ABSTRACT
Clostridioides difficile infection (CDI) is a toxin-mediated disease. Several factors have been identified that influence the production of the two major C. difficile toxins, TcdA and TcdB, but prior published evidence suggested that additional unknown factors were involved in toxin regulation. Previously, we identified a C. difficile regulator, RstA, that promotes sporulation and represses motility and toxin production. We observed that the predicted DNA-binding domain of RstA was required for RstA-dependent repression of toxin genes, motility genes and rstA transcription. In this study, we further investigated the regulation of toxin and motility gene expression by RstA. DNA pulldown assays confirmed that RstA directly binds the rstA promoter via the predicted DNA-binding domain. Through mutational analysis of the rstA promoter, we identified several nucleotides that are important for RstA-dependent transcriptional regulation. Further, we observed that RstA directly binds and regulates the promoters of the toxin genes, tcdA and tcdB, as well as the promoters for the sigD and tcdR genes, which encode regulators of toxin gene expression. Complementation analyses with the Clostridium perfringens RstA ortholog and a multi-species chimeric RstA protein revealed that the C. difficile C-terminal domain is required for RstA DNA-binding activity, suggesting that species-specific signaling controls RstA function. Our data demonstrate that RstA is a transcriptional repressor that autoregulates its own expression and directly inhibits transcription of the two toxin genes and two positive toxin regulators, thereby acting at multiple regulatory points to control toxin production.
IMPORTANCE Clostridioides difficile is an anaerobic, gastrointestinal pathogen of humans and other mammals. C. difficile produces two major toxins, TcdA and TcdB, which cause the symptoms of the disease, and forms dormant endospores to survive the aerobic environment outside of the host. A recently discovered regulatory factor, RstA, inhibits toxin production and positively influences spore formation. Herein, we determine that RstA directly represses toxin gene expression and gene expression of two toxin gene activators, TcdR and SigD, creating a complex regulatory network to tightly control toxin production. In addition, the ability for RstA to bind DNA and repress toxin production requires the species-specific domain predicted to respond to small quorum-sensing peptides. This study provides a novel regulatory link between C. difficile sporulation and toxin production. Further, our data suggest that C. difficile toxin production is regulated through a direct sensing mechanism.
INTRODUCTION
Clostridioides difficile infection (CDI) is a nosocomial and community-acquired gastrointestinal disease that affects individuals with dysbiotic gut microbiota, which commonly occurs after antibiotic treatment (1, 2). Clinical outcomes range from mild diarrhea to severe disease symptoms, including sepsis and death (1). The two glycosylating exotoxins, TcdA and TcdB, elicit CDI symptoms and are indispensable for C. difficile virulence (3). Environmental and intracellular signals, including nutrient availability and metabolic cues, strongly influence toxin production (4-7). There are numerous identified C. difficile factors that control toxin gene expression in response to these signals (8-12); however, the regulatory pathways and molecular mechanisms that directly control toxin gene expression are not fully understood (13).
Our previous work identified a novel regulator, RstA, which depresses C. difficile toxin production and motility (14). RstA inhibits transcription of the toxin genes, tcdA and tcdB, the toxin-specific sigma factor, tcdR, and the flagellar-specific sigma factor, sigD, which is essential for motility and directs tcdR expression (11, 12, 14-16). In addition to repressing motility and toxin production, RstA positively influences C. difficile spore formation, which is critical for the survival of the bacterium outside of the host and for transmission from host-to-host, indicating that RstA regulates diverse phenotypes important for C. difficile pathogenesis.
The predicted secondary structure of RstA reveals three apparent domains: an N-terminal conserved helix-turn-helix DNA-binding domain, followed by a series of multiple tetratricopeptide repeat (TPR) domains comprising a putative Spo0F-like protein-binding domain, and a C-terminal putative quorum-sensing-like domain (14). These characteristic features place RstA in the RRNPP (Rap/Rgg/NprR/PlcR/PrgX; formerly, RNPP) family of proteins. RRNPP proteins are prevalent in Gram-positive organisms and regulate competence, sporulation, toxin production and other important survival and virulence phenotypes (17-19). The DNA-binding or protein-binding activity of RRNPP proteins are controlled by the direct binding of small, quorum-sensing peptides (19). The precursor proteins encoding the quorum-sensing peptides are often adjacent to the regulatory RRNPP protein, and are translated, exported, processed and reinternalized at high cell densities (20-25). In addition, RRNPP proteins often autoregulate their own expression, as is observed for RstA (14). The presence of these conserved domains within RstA provides insight into how RstA may regulate C. difficile toxin production, motility and sporulation.
