c-di-AMP signaling is required for bile salt resistance, osmotolerance, and long-term host colonization by Clostridioides difficile

Abstract

To colonize the host and cause disease, the human enteropathogen Clostridioides difficile must sense, respond, and adapt to the harsh environment of the gastrointestinal tract. We showed that the production and degradation of cyclic diadenosine monophosphate (c-di-AMP) were necessary during different phases of C. difficile growth, environmental adaptation, and infection. The production of this nucleotide second messenger was essential for growth because it controlled the uptake of potassium and also contributed to biofilm formation and cell wall homeostasis, whereas its degradation was required for osmotolerance and resistance to detergents and bile salts. The c-di-AMP binding transcription factor BusR repressed the expression of genes encoding the compatible solute transporter BusAA-AB. Compared with the parental strain, a mutant lacking BusR was more resistant to hyperosmotic and bile salt stresses, whereas a mutant lacking BusAA was more susceptible. A short exposure of C. difficile cells to bile salts decreased intracellular c-di-AMP concentrations, suggesting that changes in membrane properties induce alterations in the intracellular c-di-AMP concentration. A C. difficile strain that could not degrade c-di-AMP failed to persist in a mouse gut colonization model as long as the wild-type strain did. Thus, the production and degradation of c-di-AMP in C. difficile have pleiotropic effects, including the control of osmolyte uptake to confer osmotolerance and bile salt resistance, and its degradation is important for host colonization.

A c-di-AMP gut check for C. difficile

Clostridioides difficile causes antibiotic-associated colitis. To colonize the gut, it must adapt to the high osmolarity of the gut lumen and resist the antibiotic actions of bile salts. Oberkampf et al. report that C. difficile produced and degraded the intracellular second messenger c-di-AMP in response to different environmental conditions. Production of c-di-AMP promoted K+ and cell wall homeostasis and biofilm formation but sensitized the cells to osmotic stress. Bile salts induced c-di-AMP degradation, which was required for the derepression of genes encoding a solute import system that protected C. difficile from hyperosmotic stress and bile salts. Mutant bacteria lacking the enzyme that degrades c-di-AMP did not colonize the guts of antibiotic-treated mice. These findings establish c-di-AMP as an important signaling molecule in C. difficile, with a role in host colonization (see Focus by Purcell).

INTRODUCTION

Clostridioides difficile is an important human enteropathogen that has become an increasing public health concern (1, 2). This strictly anaerobic, spore-forming Gram-positive bacterium is a major cause of antibiotic-associated nosocomial diarrhea in adults (3). Most virulent C. difficile strains produce two glucosylating toxins (TcdA and TcdB) that play a key role in disease pathogenesis by targeting the gut epithelium, resulting in severe inflammation and damage to the colon (4, 5). Transmission of C. difficile depends on the production of highly resistant spores, which germinate in the small intestine in response to primary bile salts (6, 7). A normal intestinal microbiota mediates colonization resistance against C. difficile, but antibiotic treatment disrupts this balance, resulting in C. difficile growth, colonization of the intestine, and toxin production (8, 9).

During the course of infection in the host gastrointestinal tract, C. difficile encounters multiple stresses, including the presence of various antimicrobial compounds and reactive oxygen species that are produced during inflammation and the host immune response to infection (10–12). C. difficile vegetative cells are also exposed to primary and secondary bile salts. Primary bile salts produced by the human liver consist mainly of cholate and chenodeoxycholate conjugated with either taurine or glycine. Secondary bile salts, including deoxycholate and lithocholate in humans, are derived from primary bile salts by modifications carried out by intestinal bacteria (13). Although the primary bile salt taurocholate induces spore germination, the secondary bile salt deoxycholate is a poor germinant and inhibits vegetative cell growth (14). The high osmolarity in the intestinal lumen [equivalent to 300 mM sodium chloride (NaCl)] constitutes another important stress (15).

Bacteria respond to osmotic stresses by adjusting their intracellular concentrations of osmolytes to limit transmembrane water fluxes and maintain turgor. The emergency response against an hyperosmotic shock is the uptake of potassium ions (K⁺), followed by the synthesis and/or import of compatible solutes, such as carnitine and glycine betaine, that act as osmoprotectants (16). These generally neutral compounds are preferred osmolytes because they can accumulate to very high concentrations without inducing severe disturbances in cellular metabolism (17). Several osmolyte transport systems have been identified in Gram-positive bacteria, and many of these transporters are controlled by the second messenger cyclic diadenosine monophosphate (c-di-AMP) (18). For example, increased concentrations of c-di-AMP inhibit the three K⁺ transport systems, KimA, KtrAB, and KtrCD, in Bacillus subtilis (19). c-di-AMP also negatively controls the activity of the carnitine transporter OpuC in Listeria monocytogenes and Staphylococcus aureus and binds to the transcriptional repressor BusR to control the expression of the glycine betaine transporter genes busAA-busAB in Lactococcus lactis and Streptococcus agalactiae (20, 21).

c-di-AMP is widely produced among Gram-positive bacteria, including prominent human pathogens. It is synthesized from two molecules of adenosine triphosphate (ATP) by diadenylate cyclase (DAC) enzymes and degraded to the linear dinucleotide phosphadenylyl adenosine or two molecules of adenosine monophosphate (AMP) by distinct c-di-AMP phosphodiesterase (PDE) enzymes. Many important human pathogens have only a single DAC domain–containing protein called CdaA for c-di-AMP production. However, spore-forming clostridia and bacilli contain one (DisA) or two (DisA and CdaS) additional DACs, respectively (22). DisA plays a role in the control of DNA integrity, and CdaS is specifically involved in sporulation-related processes (23–26). Two other DACs (CdaM and CdaZ) have been identified in only few organisms (27, 28). Four different classes of PDEs degrade c-di-AMP, but most of the reported PDEs belong to the membrane-bound GdpP protein family, which consists of a signal regulatory module linked to a GGDEF domain and a DHH-DHHA1 catalytic domain (29–32).

c-di-AMP is essential for growth under standard laboratory conditions in most of the Firmicutes (18, 20, 33–36). However, c-di-AMP becomes dispensable when the bacteria are cultivated on defined minimal media (19, 20, 35). Furthermore, intracellular accumulation of c-di-AMP is toxic and inhibits growth (37). As a second messenger, c-di-AMP initiates signal transduction by binding to receptors that regulate downstream cellular processes. c-di-AMP receptors include enzymes such as the pyruvate carboxylase in L. monocytogenes and L. lactis (38, 39), osmolyte transporters (28, 40–46), transcriptional regulators (20, 21, 47), and the KdpD sensor kinase of the KdpDE potassium-responsive two-component regulatory system (48, 49). In addition, c-di-AMP–responding riboswitches controlling the expression of genes encoding K⁺ uptake systems in B. subtilis and Bacillus thuringiensis have also been identified (19, 50, 51). Besides its main and conserved role in osmoregulation, c-di-AMP is also implicated in a wide range of cellular processes, including central metabolism, cell wall homeostasis, biofilm formation, and virulence (23, 30, 33, 36, 52–57).

