RstA Is a Major Regulator of Clostridioides difficile Toxin Production and Motility.

Clostridioides difficile infection (CDI) is a toxin-mediated diarrheal 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 multispecies 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 the host. A recently discovered regulatory factor, RstA, inhibits toxin production and positively influences spore formation. Herein, we determine that RstA directly binds its own promoter DNA to repress its own gene transcription. In addition, our data demonstrate 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. 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, species-specific sensing mechanism.

material). Corroborating our previous data (14), 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 (14), only full-length RstA restored toxin production to wild-type levels in the rstA background ( Fig. S2C and D), 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 (26)(27)(28). Additionally, we previously observed that rstA expression remains relatively unchanged throughout growth in multiple conditions and that rstA transcription is increased in an rstA mutant (14), suggesting that expression of rstA may be autoregulated. To determine whether RstA is DNA-binding protein, we first defined the rstA promoter region and probed the DNA-binding capability of RstA within its own promoter. The transcriptional start of rstA was identified at Ϫ32 bp upstream from the translational start using 5= RACE. Corresponding A Ϫ10 and Ϫ35 consensus sequences were detected immediately upstream of this transcriptional start site ( Fig. 1A and B). To verify the mapped promoter and to determine whether 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 mutants. 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, PrstA 291 and PrstA 231 , produced similar fold changes in activity in the rstA mutant and parent strains, as observed for the full-length promoter. However, reporter activity was lower in the PrstA 115 fragment compared to the longer fragments, suggesting that an enhancer sequence or an additional RstA-independent transcriptional activator is located between Ϫ231 bp to Ϫ115 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 Ϫ489 bp to Ϫ112 bp; intergenic region (IR) in Fig. 1A] was inactive, indicating that an additional promoter is not located within this region. We also tested whether RstA-dependent repression of the full-length PrstA reporter could be complemented. We expressed the rstA-FLAG construct from the nisin-inducible promoter, cprA (29), divergently from the PrstA::phoZ construct on the same plasmid in the rstA::erm background. PrstA reporter activity was reduced in a dose-dependent manner relative to the amount of nisin added to the medium (Fig. S3), further confirming the autoregulatory effect RstA exerts on its own expression. Altogether, the data demonstrate that the mapped A -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 Ϫ10 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 (30). Most of the single nucleotide substitutions did not significantly alter reporter activity compared to the wild-type PrstA reporter (Fig. 1D). However, 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 that 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 flagellum-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 A -dependent promoter (Ϫ16 bp from the translational start), a D -dependent promoter (Ϫ76 bp from the translational start), and two putative TcdR promoters farther upstream ( Fig. 2A) (11,12,(31)(32)(33). Expression of the tcdR gene is relatively low in C. difficile (11,32,34), at least in part due to repression by CodY and CcpA binding throughout the tcdR promoter region under nutrient-rich conditions (8,9,33,35,36). We examined each of the promoter elements within PtcdR to determine whether 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, 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 TcdR -dependent promoter fusions exhibited similar low reporter activities in the parent and rstA strains (Fig. 2B). However, increased reporter activity was observed in the rstA mutant for the individual A -dependent and D -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 TcdR -dependent promoters ( Fig. 2A) (34,37,38). 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 and 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 28 , is a sigma factor that coordinates flagellar gene expression and directly activates tcdR gene expression (32). The sigD gene is located in a large, early-stage flagellar operon that is transcribed from a A -dependent promoter located 496 bp upstream from the first gene of the flgB operon (39). Interestingly, the flgB promoter sequences from two different C. difficile strains, the historical epidemic  (12,(32)(33)(34) and the open reading frames of all three genes (not drawn to scale). Pale red boxes approximate CodYand CcpA-binding sites within the toxin gene promoters (8,9,36) RstA Directly Represses C. difficile Toxin ® strain, strain 630, and a current epidemic strain, strain R20291, are identical to the A promoter sequence through the translational start site but diverge considerably upstream of this region (Fig. S4). No additional promoter elements were identified in the strain 630 or R20291 sequences upstream of the A -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 strain 630Δerm and R20291 PflgB reporters were higher in the rstA mutant than in 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-labeled 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 blotting 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 T-19A/A-21C variants captured significantly less RstA-FLAG than the wild-type promoter, suggesting that at least the T-19A nucleotide facilitates RstA interaction ( Fig. 4A and Fig. S5A). 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. Finally, RstA-FLAG did not interact with unlabeled streptavidin beads nonspecifically ( Fig. 4A and Fig. S5A). 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 whether 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 and Fig. S5B). 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 A -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( A ) fragment exhibited the same RstA-dependent regulation in reporter assays as the 76-bp reporter (Fig. S6). RstA-FLAG bound to the A -dependent and D -dependent tcdR promoter fragments but was not recovered from either of the TcdR -dependent promoters ( Fig. 4C and Fig. S5B), corroborating the reporter fusion results that demonstrated RstA repression of only the A -dependent and D -dependent tcdR promoter elements.
