The Ser/Thr Kinase PrkC Participates in Cell Wall Homeostasis and Antimicrobial Resistance in Clostridium difficile

Clostridium difficile is the leading cause of antibiotic-associated diarrhea in adults. During infection, C. difficile must detect the host environment and induce an appropriate survival strategy. Signal transduction networks involving serine/threonine kinases (STKs) play key roles in adaptation, as they regulate numerous physiological processes. PrkC of C. difficile is an STK with two PASTA domains. We showed that PrkC is membrane associated and is found at the septum.

ABSTRACT

Clostridium difficile is the leading cause of antibiotic-associated diarrhea in adults. During infection, C. difficile must detect the host environment and induce an appropriate survival strategy. Signal transduction networks involving serine/threonine kinases (STKs) play key roles in adaptation, as they regulate numerous physiological processes. PrkC of C. difficile is an STK with two PASTA domains. We showed that PrkC is membrane associated and is found at the septum. We observed that deletion of prkC affects cell morphology with an increase in mean size, cell length heterogeneity, and presence of abnormal septa. A ΔprkC mutant was able to sporulate and germinate but was less motile and formed more biofilm than the wild-type strain. Moreover, a ΔprkC mutant was more sensitive to antimicrobial compounds that target the cell envelope, such as the secondary bile salt deoxycholate, cephalosporins, cationic antimicrobial peptides, and lysozyme. This increased susceptibility was not associated with differences in peptidoglycan or polysaccharide II composition. However, the ΔprkC mutant had less peptidoglycan and released more polysaccharide II into the supernatant. A proteomic analysis showed that the majority of C. difficile proteins associated with the cell wall were less abundant in the ΔprkC mutant than the wild-type strain. Finally, in a hamster model of infection, the ΔprkC mutant had a colonization delay that did not significantly affect overall virulence.

INTRODUCTION

Clostridium difficile is the leading cause of antibiotic-associated infections in adults. Severity of C. difficile infection (CDI) symptoms ranges from diarrhea to life-threatening pseudomembranous colitis. Major risks associated with CDI include antibiotic exposure leading to dysbiosis of gut microbiota, and people of advanced age and hospitalized patients are at increased risk for CDI (1). The impact of CDI is significant in terms of mortality, morbidity, disease management, and financial burden. With the emergence of new isolates, the incidence and severity of CDI have increased in both North America and Europe (2). C. difficile is acquired from the environment through the ingestion of spores. This resistant form of the bacterium is thought to be responsible for transmission, environmental persistence, and dissemination. When the normal intestinal microbiota is disrupted, the pools of metabolites present in the gut change, leading to an increased concentration of cholate conjugates, which in turn triggers spore germination (3, 4). The vegetative cells can then multiply and colonize the dysbiotic gastrointestinal tract. Toxigenic strains of C. difficile produce two toxins, TcdA and TcdB, which are its main virulence factors. These toxins cause alteration of the actin cytoskeleton of epithelial cells and neutrophil recruitment, causing local inflammation (3). Following the administration of a targeted antibiotic therapy, such as metronidazole or vancomycin, CDI relapse can occur, involving spore and possibly biofilm formation (5, 6). Additional factors, including those involved in stress adaptation and surface-associated proteins, such as adhesins, the fibronectin-binding protein (FbpA), the surface layer protein (SlpA), and the flagellum, also participate in the colonization process, allowing C. difficile to establish its intestinal niche (7).

Gut dysbiosis is an essential step that allows C. difficile to colonize the colon and cause an infection. Certain antibiotics, such as cephalosporins, clindamycin, and fluoroquinolones, are known to increase the risk of developing a CDI. Most C. difficile strains are resistant to these antibiotics, which is of major concern (1). Fluoroquinolone resistance in C. difficile strains is associated with amino acid substitution in the gyrase GyrA or GyrB (1). The mechanisms responsible for cephalosporin resistance by C. difficile remain poorly characterized (1). However, some progress has been made, as a class D β-lactamase was identified in the genomes of most C. difficile strains (8). In addition to antibiotics, vegetative cells also encounter secondary bile salts (4), host-secreted antimicrobial peptides, and reactive oxygen and nitrogen species produced during inflammation (3, 9). To survive in the host, C. difficile must detect the host environment and induce an appropriate survival strategy, which may include activating the general stress response, sporulation, biofilm formation, and the production of virulence factors (5, 6, 9).

Protein phosphorylation is a reversible posttranslational modification used to transduce signal and regulate cellular processes. Bacterial serine/threonine kinases of the Hanks family (STKs) share structural similarities with eukaryotic serine/threonine kinases and undergo substrate phosphorylation on serine or threonine residues (10). STKs and their associated phosphatases (STP) regulate numerous physiological processes, including translation, carbon and cell wall metabolism, antibiotic tolerance, cell division, developmental processes, and virulence (10, 11). Phosphoproteomic studies revealed that STKs might phosphorylate a broad spectrum of substrates (12), including enzymes, components of the cellular machinery involved in translation, division, or repair, and several transcriptional regulators that mainly form parts of two-component systems (10, 13). Therefore, STKs are key integration nodes in signaling networks that control bacterial physiology, growth, and stationary phase. In Bacillus subtilis, PrkC is a transmembrane kinase with an extracellular signal receptor domain containing penicillin-binding and STK-associated (PASTA) repeats. PrkC interacts with peptidoglycan fragments and β-lactam in vitro (14, 15), while a PASTA-STK (Stk1) homologue in Staphylococcus aureus interacts with lipid II (16). In B. subtilis, PrkC is a key signaling enzyme that controls stationary-phase survival, biofilm formation, sporulation, germination (13), and peptidoglycan (PG) metabolism (17). Inactivation of genes encoding PASTA domain-containing STKs is associated with a range of phenotypes among Gram-positive bacteria, with defects in cell division and changes in cell wall metabolism commonly observed (10, 17, 18). In S. aureus, Stk1 plays a role in virulence, antibiotic resistance, and cell wall remodeling (17), whereas in Streptococcus pneumoniae, StkP controls cell division, competence, virulence, adherence to eukaryotic cells, stress response, and cell wall metabolism (17, 19).

The role of STKs has not been studied in anaerobic firmicutes. C. difficile has two genes encoding STKs, CD2148 and prkC (CD2578), and one gene encoding an STP (CD2579). In this study, we characterized the role of the C. difficile PASTA-STK, PrkC, in the regulation of the physiological processes important in CDI. Inactivation of the prkC gene resulted in changes in the morphology and the properties of the cell envelope as well as increased sensitivity to antimicrobial compounds targeting the cell wall.

RESULTS

CD2578 is the PASTA-STK of C. difficile.