Understanding how C. difficile interacts with its host requires the elucidation of the survival mechanisms necessary for colonization and pathogenesis. In this study, we explored c-di-AMP signaling in C. difficile and identified osmolyte homeostasis as a critical function of this second messenger. We showed that c-di-AMP had pleiotropic effects in C. difficile, including mediating the tolerance to osmotic and detergent stresses as well as the tolerance to physiologically relevant bile salts. Exposure to bile salts caused a rapid decrease in intracellular c-di-AMP concentrations, resulting in the inhibition of the activity of the c-di-AMP–regulated transcriptional repressor BusR, which connects osmotic and bile salt tolerance. The main targets of BusR were busAA and busAB, which encode a functional compatible solute uptake system, BusAA-AB, that was also required for osmotolerance. Together, these data reveal that c-di-AMP is a key regulatory molecule in C. difficile that modulates osmolyte uptake in response to the perception of osmotic and bile salt stresses to regulate adaptation to the host intestinal environment during infection.

RESULTS

C. difficile produces the second messenger c-di-AMP

A bioinformatics search identified two genes encoding DAC enzymes homologous to DisA and CdaA (CD0028 and CD0110, respectively) and one gene encoding a PDE homologous to GdpP (CD3659) in the reference genome of C. difficile strain 630. The presence of such enzymes strongly suggested that C. difficile might produce c-di-AMP. To test this hypothesis, we extracted nucleotides from C. difficile 630Δerm, a commonly used spontaneous erythromycin-sensitive derivative of the strain 630, grown to exponential phase, and assessed the presence of intracellular c-di-AMP by liquid chromatography–tandem mass spectrometry (LC-MS/MS) (fig. S1). We detected about 4.7 pmol of c-di-AMP per milligram of protein in the wild-type strain (Fig. 1A). In-frame deletion strains for each of the three genes encoding putative c-di-AMP turnover enzymes were readily generated by allelic exchange in C. difficile 630Δerm (fig. S2, A and B). In contrast, attempts to create a ΔdisAΔcdaA double-mutant strain in standard culture conditions were unsuccessful, suggesting that c-di-AMP might be essential for C. difficile growth in rich medium. To confirm the role of the predicted enzymes, we determined the intracellular c-di-AMP concentration of the different gene deletion strains by LC-MS (Fig. 1A and fig. S1). As expected, deletion of gdpP significantly increased the intracellular concentrations of c-di-AMP. In contrast, the c-di-AMP concentrations were not significantly changed in a disA mutant or a cdaA mutant.

Fig. 1. c-di-AMP concentrations in strains lacking c-di-AMP turnover enzymes and their associated changes in phenotypes.

(A) Intracellular c-di-AMP concentrations in C. difficile 630Δerm wild-type (WT), ΔdisA, ΔcdaA, and ΔgdpP strains grown in TY medium were quantified by LC-MS/MS. Means and SEM are shown; n = 3 independent experiments. **P ≤ 0.01 by a one-way analysis of variance (ANOVA) followed by a Dunnett’s multiple comparisons test comparing values to the average WT value. (B) Growth curves of C. difficile WT and mutant strains in TY medium. Means and SEM are shown; n = 4 independent experiments. (C) Susceptibility of C. difficile WT and mutant strains to three β-lactam antibiotics in disk diffusion assays. The zone of inhibition is expressed as the total diameter of the clearance zone and includes the 7-mm filter paper disk. Means and SEM are shown; n = 3 independent experiments. ****P ≤ 0.0001 by a two-way ANOVA followed by a Dunnett’s multiple comparisons test comparing values with the average WT value. (D) Biofilm formation by C. difficile WT and mutant strains grown in BHISG medium for 24 hours. Means and SEM are shown; n = 4 independent experiments. ****P ≤ 0.0001 by a one-way ANOVA followed by a Dunnett’s multiple comparisons test comparing values with the average WT value. (E) Growth of C. difficile WT and mutant strains on TY agar or TY agar + 300 mM NaCl after 24-hour incubation at 37°C. Spots are from cells grown overnight in TY and serially diluted. Data are representative of three independent experiments.

Fluctuations in c-di-AMP concentrations have pleiotropic effects in C. difficile

We investigated the effects of mutations in disA, cdaA, or gdpP on several phenotypes that have been associated with fluctuations of c-di-AMP concentrations in other bacteria. We first explored the effect of disA, cdaA, or gdpP deletion on C. difficile growth. Whereas growth of the disA and cdaA mutants in rich media [tryptone yeast extract (TY) or brain-heart infusion supplemented with glucose (BHISG)] was similar to that of the wild-type strain, gdpP deletion resulted in slower growth (Fig. 1B and fig. S3). This observation is in agreement with the toxicity of intracellular accumulation of c-di-AMP reported in other bacteria (37). To determine whether c-di-AMP affects cell wall homeostasis in C. difficile, we measured susceptibility to cell wall–targeting β-lactam antibacterial drugs using antibiotic disk diffusion assays. The ΔcdaA strain was more susceptible than the wild type to all tested antibacterial drugs, whereas the ΔdisA and ΔgdpP strains had the same susceptibility as the wild type (Fig. 1C). These data suggest a role for CdaA in maintaining cell wall homeostasis. We also assessed the ability of the mutants to form biofilms. Under our experimental conditions, only a faint biofilm was observed in the wild-type, ΔcdaA, or ΔdisA strains (Fig. 1D). In contrast, a strong biofilm was obtained with the ΔgdpP strain, revealing a link between c-di-AMP and biofilm formation in C. difficile.

We next tested the tolerance of the wild-type and the mutant strains to high osmotic stress. The ΔgdpP mutant was highly susceptible to 300 mM NaCl when compared with the wild type, whereas the ΔdisA and ΔcdaA strains behaved like the wild type (Fig. 1E). Accordingly, expression of cdaA from a plasmid under control of the inducible P_tet promoter resulted in an increased susceptibility of the wild type to NaCl compared with a control strain carrying an empty vector (fig. S4). In contrast, overexpression of gdpP on a plasmid had no effect on osmotolerance. Together, these data demonstrate that c-di-AMP promotes cell wall homeostasis and biofilm formation in C. difficile and sensitizes the cells to osmotic stress.

Regulation of K⁺ uptake is the essential function of c-di-AMP in C. difficile

Because the control of K⁺ homeostasis is an essential function of c-di-AMP in B. subtilis (19), we attempted to generate a ΔdisAΔcdaA double mutant by using our gene deletion protocol using a modified C. difficile minimal medium (CDMM) containing 0.1 μM K⁺ instead of brain-heart infusion (BHI). The ΔdisAΔcdaA mutant was successfully generated (fig. S2A) and viable under these conditions. The growth of this mutant was similar to that of the wild type in CDMM containing 1 mM K⁺ but was greatly reduced in the presence of 5 mM K⁺ and abolished in the presence of 25 mM K⁺ (Fig. 2A). Examination of the ΔdisAΔcdaA cells by phase contrast microscopy revealed a strong effect of the K⁺ concentration on cell morphology characterized by a pronounced elongation and curvature of the cells when K⁺ concentration increased (Fig. 2, B and C). No c-di-AMP was detected by LC-MS/MS in nucleotide extracts from this strain, indicating that DisA and CdaA are the only two enzymes involved in the production of c-di-AMP in C. difficile. In contrast to the ΔdisAΔcdaA strain, the wild-type and the ΔgdpP strains grew on CDMM plates regardless of the K⁺ concentration (Fig. 2A). Cells of the wild-type strain became shorter as K⁺ concentration decreased, and ΔgdpP cells were consistently shorter than those of the wild type under all conditions (Fig. 2, B and C). In addition, we observed abnormalities in the ΔgdpP cell shape, such as bent cells and cells with cytokinesis defects (fig. S5). These data thus underline an essential function of c-di-AMP in controlling K⁺ homeostasis in C. difficile.