DNA pulldown assays were also performed to ascertain whether 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 and Fig. S5B). These data provide direct biochemical evidence that RstA represses flgB, 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. The biotin pulldown data suggest that RstA represses toxin gene expression through a multitiered regulatory pathway. To test whether direct repression of tcdA and tcdB transcription by RstA is physiologically relevant and independent of SigD, we created an rstA sigD double mutant and examined the impact of each mutation on toxin production. To aid in construction of an rstA sigD double mutant, we utilized the recently developed TcdRP2 -dependent (188-bp), or TcdRP1 -dependent (112-bp) promoters (C), or the full-length tcdR (446-bp), tcdA (511-bp), or tcdB (501-bp) promoters (D). All promoter fragments were bound to streptavidin-coated magnetic beads and incubated with C. difficile cell lysates grown in TY medium (pH 7.4) supplemented with 2 g/ml thiamphenicol and 1 g/ml nisin to mid-log phase (OD 600 of 0.5 to 0.7), expressing either the RstA-3XFLAG construct (MC1004) or the RstAΔHTH-3XFLAG construct (MC1028).
RstA Directly Represses C. difficile Toxin ® CRISPR-Cas9 system modified for use in C. difficile to create an unmarked, nonpolar deletion of rstA in the 630Δerm and sigD::erm backgrounds (Fig. S7) (40). TcdA protein levels were ϳ3-fold higher in the rstA sigD double mutant than in the sigD mutant ( Fig. 5A; total protein loaded shown in Fig. S8A), 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, a previously characterized sigD overexpression construct (11,41) restored TcdA to wild-type levels in the rstA sigD mutant, 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 levels in 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 the species-specific C-terminal domains. 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 cofactor 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)(24)(25)(42)(43)(44). To test this, we expressed RstA orthologs of other clostridial species (Fig. S9A) (45), including Clostridium acetobutylicum, Clostridium perfringens, and Clostridium (Paeniclostridium) sordellii in the C. difficile rstA mutant background. Only the C. perfringens RstA was stably produced in C. difficile (Fig. S9B). However, expression of the C. perfringens rstA ortholog failed to restore TcdA protein to wild-type levels ( Fig. 6A; total protein loaded shown in Fig. S8B). C. perfringens RstA may be unable to repress C. difficile toxin production because the C. perfringens DNA-binding domain cannot recognize the C. difficile DNA target sequences and/or because the DNA-binding activity of C. perfringens RstA is not functional in C. difficile. To distinguish between these possibilities, we constructed a chimeric protein containing the C. perfringens DNAbinding 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 DNAbinding domain is functional in C. difficile. To confirm these results, we performed qRT-PCR analyses of tcdR, tcdA, and tcdB genes in these strains. The full-length C. perfringens RstA did not complement toxin gene expression in the C. difficile rstA mutant, while the CpHTH-CdCterminal-FLAG chimeric RstA restored toxin gene transcript levels back to those observed in the parent strain (Fig. 6B), corroborating our previous results. These data strongly suggest that the C-terminal portion of RstA responds to species-specific 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. 6C). Unexpectedly, a hypersporulation phenotype was observed when the CpHTH-CdC-terminalFLAG RstA chimera was expressed in the rstA mutant (Fig. 6C), indicating that the chimeric C. perfringens-C. difficile RstA promotes C. difficile sporulation to even higher levels than the native C. difficile RstA does. This hypersporulation phenotype RstA Directly Represses C. difficile Toxin ® suggests that the C. perfringens 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 dependent on the strain or are indirect (8,36,(46)(47)(48). Further, additional environmental conditions and metabolic signals, such as temperature and proline, glycine, and cysteine availability (5, 6, 10, 49), 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 flagellum-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 multitiered 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, following two other transcriptional repressors, CodY and CcpA (8,9,36), 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 (50). 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 (50).