To better understand the regulatory impact RstA exerts on C. difficile toxin production and sporulation, we examined the function of the conserved DNA-binding domain. Our previous study had shown that the DNA-binding domain is required for RstA-dependent regulation of rstA expression and toxin gene expression, but is expendable for sporulation regulation. Here, we demonstrate that RstA directly binds to its promoter via an imperfect inverted repeat, and directly binds the sigD and the toxin genes promoters. Further, our data demonstrate that RstA and SigD independently control toxin expression, creating a multi-tiered regulatory pathway by which RstA represses toxin production. Finally, we show that the C. perfringens rstA ortholog does not complement toxin production or sporulation in a C. difficile rstA mutant. However, a chimeric RstA protein containing the C. perfringens DNA-binding domain and the C. difficile Spo0F-binding and quorum-sensing-binding domains restores both sporulation and toxin production, providing evidence that the ability to respond to species-specific signaling is necessary for RstA DNA-binding activity.
MATERIALS AND METHODS
Bacterial strains and growth conditions
The bacterial strains and plasmids used in the study are listed in Table S1. C. difficile strains were routinely cultured in BHIS or TY medium (pH 7.4) supplemented with 2-5 μg/ml thiamphenicol and/or 1 μg/ml nisin as needed (26). Overnight cultures of C. difficile were supplemented with 0.1% taurocholate and 0.2% fructose to promote spore germination and prevent sporulation, respectively, as indicated (26, 27). C. difficile strains were cultured in a 37°C anaerobic chamber with an atmosphere of 10% H2, 5% CO2 and 85% N2, as previously described (28). Escherichia coli strains were grown at 37°C in LB (29) with 100 μg/ml ampicillin and/or 20 μg/ml chloramphenicol as needed. Kanamycin (50 μg/ml) was used for counterselection against E. coli after conjugation with C. difficile, as previously described (30).
Strain and plasmid construction and accession numbers
Oligonucleotides used in this study are listed in Table S2. Details of vector construction are described in the Supplemental Material (Fig. S1). C. difficile strains 630 (Genbank no. NC_009089.1) and R20291 (Genbank no. FN545816.1), Clostridium acetobutylicum ATCC 824 (Genbank no. NC_003030.1), Clostridium sordellii ATCC 9714 (Genbank no. APWR00000000) and Clostridium perfringens S13 (Genbank no. BA000016.3) were used as templates for primer design and PCR amplification. The rstA ortholog from C. acetobutylicum was synthesized by Genscript (Piscataway, NJ). The Streptococcus pyogenes CRISPR-Cas9 system, which has been modified for use in C. difficile (31), was used to create a non-polar deletion of the rstA gene. The 630Δerm and RT1075 (sigD::erm) strains containing the rstA-targeted CRISPR-Cas9 plasmid (MC1133 and MC1193, respectively) were grown overnight in TY medium with 5 μg/ml thiamphenicol. The next morning the cultures were backdiluted into fresh TY medium supplemented with 5 μg/ml thiamphenicol and 100 ng/ml anhydrous tetracycline for 24 h to induce expression of the CRISPR-Cas9 system. A small aliquot of this culture was streaked onto BHIS plates, and colonies were screened by PCR for the presence or absence of the rstA allele
Mapping the rstA transcriptional start with 5’ rapid amplification of cDNA ends (5’ RACE)
DNase-I treated RNA from the rstA::erm mutant (MC391) was obtained as described above. 5’ RACE was performed using the 5’/3’ RACE Kit, Second Generation (Roche), following the manufacturer’s instructions as previously reported (32). Briefly, first strand cDNA synthesis was performed using the rstA-specific primer oMC982, followed by purification with the High Pure PCR Product Purification Kit (Roche). After subsequent poly(A)-tailing of first strand cDNA, PCR amplification was performed using an oligo(T) primer and the rstA-specific primer oMC983 with Phusion DNA Polymerase (NEB). The resulting PCR products were purified from a 0.7% agarose gel (Qiagen) and TA cloned into pCR2.1 (Invitrogen) using the manufacturer’s supplied protocols. Plasmids were isolated and sequenced (Eurofins MWG Operon) to determine the transcriptional start site (−32 bp from translational start site; n = 7).
