The RgaS-RgaR two-component system promotes Clostridioides difficile sporulation through a small RNA and the Agr1 system

The ability to form a dormant spore is essential for the survival of the anaerobic pathogen, Clostridioides difficile, outside of the mammalian gastrointestinal tract. The initiation of sporulation is governed by the master regulator of sporulation, Spo0A, which is activated by phosphorylation. Multiple sporulation factors control Spo0A phosphorylation; however, this regulatory pathway is not well defined in C. difficile. We discovered that RgaS and RgaR, a conserved orphan histidine kinase and orphan response regulator, function together as a cognate two-component regulatory system to directly activate transcription of several genes. One of these targets, agrB1D1, encodes gene products that synthesize and export a small quorum-sensing peptide, AgrD1, which positively influences expression of early sporulation genes. Another target, a small regulatory RNA now known as SpoZ, impacts later stages of sporulation through a small hypothetical protein and an additional, unknown regulatory mechanism(s). Unlike Agr systems in many organisms, AgrD1 does not activate the RgaS-RgaR two-component system, and thus, is not responsible for autoregulating its own production. Altogether, we demonstrate that C. difficile utilizes a conserved two-component system that is uncoupled from quorum-sensing to promote sporulation through two distinct regulatory pathways.


Introduction
Clostridioides difficile is an anaerobic gastrointestinal pathogen that causes severe diarrheal disease [1].The vegetative cells of this strict anaerobe undergo a dramatic morphogenesis within the mammalian gastrointestinal tract to form dormant, highly resistant endospores.This infectious spore enables long-term survival outside of the host and facilitates transmission to new hosts [2].The initiation of spore formation requires the activation of the highly conserved transcriptional regulator, Spo0A, which is encoded in all endospore-forming bacteria [2][3][4].Spo0A DNA-binding activity is controlled by the phosphorylation of a conserved aspartate residue [5,6], allowing phosphorylated Spo0A (Spo0A~P) to directly activate early sporulationspecific gene expression and triggering the entry into sporulation [5][6][7][8].
In other well-studied spore formers, most notably the soil dweller Bacillus subtilis, Spo0A phosphorylation is tightly controlled by an expansive regulatory network comprised of kinases, phosphatases, phosphotransferases, and additional regulators [9].B. subtilis Spo0A phosphorylation is achieved via a phosphorelay that begins with sensor histidine kinases that autophosphorylate and transmit a phosphoryl group through a response regulator (Spo0F) and a phosphotransferase (Spo0B) to Spo0A.Many of the key regulatory proteins that govern Spo0A activation in B. subtilis are not well conserved or are absent in C. difficile genome [10][11][12][13], suggesting that C. difficile utilizes unique factors and regulatory pathways to initiate spore formation.Three C. difficile phosphotransfer proteins (PtpA, PtpB, and PtpC), originally annotated as orphan histidine kinases, appear to inactivate Spo0A under the conditions tested [14,15].Thus, the primary driver of C. difficile Spo0A phosphorylation and activation is unknown.
Environmental cues and nutrient availability influence C. difficile sporulation initiation in the host gastrointestinal tract [16][17][18][19][20]. Additionally, bacterial global signaling systems play a significant role in the decision to initiate sporulation.The intracellular nucleotide second messenger, c-di-GMP, is an inhibitor of early sporulation in C. difficile [21].In other spore formers, quorum-sensing systems are well established for regulating sporulation initiation [22,23].The RRNPP family of multifunctional proteins in B. subtilis and Bacillus cereus sp.directly bind small quorum-sensing peptides to regulate their activity, which impacts Spo0A phosphorylation [24,25].The C. difficile RRNPP ortholog, RstA, positively influences spore formation by directly binding an inhibitor of sporulation [26,27].However, RstA differs from most members of the RRNPP family in that its cognate quorum-sensing peptide is not adjacently encoded on the genome and has yet to be discovered [26], providing additional evidence that C. difficile employs distinct regulatory pathways to initiate sporulation.
Another conserved quorum sensing system implicated in spore formation is the accessory gene regulator (Agr) system, which utilizes an extracellular, cyclic autoinducing peptide (also generically known as the AIP) as the quorum-sensing signal.The prototypical Agr system encompasses the autoinducing peptide (AgrD), a transmembrane protease (AgrB) that processes and exports AgrD, a sensor histidine kinase (AgrC), which directly senses extracellular AgrD, and the cognate response regulator (AgrA) [28].Prior studies in Clostridium botulinum, Clostridium sporogenes, and Clostridium perfringens demonstrated that disruption or knockdown of the agrBD genes resulted in decreased spore formation [29,30].The majority of C. difficile genomes encode an agrB1D1 operon with no apparent cognate agrA1 or agrC1 present [29,[31][32][33].Recent work revealed that the agrB1D1 locus positively impacts early sporulation gene expression and spore formation [34]; however, the identity of the AgrD1 receptor remains elusive.
C. difficile encodes an orphan histidine kinase (CD0576; herein, RgaS) that is annotated as an ArgC-like sensor histidine kinase.Because the sporulation factor(s) that activate Spo0A are unclear, we asked if RgaS plays a role in C. difficile sporulation initiation.Our results reveal that RgaS signals through an orphan response regulator, RgaR, to promote C. difficile spore formation.We demonstrate that the conserved residues required for phosphoryl group transfer in both RgaS and RgaR are necessary for their function as a cognate two-component system.Further, RgaSR directly activate transcription of several loci, including agrB1D1 and a small regulatory RNA, n00620 (nc063; herein referred to as spoZ).We establish that RgaSR influences spore formation through direct regulation of both agrB1D1 and spoZ transcription.Additionally, we show that the effects of AgrD1 and spoZ are independent and impact different stages of sporulation.Unlike prototypical Agr systems, RgaS does not respond to the AgrD1 peptide, indicating that AgrB1D1 has adapted distinct functions that are independent of the RgaSR two-component system.Finally, we present evidence that RgaSR promotion of spore formation through AgrD1 and SpoZ is functionally conserved in diverse C. difficile strains.

