glycosyltransferase involved in FliC glycosylation) produced a 3- to 9-fold reduction in PaLoc gene transcript levels. Insertional inactivation of a nonflagellar gene (CD3350) had no significant effect on expression levels of any of the five tcd genes at either 4 or 7 h of growth compared to that for 630Δerm tcd genes.
As the 27 genes from the F3 region of the flagellar regulon appear to be coexpressed in a single operon, we next determined if we could observe any polar effects upon insertion of the erm cassette on downstream genes within the locus. We selected the transcriptional regulator fliA gene, which is located downstream of fliF, fliG, and flhB-fliR and upstream of fliM, to determine this. Measurement of fliA transcript levels at 4 and 7 h in each of the F3 region flagellar mutants revealed changes in fliA transcript levels compared to those for 630Δerm (Fig. 4A and B) in the 630Δerm fliF::erm, 630Δerm fliG::erm, and 630Δerm flhB-fliR::erm strains. Transcriptional levels of fliA in each of these flagellar mutants were decreased by approximately 8- to 20-fold in cells grown in TY broth for 4 or 7 h, most likely due to polar effects on transcription following insertion of the erm cassette. In contrast, the level of transcription of fliA at both 4 and 7 h in the remaining F3 region mutant strain, 630Δerm fliM::erm, was decreased by less than 2- to 3-fold compared to that in the parent strain. However, a similar downregulation of the PaLoc genes was still observed in this strain compared to that in the other F3 region mutant strains, suggesting that the effect observed was not directly due to a polar effect on fliA transcript levels. In 630Δerm fliC::erm, fliA levels at both 4 and 7 h were unaffected compared to those in the parent strain.
FliA role in toxin regulation.It has been established in many Gram-positive and Gram-negative motile organisms that FliA (which is a σ28 homolog) directs the expression of late-stage genes (F1 region), including the flagellar hook components (e.g., flgL and flgK) and flagellar capping protein (fliD). We next determined if inactivation of fliA would produce a phenotype similar to that observed for 630Δerm fliC::erm, whereby transcription and TcdA and TcdB extracellular toxin levels are increased. Insertional inactivation of the fliA gene was completed using the ClosTron mutagenesis system, and motility stabs confirmed the loss of functional flagella in two independent colonies (Fig. 2, lane 8). Broth culture supernatants (4, 7, and 24 h) were then examined as described above and were shown to have significantly reduced levels of toxin activity on HLF monolayers, in a fashion similar to that for the 630Δerm fliF::erm, 630Δerm fliG::erm, 630Δerm fliM::erm, and 630Δerm flhB-fliR::erm mutant strains (Fig. 5A). In agreement with this, PaLoc gene transcript levels in the 630Δerm fliA::erm mutant strain were significantly reduced at both 4 and 7 h of growth (Fig. 5B and C).
Effect of inactivation of flagellar transcription factor fliA on toxin production. (A) Toxin levels in cell culture supernatants were determined using a cell monolayer rounding assay. Strains 630Δerm and 630Δerm fliA::erm were grown in TY broth for 4, 7, and 24 h. Sterile culture supernatants were diluted 1/10 and added to 200 μl of HLF cell monolayers, and the extent of cell rounding was monitored at 24 h using the cytopathic effect scoring system. Two independent 630Δerm fliA::erm colonies were examined, and results for one representative colony are shown. Columns and error bars represent the mean ± SEM percent cell rounding generated by supernatants from 630Δerm or 630Δerm fliA::erm (n = 4). The data were analyzed for significance by two-tailed, unpaired t test (**, P < 0.01; ***, P < 0.001) comparing 630Δerm and 630Δerm fliA::erm at each supernatant time point. (B, C) qRT-PCR demonstrating the change in PaLoc gene transcript levels in 630Δerm fliA::erm grown in TY broth for 4 h (B) and 7 h (C) compared to those in 630Δerm. Transcript levels that were significantly different from those for 630Δerm are indicated with asterisks (paired one-tailed t test; *, P < 0.05; **, P < 0.01; ***, P < 0.001).