Sporulation assays
C. difficile cultures were grown in BHIS medium supplemented with 0.1% taurocholate and 0.2% fructose until mid-exponential phase (i.e., an optical density at 600 nm (OD600) of 0.5), and 0.25 ml were spotted onto 70:30 sporulation agar as a lawn (27). After 24 h growth, ethanol resistance assays were performed as previously described (33, 34). Briefly, the cells were scraped from plates after 24 hours (H24) and suspended in BHIS medium to an OD600 of 1.0. The total number of vegetative cells per ml was determined by immediately serially diluting and applying the resuspended cells to BHIS plates. Simultaneously, a 0.5 ml aliquot was mixed with 0.3 ml 95% ethanol and 0.2 ml dH2O, to achieve a final concentration of 28.5% ethanol, vortexed and incubated for 15 min to eliminate all vegetative cells; ethanol-treated cells were subsequently serially diluted in 1X PBS + 0.1% taurocholate and applied to BHIS + 0.1% taurocholate plates to determine the total number of spores. After at least 36 h of growth, CFU were enumerated and the sporulation frequency was calculated as the total number of spores divided by the total number of viable cells (spores + vegetative cells). A spo0A mutant (MC310) was used as a negative sporulation control. Statistical significance was determined using a one-way ANOVA, followed by Dunnett’s multiple-comparison test (GraphPad Prism v6.0), to compare sporulation efficiency to the rstA mutant.
Alkaline phosphatase activity assays
C. difficile strains containing the reporter fusions listed in Table S1 were grown and harvested on either 70:30 sporulation agar at H8 or from TY liquid medium in stationary phase (T3 or H24). Alkaline phosphatase assays were performed as described previously (35) with the exception that no chloroform was used for cell lysis. Technical duplicates were averaged, and the results are presented as the mean and standard error of the mean of three biological replicates. The two-tailed Student’s t test was used to compare the activity in the rstA mutant to the parent strain.
Biotin pulldown assays
Biotin pulldown assays were performed as described in Jutras et. al. 2012 (36). Briefly, an excess of biotin-labeled DNA bait was coupled to streptavidin-coated magnetic beads (Invitrogen) in B/W Buffer, and the bead-DNA complexes were washed with TE Buffer to remove unbound DNA. To prepare cell lysates, C. difficile expressing either RstA-FLAG (MC1004) or RstAΔHTH-FLAG (MC1028) in the rstA background were grown to mid-log phase (OD600 = 0.5) in 500 ml TY medium, pH 7.4, pelleted, rinsed with sterile water and stored at −80°C overnight. The pellets were suspended in 4.5 ml BS/THES Buffer and lysed by cycling between a dry ice/ethanol bath and a 37°C water bath. The cell lysates were vortexed for 1 min to shear genomic DNA, and cell debris was pelleted at 15K rpm for 15 min at 4°C. The supernatant, along with 10 μg/ml salmon sperm DNA as a nonspecific competitor, was then applied to the bead-DNA complexes and rotated for 30 min at room temperature. This incubation was repeated once with additional supernatant for two total incubations. The bead-DNA-protein complexes were washed extensively with BS/THES Buffer supplemented with, and then without, 10 μg/ml salmon sperm DNA to remove nonspecific proteins. The remaining bound protein was eluted with 250 mM NaCl in Tris-HCl, pH 7.4, and the eluates were immediately analyzed by SDS-PAGE and Western blot using FLAG M2 antibody (Sigma; see below). Each DNA bait fragment was tested in at least two independent experiments.
Western blot analysis
The indicated C. difficile strains were grown in TY medium at 37°C and harvested at 24 h (34). Total protein was quantitated using the Pierce Micro BCA Protein Assay Kit (Thermo Scientific), and 8 μg of total protein was separated by electrophoresis on a precast 4-15% TGX gradient gel (Bio-Rad) and then transferred to a 0.45 μm nitrocellulose membrane. Western blot analysis was conducted with either mouse anti-TcdA (Novus Biologicals) or mouse anti-FLAG (Sigma) primary antibody, followed by goat anti-mouse Alexa Fluor 488 (Life Technologies) secondary antibody. Imaging and densitometry were performed with a ChemiDoc and Image Lab Software (Bio-Rad), and a one-way ANOVA, followed by Dunnett’s multiple-comparison test, was performed to assess statistical differences in TcdA protein levels between the rstA mutant and each rstA overexpression strain (GraphPad Prism v6.0). At least three biological replicates were analyzed for each strain, and a representative Western blot image is shown.