Fig. 2. Control of K⁺ uptake is the essential function of c-di-AMP in C. difficile.

(A) Growth of C. difficile 630Δerm WT, ΔgdpP, and ΔdisAΔcdaA strains on CDMM agar plates containing 0.1 μM, 1 mM, 5 mM, or 25 mM KCl after 48 hours at 37°C. Data are representative of three independent experiments. (B) Phase contrast microscopy images of C. difficile WT, ΔgdpP, and ΔdisAΔcdaA strains grown for 48 hours at 37°C on CDMM agar plates containing 0.1 μM, 1 mM, or 5 mM KCl. Scale bar, 10 μm. Data are representative of three independent experiments. (C) Scatterplots showing cell length of C. difficile WT, ΔgdpP, and ΔdisAΔcdaA, with the median and SD of each distribution indicated by black lines; n = 100 cells per genotype from three independent experiments. ****P ≤ 0.0001 by a two-way ANOVA followed by a Dunnett’s or Tukey’s multiple comparisons test. ND, not determined. (D) Autoradiography of nitrocellulose membranes from DRaCALAs testing the interaction of KtrA and KdpD with [³²P]-c-di-AMP alone or in the presence of the indicated unlabeled nucleotides. Binding of the radiolabeled ligand (1 nM) to the protein is indicated by central dark spots surrounded by a ring of signal representing diffusion of the unbound ligand. In competition assays, an excess of unlabeled nucleotides (150 μM) was added to the reaction before spotting on membrane. n = 2 independent whole E. coli protein extracts.

c-di-AMP binds to proteins likely involved in K⁺ uptake

In silico analyses revealed that two potential K⁺ transport systems, KtrAB (CD0696-CD0697) and KdpABC (CD1591-CD1593), the expression of the latter of which is likely regulated by the two-component regulatory system KdpDE (CD1829-CD1830), are encoded by the genome of C. difficile 630. In S. aureus, c-di-AMP inhibits both the expression of kdpABC and the activity of KtrAB by binding to the universal stress protein domain of the histidine kinase KdpD and to the RCK_C domain of KtrA, respectively (42, 48). These domains are conserved in the C. difficile orthologs. To assess the interaction between c-di-AMP and the proteins KdpD and KtrA of C. difficile, we performed a differential radial capillary action of ligand assay (DRaCALA). The corresponding encoding genes were expressed as recombinant proteins in Escherichia coli, and whole-cell extracts were incubated with [³²P]-labeled c-di-AMP. Radiolabeled c-di-AMP interacted with both proteins, indicating that they are c-di-AMP targets (Fig. 2D). In addition, an excess of unlabeled c-di-AMP, but not of other unlabeled nucleotides tested, outcompeted [³²P]-c-di-AMP for binding with KdpD or KtrA, demonstrating the binding specificity of c-di-AMP to these proteins. Thus, these results strongly suggest that c-di-AMP directly controls K⁺ uptake through the modulation of the activity or the expression of K⁺ transport systems in C. difficile.

c-di-AMP relieves BusR-mediated repression of genes encoding a putative osmolyte transporter in C. difficile

In response to hyperosmotic stress, the import of K⁺ into bacterial cells is followed by a secondary response involving the synthesis or the uptake of compatible solutes such as glycine betaine or carnitine (16). C. difficile has an ortholog of the c-di-AMP binding protein BusR, a transcriptional repressor of the glycine betaine uptake system BusAB in L. lactis and S. agalactiae (20, 21). Using DRaCALA, we demonstrated the interaction between c-di-AMP and the BusR protein of C. difficile (Fig. 3A). We then deleted the busR gene in the ΔgdpP strain (fig. S2B) and used RNA sequencing (RNA-seq) to compare the transcriptomes of the ΔgdpP and the ΔgdpP ΔbusR mutant strains. The ΔgdpP background was used in this analysis because BusR has been shown to exert its repressing activity in the presence of high c-di-AMP concentrations in L. lactis (58). Using a fold change cutoff of >twofold and a P value limit of 0.05, we identified 47 genes whose expression was dependent on BusR, with 15 positively and 32 negatively regulated genes (Table 1). Of these genes, the most highly repressed by BusR were two genes, CD0900 and CD0901, comprising an apparent operon (59). These genes encode a putative compatible solute ATP-binding cassette (ABC) transporter system composed of an ATP binding protein (CD0900, OpuCA) and a permease (CD0901, OpuCC) presenting a limited homology with BusAA (35.1% identity, 55.1% similarity) and BusAB (21.7% identity, 38.1% similarity) of L. lactis, respectively. For consistency, these C. difficile genes will be hereafter named busAA and busAB. The data of the RNA-seq analyses for busAA expression were confirmed by quantitative reverse transcription polymerase chain reaction (PCR) (fig. S6), supporting that the c-di-AMP binding protein BusR is a repressor of busAA-AB in C. difficile.

Fig. 3. BusR is a c-di-AMP target that controls osmotic homeostasis by repressing busAA-AB expression.

(A) Autoradiography of nitrocellulose membranes from DRaCALAs testing the interaction between BusR and [³²P]-c-di-AMP alone and in the presence of unlabeled nucleotides. Binding of the radiolabeled ligand (1 nM) is indicated by dark spots centered on the nitrocellulose. In competition assays, excess of unlabeled nucleotides (150 μM) was added to the reaction before spotting on membrane. n = 2 independent whole E. coli protein extracts. (B and C) Growth of C. difficile 630Δerm WT and mutant strains on TY agar, TY agar + 300 mM NaCl, or TY agar + 500 mM NaCl after 24-hour incubation at 37°C. Spots are from cells grown overnight in TY and serially diluted. Data are representative of three independent experiments. (D) Growth of C. difficile WT strains carrying empty vector (p) or p-busAA-AB on TY agar + thiamphenicol (Tm), or TY agar + Tm + 500 mM NaCl after 48-hour incubation at 37°C. Spots are from cells grown overnight in TY and serially diluted. Data are representative of three independent experiments.

Table 1.

Genes differentially regulated in the ΔgdpPΔbusR strain in comparison with the ΔgdpP strain.