The orphan histidine kinase RgaS and the RgaR response regulator promote C. difficile sporulation
The master response regulator of sporulation, Spo0A, is activated by phosphorylation.Multiple kinases activate sporulation in other spore-forming organisms; however, these factors are not conserved in C. difficile.Although three phosphotransfer proteins are proposed to act as phosphatases to inhibit Spo0A activity [15], a factor that directly phosphorylates Spo0A has not been discovered [13].We sought to determine if other encoded kinases could activate Spo0A to drive sporulation.We identified an additional, putative orphan histidine kinase within the C. difficile genome, rgaS (CD0576), that has limited similarity to the other sporulation-associated phosphotransfer proteins, including in the dimerization and histidine autophosphorylation domain (H-box; S1 Fig) .To examine whether RgaS contributes to C. difficile sporulation, we first utilized CRISPR interference (CRISPRi) to directly repress rgaS gene expression.As the addition of xylose to sporulation medium reduces sporulation frequency [15], we modified the C. difficile CRISPRi plasmid [35] to drive expression of the nucleasedeactivated version of the dCas9 gene with the nisin-inducible cprA promoter.A scrambled sgRNA (sgRNA-neg) was included as a control.We assessed the sporulation frequency of 630Δerm expressing either sgRNA-neg or sgRNA-rgaS by determining the ratio of spores to vegetative cells after 24 h of growth on 70:30 sporulation agar, with and without nisin.Expressing the anti-rgaS target resulted in a ~6-fold decrease in spore formation compared to the control strain (Fig 1A).To ensure that rgaS was directly targeted, rgaS transcripts were measured by quantitative reverse-transcription PCR (qRT-PCR), and as expected, rgaS transcripts were decreased ~25-fold in the presence of nisin (S2A Fig) .This data suggests that RgaS positively impacts C. difficile sporulation.
RgaS possesses a histidine kinase-like ATPase domain that is similar to the AgrC histidine kinase of Staphylococcus aureus (Fig 1B).Additional AgrC-like kinases have been identified in Clostridia, including the Clostridium perfringens VirS [36] and the C. difficile AgrC2, which is encoded only in the epidemic ribotype 027 strains [37].Recent work established that C. perfringens VirS, which directly phosphorylates its cognate response regulator, VirR [38], responds to the Agr quorum sensing peptide produced by C. perfringens to activate VirR- We hypothesized that if RgaS is a VirS-like histidine kinase, then its cognate response regulator would be similar to C. perfringens VirR.We scanned the C. difficile 630 genome for orphan response regulators and located two VirR orthologs: CD1089 (RgbR) and CD3255 (RgaR).A prior investigation showed that overexpression of rgaR, but not rgbR, in a C. perfringens virR mutant partially complemented toxin production, suggesting that RgaR functions as a VirR ortholog [40].Further, they discovered that RgaR directly activated transcription of three C. difficile operons, including the partial agrB1D1 locus, while an rgbR mutant did not exhibit any transcriptional changes [40] To further explore the impact of RgaS and RgaR on C. difficile sporulation, we generated rgaS and rgaR deletion mutants using allele-coupled exchange [41]  To ensure that the decreased sporulation frequencies of the rgaS and rgaR mutants are solely the result of the deleted genes, we complemented these mutants by expressing the wildtype rgaS and rgaR alleles from their native promoters on the chromosome via the Tn916 conjugative transposon.Expression of the rgaS and rgaR alleles in their respective mutants restored sporulation frequencies to greater than wild-type levels (Fig 2C).Our data suggest that RgaS and RgaR function in the same regulatory pathway as a twocomponent regulatory system (TCS), where the RgaS histidine kinase undergoes autophosphorylation at the conserved histidine residue and transfers the phosphoryl group to the conserved aspartate residue of the RgaR response regulator [42].To determine whether the predicted functional residues of RgaS and RgaR are essential for their activity, we replaced the conserved histidine residue of RgaS and the conserved aspartate residue of RgaR with alanine, using site-directed mutagenesis, and expressed these alleles in their respective mutants via the Tn916 conjugative transposon.Neither the rgaS-H256A nor the rgaR-D57A allele complemented the low sporulation phenotypes of the respective mutants (Fig 2C), demonstrating that these residues are required for RgaS and RgaR to promote sporulation.

Late-stage sporulation is downregulated in the rgaR mutant
To better understand how the RgaS-RgaR TCS promotes sporulation, we performed RNA-seq on the rgaR mutant during growth on sporulation agar (H 12 ) to identify differentially expressed genes during sporulation (S1 Table ).The development of spores requires the activation of Spo0A followed by the sequential activation of the sporulation sigma factors, SigF, SigE, SigG, and SigK [10,43].RNA-seq analysis revealed that only a few Spo0A-or SigF-dependent transcripts were significantly decreased in the rgaR mutant (~22% of the Spo0A regulon and ~33% of the SigF regulon; Table 1), suggesting that RgaR does not strongly influence the initiation of sporulation.To confirm that Spo0A activity is not impacted in the rgaS and rgaR mutants, we measured transcript levels of sigE, a Spo0A~P-dependent gene, by qRT-PCR (S4A Fig) .There were no significant differences in sigE transcript levels between the rgaS, rgaR, and parent strains, confirming that early sporulation gene regulation is not considerably impacted by RgaSR.In contrast, ~84% of the SigE regulon, 94% of the SigG, and 97% of the SigK regulons were significantly downregulated in the rgaR mutant (Table 1), revealing that RgaR promotes the progression to later stages of spore formation.
The RNA-seq results also revealed that expression of the toxin genes tcdA and tcdB were largely unaffected in the rgaR mutant, with a modest ~2-fold decrease in tcdB transcripts (S1 Table ).We further assessed toxin production by measuring the total TcdA and TcdB toxin present in the supernatants of the rgaS and rgaR mutants.Toxin levels were unchanged between the rgaS mutant, the rgaR mutant, and the parent strain (S4B Fig) .Altogether, these data confirm that the RgaSR TCS does not substantially impact toxin production in the conditions tested and suggests that the effects of RgaSR on sporulation regulation are independent of toxin regulation.a The Spo0A and sporulation-associated sigma factor regulons are based on those reported in [7].
b Genes whose expression were dependent on multiple sporulation-specific sigma factors were excluded from this analysis.
c These values are based on genes differentially expressed at least 2-fold and with p < 0.05 from the RNA-seq analyses of 630Δerm ΔrgaR (MC2229) compared to the 630Δerm parent strain grown on 70:30 agar at H 12 . https://doi.org/10.1371/journal.pgen.1010841.t001