FIG 7
Model of RstA-mediated repression of C. difficile toxin production. SigD, the flagellum-specific sigma factor, directly induces gene transcription of tcdR, the toxin-specific sigma factor. Toxin gene expression is then directed by TcdR. RstA inhibits production of TcdA and TcdB and its own gene expression by directly binding to and repressing transcription of sigD, tcdR, tcdA, tcdB, and rstA, creating a complex, multitiered regulatory network to ensure that the toxin gene expression is appropriately timed in response to the signal(s) that activate RstA.
CodY was found to influence the population so that fewer cells produced toxin, but CcpA and RstA were not tested (50). 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 the 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 (46,(51)(52)(53)(54). flgB expression is further regulated posttranscriptionally via a c-di-GMP riboswitch and a flagellar switch, both of which are located within the large, 496-bp 5= untranslated region (39,55,56); 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 nearperfect 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 Ϫ35 consensus sequences of the tcdA, tcdB, flgB, and A -dependent tcdR promoters, and immediately upstream of the D -dependent tcdR promoter (Fig. S10) (57), 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. Other RRNPP regulators have also been found to bind imperfect, palindromic repeats or specific, conserved sequences in target promoters, but to our knowledge, only PlcR has a defined binding motif (24,58,59). 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 the C. difficile genome seem likely.
The inability to recapitulate RstA-DNA binding with purified RstA in vitro together with the functional analysis of full-length and chimeric C. difficile and C. perfringens 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 (42,60) and orphan receptors whose cognate ligands have not yet been discovered (61)(62)(63). RstA falls into this latter category, as there are no open reading frames adjacent to rstA that encode an apparent quorum-sensing peptide precursor. Importantly, no type of ligand other than small, quorum-sensing peptides has been identified for RRNPP proteins. In addition to RstA, other quorumsensing 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 autoinducer peptide (AIP) that is sensed extracellularly (52,64,65); 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 gene expression (66), indicating that AI-2 does not signal through RstA either. Identification of the cofactor that controls RstA activity is a high priority, as this will likely provide insight into the physiological conditions and/or metabolites that influence 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.

MATERIALS AND METHODS
Bacterial strains and growth conditions. The bacterial strains and plasmids used in the study are listed in Table 1. Clostridioides difficile strains were routinely cultured in BHIS or TY medium (pH 7.4) supplemented with 2 to 5 g/ml thiamphenicol and/or 1 g/ml nisin throughout growth as needed (67). 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 (67,68). C. difficile strains were cultured in a 37°C anaerobic chamber with an atmosphere of 10% H 2 , 5% CO 2 , and 85% N 2 , as previously described (69). Escherichia coli strains were grown at 37°C in LB (70) with 100 g/ml ampicillin and/or 20 g/ml chloramphenicol as needed. Kanamycin (50 g/ml) was used for counterselection against E. coli HB101 pRK24 after conjugation with C. difficile, as previously described (71).