CD2578 (PrkC) from C. difficile contains a cytosolic N-terminal kinase domain and an extracellular motif containing repeats of a PASTA that are separated by a transmembrane segment (Fig. 1A). Based on its predicted amino acid sequence, the kinase domain of PrkC has a high level of sequence identity to that of B. subtilis (PrkC), S. aureus (Stk1), S. pneumoniae (StkP), Enterococcus faecalis (IreK), and Mycobacterium tuberculosis (PknB) (43% to 50%). The conserved lysine within the ATP-binding P loop of the kinase domain is present and corresponds to position 39 in PrkC (Fig. 1A). This residue is essential for the phosphotransfer in other STKs (20). As observed for the PASTA-STKs in other firmicutes, a transmembrane segment is predicted from amino acids 375 to 397 (21, 22). The extracellular sequence of PrkC contains 2 PASTA motifs (Fig. 1A), while up to 7 PASTA motifs are typically present in STKs of other firmicutes (17). Interestingly, C. difficile contains a very atypical Ser-Gly-Asn (SGN)-rich region of 100 amino acids at the C-terminal part of the protein (Fig. 1A). This SGN-rich region is absent from PASTA-STKs of other firmicutes. Interestingly, SGN-rich regions are found in some cell wall-anchored proteins of bacilli and in CwpV, a surface-associated protein in C. difficile, but their functions have yet to be characterized.

FIG 1.

Organization, localization, and kinase activity of the PrkC protein. (A) Organization of the domains of C. difficile CD2578-PrkC. PrkC contains a cytoplasmic kinase domain in the N-terminal part (brown), a transmembrane (TM) segment, two PASTA domains (green), and an atypical SGN (Ser, Gly, Asn)-rich domain in the C-terminal part (pink). The conserved lysine residue (K39) within the ATP-binding P loop of the kinase that is required for phosphotransfer is indicated. (B) Localization of the PrkC-HA-tagged protein. Cells expressing prkC fused to HA were grown in the presence of 15 ng/ml of ATc and harvested during exponential growth. Samples were fractionated into membrane (Mb) and cytoplasmic (Cy) fractions. Protein fractions were analyzed by Western blotting using an antibody raised against HA. (C) Localization of the SNAP-PrkC fusion during growth. The SNAP-PrkC protein was produced during the exponential growth phase in the presence of 50 ng/ml ATc. After labeling with the TMR-Star substrate, PrkC-SNAP localization was analyzed by fluorescence microscopy. AF, autofluorescence. The scale bar represents 5 μm. (D) Western blot performed after fractionation in a soluble fraction (lanes 1) and an insoluble fraction (lanes 2) obtained from exponential-phase cultures of the WT, the ΔprkC mutant, the complemented strain, and the ΔprkC mutant carrying pDIA6103-prkC-K39→A (K→A). An anti-P-Thr antibody was used to detect phosphorylated threonine. Red arrows indicate the bands detected in the WT strain that disappeared from the ΔprkC mutant.

The prkC locus of C. difficile.

The prkC gene likely belongs to a large cluster of genes, which ranges from dapF (CD2590) to CD2578-prkC (Fig. 2A). As usually observed in other firmicutes (23, 24), a gene (CD2579-stp) encoding a PP2C-type phosphatase (STP), which is probably involved in the dephosphorylation of the PrkC substrates, is adjacent to prkC. Upstream of stp, we found genes encoding proteins involved in translation (rlmN, rsmB, def, and fmt), transcription (rpoZ), DNA replication (priA), and metabolism (gmk and coaBC). This cluster is conserved in B. subtilis and Bacillus cereus, with the exception of 3 genes that are present only in C. difficile: CD2582 and CD2583, encoding membrane proteins, and dapF, encoding a diaminopimelate epimerase (Fig. 2A). To determine if prkC is cotranscribed with stp, as observed in other firmicutes and the other genes located upstream of stp, such as the rlmN gene, reverse transcription-PCR (RT-PCR) experiments were performed. PCR products were detected using primer pairs located in adjacent genes (prkC-stp or stp-rlmN) with RNAs processed by a reverse transcriptase but not with untreated RNA used as a negative control (Fig. 2B). This result indicated that the prkC gene is in an operon with stp and also rlmN. Using genome-wide transcriptional start site (TSS) mapping (25), we identified a unique TSS in the prkC locus located 23 bp upstream of dapF, suggesting that this gene is the first gene of a large operon. We found a TG-N-TATAAT-extended −10 box specific for the consensus of σ^A-dependent promoters upstream of the TSS (Fig. 2A). However, we cannot exclude the existence of additional promoters between dapF and prkC that were not identified in the genome-wide TSS mapping (25).

FIG 2.

Genetic organization of the prkC locus. (A) Schematic representation of the gene cluster present upstream of prkC. The locus between CD2590-dapF and prkC includes genes encoding proteins involved in translation, transcription, DNA replication, metabolism, and membrane. The genome-wide TSS mapping (25) indicated the presence of a promoter upstream of dapF. The extended −10 box and the TSS are indicated in boldface. CoA, coenzyme A. (B) PCR realized with primers annealing in prkC and stp (IMV936 and IMV908, lanes 1 and 2) or stp and rlmN (IMV935 and IMV907, lanes 3 and 4) either on RNA extracted from the 630Δerm strain (lanes 1 and 3) or on cDNA synthesized by reverse transcription from the same RNA using primer IMV843 (lanes 2 and 4). Smart ladder, 200 to 10,000 bp.

Localization of the PrkC protein.

To determine the cellular location of PrkC, we expressed the prkC gene fused to a hemagglutinin (HA) tag under the control of the inducible promoter P_tet (pDIA6103-P_tet-prkC-HA). A Western blot analysis of the membrane or cytoplasmic fraction with an anti-HA antibody revealed two bands corresponding to proteins of about 100 kDa and 50 kDa (see Fig. S2A in the supplemental material). These two proteins were only detected in the membrane fraction (Fig. 1B). The expected molecular weight of the PrkC-HA-tagged protein is around 75 kDa, and the band detected around 100 kDa likely corresponds to PrkC-HA. Indeed, the PrkC and StkP proteins of B. subtilis and S. pneumoniae are detected with an apparent molecular weight higher than their expected sizes (21, 22), and the phosphorylation of the kinase results in a reduced electrophoretic mobility (26). The lower band at 50 kDa could be a form of degradation product that might correspond to the extracellular domain fused to HA. It is worth noting that PrkC of B. subtilis is sensitive to protease cleavage (21).

To determine the localization of PrkC, we used a 630Δerm strain containing a plasmid encoding a SNAP^Cd-PrkC protein fusion produced under the control of the P_tet promoter. The production of the SNAP^Cd-PrkC protein is stable, as shown by Western blotting using an anti-SNAP antibody (Fig. S2B). After 2 h of induction, we detected the SNAP^Cd-PrkC fusion protein at the septum of dividing cells (Fig. 1C). Thus, the PrkC protein is membrane associated and localizes at the septum during cell growth.

Deletion of the prkC gene in C. difficile and detection of PrkC activity.