A very similar pattern of toxin production was observed in all the mutants with gene knockouts from this F3 locus, irrespective of whether the genes inactivated were structural components of the flagellar basal body, were involved in export, or were transcriptional regulatory genes. While FliA is widely recognized as the transcriptional regulator of late-stage (F1 locus) genes, the current study reveals a distinct pattern of tcd transcriptional regulation between the 630Δerm fliA::erm and the 630Δerm fliC::erm mutant strains.
Analyses of protein secretome in TY broth.To assess the role of the flagellar regulon in secretion, we next examined broth culture supernatants (24 h of growth in TY broth) using a proteomics approach. For this analysis, we selected three mutant strains which are representative of the phenotypes observed by HLF cytotoxicity and qRT-PCR. These were 630Δerm fliC::erm (in which toxin transcription is increased and toxin activity is increased), 630Δerm CD0240::erm (in which toxin transcription is unaffected and toxin activity is only slightly changed), and 630Δerm fliM::erm (in which toxin transcription is reduced, toxin activity is reduced, and the genetic location in F3 locus ensures minimal polar effects). The impact of inactivation of the fliC, CD0240, and fliM genes on the extracellular protein profile (secretome) of C. difficile was determined using a gel LC-based comparison of the proteins secreted from each strain. In this study, secretome samples were collected from replicate growths, and protein samples derived from equivalent amounts of culture were compared (i.e., samples were not normalized for total protein amount). Stringent criteria for final protein identification were used to reduce potential false-positive identification, as described in Materials and Methods. In total, about 350 proteins were detected in supernatants of strain 630Δerm, a number comparable to that reported in an earlier secretome analysis of 630Δerm (32).
Protein abundance was calculated using the exponentially modified protein abundance index (emPAI) (23), based upon the number of observable peptides and the number of observed parent ions. From the comparisons, a number of flagellar and other cell surface virulence-related proteins were observed to differ in abundance when comparing parent and mutant strains. These selected proteins are listed in Table 1, with the complete list of proteins differing in abundance shown in Table S2 in the supplemental material. The protein observed to be the most increased in abundance in supernatants of 630Δerm fliC::erm was TcdA (CD0663), observed at approximately 19-fold higher levels than in 630Δerm. In addition, the flagellum-associated transglycosylase (CD0226), flagellar hook-associated protein FlgK (CD0231), and flagellar cap protein FliD (CD0237) from the F1 flagellar locus were also observed at greater than 10-fold higher levels in the supernatants of the 630Δerm fliC::erm mutant. Bioinformatics analyses using LocateP (a subcellular location protein predictor tailored for Gram-positive bacteria [49]) indicated that a number of the proteins observed at increased levels in supernatants of 630Δerm fliC::erm were putatively involved in a Sec secretory pathway.
Selected proteins present in the secretome of 630Δerm, 630Δerm fliC::erm, 630Δerm CD0240::erm, and 630Δerm fliM::erma
Comparing the relative abundance of proteins in supernatants of 630Δerm and 630Δerm CD0240::erm also showed TcdA to be present at higher levels in the mutant strain (Table 1). Approximately 5-fold greater levels of TcdA were observed. Components of the flagellar apparatus, including flagellin (CD0239, FliC), and the flagellar cap protein FliD (CD0237) were observed at greater than 2-fold lower levels in the secretome of 630Δerm CD0240::erm than in that of 630Δerm (Table 1). In addition, other known cell surface proteins, such as the surface layer-associated protein Cwp66 (CD2789), the cell surface-associated cysteine protease Cwp84 (CD2787), and the cell surface layer protein SlpA (CD2791), were also observed at greater than 2-fold lower levels in the supernatants of 630Δerm CD0240::erm.
No proteins were observed in the supernatants of 630Δerm fliM::erm at increased levels compared to those for 630Δerm. One of the proteins observed at decreased levels included TcdA, observed at approximately 3-fold lower levels. Several flagellin-associated proteins were not observed in supernatants of 630Δerm fliM::erm, such as flagellin (CD0239), putative flagellar hook-length control protein (CD0253, FliK), and flagellar hook-associated protein (CD0231, FlgK). Of note, proteins such as the cell surface protein Cwp66 (CD2789), flagellar cap protein (CD0237), cell surface-associated cysteine protease Cwp84 (CD2787), and cell surface protein (CD2791) were all observed at reduced levels in the supernatants of this mutant compared to those for 630Δerm. TcdB was not detected by mass spectrometry analyses in any of the culture supernatants.