RgaS and RgaR have the same effects on gene expression
Prior work revealed that RgaR directly binds and activates transcription from three promoters that contain highly conserved VirR-like consensus binding sites [40].These direct RgaR targets include the promoters for agrB1D1, the hypothetical factor CD2098, and CD0587-CD0588, a two-gene operon encoding two hypothetical proteins.Bioinformatical mining uncovered two additional VirR-like binding sites located in intergenic regions within the C. difficile genome [40], which have since been annotated.One VirR-like binding site is located upstream of a small ORF encoding a hypothetical protein, CD15111, and the second site lies immediately upstream of a predicted small regulatory RNA we have dubbed SpoZ (Fig 3A).Notably, the RNA-seq data revealed that only sporulation genes were downregulated in the rgaR mutant with the exception of these known direct RgaR targets (S1 Table ), further supporting that this list encompasses all genes directly regulated by RgaR.
We next asked if these direct RgaR targets are similarly regulated in the rgaS mutant.Using qRT-PCR, we assessed transcript levels of RgaR-regulated genes in sporulating cells and observed the same decrease in transcript levels for the rgaS and rgaR mutants (Fig 3B).Altogether, these data strongly support that RgaS is the cognate histidine kinase of the RgaR response regulator and that these proteins comprise a functional two-component regulatory system.

RgaR regulates sporulation through the small regulatory RNA, SpoZ
To determine which direct RgaR target is responsible for the low sporulation phenotype in the rgaS and rgaR mutants, we used CRISPRi to individually knockdown transcription of CD0587, CD2098, CD15111, and spoZ, and assessed the impacts on sporulation.Knockdown of CD0587, CD2098, or CD15111 did not alter sporulation frequency compared to the control (Fig 4A).However, the knockdown of spoZ significantly increased spore formation (Fig 4A).This was a surprising result given that the sporulation phenotype of the spoZ knockdown is opposite of what we observe for the rgaS and rgaR mutants.However, spoZ is the only known RgaR target that differentially impacts sporulation when targeted by CRISPRi, suggesting that the RgaS and RgaR influence sporulation frequency through the SpoZ small regulatory RNA.As before, we confirmed direct targeting of transcription by CRISPRi using qRT-PCR (S5 Fig).
Because the spoZ locus was not annotated when the original RgaR regulon was elucidated by bioinformatics, RgaR-dependent regulation of this locus had not been confirmed [40].We created a reporter fusion of the spoZ promoter region to the alkaline phosphatase (AP) gene, phoZ [44].We expressed PspoZ::phoZ in 630Δerm and the rgaR mutant and found that AP activity was virtually eliminated in the rgaR mutant compared to the parent strain (Fig 4B), indicating that RgaR is required for efficient spoZ expression.To confirm that the identified VirR/RgaR direct repeats are responsible for direct RgaR-dependent transcription of spoZ, we mutated the critical "G" residue within each direct repeat of the PspoZ::phoZ reporter.AP activity from the PspoZ G-51A/G-72A ::phoZ reporter was reduced ~5-fold in the rgaR mutant compared to the parent strain (Fig 4B), indicating that these two nucleotides within the direct repeats are necessary for direct RgaR binding to the spoZ promoter.Finally, AP activity from the PspoZ G-51A/G-72A ::phoZ reporter was reduced even further in the rgaR background, suggesting that RgaR still has some affinity to the mutated spoZ promoter region.Altogether, these data affirm that RgaR directly activates transcription from the spoZ promoter.

The regulatory sRNA SpoZ influences C. difficile sporulation through multiple regulatory pathways
To determine how SpoZ impacts C. difficile sporulation, we first characterized the spoZ locus.Immediately downstream from the spoZ locus is a small ORF (58 amino acids), CD16671, which is encoded on the antisense strand (Fig 5A).To confirm the expression and determine the length of the spoZ transcript, we performed near-infrared northern blotting and detected a transcript corresponding to approximately 560 nt in the 630Δerm background (Fig 5B).A similarly sized transcript, although less abundant, was observed in the R20291 background.No corresponding spoZ transcript was detected in the rgaR mutant, which was expected as RgaR is required for efficient activation of spoZ transcription.We further analyzed the cDNA synthesized from this locus in the parent strain (S6A, S6B Fig) , and these data confirmed that the spoZ transcript terminates within the CD16671 promoter region (S6B Fig) .Altogether, these data corroborate the previously mapped transcription termination site for spoZ, slightly beyond the CD16671 ORF (528 nt) [45], indicating that spoZ is encoded antisense to the entire CD16671 ORF.The predicted secondary structure of SpoZ was determined using UNAFold [46] and is shown in S6C Fig.
SpoZ was originally identified as a non-coding intergenic RNA and is predicted to act as an antisense target for the mRNA transcript of the neighboring ORF, CD16671 [45,47,48].Subsequent studies discovered that SpoZ copurifies with Hfq, an RNA-binding protein that mediates post-transcriptional RNA interactions [48].We hypothesized that SpoZ functions as an antisense RNA that targets the CD16671 transcript for Hfq-dependent turnover.We assessed CD16671 transcript levels in the rgaS and rgaR mutants and found that the CD16671 transcript is significantly increased in these mutants (Fig 5C).These results suggest that SpoZ may promote sporulation by inhibiting accumulation of the CD16671 transcript and subsequent protein product.
To test this hypothesis, we generated deletion mutants of either the entire spoZ-CD16671 locus or only CD16671 (see Methods; Because the rgaS and rgaR mutants have fewer spoZ transcripts and lower sporulation frequencies, and the spoZ-CD16671 mutant and spoZ knockdown strains also have fewer spoZ transcripts but hypersporulate, we asked if the difference in the sporulation phenotypes between these strains is due to differences in CD16671 transcript levels.We revisited the spoZ CRISPRi knockdown strain, which hypersporulated ( These data support the hypothesis that CD16671 accumulation inhibits sporulation, and provides additional evidence that SpoZ has an additional unidentified target(s) that mediates its effects on sporulation when CD16671 is not expressed at high levels.Interestingly, 630Δerm encodes a CD16671 ortholog, CD19903, which shares 80% nucleotide identity and 81% amino acid identity (S8A Fig) .We asked whether SpoZ also affects CD19903 transcript levels and found that there was no significant difference in CD19903 transcript levels in the rgaR mutant (S8B Fig) .We observed a statistically significant, but highly variable, ~1.8-fold increase in CD19903 transcript levels in the spoZ-CD16671 mutant.These data indicate that SpoZ does not influence CD19903 in the conditions tested.It is likely that CD19903 functions similarly to CD16671 but is under different regulatory control.
Finally, the spoZ-CD16671 mutant was confirmed via whole genome sequencing (see Methods).However, despite numerous attempts with various constructs and approaches, we were unable to complement the spoZ-CD16671 mutant.In all, the regulatory portion of SpoZ is not well defined, and the regulation of the spoZ-CD16671 locus is complex.Additional studies are needed to delineate the multifaceted molecular roles SpoZ plays in C. difficile spore formation.