Quantitative reverse transcription PCR analysis (qRT-PCR)
C. difficile was cultivated in TY medium and harvested at T3 (defined as three hours after the start of transition phase; OD600 = 1.0). Aliquots of 3 ml culture were immediately mixed with 3 ml ice-cold ethanol:acetone (1:1) and stored at −80°C. RNA was purified and DNase-I treated (Ambion) as previously described (37-39), and cDNA was synthesized using random hexamers (39). Quantitative reverse-transcription PCR (qRT-PCR) analysis, using 50 ng cDNA per reaction and the SensiFAST SYBR & Fluorescein Kit (Bioline), was performed in technical triplicates on a Roche Lightcycler 96. cDNA synthesis reactions containing no reverse transcriptase were included as a negative control to ensure no genomic DNA contamination was present. Results are presented as the mean and standard error of the means of three biological replicates. Statistical significance was determined using a one-way ANOVA, followed by Dunnett’s multiple-comparison test (GraphPad Prism v6.0), to compare transcript levels between the rstA mutant and each rstA overexpression strain.
RESULTS
RstA autoregulates its gene transcription via an inverted repeat overlapping the promoter
Our previous work provided preliminary genetic evidence that the N-terminal putative helix-turn-helix DNA-binding domain was necessary for regulation of toxin gene expression but was dispensable for sporulation initiation (14). However, further work with the recombinant His-tagged RstA proteins revealed that the constructs were expressed at low levels and were not detectable by Western blot in C. difficile lysates (data not shown). We created a new series of tagged proteins, possessing the 3XFLAG tag on the C-terminal end, and found that these were stably expressed and easily detected in C. difficile rstA::erm lysates (Fig. S2A). Corroborating our previous data, expression of the wild-type RstA, the full-length FLAG-tagged RstA and the truncated RstAΔHTH-FLAG-tagged allele complemented sporulation in the rstA mutant (Fig. S2B). As previously observed, only full-length RstA restored toxin production to wild-type levels in the rstA background (Fig. S2C), confirming that the helix-turn-helix motif within the DNA-binding domain is essential for RstA-dependent control of toxin production.
We hypothesized that RstA directly binds to DNA to control toxin gene expression and transcription of additional target genes. This interaction is predicted to occur via the putative DNA-binding domain, as observed for other RRNPP transcriptional regulators (40-42). To test this, we first defined the rstA promoter region and probed the DNA-binding capability of RstA within the promoter. The transcriptional start of rstA was identified at −32 bp upstream from the translational start using 5’ RACE. Corresponding σA-10 and −35 consensus sequences were detected immediately upstream of this transcriptional start site (Fig. 1A, B). To verify the mapped promoter and to determine if any additional promoters are present that drive rstA transcription, a series of promoter fragments fused to the phoZ reporter gene was created, and alkaline phosphatase (AP) activity was measured in the 630Δerm and rstA::erm mutant. As previously observed, the full-length 489 bp rstA promoter fragment exhibited a 1.8-fold increase in activity in the rstA mutant compared to the parent strain, indicating RstA-dependent repression (Fig. 1C, (14)). The truncated promoter fragments, PrstA291 and PrstA231, produced a similar fold change in activity in the rstA mutant and parent strain, as was observed for the full-length promoter. However, reporter activity was lower in the PrstA115 fragment compared to the longer fragments, suggesting that an enhancer sequence or an additional RstA-independent transcriptional activator is located between −231 bp to −115 bp upstream of the rstA open reading frame. A promoter fragment reporter fusion containing 380 bp of sequence upstream from the mapped rstA promoter (from −489 bp to −112 bp; IR; Fig. 1A) was inactive, indicating that an additional promoter is not located within this region. Altogether, the data demonstrate that the mapped σA-dependent promoter drives rstA expression and that RstA can repress transcription from this promoter.
The results obtained from the promoter-reporter fusions suggested that RstA binding was likely to occur within the 115 bp upstream of the translational start site. A 29 bp imperfect inverted repeat was identified within the predicted PrstA −10 consensus sequence, suggesting a possible regulatory binding site within this region (Fig. 1B). To determine whether this sequence serves as an RstA recognition site, we created a series of single nucleotide substitutions within the inverted repeat in the 489 bp PrstA reporter fusion, avoiding conserved residues required for RNAP-holoenzyme recognition (43). Most of the single nucleotide substitutions did not significantly alter reporter activity compared to the wild-type PrstA reporter (Fig. 1D). However, three nucleotide substitutions in two positions, A-21 and T-19, abolished RstA repression in the parent strain, increasing reporter activity to match that of the rstA::erm mutant. These data suggest that the A-21 and T-19 nucleotides are important for RstA binding to the rstA promoter.