To study the role of PrkC in the physiology of C. difficile, we inactivated the prkC gene by allelic exchange in the 630Δerm strain (27). Deletion of the prkC gene from codon two to the stop codon was confirmed by PCR (Fig. S1). The prkC mutant was complemented with the wild-type prkC gene (pDIA6103-P_tet-prkC) or a modified copy of prkC (pDIA6103-P_tet-prkC-K39→A) with the lysine residue required for phosphotransfer replaced by an alanine.

The kinase activity of PrkC was demonstrated by comparing the profiles of phosphorylation of the wild-type (WT) strain or the ΔprkC mutant carrying pDIA6103, pDIA6103-P_tet-prkC, or pDIA6103-P_tet-prkC-K39→A. We assumed that PrkC and PrkC-K39→A were produced to the same level and were located in the membrane, because we detected similar levels in the membrane fraction of these proteins with an HA tag fusion (PrkC-HA or PrkC-K39→A-HA) (Fig. S2A). We detected proteins phosphorylated on threonine in the WT strain both in soluble and in insoluble fractions by Western blot analysis with an antibody against phosphorylated threonine residues (anti-PThr) (Fig. 1D). Some bands that were detected in the WT strain and with a lower intensity in the complemented strain disappeared from the ΔprkC mutant and the strain producing a modified PrkC-K39→A protein. This result strongly suggested that PrkC has a kinase activity in vivo and that the lysine at position 39 (Fig. 1A) is essential for the phosphorylation of PrkC targets, as observed in the PASTA-STKs of other firmicutes (17, 20).

Impact of prkC deletion on growth, sporulation, and germination.

The 630Δerm strain and the ΔprkC mutant exhibited nearly identical doubling times and growth yields in TY broth (30 g·liter⁻¹ Bacto tryptone, 20 g·liter⁻¹ yeast extract, pH 7.4) (Fig. S3A). No difference in number of CFU was observed between the two strains (Fig. S4A and data not shown). We also performed coculture of the WT and ΔprkC mutant strains in TY (Fig. S3C). We observed that the ΔprkC mutant was less abundant than the wild-type strain by 2-fold after 8 h or 24 h of growth. This suggests that there is no drastic difference in fitness associated with the inactivation of prkC. The growth of the ΔprkC mutant containing pDIA6103-P_tet-prkC was similar to that of the WT strain in the presence of 15 ng/ml of anhydrotetracycline (ATc) but was affected when we added 50 ng/ml of ATc (Fig. S3B). This growth defect was also observed in the ΔprkC mutant expressing prkC-K39→A (Fig. S3B). In contrast, the addition of increasing concentrations of ATc had no effect on the growth of the WT strain harboring an empty plasmid (Fig. S3B). A quantitative RT-PCR analysis also showed that expression of prkC was similar in the WT and complemented strains in the presence of 15 ng/ml ATc (Fig. S3D). In the presence of 50 ng/ml ATc, the expression of prkC probably increased in the ΔprkC strain containing pDIA6103-P_tet-prkC, because we detected a drastic overproduction of the SNAP-PrkC fusion in the presence of 50 ng/ml compared to production at 20 ng/ml (Fig. S2). Overproduction of PrkC is likely responsible for the growth defect observed in the presence of 50 ng/ml ATc, suggesting that a proper level of prkC expression is important for the fitness of C. difficile. Since the kinase activity of PrkC apparently is not responsible for the growth defect when the protein is overexpressed, the transmembrane segment or the extracellular domain is probably involved in this toxicity.

To determine if PrkC is involved in sporulation, we measured sporulation levels of the WT and the ΔprkC mutant strains after 24 h or 72 h of growth in sporulation medium (SM). No significant differences were observed between the two strains for the quantity of spores produced (Fig. S4A). When we tested the ability of purified spores to germinate, we observed that the drop in the optical density at 600 nm (OD₆₀₀) of spore suspensions after the addition of taurocholate was similar for the WT and the ΔprkC mutant (Fig. S4B). In B. subtilis, PrkC is involved in the germination of spores in the presence of B. subtilis muropeptides (28). However, purified muropeptides from C. difficile (0.1 to 0.01 mg/ml) failed to induce spore germination of the WT or the ΔprkC mutant (data not shown). Bryostatin is a molecule involved in STK activation that induces germination of B. subtilis spores (28). This compound at 1 μM or 10 μM had no effect on C. difficile spores (data not shown). All these results indicated that PrkC is not involved in the control of sporulation and spore germination in C. difficile.

prkC deletion affects cell morphology.

To determine if deletion of prkC had an effect on cell morphology, we analyzed by phase-contrast microscopy cells of the 630Δerm strain, the ΔprkC mutant, and the complemented strain during exponential growth phase (Fig. S5A). The ΔprkC mutant cells were more elongated than those of the WT strain and the complemented strain (Fig. S5A). We then measured the cell size using cells labeled with 4′,6-diamidino-2-phenylindole (DAPI) and FM4-64 (Fig. 3A and B). The ΔprkC mutant cells had an average cell size of 9.2 μm, whereas the WT and the complemented strains had an average cell size of 4.5 μm and 6.4 μm, respectively (Fig. 3B). We detected elongated cells (Fig. 3A, blue arrows) and undivided cells with several septa (Fig. 3A, green arrows). Interestingly, we observed septation defects for the ΔprkC mutant (Fig. 3A, white box, C, and Fig. S5B). Our analysis revealed that 4% of the cells had abnormal septation (Fig. 3C and Fig. S5B, yellow arrows) and that 1% of the cells had adjacent septa which created a minicell lacking DAPI-stained DNA (Fig. 3C and Fig. S5B, white arrows). We then analyzed by transmission electron microscopy (TEM) the cell morphology and the structure of the septa for the WT, the ΔprkC mutant, and the complemented strains. Cells with normal septation were observed for the WT strain, complemented strain, and ΔprkC mutant (Fig. 4A), while some cells of the ΔprkC mutant displayed aberrant septation, with the presence of multiple and adjacent septa (Fig. 4B). Minicells attached to each other were also observed (Fig. 4B, left), and these probably arise from the formation of several septa in close proximity (Fig. 4B, right, white arrows). These results indicated that deletion of the prkC gene in C. difficile affects bacterial cell morphology and septum formation.

FIG 3.

Morphology of the ΔprkC mutant. (A) Fluorescence microscopy was carried out on the WT, the ΔprkC mutant, and the complemented (Comp) strains. Membranes and DNA were visualized with FM4-64 (red) and DAPI (blue), respectively. A blue arrow shows an elongated cell and green arrows show nonseparated cells. The white square indicates the presence of a septum with an aberrant structure. Scale bars, 5 μm. (B) Cell size distribution for the WT, the ΔprkC mutant, and the complemented (Comp) strains. The measurement was done on cells labeled with FM4-64 and DAPI. Means and errors of the means were calculated after the measurement of at least 600 cells for each strain. The analysis was performed with the software ImageJ. (C) Presence of aberrant septation and anucleated minicells in ΔprkC mutant cells. Fluorescence microscopy of cells stained with FM4-64 and DAPI revealed the presence of aberrant septum (yellow arrows) and anucleated minicells (white arrows) in the ΔprkC mutant. Scale bars represent 5 μm.