Strain and plasmid construction and accession numbers. Oligonucleotides used in this study are listed in Table 2. Details of vector construction are described in the supplemental material (see Fig. S1 in the supplemental material). C. difficile strains 630 (GenBank accession no. NC_009089.1) and R20291 (GenBank accession no. FN545816.1), Clostridium acetobutylicum ATCC 824 (GenBank accession no. NC_003030.1), Clostridium sordellii ATCC 9714 (GenBank accession no. APWR00000000), and Clostridium perfringens S13 (GenBank accession no. BA000016.3) were used as the 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 (40), was used to create a nonpolar 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 g/ml thiamphenicol. The next morning, the cultures were backdiluted into fresh TY medium supplemented with 5 g/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 (72). 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 (Ϫ32 bp from translational start site; n ϭ 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 [OD 600 ] of 0.5), and 0.25-ml portions were spotted onto 70:30 sporulation agar supplemented with 2 g/ml thiamphenicol and 1 g/ml nisin as a lawn (68). After 24 h growth, ethanol resistance assays were performed as previously described (73,74). Briefly, the cells were scraped from plates after 24 h (H 24 ) and suspended in BHIS medium to an OD 600 of 1.0. The total number of vegetative cells per milliliter 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 of 95% ethanol and 0.2 ml of dH 2 O 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 1ϫ PBS plus 0.1% taurocholate and applied to BHIS plus 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 plus vegetative cells). A spo0A mutant (MC310) was used as a negative sporulation control. Statistical significance was determined using a one-way ANOVA,   presented as the means and standard errors of the means for three biological replicates. The two-tailed Student's t test was used to compare the activity in the rstA mutant to the activity in the parent strain.
Biotin pulldown assays. Biotin pulldown assays were performed as described by Jutras et al. (76). Briefly, a threefold excess of biotin-labeled DNA bait (30 g) was coupled to streptavidin-coated magnetic beads (Invitrogen; binding capacity of 10 g) in B/W buffer, and the bead-DNA complexes were washed with TE buffer to remove unbound DNA. In addition, an unbiotinylated PrstA (30 g) negative control and a beads-only (dH 2 O) negative control were treated alongside the test DNA fragments to ensure that RstA did not interact nonspecifically with the streptavidin-coated magnetic beads. To determine the total amount of biotinylated-DNA bound to each bead preparation, each incubation and subsequent washes were quantitated via a Nanodrop 1000 and subtracted from the initial amount of 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 (OD 600 of 0.5 to 0.7) in 500 ml TY medium (pH 7.4) supplemented with 2 g/ml thiamphenicol and 1 g/ml nisin, pelleted, rinsed with sterile water, and stored at Ϫ80°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 g 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 and 10 g salmon sperm DNA for two total incubations. The bead-DNA-protein complexes were washed seven times with BS/THES buffer supplemented with 10 g/ml salmon sperm DNA and then without salmon sperm DNA to remove nonspecific proteins. The beads were transferred to clean microcentrifuge tubes twice during the washes to eliminate carry-over contamination. 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 blotting using FLAG M2 antibody (Sigma; see below). Each DNA bait fragment was tested in at least three independent experiments. As a control following each experiment, bait DNA was recovered by incubating the labeled beads in dH 2 O at 70°C for 10 min and analyzed on a 1.5% agarose gel to ensure that no cross-contamination occurred (data not shown). Densitometry was performed using Image Lab Software (Bio-Rad), and subsequent statistical analyses included a one-way ANOVA, followed by Dunnett's multiple-comparison test (GraphPad Prism v6.0).
Western blot analysis. The indicated C. difficile strains were grown in TY medium (pH 7.4) supplemented with 2 g/ml thiamphenicol and 1 g/ml nisin at 37°C and harvested at H 24 (24 h) (74). Total protein was quantitated using the Pierce Micro BCA protein assay kit (Thermo Scientific), and 8 g of total protein was separated by electrophoresis on a precast 4 to 15% TGX stain-free gradient gel (Bio-Rad), and total protein was imaged using a ChemiDoc (Bio-Rad). Corresponding gel images for each Western blot are included in the supplemental material as indicated in the text. Protein was then transferred to a 0.45-m nitrocellulose membrane, and 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. C. difficile was cultivated in TY medium (pH 7.4) supplemented with 2 g/ml thiamphenicol and 1 g/ml nisin and harvested at T 3 (defined as three hours after the start of transition phase; OD 600 of 1.0 [approximately equivalent to H 8 on plates]). Aliquots (3 ml) of culture were immediately mixed with 3 ml of ice-cold ethanol-acetone (1:1) and stored at Ϫ80°C. RNA was purified and DNase I treated (Ambion) as previously described (29,35,77), and cDNA was synthesized using random hexamers (77). 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 reaction mixtures containing no reverse transcriptase were included as a negative control to ensure that no genomic DNA contamination was present. Results are presented as the means and standard errors of the means for 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.