Gene name*	Function†	Fold change‡	q value§
BusR-repressed genes
CD0900–0901 (busAA-AB)	ABC transporter	13.32 to 15.54	1.31 × 10−128
CD2348/CD2351/CD2354–2357	Glycine reductase pathway	2.02 to 3.38	0.035
CD0617	CAAX proteases and bacteriocin-processing enzymes (CPBP) family intramembrane metalloprotease	3.23	1.07 × 10−11
CD0519	Hypothetical protein	3.21	8.42 × 10−08
CD0832–0833 (aksA-acnB)	Aconitate metabolism	2.30 to 2.67	2.34 × 10−03
CD0209	Tagatose-6-phosphate kinase	2.45	6.78 × 10−08
CD0724–0728	Acetyl-CoA decarbonylase/synthase complex	2.21 to 2.39	1.31 × 10−04
CD0616	MerR family transcriptional regulator	2.35	2.67 × 10−07
CD0207	Phosphotransferase system (PTS), fructose-like IIC component	2.16	1.00 × 10−03
CD2883 (celB)	PTS system, cellobiose-specific IIC component	2.16	1.01 × 10−03
CD0834 (icd)	Isocitrate dehydrogenase	2.15	8.14 × 10−03
CD0723 (lpdA)	Dihydrolipoyl dehydrogenase	2.13	5.02 × 10−05
CD0722 (metF)	5,10-Methylenetetrahydrofolate reductase	2.11	5.32 × 10−05
CD0203 (uvrA)	Excinuclease ABC subunit	2.07	3.38 × 10−04
CD0528	Aminohydrolase	2.06	3.36 × 10−03
CD3089	PTS transporter subunit enzyme IIC (EIIC)	2.03	3.21 × 10−06
CD1768	Hypothetical protein	2.03	1.64 × 10−04
CD2401 (cotD)	Manganese catalase family protein	2.03	4.28 × 10−03
CD0719 (fchA)	Cyclodeaminase/cyclohydrolase family protein	2.02	9.10 × 10−05
BusR-induced genes
CD0324–0327 (cbi)	Cobalamin/cobalt synthesis and transport	−4.76 to −4.17	1.57 × 10−11
CD1797	Flavin adenine dinucleotide (FAD)–dependent oxidoreductase	−4	4.73 × 10−09
CD1796	4Fe-4S binding domain protein	−3.57	9.49 × 10−07
CD1595 (cysE)	Serine O-acetyltransferase	−2.22	4.41 × 10−04
CD2004 (effR)	MarR family transcriptional regulator	−2.13	2.92 × 10−03
CD2997–2999	ABC-type transport system, iron family	−2.04 to −2.08	2.55 × 10−04
CD2479	tRNA threonylcarbamoyladenosine dehydratase	−2.08	5.49 × 10−04
CD2993	Hypothetical protein	−2.04	1.87 × 10−04
CD3370	Recombinase family protein	−2.04	6.31 × 10−03
CD0581	TetR/AcrR family transcriptional regulator	−2	1.73 × 10−02

^*From GenBank.

^†Putative functions as determined by current annotation of the C. difficile 630 genome.

^‡The fold changes listed are averages from four biological replicates. Fold changes signify the differential expression in C. difficile ΔgdpPΔopuR compared with the ΔgdpP strain. For genes in a putative operon, the range of fold change is reported.

^§The q value is an adjusted P value, taking into account the false discovery rate. The q values listed are the averages from four biological replicates. For genes in a putative operon, the highest q value is reported.

BusR and BusAA-AB are required for osmotic homeostasis in C. difficile

To evaluate the role played by BusR and BusAA-AB in osmotolerance, we generated single and double deletion strains for busAA and busR (ΔbusAA, ΔbusR, and ΔbusAAΔbusR) (fig. S2C) and analyzed the ability of these mutants to grow in the presence of NaCl. Whereas ΔbusAA was more susceptible to hyperosmotic stress than the wild type but less so than ΔgdpP, deletion of busR resulted in a strong increase in resistance to 500 mM NaCl (Fig. 3B). In addition, the ΔbusRΔbusAA double mutant was as susceptible to NaCl as the busAA single-mutant strain, and deletion of busR in the ΔgdpP background partially rescued osmoresistance (Fig. 3, B and C). This indicates that the osmoresistance phenotype of ΔbusR is fully mediated by the derepression of busAA-AB expression but that the osmosusceptibility of ΔgdpP is not caused only by the inhibition of busAA-AB expression. To confirm that higher amounts of BusAA-AB were sufficient to increase osmoresistance, we constructed the vector p-busAA-AB, carrying the busAA-AB genes under the control of the strong and constitutive P_cwp2 promoter (60), and introduced it into the wild-type strain. As expected, C. difficile overexpressing busAA-AB was more resistant to an osmotic stress compared with the vector control strain (Fig. 3D). Together, these data suggest that c-di-AMP binding to BusR promotes BusR-mediated repression of busAA-AB, indicating that high c-di-AMP concentrations inhibit adaptation to osmotic stress. Therefore, cells are most able to respond to osmotic stress by taking up osmolytes when c-di-AMP concentrations are low. In addition, our data reveal that c-di-AMP also inhibits responses to osmotic stress through other mechanisms because ΔgdpP, which has high c-di-AMP concentrations, is more susceptible to osmotic stress than ΔbusRΔbusAA.

BusAA-AB is an osmolyte transporter repressed by c-di-AMP

To investigate the function of BusAA-AB, we tested the growth of our deletion mutant panel on CDMM agar plates containing 6.5 mM K⁺ and supplemented with 200 mM NaCl. Under these conditions, the wild-type, ΔbusR, ΔbusAA, and ΔbusRΔbusAA strains were equally susceptible to hyperosmotic stress (Fig. 4A). This suggested that the osmoprotective compound transported by BusAA-AB when cells were grown in TY (Fig. 3B) was not present in CDMM. In contrast, the ΔgdpP and ΔgdpPΔbusR strains were highly susceptible when compared with the wild type, consistent with the presence of K⁺ in the medium. Addition of 0.4 mM glycine betaine or carnitine in the culture medium improved the growth of the wild type and the busR mutant in the presence of NaCl but had no effect on the growth of ΔbusAA and ΔbusRΔbusAA (Fig. 4A). Osmotolerance of ΔgdpPΔbusR was also increased when compared with ΔgdpP. These results indicate that the phenotypes of ΔbusR and ΔbusAA in response to osmotic stress result from a dysregulated compatible solute uptake and that the BusAA-AB transporter can take up both betaine and carnitine.

Fig. 4. BusAA-AB is an osmolyte transporter.

(A) Growth of C. difficile 630Δerm WT and mutant strains on CDMM agar, CDMM agar + 200 mM NaCl, CDMM agar + 200 mM NaCl + 0.4 mM glycine betaine, or CDMM agar + 200 mM NaCl +0.4 mM carnitine after 48-hour incubation at 37°C. Spots are from log-phase cells grown in TY, washed once, and serially diluted. Data are representative of three independent experiments. (B) C. difficile WT cells grown in CDMM were treated with water, 0.4 mM glycine betaine, 200 mM NaCl, or 200 mM NaCl + 0.4 mM glycine betaine for 30 min at 37°C. c-di-AMP concentrations were then measured using a competitive c-di-AMP ELISA. Means and SEM from three biological replicates were plotted as picomol c-di-AMP per milligram of protein. **P ≤ 0.01 and ***P ≤ 0.001 by a one-way ANOVA followed by a Tukey’s multiple comparisons test. (C) Intracellular c-di-AMP concentrations in C. difficile WT, ΔbusR, and ΔbusAA strains grown in TY medium were quantified by LC-MS/MS. Means and SEM are shown; n = 3 independent experiments. ***P ≤ 0.001 by a one-way ANOVA followed by a Dunnett’s multiple comparisons test comparing values with the average WT value.

To further assess the effect of a hyperosmotic stress on the intracellular c-di-AMP concentration, we exposed the wild-type strain grown in liquid CDMM to 200 mM NaCl for 30 min. Incubation with an equivalent volume of water was used as a control. Using a competitive enzyme-linked immunosorbent assay (ELISA) assay, we found that c-di-AMP concentrations were significantly lower in bacteria challenged with NaCl compared with the control condition (Fig. 4B). Cell exposure to 0.4 mM glycine betaine did not affect c-di-AMP concentration, and incubation with both NaCl and glycine betaine led to the same effect as with NaCl alone. This suggests that the defense of the bacterium against an osmotic stress in liquid medium requires continuous compatible solute uptake in our experimental conditions.