RstA inhibits toxin and motility gene transcription
Regulatory control of toxin gene expression in C. difficile involves multiple sigma factors and transcriptional regulators, which ensure toxin production occurs in the appropriate environmental conditions (13). Our previous work (14) demonstrated that an rstA::erm mutant has increased transcription of the C. difficile toxin genes, tcdA and tcdB, the toxin-specific sigma factor, tcdR, and the flagellar-specific sigma factor, sigD, which is required for motility and directs tcdR transcription (11, 12). To determine whether RstA is involved directly in repressing transcription of these genes, we first constructed phoZ reporter fusions with the promoter regions for each gene and examined RstA-dependent transcriptional activity.
The tcdR promoter region contains four identified independent promoter elements: a σA- dependent promoter (−16 bp from the translational start), a σD-dependent promoter (−76 bp from the translational start) and two putative σTcdR promoters farther upstream (Fig. 2A; (11, 12, 44- 46)). Expression of the tcdR gene is relatively low in C. difficile (11, 45, 47), at least in part due to repression by CodY and CcpA binding throughout the tcdR promoter region under nutrient-rich conditions (8, 9, 37, 46, 48). We examined each of the promoter elements within PtcdR to determine if RstA affects transcription from these promoters. A series of reporter fusions was created for each of the promoter elements, which were examined in the rstA::erm mutant and parent strain, respectively, and activity was measured after 24 h of growth in TY medium (Fig. 2A). A full-length 517 bp PtcdR::phoZ reporter and the two σTcdR-dependent promoter fusions exhibited similar low reporter activities in the parent and the rstA strains (Fig. 2B). But, increased reporter activity was observed in the rstA mutant for the individual σA-dependent and σD-dependent promoter fusions. These results indicate that RstA impacts the function of these promoter elements and contributes to repression of tcdR transcription.
We also examined RstA-dependent regulation of tcdA and tcdB transcription, both of which are expressed solely from σTcdR-dependent promoters (Fig. 2A; (47, 49, 50)). PtcdA reporter activity was increased 3.6-fold and PtcdB activity was 2.1-fold greater in the rstA strain compared to the parent (Fig. 2C, D). Altogether, these data indicate that RstA represses toxin gene transcription at the individual gene level and through repression of tcdR.
SigD, also known as FliA or σ28, is a sigma factor that coordinates flagellar gene expression and directly activates tcdR gene expression (45). The sigD gene is located in a large, early-stage flagellar operon, that is transcribed from a σA-dependent promoter located 496 bp upstream from the first gene of the operon, flgB (51). Interestingly, the flgB promoter sequences from two different C. difficile strains, the historical epidemic strain, 630, and a current epidemic strain, R20291, are identical from the σA promoter sequence through the translational start site, but diverge considerably upstream of this region (Fig. S3). No additional promoter elements were identified in the 630 or R20291 sequences upstream of the σA-dependent promoter (Fig. 3A). To determine whether RstA influences sigD transcription through repression of PflgB, promoter reporter fusions representing each strain were constructed. As anticipated, activity of the 630Δerm and R20291 PflgB reporters were higher in the rstA mutant compared to the parent strain (1.7-fold and 1.5-fold, respectively; Fig. 3B), indicating that RstA represses flgB, and consequently, sigD transcription.
RstA directly binds the rstA, tcdR, flgB, tcdA and tcdB promoters via the conserved helix-turn-helix DNA-binding domain
To determine whether RstA directly binds target DNA, a variety of in vitro electrophoretic gel shift assays were attempted, but no binding was observed in any condition tested. We considered that the lack of RstA-DNA interaction by gel shift may occur because of the absence of a cofactor, such as a quorum-sensing peptide, or because of a transient complex or oligomerization state. To overcome this obstacle, we performed biotin-labelled DNA pulldown assays to assess the DNA-binding capacity of RstA under native conditions. Biotinylated DNA was coupled to streptavidin beads as bait and incubated with cell lysates expressing either full-length RstA-FLAG or RstAΔHTH-FLAG protein. Specifically-bound proteins were eluted and analyzed by western blot using FLAG M2 antibody.