FIG 4.

Presence of abnormal septa in the ΔprkC mutant cells. (A) TEM pictures showing normal septal structure in the 630Δerm (WT), the ΔprkC mutant, and the complemented (Comp) strains. Scale bars represent 100 nm. (B) TEM pictures showing aberrant septal structure in the ΔprkC mutant cells. In the right panel, we can distinguish that the synthesis of two septa (white arrows) has begun near an apparently normal septum. Scale bars represent 200 nm.

prkC deletion increases sensitivity to detergents and autolysis.

To determine if prkC deletion also affects the cell envelope, we tested the sensitivity of the prkC mutant to compounds with detergent activity. The presence of 0.006% SDS affected the growth of the ΔprkC mutant but did not affect the WT and the complemented strains (Fig. 5A). It is worth noting that in the gastrointestinal tract, C. difficile is subjected to the bactericidal effects of secondary bile salts that solubilize phospholipids, impair membrane integrity, and weaken the cell wall (29). We therefore tested the sensitivity of the ΔprkC mutant to the secondary bile salt deoxycholate (DOC) and to the primary bile salt cholate. While the sensitivity to cholate (0.03% or 0.04% in TY) was similar for all the strains (data not shown), we observed that the growth of the ΔprkC mutant was reduced in the presence of 0.03% DOC (Fig. 5B), a physiological concentration encountered in the gut (30, 31). In addition, the autolysis in the presence of 0.01% of the nonionic detergent Triton X-100 is more rapid for the ΔprkC mutant than for the WT and the complemented strains (Fig. 5C), while no difference in autolysis was observed in the absence of Triton X-100 (Fig. S6A). This result suggested that the ΔprkC mutant is more sensitive to the PG hydrolysis than are the WT and the complemented strains.

FIG 5.

prkC deletion increases sensitivity to detergents and autolysis. (A) Resistance to SDS stress of the ΔprkC mutant, the 630Δerm (WT), and the complement (Comp) strains was tested on BHI plates containing 0.006% SDS. This experiment was performed in triplicate, and this plate is representative of the results obtained. (B) Growth of the WT strain (black circle), the ΔprkC mutant (white circle), and the complemented strain (white square) in 24-well microplates containing TY medium in the presence of 0.03% DOC. A growth curve without DOC is presented in Fig. S3. Four independent cultures were done. (C) Autolysis of the WT (black circle), ΔprkC (white circle), and complemented (black square) strains in the presence of 0.01% Triton X-100. The OD₆₀₀ of the samples incubated at 37°C was determined every 5 min until complete cell lysis was reached. Four independent experiments were done.

PrkC controls sensitivity to antimicrobial compounds targeting the cell envelope.

Given that the deletion of prkC affected cell morphogenesis, we tested the sensitivity of the prkC mutant to antibiotics targeting cell wall biosynthesis (26, 32). Using disk assays, we showed that the size of the zone of inhibition increased for the ΔprkC mutant compared to those of the WT and the complemented strains for all β-lactams tested (Fig. 6A). Thus, the prkC mutant is more susceptible to ticarcillin, amoxicillin, imipenem, and several cephalosporins (cefoxitin, ceftazidime, and cefepime). In contrast, we did not observe any differences for antibiotics that target transcription or translation (Fig. 6A and Fig. S6B). We also observed an important reduction in the MIC of the ΔprkC mutant compared to that of the WT strain for the glycopeptide teicoplanin (>12-fold) but not for vancomycin (2-fold) (Table 1). In addition, we observed a slight increase in the susceptibility of the mutant to amoxicillin and imipenem (4-fold) and a more substantial increase for the cephalosporins tested (Table 1). The MIC for the second-generation cephalosporin cefoxitin was reduced by more than 6-fold for the ΔprkC mutant compared to that for the WT strain, whereas the MIC for the third-generation cephalosporins ceftazidime, cefepime, and cefotaxime was reduced for the ΔprkC mutant by 20-, 17.5-, and 8-fold, respectively. The same increase in sensitivity to cefoxitin, ceftazidime, and cefepime was seen for the ΔprkC mutant carrying pDIA6103-P_tet-prkC-K39→A (Fig. 6A), suggesting that the kinase activity of PrkC is required for C. difficile to express an intrinsic high level of resistance to cephalosporins (1).

FIG 6.

Sensitivity of the ΔprkC mutant to antibiotics targeting cell wall and to lysozyme. (A) Sensitivity to antibiotics targeting the cell wall. Histograms representing the diameters of growth inhibition area after 24 h of incubation on BHI plates for the 630Δerm strain pDIA6103 (WT; black), the ΔprkC mutant pDIA6103 (medium gray), the complemented strain (Comp; dark gray), and the ΔprkC mutant expressing PrkC-K39A (pale gray). We used antibiogram disks containing 75 μg ticarcillin, 25 μg amoxicillin, 10 μg imipenem, 30 μg ceftazidime, 30 μg cefepime, or 15 μg erythromycin. Cefoxitin was tested at 100 μg. The results presented correspond to 7 experiments (ticarcillin, amoxicillin, and imipenem) or 4 experiments (cephalosporins). Data were analyzed by t test. *, P < 0.05. (B) Sensitivity to lysozyme was determined on Pep-M plates. Lysozyme (800 μg) was added to a 6-mm disk. Histograms representing the diameter of growth inhibition measured for the 630Δerm pDIA6103 (WT), the ΔprkC mutant pDIA6103 (∆prkC), and the complemented (Comp) strains. The experiment was performed in quadruplicate. Data were analyzed by t test. *, P < 0.05.

TABLE 1.

MICs of the 630Δerm strain and of the ΔprkC mutant for antibiotics targeting cell wall and CAMPs

Compound tested	MICa (μg/ml)		Median fold sensitization
	630Δerm	ΔprkC
Antibiotics targeting cell wall
Vancomycin	1.5	0.75	2
Teicoplanin	0.19	<0.016	>12
Amoxicillin	4	1	4
Imipenem	6	1.3	4
Cefoxitin	>256	44	>6
Ceftazidime	60	3	20
Cefepime	70	4	17.5
Cefotaxime	128	16	8
Antimicrobial peptides
Polymyxin B	340	55	6
Bacitracin	550	25	22
Nisin	140	30	4.5

^aMICs for antibiotics were determined using Etest, with the exception of cefotaxime. The MICs for cefotaxime and antimicrobial peptides were determined by the method of dilution.