Next, we sought to determine whether changes in compatible solute uptake caused by loss of BusR or BusAA affected cellular c-di-AMP concentrations (Fig. 4C). Measurement of intracellular c-di-AMP concentrations by LC-MS/MS (fig. S1) revealed an increased amount of c-di-AMP in ΔbusR grown in TY compared with the wild-type strain. In contrast, c-di-AMP concentrations were not significantly changed in ΔbusAA. These results suggest that C. difficile cells adjust their c-di-AMP concentrations in response to fluctuations of the intracellular glycine betaine or carnitine concentrations in unstressed conditions. This in turn modulates BusR-mediated repression of the BusAA-AB osmolyte transporter genes in the wild-type strain and maintains compatible solute homeostasis.

BusR mediates resistance to the detergent activity of bile salts

Several osmolyte uptake systems are involved in bile tolerance in L. monocytogenes (61). These data prompted us to investigate the role of BusR and BusAA in bile salt tolerance in C. difficile. We first analyzed the phenotypes of the ΔbusR, ΔbusAA, and ΔbusRRΔbusAA strains grown in TY in response to a commercial bile salt extract. The ΔbusR strain was highly resistant to this stress compared with the wild-type strain, whereas ΔbusAA and ΔbusRΔbusAA were more susceptible (Fig. 5A and fig. S7A). Because bile salt extract is a mixture of primary and secondary bile salts, we tested the growth of the mutant strains in the presence of the primary bile salt cholate or the secondary bile acid deoxycholate. Results were similar for both bile salts and in line with those obtained with the bile salt extract (Fig. 5A). Because bile salts have detergent properties, we tested the tolerance of the mutant strains to the ionic detergent SDS and the nonionic detergent Triton X-100 (Fig. 5B). Again, ΔbusR was found to be highly resistant to both compounds, whereas ΔbusAA and ΔbusRΔbusAA were more sensitive than the wild-type strain.

Fig. 5. c-di-AMP promotes resistance to the detergent action of bile salts by controlling the BusR-dependent, BusAA-AB–mediated import of osmolytes.

(A) Growth of C. difficile 630Δerm WT and mutant strains on TY agar, TY agar + 1% bile salts, TY agar + 0.03% deoxycholate (DOC), or TY agar + 0.4% cholate after 48-hour incubation at 37°C. Spots are from cells grown overnight in TY and serially diluted. Data are representative of three independent experiments. (B) Growth of C. difficile WT and mutant strains on TY agar, TY agar + 0.01% Triton X-100, or TY agar + 0.003% SDS after 48-hour incubation at 37°C. Spots are from cells grown overnight in TY and serially diluted. Data are representative of three independent experiments. (C) Growth of C. difficile WT and mutant strains on CDMM agar, CDMM agar + 0.009% Triton X-100, or CDMM agar + 0.009% Triton X-100 + 0.4 mM glycine betaine after 48-hour incubation at 37°C. Spots are from log-phase cells grown in TY, washed once in 0.9% saline, and serially diluted. Data are representative of three independent experiments.

We sought to determine whether the import of compatible solutes into the cells by BusAA-AB directly drove detergent tolerance. Wild-type, ΔbusR, ΔbusAA, and ΔbusRΔbusAA mutants were grown on CDMM agar plates supplemented with 0.008 to 0.01% Triton X-100 in the absence or presence of 0.4 mM glycine betaine (Fig. 5C and fig. S7B). The growth defect caused by the detergent was in the same range for all strains in the absence of compatible solutes. Addition of glycine betaine in the medium partially restored the growth of the wild type and, to a greater extent, that of ΔbusR. Strains ΔbusAA and ΔbusRΔbusAA remained more susceptible to Triton X-100 than the wild type, but their growth was also slightly improved by glycine betaine, suggesting the involvement of another betaine transporter. Together, these results show that BusR plays an important role in the resistance to the detergent activity of bile salts by controlling the transport of compatible solutes by BusAA-AB into C. difficile cells.

Intracellular c-di-AMP concentrations modulate bile salt resistance and are reduced by bile salt exposure

To explore the potential functional link between c-di-AMP and bile salt resistance, we evaluated the phenotypes of the ΔgdpP and ΔgdpPΔbusR strains in response to mixed bile salt extract, cholate, or deoxycholate (Fig. 5A). For each tested bile salt, ΔgdpP showed the same susceptibility phenotype as the ΔbusAA strain, and the deletion of busR in the ΔgdpP mutant abolished the susceptibility. These data indicate that c-di-AMP modulates bile salt resistance primarily by promoting the repressive activity of BusR on busAA-AB expression.

We examined the effect of bile salts on c-di-AMP concentrations in the wild-type strain. Cells grown to the exponential phase were washed and resuspended in a low osmolarity buffer with glucose. After a 10-min incubation, water, NaCl, deoxycholate, or cholate was added, and cells were incubated for another 10 min before c-di-AMP was quantified using a competitive ELISA assay (Fig. 6). In our conditions, exposure to NaCl resulted in a slight decrease in the c-di-AMP concentration in comparison with the water control. In contrast, a rapid and statistically significant decrease in c-di-AMP concentration was observed in the samples treated with deoxycholate and cholate relative to the water control. This demonstrates that bile salt exposure promotes c-di-AMP degradation in C. difficile, leading to higher busAA-AB expression and, in turn, conferring bile salt resistance.

Fig. 6. A bile salt treatment decreases c-di-AMP concentrations in C. difficile.

C. difficile 630Δerm cells suspended and incubated for 10 min in a low-osmolarity buffer with glucose (t₀) were treated with water, 300 mM NaCl, 0.028% deoxycholate, or 0.4% cholate for 10 min at 37°C (t₁₀). c-di-AMP concentrations were measured at t₀ and t₁₀ using a competitive c-di-AMP ELISA. Means and SEM from three biological replicates were plotted as picomol c-di-AMP per milligram of protein. ***P ≤ 0.001 by a one-way ANOVA followed by a Dunnett’s multiple comparisons test comparing values with the average of the water-treated samples.

GdpP, but not BusR or BusAA, is required for C. difficile persistence in the murine gut

Given the importance of GdpP, BusR, and BusAA for resistance to osmotic and bile salt stresses, we hypothesized that they would affect initial colonization of the host, persistence in the gut, or both. We therefore examined the ability of the corresponding mutants to colonize the gut in an antibiotic-treated mouse model of C. difficile infection. Mice were pretreated with clindamycin and subsequently infected orally with 5 × 10⁴ spores of either the wild-type, ΔgdpP, ΔbusR, or ΔbusAA strain. The intestinal burden of C. difficile was monitored by collecting feces from the mice over a 17-day period and enumerating colonies on selective medium (Fig. 7). The numbers of total colony-forming units (CFUs) recovered from the feces of mice infected with the wild type or any of the mutant strains were all similar at 2 days after inoculation, suggesting that none of these genes is required to establish colonization of the gastrointestinal tract. At day 6, mice infected with the wild-type strain showed a 1.5 log decrease in total CFUs compared with day 2 and then maintained this level of colonization throughout the remainder of the experiment. A similar pattern of colonization was observed for both the ΔbusR and ΔbusAA mutant strains. In contrast, mice infected with the ΔgdpP mutant showed a sharp decrease in CFU at days 6, 9, and 13 and completely cleared the bacteria at day 17. These data indicate that, under our experimental conditions, GdpP, unlike BusR and BusAA, is essential for the persistence of C. difficile in the intestinal environment. c-di-AMP therefore appears to be an important signaling molecule controlling the ability of C. difficile to colonize the host.