We first tested the ability of RstA to directly interact with its own promoter. RstA-FLAG protein was recovered using the wild-type rstA promoter region as bait, demonstrating specific interaction of the RstA protein (Fig. 4A). However, the PrstA fragment did not capture RstAΔHTH-FLAG protein, indicating that the conserved HTH domain of RstA is essential for DNA binding. In addition to the wild-type rstA promoter, the PrstA T-19A and PrstA A-21C variants that eliminated RstA-dependent regulation in vivo were used as bait (Fig. 1D). Both the PrstA T-19A and PrstA A-21C variants captured less RstA-FLAG than the wild-type promoter, suggesting that these nucleotides facilitate RstA interaction (Fig. 4A). Finally, the intergenic region upstream of the rstA promoter (Fig. 1A; IR) did not recover the full-length RstA-FLAG, indicating that RstA recognizes a specific DNA sequence within the promoter region. Altogether, these data demonstrate that RstA functions as a DNA-binding protein that directly and specifically binds its own promoter to repress transcription.
To determine if RstA directly binds DNA to repress the transcription of genes encoding toxin regulators, we examined RstA binding to the flgB and tcdR promoter regions. RstA-FLAG protein bound specifically to the full-length tcdR promoter region, as well as the 630 and R20291 flgB promoters (Fig. 4B). Again, the HTH domain was required for these RstA-promoter interactions. To identify which internal promoter elements directly interact with RstA, previously characterized tcdR promoter fragments were used as bait (Fig. 2B), with the exception of a longer σA-dependent promoter fragment (92 bp rather than 76 bp) to limit potential steric hindrance of RstA binding due to the 5’ biotin label. This longer 92 bp PtcdR(σA) fragment exhibited the same RstA-dependent regulation in reporter assays as the 76 bp reporter (Fig. S4). RstA-FLAG bound to the σA-dependent and σD-dependent tcdR promoter fragments but was not recovered from either of the σTcdR-dependent promoters (Fig. 4C), corroborating the reporter fusion results that demonstrated RstA repression of only the σA-dependent and σD- dependent tcdR promoter elements.
DNA pulldown assays were also performed to ascertain if RstA directly binds to the tcdA and tcdB promoters. Both of the toxin promoters captured the full-length RstA-FLAG protein and failed to recover the RstAΔHTH-FLAG protein (Fig. 4D). These data provide direct biochemical evidence that RstA represses tcdR, tcdA and tcdB transcription by binding to the promoter regions of these genes.
RstA represses toxin gene expression independently of SigD-mediated toxin regulation
Our data indicate that RstA represses toxin gene expression directly by binding to the tcdA and tcdB promoter regions, and indirectly by repressing transcription of the sigma factors tcdR and sigD, which activate toxin gene expression. To confirm genetically that RstA represses toxin gene expression independently of sigD, we created an rstA sigD double mutant and examined the impact of each mutation on toxin production. To aid in construction of a rstA sigD double mutant, we utilized the recently developed CRISPR-Cas9 system modified for use in C. difficile to create an unmarked, non-polar deletion of rstA in the 630Δerm and sigD::erm backgrounds (Fig. S5) (31). TcdA protein levels were ~3-fold higher in the rstA sigD double mutant compared to the sigD mutant (Fig. 5A), indicating that RstA represses toxin production independently of SigD. Overexpression of rstA in the rstA sigD mutant returned TcdA protein to the levels found in the sigD mutant. Likewise, sigD overexpression in the rstA sigD mutant restored TcdA to wild-type levels, further supporting that SigD and RstA regulate toxin production independently (Fig. 5A). In addition, transcript levels of tcdA, tcdB, and tcdR were increased in the rstA sigD mutant compared to the sigD mutant (Fig. 5B), mirroring the TcdA protein results. Altogether, these data provide further evidence that RstA is major regulator of toxin production that directly and indirectly represses toxin gene expression independently of SigD.