Cationic antimicrobial peptides (CAMPs) that are produced by both bacteria and the host also target the cell envelope and/or membrane of bacteria (33). Thus, we tested the resistance of the WT and mutant strains for several CAMPs of bacterial origin. The MICs of the ΔprkC mutant for polymyxin B, bacitracin, and nisin were 6-, 22-, and 4.5-fold lower, respectively, than the MICs for the WT strain (Table 1). In the gastrointestinal tract, host-produced antimicrobial proteins such as lysozyme inhibit colonization of many Gram-positive pathogens by hydrolyzing the PG and breaking the integrity of the cell wall. However, C. difficile is known for its intrinsic resistance to lysozyme that might be associated with high levels of deacetylation of its PG (34). Interestingly, using an antimicrobial disk assay realized on peptone-containing medium (Pep-M) plates, we observed a greater size of the zone of inhibition for the ΔprkC mutant (17 mm) than the WT strain (7.25 mm) in the presence of 800 μg of lysozyme (Fig. 6B). These results indicated that prkC deletion is involved in resistance to CAMPs and lysozyme.

prkC deletion affects motility, sedimentation, and biofilm formation.

To identify other phenotypes associated with cell envelope properties, we first tested motility on semisolid brain heart infusion (BHI) plates and observed a reduced motility for the ΔprkC mutant compared to that of the WT and complemented strains (Fig. 7A). After 48 h, the motility of the WT strain was 1.35 mm ± 0.2 mm, while the motility of the ΔprkC mutant was 0.8 mm ± 0.1 mm. We also observed that the ΔprkC mutant formed more aggregates at the bottom of the tube than the WT and complemented strains (Fig. S7A). We also tested the ability of these strains to form biofilm after an exposure to compounds triggering an envelope stress, such as DOC or polymyxin B. In the presence of subinhibitory concentrations of polymyxin B (20 μg/ml) or DOC (0.01%), the ΔprkC mutant formed 6- and 10-fold more biofilm than the WT and complemented strains, respectively, after 24 h (Fig. 7B and C). However, no difference in biofilm formation was observed without these compounds (data not shown). When we quantified the biofilm formed in the presence of DOC at 24 h, 48 h, and 72 h (Fig. S7B), we found that the ΔprkC mutant formed biofilm earlier than the WT strain (at 24 h), while more biofilm was produced by the WT strain than the ΔprkC mutant at 48 h. A previous work showed that extracellular DNA (eDNA) is a major component of the matrix in DOC-induced biofilms (35). In both the ΔprkC mutant and the WT strain, DOC-induced biofilms were rapidly dispersed when treated with DNase (Fig. S7C). Moreover, we detected eDNA in the extracellular matrix for the ΔprkC mutant at 24 h and 48 h but only at 48 h for the WT strain (Fig. 7D). Since eDNA is required for biofilm stability (35), the early release of eDNA detected in the ΔprkC mutant might be responsible for the premature biofilm formation. Thus, deletion of prkC seems to affect motility and the ability of C. difficile to sediment and to form biofilm under conditions that induce cell envelope stress.

FIG 7.

Sedimentation, biofilm formation, and motility of the ΔprkC mutant. (A) Motility test on 0.3% agar BHI plates. The plates shown are representative of 3 independent tests. (B and C) Mean values of the OD₅₇₀ measured after crystal violet staining of the mass of biofilm obtained after 24 h in the presence of 20 μM polymyxin B (B) or 0.01% DOC (C). Error bars show standard deviations from three independent experiments performed in triplicate. Data were analyzed by t test. P values were 0.031 and 0.0045 for polymyxin B (B) and DOC (C). Crystal violet staining reveals the amount of biofilm formed after 24 h corresponding to the values represented in the graphs. (D) Detection of eDNA in the matrix of 24 h and 48 h DOC-induced biofilms formed by the WT strain and the ΔprkC mutant. The experiment was done in triplicate. Presented is a gel representative of the results obtained.

Comparison of the PG structure and composition between the WT strain and the ΔprkC mutant.

Based on the changes in cell shape and increased susceptibility to β-lactams and lysozyme of the ΔprkC mutant, we hypothesized that these phenotypes could be due to modification of the PG structure or composition. However, we showed that the muropeptide profiles of the WT strain (black) and the ΔprkC mutant (red) are almost identical (Fig. S8). Indeed, we found no difference in the abundance of dimers containing a 3-3 cross-link (peaks 9, 10, 11, 13, 14, 15, and 17) generated by l,d-transpeptidases that might be associated with the high level of intrinsic resistance of C. difficile to some β-lactams (36) or containing a 4-3 cross-link (peaks 18, 19, and 21) catalyzed by d,d-transpeptidases, a target of β-lactam (Fig. S8). Moreover, the ratio of 4-3 to 3-3 cross-linking was not affected under our conditions. In addition, the amounts of deacetylated muropeptides (peaks 4, 7, 9, 10, 11, 13, 14, 15, 17, 18, 19, and 21) (36) were similar in the WT strain and the ΔprkC mutant (Fig. S8). As muropeptide deacetylation contributes to a high level of resistance to lysozyme in C. difficile (34), the increased sensitivity of the ΔprkC mutant to lysozyme cannot be explained by changes in the level of PG deacetylation (Fig. S8). Our results suggested that susceptibility of the prkC mutant to some β-lactams or lysozyme is not related to changes of its PG structure or composition.

Impact of the prkC deletion on PSII.

In firmicutes, the wall teichoic acids (WTA) are involved in cell division, maintenance of cell shape, and susceptibility to CAMPs or β-lactam (37). The 630Δerm strain of C. difficile has an atypical wall teichoic acid, known as polysaccharide II (PSII), composed of hexaglycosylphosphate repeats (38, 39). To determine whether the phenotypes observed for the ΔprkC mutant were associated with changes in the abundance, localization, and/or structure of PSII, we first purified PSII of the WT and ΔprkC mutant strains. Their structures were analyzed by nuclear magnetic resonance (NMR), which showed that the ¹H NMR and ¹³C heteronuclear single quantum coherence spectra from the 630Δerm mutant PSII were in good agreement with previous data (39, 40) and that no difference in PSII structure was detected between the prkC mutant and the WT strain (Fig. S9).

We also investigated a possible role for PrkC in controlling PSII localization. Using an immunoblot analysis with an antibody raised against PSII (38), we observed that more PSII was shed in the supernatant during growth by the ΔprkC mutant than by the parental and complemented strains during growth (Fig. 8A). In contrast, we detected similar quantities of PSII in total cell extracts for all strains (Fig. S10). Thus, this result suggests that the deletion of prkC leads to an enhanced release of PSII into the supernatant.

FIG 8.

Impact of prkC deletion on PSII localization and PG and PSII production. (A) Immunoblot detection of PSII using a serum antibody raised against this glycopolymer (38) in supernatants of the WT, ΔprkC, and complemented strains. For each sample, we normalized by using the OD₆₀₀ of the corresponding culture. ND, not diluted. All immunoblots are representative of at least four replicates. (B and C) Quantification of PG (B) and PSII (C) present at the surface of the WT and the ΔprkC mutant strains. The values were normalized to the WT levels, considered 1. Error bars represent standard deviations from at least 4 independent experiments.