Fig. 7. GdpP, but not BusR or BusAA, is required for C. difficile to sustain infection of the murine gut.

Mice were inoculated with 5 × 10⁴ spores of C. difficile 630Δerm (WT), ΔbusR, ΔbusAA, or ΔgdpP. Feces were collected on the indicated days and plated on BHI + 3% defibrinated horse blood + 0.1% taurocholate + cefoxitin (25 μg/ml) + cycloserine (250 μg/ml) to assess the total number of CFUs (vegetative cells + spores). n = 6 mice inoculated for each C. difficile genotype. *P < 0.05, **P < 0.01, and ***P < 0.001 by a two-way ANOVA followed by a Dunnett’s multiple comparisons test.

DISCUSSION

Cyclic nucleotides are second messengers broadly used by bacteria to regulate diverse cellular processes in response to environmental stresses. Cyclic diguanylate monophosphate (c-di-GMP) has been found to regulate many important functions in C. difficile, including motility, adhesion, biofilm formation, and toxin expression (60, 62–68). On the other hand, C. difficile was previously shown to have an active DAC (69), but the production and roles of c-di-AMP had not been further investigated in this important enteropathogen. Here, we showed that deletion of the gene encoding the DAC, cdaA, resulted in increased susceptibility to peptidoglycan-targeting β-lactam antibacterial drugs. A direct connection between c-di-AMP synthesis and peptidoglycan biosynthesis has been established in L. lactis, B. subtilis, and S. aureus (70–73). In these bacteria, the glucosamine-6-phosphate mutase GlmM, an enzyme for the production of the essential peptidoglycan synthesis intermediate glucosamine-1-P, has been shown to form a complex with CdaA and to regulate CdaA activity (58). In numerous Firmicutes and in S. aureus, cdaA and glmM are transcriptionally coregulated (74). In C. difficile, glmM is separated from the cdaA-cdaR gene cluster by a seven-gene operon (CD0112 to CD0118), but further studies are needed to investigate potential links between cdaA-cdaR and glmM. We also found that increased c-di-AMP concentrations promoted biofilm formation in C. difficile as previously shown in several streptococci (43, 75, 76). In Streptococcus mutans, control of biofilm formation by c-di-AMP is mediated through the stimulation of the expression of gtfB, encoding an enzyme responsible for the production of water-insoluble glucans (76). In contrast, accumulation of c-di-AMP in B. subtilis inhibits biofilm formation by affecting the activity of SinR (56). Factors connecting c-di-AMP and biofilm formation remain to be determined in C. difficile.

Here, we demonstrated that the essential function of c-di-AMP in regulating osmotic homeostasis is conserved in C. difficile. We identified two putative K⁺ transport systems, Ktr and Kdp, in C. difficile and showed that proteins belonging to these systems, KtrA and KdpD, bind c-di-AMP in vitro. Whereas the involvement of these two proteins in K⁺ uptake remains to be experimentally validated, our data strongly suggest that they might be c-di-AMP binding effectors regulating K⁺ homeostasis. The mechanism also involves the binding of c-di-AMP to the transcriptional regulator BusR, a repressor of the busAA-AB operon encoding a compatible solute transporter. The involvement of this system in the response to hyperosmolarity is consistent with transcriptomic and proteomic analyses of C. difficile experiencing hyperosmotic shock (77, 78). Our findings that uncontrolled compatible solute influx through BusAA-AB (due to the deletion of busR) led to an increase in c-di-AMP concentration, whereas the loss of BusAA slightly decreased c-di-AMP concentration, are in agreement with those recently reported in L. lactis (21) and reinforce the hypothesis that sensing of fluctuations of the cellular turgor constitutes a signal to adjust the intracellular concentration of c-di-AMP.

An important and previously unreported finding from our study is that bile salt exposure rapidly stimulated c-di-AMP degradation, and we demonstrated that the BusAA-AB–mediated osmolyte uptake plays an important role in bile salt tolerance. The mechanism involved seems to be identical to the one conferring resistance to NaCl. Thus, cells lower their c-di-AMP concentrations in response to hyperosmotic or bile salt stress, which triggers a relief of busAA-AB repression by BusR and an increased uptake of protective compatible solutes (Fig. 8). Consistent with the role of BusAA-AB in bile salt resistance in C. difficile, expression of busAA was induced 12-fold in deoxycholate-induced biofilms compared with noninduced biofilms (79). In addition, the abundance of BusAA and BusAB proteins in C. difficile 630Δerm was increased after a 90-min exposure to either primary or secondary bile salts (80). In L. monocytogenes, the carnitine transporter OpuC is also involved in bile tolerance, and opuC transcription is induced in response to bile salts (61). Moreover, c-di-AMP was shown to bind the cystathionine beta-synthase domain of the OpuCA subunit in this bacterium, but no link between c-di-AMP and bile tolerance has been established to our knowledge (46).

Fig. 8. Cyclic di-AMP controls osmolyte uptake to confer protection against osmotic and bile salts stresses.

(A) In standard growth conditions, c-di-AMP binds to BusR to promote its repressive activity on busAA-AB, encoding a transporter of the osmoprotectants glycine betaine and carnitine. (B) When cells are exposed to hyperosmotic stress or the detergent activity of bile salts, intracellular c-di-AMP concentrations rapidly drop. Because both stresses alter membrane integrity, this alteration might be sensed by the membrane-bound phosphodiesterase GdpP, which degrades c-di-AMP. When c-di-AMP concentrations are low, BusR is not bound to c-di-AMP and does not exert its repressive activity on busAA-AB. This results in an increased uptake of the osmoprotectants by the BusAA-AB transporter to confer protection against osmotic and bile salt stresses.

The busAA-AB operon of C. difficile is positively controlled by the stress-responsive alternative sigma factor σ^B, and there is a σ^B-dependent promoter upstream of busAA (59, 81), but a strain lacking σ^B showed no growth defect compared with the wild type in the presence of NaCl or bile salts (81). Only a putative σ^A promoter was identified upstream busR of C. difficile (82), suggesting that the c-di-AMP signaling pathway can provide a specific response to osmotic and bile salt stresses independently of the activation of the σ^B-dependent general stress response in C. difficile. Expression of opuC genes is also regulated by σ^B in L. monocytogenes, with putative σB promoter motifs identified upstream of the opuC operon (83, 84). However, the osmotic induction of the opuC operon has been shown to be strongly σ^B dependent, and the tolerance to bile salts mediated by σ^B in this bacterium (85, 86).

Because of their amphipathic nature, bile salts are antibacterial compounds that act as natural detergents and disrupt bacterial membranes (87, 88). In Enterococcus faecalis and Propionibacterium freudenreichii, a pretreatment with a sublethal dose of bile salts or the detergent SDS conferred a similar protection against lethal concentrations of bile salts, demonstrating that the physiological responses to bile salts and SDS are closely related (89, 90). We showed in this study that BusR and BusAA play an important role in SDS and Triton X-100 tolerance, suggesting that intracellular c-di-AMP concentrations are decreased in response to the detergent activity of bile salts to confer protection. Supporting our data, c-di-AMP–specific PDE gene mutants with elevated c-di-AMP concentrations were more sensitive to Triton X-100 than their respective wild type in Bacillus anthracis and Streptococcus mitis (91, 92).