AgrD1 does not signal through the RgaS-RgaR two component system
The prototypical Agr signaling systems form a regulatory feedback loop, in which the quorum peptide signal (AgrD) activates the sensor kinase to initiate Agr-dependent regulation.As RgaS-RgaR directly activates agrB1D1 transcription and a previously published agrB1D1 mutant exhibits lower sporulation frequency [34], we hypothesized that the AgrD1 quorumsensing peptide is recognized by RgaS to activate RgaR-dependent gene expression.To test this, we created an agrB1D1 deletion mutant by allelic exchange (see Methods; S9A Fig) .The agrB1D1 mutant made fewer spores, similar to the rgaS and rgaR mutants, and expressing agrB1D1 from an inducible promoter restored the sporulation frequency to greater than wildtype levels (Fig 6A).
We next asked whether the RgaS-RgaR-dependent gene expression was regulated by AgrD1.To answer this question, we assessed RgaR-dependent gene expression in the agrB1D1 mutant.Surprisingly, qRT-PCR analysis revealed that transcript levels of CD0587, CD2098, CD15111, and spoZ were not dramatically reduced in the agrB1D1 mutant (Fig 6B).Thus, RgaS is not the sensor for the AgrD1 peptide, as expression of RgaR-dependent genes is not affected by AgrD1.This unconnected decrease in sporulation for the agrB1D1 mutant indicates that AgrB1D1 promotes sporulation through a mechanism that is independent of RgaSR.
Prior analyses of transcription in agrB1, agrD1, and agrB1D1 mutants revealed that AgrD1 promotes transcription of early sporulation genes [34].In contrast, disruption of rgaR most significantly impacted expression of late stage sporulation genes (Table 1), implying that SpoZ affects later stages of sporulation.As the rgaR mutant has significantly decreased transcript levels of both agrB1D1 and spoZ (Fig 3B), we asked whether AgrD1 or SpoZ promote different stages of sporulation by performing an epistasis study in the rgaR mutant.We overexpressed either the agrB1D1 or the spoZ-CD16671 locus in the rgaR mutant and assessed sporulation frequency.We anticipated that if SpoZ is epistatic to AgrD1 (that is, if SpoZ functions downstream from AgrD1 in the sporulation regulatory pathway) that only spoZ overexpression will restore sporulation in the rgaR mutant.As expected, overexpression of agrB1D1 did not increase sporulation of the rgaR mutant; however, overexpression of spoZ-CD16671 in the rgaR mutant doubled spore formation (Fig 6C), demonstrating that spoZ is able to partially complement the low sporulation phenotype of the rgaR mutant.Since AgrD1 supports early sporulation, overexpression of agrB1D1 cannot compensate for an rgaR mutant, as SpoZ must be required for later sporulation.

The RgaS-RgaR regulatory network is conserved in the epidemic R20291 background
As the agrB1D1 locus is highly conserved in almost all sequenced C. difficile genomes, we asked whether the RgaS-RgaR system is present and functions similarly in the epidemic ribotype 027 strain, R20291.The R20291 genome encodes highly conserved RgaS and RgaR orthologs, CDR20291_0503 and CDR20291_3113.Null rgaS and rgaR mutants were created in the R20291 background (see Methods; S9B Fig) .We assessed sporulation frequency and observed that the rgaS and rgaR mutants produced fewer spores (~1.6-fold and ~2.1-fold, respectively) than the R20291 parent strain (Fig 7A).Transcriptional analysis of the direct RgaR targets, agrB1D1, the CD2098 ortholog (CDR20291_2005), the CD15111 ortholog (unannotated in R20291; herein referred to as CD15111^), and spoZ, revealed significantly decreased transcript levels in the rgaS and rgaR mutants (Fig 7B), indicating that the RgaS-RgaR orthologs in R20291 function similarly as in 630Δerm.
To further evaluate the RgaS-RgaR and AgrD1 regulatory circuitry in the R20291 background, we next asked whether the R20291 RgaS-RgaR system was responsive to the AgrD1 quorum-sensing peptide.To do this, we utilized CRISPRi to knockdown agrB1 transcription, which significantly decreased transcript levels of both agrB1 and agrD1 (Fig 7C).We then assessed transcript levels of the direct RgaR targets, CDR20291_2005, CD15111^, and spoZ, which remained unchanged in the agrB1D1 knockdown (Fig 7C).These data show that the RgaS-RgaR system also does not respond to AgrD1 in the R20291 background, indicating that this uncoupled signaling circuitry is conserved in diverse C. difficile strains.
To determine whether the regulatory sRNA SpoZ also impacts sporulation in R20291 through CD16671 (again, unannotated in R20291; herein referred to as CD16671^), we first examined CD16671^transcript levels in the R20291 rgaS and rgaR mutants.The R20291 rgaS and rgaR mutants had modestly increased CD16671^transcript levels (~3.8-fold and ~5.6-fold, respectively; Fig 7D), which was not as elevated as CD16671 transcript levels were in the 630Δerm background.As overexpression of CD16671 inhibits spore formation in 630Δerm, the modest increase in CD16671^transcript levels in R20291 likely explain why the R20291 rgaS and rgaR mutants do not exhibit as dramatic sporulation defects as the 630Δerm  7E; S10B).This difference in sporulation phenotypes observed between R20291 and 630Δerm when spoZ is overexpressed may be because of the relative abundance of native spoZ is less in R20291 than in 630Δerm (Fig 5B).Overall, these data indicate that the unusual RgaS-RgaR, AgrD1, and SpoZ regulatory circuitry uncovered in 630Δerm are conserved and functional in the R20291 background.