RstA DNA-binding activity requires a species-specific predicted quorum-sensing domain
The observation that RstA does not bind to target DNA in the tested in vitro conditions, but does bind DNA in cell lysates, suggests that a co-factor is required for RstA DNA-binding activity. We hypothesize that a small, quorum-sensing peptide serves as an activator for RstA DNA-binding, as has been observed for other members of the RRNPP family (23-25, 52-54). To test this, we expressed RstA orthologs of other clostridial species (Fig. S6), including Clostridium acetobutylicum (C.a.), Clostridium perfringens (C.p.), and Clostridium (Paeniclostridium) sordellii (C.s.) in the C. difficile rstA mutant background. Only the C.p. RstA was stably produced in C. difficile (Fig. S7). However, expression of the C. perfringens rstA ortholog failed to restore TcdA protein to wild-type levels (Fig. 6A). C.p. RstA may be unable to repress C. difficile toxin production because the C.p. DNA-binding domain cannot recognize the C. difficile DNA target sequences and/or because the DNA-binding activity of C.p. RstA is not functional in C. difficile. To distinguish between these possibilities, we constructed a chimeric protein containing the C. perfringens DNA-binding domain (M1-Y51) fused to the C-terminal domains of the C. difficile RstA protein (herein known as CpHTH-CdCterminalFLAG) and examined the function of this chimeric RstA in the C. difficile rstA mutant. The RstA chimera restored C. difficile TcdA levels to those observed in the parent strain (Fig. 6A), indicating that the C. perfringens DNA-binding domain is functional in C. difficile. These data strongly suggest that the C-terminal portion of RstA responds to species-specific intracellular signals to control the N-terminal DNA-binding activity.
Finally, we assessed the ability of a C. perfringens RstA to complement the low sporulation frequency of the C. difficile rstA mutant. Overexpressing the full-length C. perfringens RstA did not complement sporulation in the C. difficile rstA mutant (Fig. 6B). Unexpectedly, a hypersporulation phenotype was observed when the CpHTH-CdC-terminalFLAG RstA chimera was expressed in the rstA mutant (Fig. 6B), indicating that the chimeric C.p.-C.d. RstA promotes C. difficile sporulation to even higher levels than the native C. difficile RstA. This hypersporulation phenotype suggests that the C.p. HTH portion of the chimeric RstA protein alters the structure or activity of RstA to increase the positive effect on early sporulation events. These data warrant further investigation into the molecular mechanisms by which the C-terminal domains of RstA cooperate with the DNA-binding domain to promote sporulation.
DISCUSSION
The production of exotoxins and the ability to form quiescent endospores are two essential features of C. difficile pathogenesis. The regulatory links between toxin production and spore formation are complex and poorly understood. Some conserved sporulation regulatory factors, including Spo0A, CodY, and CcpA, strongly influence toxin production, yet some of these regulatory effects appear to be strain-dependent or are indirect (8, 48, 55-57). Further, additional environmental conditions and metabolic signals, such as temperature and proline, glycine and cysteine availability (5, 6, 10, 58), impact toxin production independently of these regulators, revealing the possibility that additional unknown factors are directly involved in toxin regulation (13). The recently discovered RRNPP regulator, RstA, represses toxin production and promotes spore formation, potentially providing a direct and inverse link between C. difficile spore formation and toxin biogenesis (14).
In this study, we show that RstA is a major, direct transcriptional regulator of C. difficile toxin gene expression. RstA inhibits toxin production by directly binding to the tcdA and tcdB promoters and repressing their transcription. RstA reinforces this repression by directly downregulating gene expression of tcdR, which encodes the sole sigma factor that drives tcdA and tcdB transcription. Finally, RstA directly represses the flgB promoter, inhibiting gene expression of the flagellar-specific sigma factor, SigD. SigD activates motility gene transcription but is also required for full expression of tcdR (11, 12). RstA repression of each major component in the toxin regulatory pathway creates a multi-tiered network in which RstA directly and indirectly controls tcdA and tcdB gene expression (Fig. 7).
RstA is the third characterized transcriptional repressor that directly binds to promoter regions for tcdR, tcdA, and tcdB. The two others, CodY and CcpA, exhibit ~10-fold higher affinity in vitro for the tcdR promoter than for the tcdA and tcdB promoters (8, 9, 48), suggesting that CodY and CcpA repress toxin gene expression primarily through repression of tcdR transcription. Similarly, RstA appears to bind with greater affinity to the tcdR promoter than to the tcdA and tcdB promoters, although we caution that biotin pulldown results are semi-quantitative. The in vivo contribution of this reinforced repression of tcdA and tcdB transcription by CodY, CcpA, and RstA remains unknown. Interestingly, recent evidence has demonstrated that tcdR gene expression serves as a bistable switch that determines whether individual C. difficile cells within a population produce TcdA and TcdB, creating a divided population of toxin-OFF and toxin-ON cells (59). TcdR governs this bistability state by maintaining low basal expression levels, allowing for small changes to result in stochastic gene expression, and by positively regulating its own expression, establishing a positive feedback loop that bolsters the toxin-ON state (59). CodY was found to influence the population so that fewer cells produced toxin, but CcpA and RstA were not tested (59). We predict that both CcpA and RstA would bias the population of cells to a toxin-OFF state. Altogether, the tight control of tcdR transcription, reinforced by direct repression of tcdA and tcdB transcription by CcpA, CodY, and RstA, results in the convergence of multiple regulatory pathways at the bistable tcdR promoter to coordinate toxin production in response to nutritional and species-specific signals. This complex regulation ensures that the energy-intensive process of toxin production is initiated only to benefit the bacterium under the appropriate conditions.