In L. monocytogenes, cell adaptation to either SDS or NaCl conferred a similar high cross-protection against lethal concentrations of bile salts (93). Likewise, a pretreatment of E. faecalis cells with subinhibitory concentrations of NaCl induced tolerance against lethal concentrations of SDS or bile salt challenges (94). Both osmotic and detergent stresses alter membrane characteristics, the former by modifying the cellular turgor (18). This leads us to hypothesize that the molecular mechanisms by which the c-di-AMP concentrations are modulated in response to osmotic and bile salt stresses are likely the same and associated with the maintenance of membrane integrity. As previously discussed by Pham et al. (21), an attractive possibility to link membrane alterations to the fluctuations of c-di-AMP concentrations would be that changes in membrane characteristics are directly sensed by the membrane-bound enzymes CdaA and GdpP through their hydrophobic domains.

The colonization and virulence of several c-di-AMP–producing pathogens are affected by altered c-di-AMP homeostasis (30, 31, 36, 75, 91, 95, 96). Likewise, C. difficile lacking gdpP was deficient in long-term colonization in mice, presumably because of the exacerbated susceptibility of this mutant strain to environmental stresses. In contrast, BusAA was shown to not be required for C. difficile maintenance in the gut in our experimental conditions. Our data revealed that c-di-AMP modulates osmotic stress adaptation through the control of both K⁺ and compatible solute transporters and bile salt resistance exclusively through the BusR-mediated control of busAA-AB expression. This suggests that the gut persistence defect of ΔgdpP is caused by its inability to adapt to a high osmotic pressure rather than the presence of bile salts. However, we cannot exclude that susceptibility to other stresses unexplored in this study or an indirect effect of the gdpP deletion on the cell physiology contributes to this phenotype. It is now recognized that the biotransformation by members of the gut microbiota of primary bile acids produced by the liver into secondary bile acids plays an important role in the mechanism of colonization resistance (97–100). Before antibacterial drug treatment, secondary bile salts inhibit C. difficile growth. However, the gut microbiota is altered by antibiotic treatment, leading to an increased concentration of the primary bile salts, which supports C. difficile spore germination and outgrowth (7, 97, 101, 102). In our study, mice were treated with clindamycin to make them susceptible to C. difficile 630Δerm infection, which decreases gut bacteria diversity and affects the bile salt profile (103). Thus, our experimental conditions cannot evaluate the full effect of the busAA deletion on the colonization and persistence of C. difficile given the lack of secondary bile salts. Nevertheless, our work identified c-di-AMP as a crucial signaling molecule regulating adaptation of C. difficile to the host environment, colonization, and tolerance to bile salts.

MATERIALS AND METHODS

Bacterial strains and culture conditions

C. difficile and E. coli strains used in this study are presented in table S1. E. coli strains were grown in LB broth and, when needed, with ampicillin (100 μg/ml) or chloramphenicol (15 μg/ml). C. difficile strains were grown in an anaerobic chamber (Jacomex) under an anaerobic atmosphere (5% H₂, 5% CO₂, and 90% N₂) in TY (104), CDMM (105), BHI (Difco), or BHISG [BHI supplemented with 0.1% l-cysteine, yeast extract (5 mg/ml), and 100 mM glucose] media. When necessary, cefoxitin (25 μg/ml), cycloserine (250 μg/ml), and thiamphenicol (7.5 μg/ml) were added to C. difficile cultures. A potassium-free medium derived from CDMM was used to study potassium requirements of C. difficile strains (table S2). Potassium chloride was added as indicated. The nonantibiotic analog anhydrotetracycline (Sigma-Aldrich) was used for induction of the P_tet promoter of pRPF185 vector derivatives in C. difficile (106). Growth curves were obtained in 96-well microplates at 37°C, and automatic recording of OD₆₀₀ (optical density at 600 nm) was done every 30 min using a plate reader (Promega GloMax Explorer).

Plasmid and strain construction

All primers used in this study are listed in table S3. For deletions, allele exchange cassettes were designed to have between 800 and 1050 base pairs of homology to the chromosomal sequence in both up- and downstream locations of the target gene. The homology arms were amplified by PCR from C. difficile strain 630 genomic DNA, and purified PCR products were directly cloned into the Pme I site of the pMSR vector using NEBuilder HiFi DNA Assembly (New England Biolabs). All pMSR-derived plasmids were initially transformed into E. coli strain NEB10β (New England Biolabs), and all inserts were verified by sequencing. Plasmids were then transformed into E. coli HB101(RP4) and transferred by conjugation into the appropriate C. difficile strains. Transconjugants were selected on BHI supplemented with cycloserine, cefoxitin, and thiamphenicol. Allelic exchange was performed following the procedure previously published (107). For the expression of recombinant KtrA, KdpD, and BusR, the DNA sequence of the corresponding genes was amplified by PCR from C. difficile strain 630 genomic DNA and cloned into Bam HI and Pst I sites of the pQE30 expression vector (Qiagen). All pQE30-derived plasmids were initially transformed into E. coli strain NEB10β (New England Biolabs), and all inserts were verified by sequencing. Plasmids were then transformed into E. coli strain XL1-Blue for protein expression. For inducible expression of cdaA and gdpP in C. difficile, the DNA sequence of the corresponding genes and their ribosome binding site (RBS) was amplified by PCR from C. difficile strain 630 genomic DNA and cloned into Xho I and Bam HI sites of the pDIA6103 vector. For constitutive expression of busAA-AB in C. difficile, the P_tet promoter and regulatory region were excised from pDIA6103 by inverse PCR. The P_cwp2 promoter and the busAA-AB coding sequence with its RBS were amplified by PCR from C. difficile strain 630 genomic DNA and cloned into the modified pDIA6103 using NEBuilder Hifi DNA Assembly (New England Biolabs). The resulting pDIA6103 derivative plasmid was transformed into the E. coli HB101 (RP4) and subsequently mated with C. difficile 630Δerm strain. Transconjugants were selected on BHI supplemented with cycloserine, cefoxitin, and thiamphenicol.

RNA isolation and quantitative reverse transcriptase PCR

Total RNAs were isolated from C. difficile strains after 4 hours of growth in TY medium. Total RNA extraction, complementary DNA (cDNA) synthesis, and real-time quantitative PCR were performed as previously described (108). In each sample, the quantity of cDNAs of a gene was normalized to the quantity of cDNAs of the dnaF gene (CD1305) encoding DNA polymerase III. The relative change in gene expression was recorded as the ratio of normalized target concentrations [threshold cycle (ΔΔC_t) method] (109).

RNA sequencing

Transcriptomic analysis for each condition was performed using four independent total RNA preparations using methods described before (79). Briefly, the RNA samples were first treated using an Epicenter Bacterial Ribo-Zero kit. This depleted ribosomal RNA fraction was used to construct cDNA libraries using a TruSeq Stranded Total RNA sample prep kit (Illumina). Libraries were then sequenced by Illumina NextSeq 500. Cleaned sequenced reads were aligned to the reannotated C. difficile strain 630 (110) for the mapping of the sequences using Bowtie (version 2.1.0). DESeq2 (version 1.8.3) was used to perform normalization and differential analysis using values of the ΔgdpP strain as a reference for reporting the expression data of the ΔgdpPΔbusR strain. Genes were considered differentially expressed if they had a ≥twofold increase or decrease in expression and an adjusted P value (q value) ≤ 0.05.

Phase-contrast microscopy

Bacterial cells were observed at ×100 magnification on an Axioskop Zeiss light microscope. The cell length of 100 cells was measured for each strain using ImageJ software (111).