Discussion
Although the morphological changes throughout spore formation are conserved between clostridia and the well-studied bacilli, the regulatory pathways and molecular mechanisms that govern sporulation have significantly diverged.Further, the mechanism of Spo0A phosphorylation, which activates sporulation, is unclear in C. difficile.While searching for sporulation factors that facilitate C. difficile sporulation initiation, we uncovered a complex regulatory circuit that promotes sporulation through an orthologous accessory gene regulator (Agr) system.We discovered a non-contiguous, cognate two-component system, comprised of the transmembrane histidine kinase RgaS and the response regulator RgaR.RgaS-RgaR directly activate transcription of several loci, including the agrB1D1 operon and the regulatory small RNA, spoZ.We further demonstrated that the gene products of these loci, the quorum-sensing peptide, AgrD1, and the regulatory sRNA, SpoZ, impact C. difficile spore formation at different stages of sporulation.Finally, our investigation revealed that AgrD1 does not activate the RgaS-RgaR two-component system, breaking the conventional positive feedback featured in many characterized Agr systems.
To our knowledge, this is the first identification of an Agr-like two-component system that activates the transcription of the gene encoding the AIP precursor but does not serve as the receptor for the AIP.Classical Agr systems are activated through a positive feedback loop in which the histidine kinase (AgrC) autophosphorylates in response to the accumulation of AIP during high cell density conditions and subsequently transfers the phosphoryl group to the response regulator (AgrA), which in turn, activates transcription of target genes, including the genes that encode the Agr system components [49][50][51][52].Our findings reveal an unusual Agr system in C. difficile that is not autoregulatory.Previous work demonstrated that an agrD1 mutant does not impact agrB1 transcript levels [34], corroborating that the AgrD1 peptide does not influence its own transcription.C. perfringens encodes a single agrB1D1 locus, and the VirS-VirR two component system is the AIP receptor [53,39]; however, VirR does not activate agrB1D1 transcription [54], also breaking the Agr autoregulatory loop.These examples suggest that clostridial Agr systems are not autoregulatory in general, although different mechanisms have evolved to unlink AIP accumulation to its synthesis.Considering that the C. difficile AgrD1 AIP is important for promoting spore formation independently of RgaS-RgaR, there must be another unidentified transmembrane sensor that serves as the AgrD1 receptor.Discovery of the C. difficile AgrD1 receptor will be important in elucidating the regulatory pathway through which AgrD1 promotes sporulation.
A striking feature of the RgaS-RgaR regulon is the direct transcriptional activation of a regulatory RNA.The regulon of Agr systems generally includes a gene encoding a small regulatory RNA.The most notable examples include the RNAIII regulatory RNA of Staphylococcus aureus, which is activated by AgrC-AgrA, and the VR-RNA, VirT, and VirU small regulatory RNAs in C. perfringens, which are activated by VirS-VirR [55][56][57].These regulatory RNAs function as global regulators in both S. aureus and C. perfringens by influencing transcript stability and translational efficiency of target mRNAs [58][59][60][61][62].Our discovery that RgaS-RgaR directly activates spoZ transcription suggests that C. difficile also coordinately regulates multiple targets through a single regulatory sRNA.
The function of SpoZ during C. difficile sporulation appears to be dependent on the accumulation of the hypothetical protein, CD16671.Low levels of spoZ transcript levels in the rgaS and rgaR mutants (~5% compared to the parent strain) led to low sporulation frequencies, while deletion or knockdown of spoZ resulted in hypersporulation.These sporulation phenotypes correlate with the abundance of the CD16671 transcript: strains with low abundance of spoZ and high levels of CD16671 exhibit low sporulation frequencies while those with low spoZ transcript levels and wild-type levels or lower of CD16671 transcripts hypersporulate (Table 2).Altogether, these data suggest that SpoZ function is hierarchal, and the inhibitory effects of CD16671 mask the regulatory functions of other SpoZ targets.The inability to complement the spoZ-CD16671 mutant may also be due to complex interplay between the spoZ and CD16671 transcripts.Additionally, a recent study employing Hfq RIL-seq identified three direct sRNA/mRNA interactions with SpoZ [63], indicating that SpoZ has multiple direct targets.These direct targets include the mRNAs of two genes, CDIF630_01110 and CDIF630_02206), encoding hypothetical proteins, and another small regulatory RNA, CDIF630nc_165, which, in turn, binds to dozens of additional targets [63].Intriguingly, only one of these direct SpoZ targets from the Hfq RIL-seq is encoded in R20291, perhaps offering another explanation of why overexpression of the spoZ-CD16671 locus resulted in different sporulation phenotypes between the R20291 and 630Δerm background (Fig 7E).Future directions will focus on the molecular mechanisms utilized by SpoZ to influence C. difficile sporulation.
Surprisingly, we found that our 630Δerm agrB1D1 mutants exhibited a variable sporulation phenotype that was selectable on plates.However, this variation was not noted with the previously published agrB1D1 mutant, which was also constructed in the 630Δerm background.Since both low and high sporulation phenotypes can be differentiated on plates after passaging a colony exhibiting a single colony morphology, we propose that there is a phase variation event occurring with higher frequency in the agrB1D1 mutants that influences sporulation.Phase variation is a common strategy used by C. difficile to exhibit significant phenotypic heterogeneity [64][65][66][67][68][69][70], and there is a possibility that the Agr1 system may regulate a recombinase or other regulatory factor that influences the phase variation of sporulation.
The components of the RgaS-RgaR-SpoZ-AgrB1D1 regulatory circuitry are highly conserved in C. difficile genomes, suggesting these regulatory pathways are important in controlling earlier and later sporulation events across the species.The universal function of this system in C. difficile is further supported by data from the R20291 strain, which demonstrated conservation of the RgaS-RgaR regulon and sporulation phenotypes.In addition, bioinformatic analysis of the R20291 genome revealed the same RgaR targets, with no additional RgaR binding sites compared to the 630Δerm strain.
Our working model of the Rga regulatory system (Fig 8) begins with RgaS autophosphorylation and phosphoryl group transfer to RgaR, which then directly activates both agrB1D1 and spoZ transcription.The resulting pro-AgrD1 is processed and exported by AgrB1, accumulates extracellularly in high cell density conditions and promotes the initiation of sporulation through an unknown factor.The small regulatory RNA, SpoZ, targets one or more sRNA/ mRNA(s) to influence late stage sporulation.One of these targets includes CD16671, whose expression and/or transcript stability is increased in the absence of SpoZ.CD16671 is proposed to inhibit late stage sporulation through an unknown mechanism.Finally, in the absence of CD16671 accumulation, SpoZ affects at least one additional target (denoted as X) that likely positively impacts spore formation.In all, the RgaS-RgaR regulon integrates at multiple points within the sporulation pathway to control C. difficile spore formation.Understanding when RgaSR is most active and which signals activate (or inhibit) this system will aid in understanding the environmental cues that promote C. difficile sporulation, permitting its survival in the environment and efficient transmission to new hosts.