Importantly, RstA is the first transcriptional regulator demonstrated to directly control flgB transcription initiation. To date, none of previously identified regulators of flgB expression, including Spo0A, SigH, Agr, Hfq, SinR and SinR’, have been shown to bind promoter DNA and regulate flagellar gene expression through transcription initiation (55, 60-63). flgB expression is further regulated post-transcriptionally via a c-di-GMP riboswitch or flagellar switch, both of which are located within the large, 496 bp 5′ untranslated region (51, 64, 65); however, the impact of RstA-mediated repression of flgB gene expression through additional pathways has not yet been explored.
Although we have identified several direct RstA targets, the sequence required to recruit RstA to target promoters remains unclear. The rstA promoter contains a near-perfect inverted repeat; however, this sequence is AT-rich, as is the case for many C. difficile promoters. Imperfect inverted repeats were also found overlapping the −35 consensus sequences of the tcdA, tcdB, flgB and σA-dependent tcdR promoters, and immediately upstream of the σD- dependent tcdR promoter (Fig. S8), suggesting that RstA inhibits transcription at these promoters by sterically obstructing RNA polymerase docking. No clear consensus sequence defining an RstA-box is delineated from these sequences. Exhaustive attempts at ChIP-seq analysis to identify the C. difficile RstA regulon proved unsuccessful; however, our data imply that RstA is a transcriptional repressor that directly controls multiple C. difficile phenotypes, and additional targets within in the C. difficile genome seem likely.
The inability to recapitulate RstA-DNA-binding with purified RstA in vitro together with the functional analysis of C. difficile and C. perfringens full-length and chimeric proteins suggest that i) RstA DNA-binding activity requires a cofactor and ii) this cofactor is species-specific. Most RRNPP members are cotranscribed with their cognate quorum-sensing peptide precursor (19), but there are notable exceptions, including those encoded by unlinked genes (52, 66) and orphan receptors whose cognate ligands have not yet been discovered (67-69). Importantly, no type of ligand other than small, quorum-sensing peptides has been identified for RRNPP proteins. In addition to RstA, other quorum-sensing factors have been implicated in C. Difficile toxin production. The incomplete Agr1 and conserved Agr2 quorum-sensing systems induce toxin production through the production of a cyclic auto-inducer peptide (AIP) that is sensed extracellularly (61, 70, 71); however, it is highly unlikely that the extracellular AIP molecule directly interacts with the cytosolic RstA protein. In addition, the interspecies LuxS-derived autoinducer-2 (AI-2) quorum-sensing molecule was found to increase C. difficile tcdA and tcdB gene expression, but not tcdR (72), indicating that AI-2 does not signal through RstA either. Although there are no open reading frames adjacent to RstA that encode an apparent quorum-sensing peptide precursor, taken altogether, it is reasonable to presume that an unidentified quorum-sensing mechanism controls RstA-dependent regulation of C. difficile TcdA and TcdB production.
Finally, as RstA is necessary for efficient C. difficile spore formation, the possibility remains that species-specific signaling is required for RstA-dependent control of early sporulation and that RstA coordinates C. difficile toxin production and spore formation in response to the same signal(s). Elucidating the molecular mechanisms that govern RstA activity will provide important insights into the regulatory control between sporulation and toxin production, reveal host cues and conditions that lead to increased toxin production, and help delineate the early sporulation events that control C. difficile Spo0A phosphorylation and activation.
ACKNOWLEDGEMENTS
We are grateful for the gift of C. perfringens S13 genomic DNA from Dr. Jorge Vidal. We give special thanks to Dr. Charles Moran and the members of McBride lab for helpful suggestions and discussions during the course of this work. This research was supported by the U.S. National Institutes of Health through research grants AI116933 and AI121684 to S.M.M, AI107029 to R.T., and AI120613 to B.R.A-F. The content of this manuscript is solely the responsibility of the authors and does not necessarily reflect the official views of the National Institutes of Health.