To precisely quantify the amount of PSII and PG in the cell wall of the WT and the ΔprkC mutant strains, we fractionated cells to recover the cell wall containing the PSII covalently linked to PG. We then separated these two compounds and quantified the amount of each product. Interestingly, we observed that the amount of PG purified from the ΔprkC mutant was lower than that of the WT strain, which represented 55% of the quantity purified from the WT strain (Fig. 8B). Similarly, but to a lesser extent, a reduction of 33% in the quantity of PSII was observed for the ΔprkC mutant (Fig. 8C). It is possible that the reduction in PG for the ΔprkC mutant could interfere with the localization of PSII and could explain the increased shedding of PSII in the supernatant (Fig. 8A).

prkC deletion affects cell wall-associated proteins.

Several properties of the bacterial cell envelope can be partly attributed to proteins localized at the cell surface. We extracted proteins noncovalently anchored to the cell wall of the ΔprkC mutant and the WT strains (41) and compared their relative abundance by mass spectrometry (MS) (Table 2 and Fig. S11). A first family of proteins is the CWPs, which contain a CWB2 motif responsible for their anchorage to the cell wall by interacting with PSII (42, 43). Indeed, we found that 27 CWP proteins with CWB2 (out of 29 in strain 630), including the S-layer protein SlpA, were more abundant in the cell wall of the WT strain than in the ΔprkC mutant, while a unique CWP protein, Cwp7, was more abundant in the ΔprkC mutant. Other noncovalently anchored proteins present on the cell surface (with SH3_3 or PG4 motifs) were also found in different amounts in the WT strain and the ΔprkC mutant. Interestingly, some of them are involved in cell wall metabolism, including l,d-transpeptidases, carboxypeptidases, and putative cell wall hydrolases.

TABLE 2.

Change in cell wall-associated proteins in C. difficile between the 630Δerm strain and the ΔprkC mutant

Gene IDd	Name	Function	Fold change between 630Δerm and ΔprkCa	−Log (P value)	Detected in surface proteome
CWB2 motif
Down in ΔprkC mutant
CD1233	cwp26	Cell wall-binding protein of skin element	31.3	2.7
CD1469	cwp20	Cell wall penicillin-binding protein	6.7	6.5
CD2518	cwp29	Cell wall-binding protein	6.4	4.6
CD2795	cwp11	Cell wall-binding protein	6.1	6.5
CD0440	cwp27	Cell wall-binding protein	5.1	4.3
CD2735b	cwp14	Cell wall-binding protein (also SH3)	4.8	5.2
CD2798	cwp9	Cell wall-binding protein	4.7	4.5
CD2767	cwp19	Cell wall-binding protein, autolysin	4.6	5.9
CD1803	cwp23	Cell wall-binding protein	4.4	4.9
CD2786	cwp5	Cell wall-binding protein	4.4	5.4
CD1751	cwp13	Cell wall-binding protein, protease	4.2	5.4
CD2787	cwp84	Cell wall-binding protein, protease	4.0	5.4	+
CD2193	cwp24	Cell wall-binding protein, glucosaminidase domain	3.9	4.4	+
CD0844	cwp25	Cell wall-binding protein	3.9	5.9	+
CD3192	cwp21	Cell wall-binding protein	3.9	6.8
CD2789	cwp66	Cell wall-binding protein	3.8	4.7
CD2791	cwp2	Cell wall-binding protein	3.7	6.6	+
CD2793	slpA	Precursor of the S-layer proteins	3.6	4.90	+
CD2713b	ldtcd2, cwp22	Cell wall-binding protein, l,d-transpeptidases	3.6	5.1
CD2794	cwp12	Cell wall-binding protein	3.6	5.6
CD0514	cwpV	HA/adhesin	3.4	4.8	+
CD1987	cwp28	Cell wall-binding protein	3.1	1.5
CD1035	cwp16	Cell wall-binding protein, amidase domain	3.1	5.7
CD1036	cwp17	Cell wall-binding protein, amidase domain	2.9	3.3
CD2796	cwp10	Cell wall-binding protein	2.8	4.8
CD2784	cwp6	Cell wall-binding protein, amidase domain	2.8	3.6	+
CD1047	cwp18	Cell wall-binding protein	2.4	3.5
CD2799	cwp8	Cell wall-binding protein	2.1	4.0
Up in ΔprkC mutant
CD2782	cwp7	Cell wall-binding protein	0.2	2.1
SH3_3 motif (down in ΔprkC mutant)
CD0183b		Putative cell wall hydrolase	10.2	1
CD2768b		Putative cell wall hydrolase	3.8	2.9
CD1135b		Putative endopeptidase	3	4.8
CD2402b		Putative cell wall hydrolase phosphatase-associated protein	2.3	2.6
PG4 motif (up in ΔprkC mutant)
CD1436		Putative hydrolasec	0.2	2.4
CD2963b	ldtcd1	l,d-transpeptidasesc	0.3	2.5
CD2149		Putative vancomycin resistance protein, VanW familyc	0.3	1.1

^aFalse discovery rate, <0.05.

^bCell wall metabolism.

^cProtein containing one transmembrane domain.

^dSH3 domain, PF08239; CWB2 motif, PD04122; PG4 motif, PF12229.

Virulence and colonization of the ΔprkC mutant in the hamster model.

The different phenotypes of the prkC mutant suggest that PrkC can play a role during critical steps of the infectious cycle of C. difficile (4). To determine a possible role of PrkC in CDI, we compared the virulence of the WT strain and the ΔprkC mutant in the acute Golden Syrian hamster model of infection. Despite a trend toward a delay in the death of the hamsters infected with the prkC mutant, no significant difference was observed for the average time of postchallenge survival between hamsters infected with the WT and the ΔprkC mutant strains (Fig. 9A). When we monitored the daily level of gut colonization of the WT and the ΔprkC mutant, we found no significant difference in the average level of gut colonization the day the hamster died, as determined by the average number of CFU per gram of feces for each hamster (Fig. 9C). However, a significant change in gut colonization was observed between the two strains 40 h postinfection (Fig. 9B). Despite the fact that the ΔprkC mutant has a delay in gut colonization, this delay has no significant effect on virulence in this infection model.

FIG 9.

prkC deletion affects gut establishment and not virulence. (A) Average survival time (h) postinfection for hamsters challenged with 630Δerm or ΔprkC spores. (B and C) Average quantity of CFU per g of feces determined 40 h postinfection (B) or the day of animal death (C) for hamsters challenged with 630Δerm or ΔprkC spores. The average values in panel B were calculated with 16 and 12 hamsters for the WT strain and the ΔprkC mutant, respectively, due to the absence of feces at this time point for four hamsters. The average values in panel C were calculated with the totality of hamsters for both groups. Data were analyzed by a Mann-Whitney test. *, P < 0.05. The sensitivity threshold of the method is 10³ g/feces sample.