Bacterial strains and growth conditions
The bacterial strains and plasmids constructed and used in this study are listed in S2 Table.
Plasmid construction details are in S3 Table .C. difficile was routinely cultured in a Coy anaerobic chamber at 37˚C in an atmosphere of 10% H 2 , 5% CO 2 , and balanced with N 2 , as previously described and grown in brain-heart infusion (BHI) broth supplemented with yeast extract [71][72][73].For plasmid maintenance or selection in C. difficile, 2-10 μg/ml thiamphenicol, 100-500 μg/ml spectinomycin, or 2-5 μg/ml erythromycin were supplemented in the media as needed.Overnight cultures were supplemented with 0.1% taurocholate to allow present spores to germinate and 0.2% fructose to prevent sporulation within the culture, as indicated [72,74].Escherichia coli were grown at 37˚C in LB medium supplemented with 100 μg/ml ampicillin, 100 μg/ml spectinomycin, 300 μg/ml erythromycin, and/or 20 μg/ml chloramphenicol as needed, and Bacillus subtilis were grown at 37˚C in LB [75] supplemented with 100 μg/ml spectinomycin or 1 μg/ml erythromycin as needed.B. subtilis was also supplemented with 3.5 mM KNO 3 to support anaerobic growth.Kanamycin (100 μg/ml) was used to counterselect against E. coli and B. subtilis after conjugation with C. difficile [76].

Strain construction
C. difficile 630 (GenBank accession no.NC_009089.1)and R20291 (GenBank accession no.NC_013316.1)were used as the templates for primer design, and C. difficile 630Δerm and R20291 genomic DNA were used as the template for PCR amplification and mutant construction.The C. difficile 630Δerm ΔrgaS, 630Δerm ΔrgaR, 630Δerm ΔrgaS ΔrgaR, 630Δerm ΔspoZ-CD16671, 630Δerm ΔCD16671, 630Δerm ΔagrB1D1, R20291 ΔrgaS, and R20291 ΔrgaR strains were constructed using allele-coupled exchange using either the pMSR or pMSR0 pseudo-suicide vectors for the 630 and R20291 backgrounds, respectively [41].Briefly, pMSR/ pMSR0 vectors containing the targeted gene's 5 0 and 3 0 homology arms, with a selective antibiotic resistance cassette in between, were conjugated into the respective parent strain and were selected for growth on 10 μg/ml thiamphenicol and 100 μg/ml kanamycin.Colonies were then passaged to either 500 μg/ml spectinomycin or 5 μg/ml erythromycin and subsequently grown in 10 ml BHIS with the selective antibiotic and 100 μg/ml anhydrotetracycline (ATC) to induce the double recombination event.After ATC treatment, isolated colonies were screened for successful allelic exchange by PCR.Complementation strains for the rgaS and rgaR mutants were constructed by either expressing the wild-type or site-directed mutated rgaS or rgaR alleles via the conjugative transposon Tn916.rgaR mutants expressing either PcprA-agrB1D1 or PcprA-spoZ-CD16671 and the 630Δerm and R20291 strains expressing PcprA-spoZ-CD16671 were also constructed using the conjugative transposon Tn916.All strains were confirmed by PCR analysis, and most were further confirmed by whole genome sequencing (SeqCenter, Pittsburgh, PA; see below).The Benchling CRISPR Guide RNA Design tool was used to create sgRNAs targeting rgaS, rgaR, CD0587, CD2098, CD15111, spoZ, and agrB1 [35], which were generated by PCR and subsequently cloned into pMC1123.The details of vector construction are given in the supplemental material (see S1 Text).Plasmids pMC1249 and pMC1250 were synthesized by Genscript (Piscataway, NJ).