To confirm the variation in abundance of toxins in the secretomes of each strain, Western blotting of NuPAGE-separated TCA-precipitated samples was carried out, and the blots were probed with monoclonal antibodies against TcdA and TcdB. Figure 6 shows that both TcdA and TcdB were detected at higher levels in supernatants from 630Δerm fliC::erm and 630Δerm CD0240::erm than in those from 630Δerm. Faint bands corresponding to TcdA were present in the 630Δerm and 630Δerm fliM::erm supernatants. TcdB was not detected by Western blotting in the secretomes of 630Δerm or 630Δerm fliM::erm. We next attempted to quantify the amounts of TcdA and TcdB present in 7- and 24-h culture supernatants from 3 representative stains (630Δerm, 630Δerm fliC::erm, and 630Δerm fliM::erm), using sandwich ELISAs (Table 2). From 7-h cultures, we determined that 630Δerm fliC::erm supernatants contained 44.6 ± 4.5 ng/ml TcdA. The amount of TcdA in 630Δerm and 630Δerm fliM::erm supernatants was below the limit of detection. In 24-h cultures, we found that 630Δerm and 630Δerm fliC::erm supernatants contained 175.5 ± 9.9 ng/ml and 2,123 ± 46.2 ng/ml TcdA, respectively, further demonstrating that the 630Δerm fliC::erm mutant is producing/secreting approximately 10-fold more TcdA than the parent strain. In contrast, TcdA could not be detected in 24-h cultures of 630Δerm fliM::erm, reflecting the reduced levels of cell rounding seen for this strain (Fig. 3). TcdA toxin levels in supernatants of 630Δerm fliC::erm containing pMTL-pfliC were restored to levels observed for the 630Δerm parent strain at both 7 and 24 h of growth (data not shown). Interestingly, TcdB could not be detected in any of the culture supernatants, despite our ELISA having a limit of detection approaching 5 ng/ml TcdB.
Western blot comparing toxin levels in C. difficile 630Δerm and flagellar mutants. Broth culture supernatants were analyzed for the presence of TcdA (A) and TcdB (B) following TCA precipitation. Lane 1, purified TcdA; lane 2, purified TcdB; lane 3, 630Δerm supernatant; lane 4, 630Δerm fliC::erm supernatant; lane 5, 630Δerm CD0240::erm supernatant; lane 6, 630Δerm fliM::erm supernatant.
Quantification of toxin levels in cell culture supernatants
Effect of flagellar mutations on virulence.While toxin expression levels are clearly affected in flagellar mutants during logarithmic growth in TY medium, we next examined the effect of each mutation on in vivo virulence in the Golden Syrian hamster model of infection. Approximately 100 spores of each strain were used to infect groups of 5 to 6 hamsters, and the hamsters were observed for clinical signs of disease (Fig. 7). Hamsters infected with the C. difficile 630 strain all succumbed to infection within 72 h of infection, with a median time to death (MTD) of 45.5 h after infection. No significant differences in the virulence of 630Δerm fliC::erm (MTD = 43 h) and 630Δerm fliF::erm (MTD = 48 h) compared to that of C. difficile 630 were observed. In contrast, hamsters infected with spores from 630Δerm fliM::erm and 630Δerm flhB-fliR::erm survived significantly longer (P < 0.05, log-rank test) than hamsters infected with spores from C. difficile 630, with MTDs of 90 h and 96 h, respectively (Fig. 7).
Virulence of C. difficile 630Δerm flagellar mutants in hamsters. Kaplan-Meier survival curves demonstrating time to irreversible moribundity after oral infection with approximately 100 spores of C. difficile 630, 630Δerm fliF::erm, 630Δerm fliM::erm, 630Δerm fliC::erm, and 630Δerm flhB-fliR::erm. Data are representative of at least two independent experiments, with n = 5 to 6 hamsters per group.