DISCUSSION

In this work, we showed that the prkC mutant of C. difficile has pleiotropic phenotypes, such as an increased sensitivity to various antimicrobial compounds (CAMPs, lysozyme, DOC, and β-lactams), modifications in motility, cell aggregation, and biofilm formation, and changes in cell morphology and septum formation or localization. The increased sensitivity to several antimicrobial compounds detected for the C. difficile ΔprkC mutant has been observed in other firmicutes. The prkA mutant of Listeria monocytogenes is more sensitive to lysozyme and several CAMPs (44), while the ireK mutant of E. faecalis has an increased sensitivity to bile and cholate (24) but not to DOC, as observed for the C. difficile ΔprkC mutant. The inactivation of the PASTA-STKs in S. pneumoniae, S. pyogenes, E. faecalis, L. monocytogenes, and S. aureus results in an increased susceptibility toward β-lactams, but the extent and pattern of the effects vary among species and strains (18, 24, 44, 45). These specific patterns of sensitivity observed for the PASTA-STK mutant for each firmicute might be associated with differences in the targets phosphorylated by the PASTA-STK, in the penicillin-binding proteins (PBPs), and in their affinity for the different β-lactams. As observed in B. subtilis and S. aureus (18, 21), the PrkC kinase of C. difficile controls biofilm formation under cell membrane stress conditions. Several factors probably contribute to the increased biofilm formation in the ΔprkC mutant in the presence of DOC or polymyxin B. These may include decreased motility, increased sedimentation, and changes in the amount of CWPs that modify adhesive properties of the cell surface. For example, the Cwp84 protein, which is less abundant in the ΔprkC mutant, is known to negatively control biofilm formation (46). In addition, premature autolysis by the ΔprkC mutant probably releases eDNA earlier in the presence of DOC, resulting in the increased biofilm formation observed at 24 h for the mutant (35). By controlling cell lysis, PrkC seems able to affect biofilm formation in response to cell membrane stresses.

Changes in the homeostasis and the integrity of the cell envelope might explain most of the phenotypes of the C. difficile ΔprkC mutant, including modification of cell shape that is determined by the orderly processing and assembly of each cell wall component (37, 47, 48). The absence of PrkC probably modifies the abundance, composition, and/or structure of a component(s) of the cell envelope, including the PG, the glycopolymers (PSII and/or LTA-PSIII), and/or cell wall-associated proteins (42). In other Gram-positive bacteria, PASTA-STK phosphorylates enzymes involved in PG and/or teichoic acid synthesis, modification, assembly, and/or turnover. This includes enzymes of the Glm pathway, Mur enzymes, MviM, a flippase involved in the transport of the lipid II-anchored compounds across the membrane, a penicillin-binding protein, LTA synthetases, and also transcriptional regulators, such as WalR-WalK and GraR, which control cell wall metabolism (17, 18, 49). In C. difficile, we did not detect a transcriptional effect of the prkC deletion on genes encoding enzymes involved in the synthesis of envelope components (E. Cuenot, unpublished data). These results strongly suggest that PrkC does not act by modifying the activity of a transcriptional regulator but rather by phosphorylating one or several proteins directly or indirectly controlling the synthesis, assembly, or turnover of at least one component of the envelope.

Teichoic acids play a role in protecting bacteria from stressful conditions by modifying the properties of the cell surface. Changes in teichoic acids resulted in phenotypes similar to those we observed in the ΔprkC mutant. These include increased β-lactam susceptibility, autolysis, modification of cell morphology, and a defect in septum positioning and numbers (37, 50). In C. difficile, the biosynthesis of PSII, its properties, and the phenotypes associated with defects in PSII have been poorly studied (38, 40). Furthermore, less information is available on the atypical LTA (40). However, we noted that the ΔprkC mutant shares common phenotypes with the lcpB mutant, encoding a protein involved in tethering PSII (38), including elongated cells, the presence of multiple septa, and increased ability to form biofilms (38). While the structure of PSII is not altered in the ΔprkC mutant, we observed a slight reduction in the amount of PSII anchored to the cell wall and an increased quantity of PSII released in the supernatant of the ΔprkC mutant. These changes in PSII localization might be caused by the dysregulation of PSII synthesis/anchorage or the production/assembly of another cell envelope component, such as PG, in the absence of PrkC. Indeed, anchoring of PSII to PG is probably affected by the reduced amount of PG detected in the ΔprkC mutant.

As observed for L. monocytogenes and E. faecalis, C. difficile is inherently resistant to cephalosporin, and PASTA-STK inactivation leads to a robust increase in sensitivity toward these antibiotics that target PG synthesis by inactivating PBPs (24, 51). In C. difficile, we showed that the kinase activity of PrkC is required for the high level of resistance to cephalosporin. The sensitivity of the ΔprkC mutant to teicoplanin, an antibiotic that affects transglycosylation by PBPs, and to bacitracin, which interferes with PG synthesis through its role in lipid II recycling, also suggests a possible role for this kinase in controlling PG metabolism (52). Other phenotypes, such as lysozyme resistance or autolysis, are also related to PG structure or metabolism. However, we failed to detect any modification in the composition of PG, reticulation of PG, 3-3 to 4-3 cross-link ratio, or PG deacetylation under our experimental conditions. Changes to these would offer, to a certain extent, an explanation for the increased sensitivity to cephalosporin and lysozyme. Nevertheless, it is possible that localized and/or subtle changes in PG composition occur at the septum, which is where PrkC is localized during growth. These changes would be difficult to detect. The main difference between the ΔprkC mutant and the WT is a decrease in the total amount of PG. This might be linked to a reduced size of the glycan chains formed. It is interesting that proteins potentially involved in PG metabolism are also detected in different amounts in the cell wall of the WT and ΔprkC mutant strains (Table 2). All these results suggest that PrkC has an effect directly or indirectly on PG synthesis or turnover in C. difficile. In L. monocytogenes, it has been suggested that the PASTA kinase specifically regulates PBPs, resulting in cephalosporin resistance (44). In Enterococcus faecium, mutants in class A PBPs are sensitive to cephalosporins, and suppressor mutations that restore cephalosporin resistance are found in ireK and ireP, encoding an STK and an STP, respectively (53). In E. faecalis, IreB negatively controls cephalosporin resistance, and IreB is a small protein of unknown function and the only substrate of IreK kinase identified to date (54). However, the molecular mechanisms linking IreB, IreK, and maybe PBPs that could explain their role in the control of cephalosporin resistance remain unknown. Interestingly, CD1283, the IreB-like protein of C. difficile, contains one threonine (T7), and this residue is phosphorylated in IreB of E. faecalis. CD1283 might contribute to the regulatory pathway downstream of PrkC. In addition, the PrkC-mediated phosphorylation of proteins controlling cell division and the synthesis of cell envelope components is another interesting possibility. Indeed, PrkC is localized at the septum, and the deletion of the prkC gene affects C. difficile cell morphology, with elongated cells, abnormal septum localization, and defects in cell separation. In S. pneumoniae and other streptococci, ΔstkP mutants have longer cells than the WT and a modified shape and exhibit cell division and separation defects (55). Several proteins involved in cell division, such as DivIVA, MapZ, GspB, and FtsZ, are phosphorylated by the PASTA-STKs in B. subtilis, S. pneumoniae, or S. aureus (17, 18, 55). In S. pneumoniae, phosphorylation of DivIVA controls cell shape and the localization of PG synthesis machinery required for cell elongation and cell constriction (55). The phosphorylation by PrkC of an orthologue of one of these proteins, especially DivIVA, could explain several cell morphology-associated phenotypes observed for the prkC mutant. Altogether, our results indicate that the PASTA-STK of C. difficile, PrkC, is involved in the control of envelope biogenesis and/or cell division.