Illumina whole genome sequencing
C. difficile strains were grown overnight in and harvested from 10 ml BHIS.After storing cell pellets overnight at -80˚C, genomic DNA was isolated using a modified Bust 'N' Grab protocol [26,78].Contaminating RNA was removed by incubation with RNase A (Ambion) for 1 h at 37˚C.Library prep and Illumina sequencing were performed by SeqCenter (Pittsburgh, PA).Briefly, sample libraries were prepared using the Illumina DNA Prep Kit and IDT 10bp UDI indices, and sequenced on an Illumina NextSeq 2000, producing 2x151bp reads.Demultiplexing, quality control, and adapter trimming was performed with bcl-convert (v3.9.3; Illumina; https://support-docs.illumina.com/SW/BCL_Convert/Content/SW/FrontPages/BCL_Convert.htm).Using Geneious Prime v2022.3, the resulting reads were paired and trimmed using the BBDuk plug-in and subsequently mapped to the reference genome (630Δerm; NC_009089.1).The Bowtie2 plug-in was used to search for the presence of SNPs and InDels under default settings with a minimum variant frequency set at 0.85.Genome sequence files were deposited to the NCBI Sequence Read Archive (SRA) BioProject PRJNA986905.

Sporulation assays and phase-contrast microscopy
C. difficile cultures were grown overnight in BHIS medium supplemented with 0.1% taurocholate, to promote germination of existing spores, and 0.2% fructose, to prevent sporulation during growth [72].C. difficile cultures were subsequently diluted with fresh BHIS, grown to midexponential phase (OD 600 ffi 0.5), and plated onto 70:30 agar supplemented with 2 μg/ml thiamphenicol and 1 μg/ml nisin as needed.After 24 h (H 24 ), ethanol-resistant sporulation assays were performed as previously described [14,79].Briefly, cells were scraped from 70:30 agar and suspended in BHIS medium to an OD 600 ffi 1.0.Total vegetative cells per milliliter were determined by immediately serially diluting suspended cells in BHIS, plating onto BHIS plates, and enumerating colony forming units (CFU) after at least 36 h of growth.Simultaneously, spores per milliliter were enumerated after a 15 min incubation in 28.5% ethanol by serially diluting in 1X PBS with 0.1% taurocholate, plating onto BHIS with 0.1% taurocholate, and counting after at least 36 h of growth.The sporulation frequency was calculated as the total number of spores divided by all viable cells (spores plus vegetative cells).A spo0A mutant was used as a negative sporulation control.Statistical significance was determined using a oneway ANOVA, followed by a Dunnett's or Tukey's multiple-comparison test, as indicated (GraphPad Prism 9.5.0).Phase-contrast microscopy was performed H 24 , using the suspended cells from 70:30 agar, with a Ph3 oil immersion objective on a Nikon Eclipse Ci-L microscope.At least two fields of view were captured with a Ds-Fi2 camera.Results represent a minimum of three independent biological experiments.

Enzyme-linked immunosorbent assay (ELISA)
C. difficile TcdA and TcdB from culture supernatants was quantified after 24 h growth of C. difficile strains in TY, pH 7.4, as previously described [80].Supernatants, diluted in the provided dilution buffer, were assayed in technical duplicates using the tgcBIOMICS kit for simultaneous detection of both TcdA and TcdB, according to the manufacturer's instructions.The results represent the means and standard error of the means of four independent biological replicates, and statistical significance was assessed via a one-way ANOVA, followed by a Dunnett's multiple-comparison test (GraphPad Prism 9.5.0).

Quantitative reverse transcription PCR analysis (qRT-PCR)
C. difficile strains were cultured on 70:30 agar as described above, and cells were harvested at H 12 (defined as 12 h after the cultures are applied to the plate) into 6 ml of 1:1:2 ethanol:acetone:dH 2 O solution and stored at -80˚C.RNA was isolated and Dnase-I treated (Ambion), and cDNA was synthesized (Bioline) using random hexamers as previously described [16,18,81].qRT-PCR analysis was performed in triplicate on 50 ng of cDNA using the Sensi-FAST SYBR & Fluorescein kit (Bioline) with a Roche Lightcycler 96.The results were analyzed by the comparative cycle threshold method [82], using the rpoC transcript as the normalizer, and are presented as the means and standard error of the means for a minimum of three independent biological replicates.An one-way ANOVA, followed by a Student's t-test or a Dunnett's multiple-comparison test as applicable (GraphPad Prism v9.5.0), was used to assess statistical significance.

RNA sequencing (RNA-seq) analysis
C. difficile strains were cultured on 70:30 sporulation agar as described above and harvested at H 12 into 6 ml of 1:1:2 ethanol:acetone:dH 2 O solution and stored at -80˚C.RNA was isolated and Dnase-I treated (Ambion).Samples were sent to Microbial Genomics Sequencing Center (MiGS; Pittsburgh, PA) where library preparation was performed using Illumina's Stranded Total RNA prep Ligation with Ribo-Zero Plus kit and 10bp IDT for Illumina indices.Sequencing was done on a NextSeq200 giving 2x50bp reads.Demultiplexing, quality control, and adapter trimming was performed with bcl-convert (v3.9.3; Illumina; see reference above).Using Geneious Prime v2022.2.2, the reads were mapped to the reference genome (630Δerm; NC_009089.1).The expression levels were calculated and then subsequently compared using DESeq2 [83].DESeq2 utilizes the Wald test to calculate P values which are then adjusted using the Benjamini-Hochberg test [83].RNA-seq raw sequence reads were deposited to the NCBI Sequence Read Archive (SRA) BioProject PRJNA986905.

Alkaline phosphatase activity assays
C. difficile strains were cultured on 70:30 sporulation agar as described above and harvested at H 8 (defined as eight hours after the cultures are applied to the plates).Alkaline phosphatase (AP) assays were performed as previously described [44] excepting the exclusion of chloroform.Technical duplicates were averaged and results represent the means and standard error of the means of three independent biological replicates.An one-way ANOVA, followed by a Tukey's multiple-comparison test (GraphPad Prism v9.5.0), was used to assess statistical significance.