Antibiotic susceptibility tests

Susceptibility tests for antibacterial drugs were conducted using disk diffusion assay. Overnight cultures of C. difficile strains were diluted to a starting OD₆₀₀ of 0.05 and grown to an OD₆₀₀ of 1, and 100 μl of the culture was spread on a TY agar plate. A disk of 10 μg of imipenem, 30 μg of cefepime, or 30 μg of moxalactam (Bio-Rad) was then placed on top of the plate. The zone of growth inhibition was measured after incubation for 24 hours at 37°C.

Biofilm formation

To generate biofilms, we diluted overnight cultures of C. difficile 1:100 into fresh BHISG, deposited 1 ml per well in 24-well polystyrene tissue culture–treated plates (Falcon, USA), and incubated the plates at 37°C in anaerobic environment for 24 hours. Biofilm biomass was measured using a crystal violet as described elsewhere (79). Briefly, spent media were removed by inversion, and wells were washed twice with phosphate-buffered saline (PBS), dried, stained with crystal violet for 2 min, and washed twice again with PBS. Crystal violet was solubilized with a 75% ethanol solution, and the OD₆₀₀ was measured with a plate reader (Promega GloMax Explorer, Promega, France).

Measurement of intracellular c-di-AMP concentrations

For quantification of the c-di-AMP concentrations in the mutant strains, overnight cultures of C. difficile strains were diluted in TY to a starting OD₆₀₀ of 0.05 and grown for 4 hours at 37°C. A 10-ml culture aliquot was harvested by centrifugation at 5000g for 10 min at 4°C. Nucleotides were extracted from the pellet with methanol/acetonitrile/Milli-Q water (40:40:20), and the c-di-AMP was detected and quantified by LC-MS/MS as described previously (112). For normalization purposes, a 1-ml aliquot from the same culture was taken and pelleted for 10 min at 8000g at 4°C. Bacterial pellets were frozen and then thawed, resuspended in 1 ml of PBS, and incubated at 37°C for 45 min to lyse the cells. The samples were centrifuged for 5 min at 10,000g, and the protein content of the supernatant was determined using a bicinchoninic acid (BCA) assay kit (Thermo Fisher Scientific). The c-di-AMP concentrations are presented as picomol c-di-AMP per milligram of C. difficile protein.

To determine the effect of osmostress on the c-di-AMP concentration, we split a 50-ml culture of C. difficile 630Δerm grown in CDMM until OD₆₀₀ of 0.5 into 12-ml aliquots. Aliquots were treated with either water, 0.4 mM glycine betaine, 200 mM NaCl, or 200 mM NaCl and 0.4 mM glycine betaine for 30 min at 37°C. Cells retrieved by centrifugation were resuspended in 750 μl of immunoassay buffer (137 mM sodium chloride, 2.7 mM potassium chloride, 10 mM sodium phosphate, 1.76 mM potassium phosphate, and 0.1% BSA).

The previously described energized cell suspension method was used to determine the effect of bile salt stress on the c-di-AMP concentrations (21). A 25-ml culture of C. difficile 630Δerm grown in TY until OD₆₀₀ of 0.7 was harvested by centrifugation at 5000g for 10 min and washed twice with low osmolarity ¹/₁₀ KPM buffer (0.01 M K₂HPO₄ adjusted to pH 6.5 with H₃PO₄ and 1 mM Mg-SO₄·7H₂O). Cells were then resuspended in 3.75 ml of ¹/₁₀ KPM, and 20 mM d-glucose was added to the cell suspension before incubating for 10 min at 37°C to stimulate the ATP production required for c-di-AMP synthesis. The cell suspension was then split into 750-μl aliquots. An aliquot was collected to represent t₀, and 50 μl of either water, NaCl (300 mM final), deoxycholate (0.028% final), or cholate (0.4% final) was added to the other aliquots before an additional 10 min incubation at 37°C.

Samples were mechanically lysed using beads and a FastPrep 5G system (MP Biomedicals), and the supernatant was collected after centrifugation (5 min at 16,000g). A small aliquot was removed to measure protein concentration using a BCA assay kit. The remainder of the sample was heated to 95°C for 10 min and used to quantify c-di-AMP using a c-di-AMP ELISA kit (Cayman Chemical) according to the recommendations of the supplier. The c-di-AMP concentrations are presented as picomol c-di-AMP per milligram of C. difficile protein.

Differential radial capillary action of ligand assay

Interaction between c-di-AMP and target proteins was tested by DRaCALA on whole E. coli protein extract (20). Expression of the KtrA-, KdpD-, and BusR-tagged proteins was induced with 1 mM isopropyl-β-d-thiogalactopyranoside for 6 hours at 30°C. Bacterial pellets from 1-ml culture were resuspended in 100 μl of binding buffer [40 mM tris (pH 7.5), 100 mM NaCl, 10 mM MgCl₂, lysozyme (0.5 mg/ml), and deoxyribonuclease (20 μg/ml)] and lysed by three freeze-thaw cycles. For DRaCALA, 1 nM [³²P]-c-di-AMP, synthetized as previously described (20), was added to the whole protein extract and incubated at room temperature for 5 min, and 2.5 μl was spotted onto nitrocellulose membrane. For competition assays, reactions were incubated with 150 μM nonlabeled nucleotides (c-di-AMP, c-di-GMP, cAMP, cGMP, AMP, and ATP; BioLog Life Science Institute, Germany) for 5 min at room temperature before addition of 1 nM [³²P]-c-di-AMP. Samples were spotted on nitrocellulose after 5-min reaction at room temperature. Radioactive signal was detected with a Typhoon system (Amersham).

Conventional mouse infection studies

C. difficile spore inoculums were generated by plating 200 μl of an overnight culture of C. difficile grown on a sporulation medium for Clostridium difficile (SMC) medium (9% Bacto peptone, 0.5% proteose peptone, 0.15% tris base, and 0.1% ammonium sulfate) in SMC agar and incubating them at 37°C for 7 days in an anaerobic chamber. Spores were then harvested in 2 ml of ice-cold sterile water and purified by centrifugation using a HistoDenz (Sigma-Aldrich) gradient (113).

Six-week-old conventional male C57BL mice (Janvier Labs) were singly housed and acclimated for a week before treatment with clindamycin (10 mg/kg; Sigma-Aldrich) by intraperitoneal injection. Twenty-four hours after clindamycin treatment, groups of six mice were challenged with 5 × 10⁴ wild-type or mutant C. difficile spores by oral gavage. To assess bacterial persistence, fecal pellets were collected over a 17-day period (days 2, 6, 9, 13, and 17). Fecal pellets were homogenized in the anaerobic hood in 1× prereduced PBS at a concentration of 100 mg/ml, serially diluted, and plated on BHI agar containing 3% defibrinated horse blood, 0.1% taurocholate, cefoxitin (25 μg/ml), and cycloserine (250 μg/ml) to assess the total number of CFUs.

Ethics statement

Animal studies were performed in agreement with European and French guidelines (Directive 86/609/CEE and Decree 87–848 of 19 October 1987). The study received the approval of the Institut Pasteur Safety Committee (Protocol 11.245) and the ethical approval of the local ethical committee “Comité d’Ethique en Experimentation Animale Institut Pasteur no. 89 (CETEA)” (CETEA dap190131).

Data and materials availability:

The RNA-seq data have been deposited into the NCBI-GEO database (www.ncbi.nlm.nih.gov/geo/) with the accession number GSE192529. All other data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.