DISCUSSION
Flagellar regulon gene inactivation clearly affects the levels of the tcdA and tcdB transcripts as well as the level of toxin which accumulates in broth culture supernatants during logarithmic phase of the C. difficile vegetative cell growth cycle. Inactivation of early-stage flagellar genes, including the transcriptional regulator fliA, results in decreased levels of tcdR, tcdB, tcdE, and tcdA transcripts and correspondingly lower toxin levels in cell culture supernatants, while inactivation of fliC, a late-stage flagellar regulon gene, results in increased levels of tcdR, tcdB, tcdE, and tcdA transcripts, as well as increased activity of the corresponding toxin in cell culture supernatants. A similar observation on Vero cell cytotoxicity has recently been made using culture supernatants from fliC and fliD mutant strains grown for 24 h (11). No effect on tcdC (negative regulator of tcdR) transcript levels was observed in any of the flagellar mutants. It has been clearly established that toxin synthesis increases as cells enter stationary phase (12, 20), and the transition between exponential and stationary phase occurs at about 8 h in TY medium for strain 630 (36). In the current study, we demonstrate that toxin gene expression and toxin activity in culture supernatants are detectable during mid- and late logarithmic phases of growth (4 and 7 h) and that inactivation of genes of the flagellar regulon has a clear impact on both transcription and toxin activity at these earlier stages in the growth phase. This trend holds true for cultures grown to stationary phase (24 h) as well. Using distinct, yet complementary assays to assess toxin production/activity (HLF cell rounding assay, qRT-PCR of PaLoc genes, proteomics analysis of the secretome, toxin immunoblotting/ELISA), we have demonstrated that TcdA and TcdB levels are indeed regulated by the flagellar regulon during vegetative cell growth. As such, it appears that regulation of toxin gene expression may be a process much more dynamic than previously reported. Proteomic analysis demonstrated that perturbation of the flagellar regulon is not solely limited to affecting toxin production and that a more global effect on protein secretion appears to be occurring. While it is acknowledged that proteomic analysis of the secretome from a 24-h TY broth culture would be complex and the secretome likely contains many cytosolic proteins due to vegetative cell lysis during the sporulation process, distinct proteomic differences between parent and mutant strains were still observed in this analysis. The identification of a change in levels of recognized virulence factors (e.g., Cwp84, Cwp66) as well as of a number of proteins of unknown function (e.g., CD2686, CD0323, CD1156) as a consequence of flagellar gene inactivation now warrants further investigation.
This study has also revealed unique patterns of regulation of the toxin genes by the genes of the flagellar regulon. The distinct difference in levels of toxin transcription between 630Δerm fliC::erm and the flagellar glycosyltransferase CD0240 mutant strain, 630Δerm CD0240::erm, is particularly intriguing. Previous studies demonstrated that inactivation of either fliC or CD0240 resulted in cells which were no longer capable of assembling a functional flagellar filament, and consequently, in both cases, cells were no longer motile (11, 41). However, in 630Δerm CD0240::erm, unglycosylated flagellin protein was still produced and secreted extracellularly (41). In the current study, although both strains were clearly nonmotile, the regulation of toxin gene expression was significantly altered only in the fliC mutant strain compared to that in the 630Δerm parent. As such, it appears that the loss of flagellin protein (FliC) rather than ability to form a functional flagellar filament (CD0240) is somehow necessary for the observed changes in toxin gene expression.
While this study has focused solely on flagellar regulation of toxin production in C. difficile 630, it should be noted that hypervirulent strains have been shown to produce increased levels of toxin compared to those produced by strain 630. While the flagellar locus of these strains has been shown to have a distinct genetic content in the F1 and F2 regions (38), the mechanistic basis of toxin upregulation remains to be determined. A future focus will be to expand the current study of the flagellar regulon to these hypervirulent strains to determine if a distinct pattern of toxin regulation is occurring in these strains compared to that occurring in strain 630.
This study is the first to directly implicate the FliA transcription factor in the expression of TcdA and TcdB. We also show in this study that inactivation of FliA, which is required for activation of the F1 region genes, produces a distinct pattern of regulation of the tcd genes compared to that for 630Δerm fliC::erm. This is the first evidence that two distinct mechanisms involving the F1 late-stage gene transcriptional activator FliA and individual F1 gene products (FliC) affect PaLoc gene transcription.