Northern blot analysis
Near-infrared fluorescent northern (irNorthern) blot analysis was performed as previously described [84].Briefly, 2.5 nmol of a spoZ DNA probe (oMC3901), containing an internal azide modification (IDT, Coralville, IA), was fluorescently labeled with 50 nmol IRdye 800 CW (DBCO) in 1X phosphate buffered saline, pH 7.4 for 6 h at 25˚C.IRdye-labeled oligonucleotides were purified with Ampure XP beads (Beckman Coulter) as described [84], eluted in 50 μl nuclease-free dH 2 O, and stored at -70˚C.For the irNorthern blot, 20 μg of RNA was diluted 1:1 with Glyoxal Load Dye (ThermoFisher Scientific), heated to 50˚C for 30 min, and electrophoresed on a 1.4% agarose gel.The RNA was transferred to a BrightStar-Plus positively charged nylon membrane (ThermoFisher Scientific) using downward capillary transfer, and the RNA was UV crosslinked to the membrane at 120 mJ/cm 2 using a Stratagene UV Stratalinker 1800.The membrane was prehybridized overnight at 37˚C in ULTRAhyb Hybridization Buffer (ThermoFisher Scientific), and hybridized with 1.3 nM IRdye-labeled DNA probe for 4 h at 37˚C.The membrane was washed once with Low Stringency Wash Solution 1 (Thermo-Fisher Scientific) for 10 min at room temperature, twice for 3 min at 37˚C. and subsequently visualized on a BioRad ChemiDoc MP.The size of the SpoZ transcript was estimated based on a standard curve calculated by the migration of the RNA Century Plus Marker (Ambion).
(see Methods; S3 Fig).Sporulation phenotypes of the mutants were assessed after 24 h growth on sporulation agar.The rgaS and rgaR single mutants sporulated at lower frequencies (~7%) than the 630Δerm parent (~21%; Fig 2A), corroborating the CRISPRi knockdown results.These sporulation phenotypes were also confirmed by phase contrast microscopy (Fig 2B), as fewer phase bright spores are visible in the rgaS and rgaR mutants.We also created a double rgaS rgaR mutant and found that the double mutant exhibited similar sporulation frequencies as the single rgaS and rgaR mutants (Fig 2C), suggesting that RgaS and RgaR function in the same regulatory pathway to affect sporulation.

Fig 3 .
Fig 3. RgaR-dependent gene expression is similarly decreased in both the rgaS and rgaR mutants.(A) Alignment of the putative RgaR binding sites with the promoters of confirmed and predicted direct RgaR targets.The direct repeats are in bold.(B) qRT-PCR analysis of RgaR-dependent transcripts at H 12 in 630Δerm, 630Δerm ΔrgaS (MC2228), and 630Δerm ΔrgaR (MC2229) grown on 70:30 agar.The means and standard error of the means of at least three independent biological replicates are shown.*, P < 0.0001 by a one-way ANOVA followed by Dunnett's multiple comparisons test.https://doi.org/10.1371/journal.pgen.1010841.g003

Fig 4 .
Fig 4. CRISPRi knockdown of RgaR-dependent genes reveal that RgaR impacts sporulation through the regulatory sRNA, SpoZ (encoded by CD630_n00620).(A) Ethanol-resistant spore formation of 630Δerm expressing psgRNA-neg (MC2065) or CRISPRi targets for CD0587 (MC2269), CD2098 (MC2270), CD15111 (MC2271), or spoZ (MC2272) grown on 70:30 agar supplemented with 2 μg/ml thiamphenicol and 1 μg/m nisin at H 24 .The means and standard errors of the means for four biological replicates are shown.(B) Alkaline phosphatase activity of either the wildtype PspoZ::phoZ or a site-directed mutagenized PspoZ G-51A/G-72A ::phoZ reporter fusion expressed from a plasmid in 630Δerm (MC2329 and MC2357, respectively) or 630Δerm ΔrgaR (MC2330 and MC2358, respectively).The promoterless phoZ reporter carried by 630Δerm (MC448) was included as a negative control.The means and standard deviations for at least three biological replicates are shown.*, P < 0.01 by one-way ANOVA followed by Dunnett's multiple comparisons test (A) or Tukey's multiple comparisons test (B).https://doi.org/10.1371/journal.pgen.1010841.g004 Figs 5B; S7A and S7B) and assessed their sporulation phenotypes.Like the spoZ knockdown strain, deletion of spoZ-CD16671 resulted in increased sporulation frequency (~2.7-fold) compared to the parent strain (Fig 5D).However, deletion of CD16671 did not impact sporulation frequency (Fig 5D).Because the decreased sporulation phenotype of the rgaS and rgaR mutants coincided with increased CD16671 transcript levels, it is possible that the absence of CD16671 does not affect C. difficile spore formation but rather the accumulation of CD16671 inhibits sporulation.To explore this possibility, we overexpressed CD16671 from an inducible promoter in the 630Δerm background and found that sporulation frequency was reduced (~1.8-fold;Fig 5E).CD16671 transcript levels were increased ~10-fold over wild-type levels (S7C Fig), suggesting that accumulation of CD16671 prevents the formation of spores.We also measured transcript levels of sigE and sigG, encoding sporulation-specific sigma factors, and found sigE and sigG transcript levels were also decreased ~1.6-fold when CD16671 was overexpressed (Fig 5F), Altogether, this data indicates that increased CD16671 abundance inhibits C. difficile sporulation-specific gene expression and spore formation.

Table 2 . CD16671 transcript levels correlate with the C. difficile sporulation phenotype in various spoZ mutants. Genotype spoZ transcript levels a CD16671 transcript levels b Sporulation frequency a
Values are based on qRT-PCR transcript levels or ethanol-resistant sporulation assays compared to the parent strain.WT, wild-type; ND, not detected. a