For many pathogenic bacteria, it is critical that organisms be able to sense arrival at the site of infection and then to respond by regulating the expression of a number of key virulence genes. These responses are an important feature of the ability of many pathogenic species to successfully transition from life in environmental reservoirs to life in their respective host and are critical for success of the pathogen. It has recently been demonstrated for V. cholerae that the coordinated regulation of virulence by quorum sensing and motility pathways occurs during the initial stages of infection (25), and it has long been recognized that motility and virulence gene expression in V. cholerae are inversely regulated (15, 39). Previous studies in C. difficile demonstrated that luxS-dependent signaling controlled toxin transcriptional regulation in cells grown to mid- and late log phase (24), while more recently, it has been shown that virulence of fliC and fliD flagellum mutants in hamsters was increased in comparison to that of the parent strain (11). We show in the current study that the flagellar regulatory hierarchy of C. difficile can control expression of the major toxin genes in two distinct patterns. Both upregulation and downregulation of toxin transcription were observed following mutations in late-stage or early-stage genes of the flagellar regulatory hierarchy, respectively. Correspondingly, a similar pattern was observed for toxin activity on HLF cells and in secreted toxin protein levels. The inability to detect TcdB in secretome samples or by ELISA is presumably due to the relatively low abundance of this protein compared to that of TcdA, as has been previously reported (44). While it appears that levels of TcdA are more significantly affected, it is important to note that TcdB has been shown to be 1,000 times more potent than TcdA (7). As such, a small increase or decrease in TcdB level could still have a profound effect in a biological assay.
Not only did mutations in genes of the flagellar regulatory hierarchy impact toxin transcription, but the virulence of several flagellar mutants was also affected in vivo in a hamster model of C. difficile infection. Hamsters infected with spores of C. difficile 630Δerm fliM::erm and 630Δerm flhB-fliR::erm (both of which were shown to have decreased levels of toxin transcript and decreased levels of toxin secreted into the growth medium) survived significantly longer than hamsters infected with C. difficile 630. However, the patterns of virulence observed for these mutants cannot be explained by toxin activity alone, since the virulence in hamsters infected with spores of 630Δerm fliF::erm was not significantly different from that in hamsters infected with spores of the parent strain, despite the fact that all three early-stage flagellar mutants displayed a similar level of reduction in toxin transcription and toxin secretion levels when grown in TY broth. Also, hamsters infected with 630Δerm fliC::erm did not succumb significantly sooner than hamsters infected with C. difficile 630 (MTDs, 43 h versus 45.5 h), as would be expected from the increased toxin transcript/expression levels in 630Δerm fliC::erm. This could be explained by the highly sensitive nature of Golden Syrian hamsters to C. difficile infection. Even with C. difficile 630, hamsters have an extremely short MTD of 45.5 h after spore inoculation, and this may represent the minimal time needed from spore inoculation to irreversible moribundity. Thus, any increase in toxin production (fliC mutant strain) may simply not be able to further shorten MTD in the hamster model. It should also be considered that in addition to TcdA and TcdB toxins, additional virulence factors may be regulated in unique patterns in each of these strains, and these factors may have an impact on the degree of virulence of each mutant strain observed in hamsters.
In conclusion, the current study demonstrates the complexity in coordination of transcription of the PaLoc via the flagellar regulon. Toxin transcription appears to be positively regulated by transcription of early-stage genes from the F3 region of the flagellar regulon. When genes from this region are inactivated, transcription of the tcdR, tcdB, tcdE, and tcdA genes is reduced. In contrast, inactivation of the flagellin structural gene fliC from the F1 region of the flagellar regulon led to a significant increase in both the levels of transcription and the levels of secreted toxin in culture supernatants. It remains to be determined if this flagellar regulatory system is part of a more complex system for the control of multiple virulence factors and whether such a system ensures the success of C. difficile cells for both initial colonization and subsequent persistence.
ACKNOWLEDGMENTS
We thank Nigel Minton and John Heap for development and provision of the ClosTron gene-knockout system, pMTL8000 plasmids, and Clostridium difficile strain 630Δerm. We thank Brendan Wren for provision of Clostridium difficile strain 630. We thank Luc Tessier for assistance with mass spectrometry instrumentation, Sara Kilmury for assistance with proteomics work, and Tom Devecseri for help with figure production.
FOOTNOTES
- Received 7 March 2012.
- Returned for modification 21 April 2012.
- Accepted 20 July 2012.
- Accepted manuscript posted online 30 July 2012.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.00224-12.
- Copyright © 2012, American Society for Microbiology. All Rights Reserved.