Finally, the virulence of the ΔprkC mutant is similar to that of the WT strain, while a colonization delay of the hamster gut is observed for the mutant. This can be explained by a global increase in sensitivity of the ΔprkC mutant to antimicrobial compounds and the possible changes in its cell envelope properties. Our results highlight the involvement of PrkC in controlling several processes corresponding to critical steps of CDI, including resistance to DOC, known to inhibit growth of vegetative cells in the gastrointestinal tract (4); to lysozyme, a critical component of the innate immune system; and to CAMPs produced by the microbiota and/or by the host (3), which are compounds present in the hamster model. Furthermore, dissecting the role of PrkC in controlling the resistance to antibiotics promoting CDI such as cephalosporins (1) could pave the way for new strategies for the prevention of these infections.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

C. difficile strains and plasmids used in this study are listed in Table S1 in the supplemental material. C. difficile strains were grown anaerobically (5% H₂, 5% CO₂, 90% N₂) in TY broth (30 g·liter⁻¹ Bacto tryptone, 20 g·liter⁻¹ yeast extract, pH 7.4), in brain heart infusion (BHI) broth, or in a peptone-containing medium (Pep-M) (56). Sporulation medium (SM) (57) was used for sporulation assays, and spores were produced in SMC medium (58). Solid media were obtained by adding agar to a final concentration of 17 g·liter⁻¹. When necessary, cefoxitin (Cfx; 25 μg·ml⁻¹) or thiamphenicol (Tm; 15 μg·ml⁻¹) was added. E. coli strains were grown in LB broth. When indicated, ampicillin (100 μg·ml⁻¹) or chloramphenicol (15 μg·ml⁻¹) was added. Anhydrotetracycline (ATc) was used to induce expression of the prkC gene from the P_tet promoter of pDIA6103 (59).

Construction of plasmids and strains.

The allelic exchange cassette used to delete the CD2578 gene (prkC) was constructed in a derivative of pMTLSC7315 (27) lacking the multicloning site. DNA sequences flanking the prkC gene (1,038 and 1,117 bp) were amplified by PCR using primers IMV854 and IMV855 or IMV856 and IMV857 (Table S2 and Fig. S1). A DNA fragment obtained by overlapping PCR was digested by XhoI and BamHI and then ligated into pMTLSC7315ΔMCS, resulting in pDIA6401. E. coli HB101 (RP4) containing pDIA6401 was mated with the C. difficile 630Δerm strain. Transconjugants were selected on BHI-Cfx-Tm plates. Colonies were restreaked onto the same medium to identify faster-growing single-crossover integrants that were then streaked onto C. difficile minimal medium supplemented with 50 μg/ml fluorocytosine (FC) to select for second crossover events. FC^r Tm^s clones that have lost pDIA6401 were analyzed by PCR using IMV867 and IMV868 to distinguish between wild-type and ΔprkC clones (Fig. S1). For complementation, the prkC gene (positions −22 to +2205 from the translational start site) was amplified using primers IMV879 and IMV880. The PCR fragment was cloned between the StuI and BamHI sites of pDIA6103 (25) to produce pDIA6413. To construct a plasmid expressing PrkC-HA (pDIA6713), a reverse PCR was performed using pDIA6413 as the template and primers EC01 and EC02. The same strategy was used to introduce a point mutation into the prkC gene (lysine at position 39 replaced by an alanine [K39→A]). The resulting plasmid, pDIA6714, was then used to construct pDIA6716 expressing HA-tagged prkC(K39→A). These plasmids were transferred into C. difficile strains by conjugation (Table S1).

To construct a translational SNAP-PrkC fusion, we amplified the SNAP coding sequence fused to a linker in 3′ orientation (GGATCCGCAGCTGCT) using pFT58 as a template and primers EC238 and EC239. pDIA6413 (pDIA6103-prkC) was amplified by inverse PCR using the primers EC236 and EC237. Using these two fragments, pDIA6855 was obtained by Gibson Assembly. pDIA6855 was conjugated in the 630Δerm strain to obtain CDIP1357 (Table S1).

To determine if prkC is cotranscribed with upstream genes (CD2579 and rlmN), we carried out an RT-PCR experiment using RNA extracted from exponentially growing cells. After cDNA synthesis using IMV843, a primer targeting the prkC gene, PCR amplification was tested with pairs of primers located in adjacent genes (prkC-CD2579 or CD2579-rlmN). Nontreated RNAs were used as a negative control.

Antimicrobial sensitivity tests.

Cultures of C. difficile strains (OD₆₀₀ of 0.3) were plated on BHI agar plates. Six-mm paper disks containing antibiotics were placed onto the agar surface. Lysozyme (800 μg) was also placed on a 6-mm paper disk using Pep-M agar plates. The growth inhibition diameter was measured after 24 h of incubation at 37°C. MICs were determined on BHI plates by Etest (bioMérieux) after 24 h of incubation at 37°C. To test susceptibility for antimicrobial peptides, protocols were modified from previously published works (34, 60). Each well of a 24-well microplate was inoculated with 1 ml of a BHI bacterial inoculum (OD₆₀₀ of 0.01). After addition of bacitracin (Sigma-Aldrich), polymyxin B (Sigma-Aldrich), or nisin (Sigma-Aldrich), microplates were incubated at 37°C for 20 h. The MIC was defined as the lowest concentration of antimicrobial peptide preventing growth.

Five μl of an overnight culture of C. difficile strains was streaked on BHI plates without or with SDS (0.005% to 0.009%). The plates were incubated at 37°C for 48 h. Deoxycholate (DOC) resistance tests were performed in 24-well microplates. Each well was inoculated with 1 ml of TY with or without DOC (0.03%). The OD₆₀₀ of each culture was monitored. For autolysis, cells (OD₆₀₀ of 1) were resuspended in 50 mM potassium phosphate buffer (pH 7.0) containing 0.01% Triton X-100 and incubated at 37°C. The OD₆₀₀ was determined every 5 min.

Sporulation, germination, motility, and biofilm formation assays.

Sporulation assays were performed as described previously (61). To purify spores, 100 μl of culture plated on solid SMC was grown at 37°C for 7 days. Spores were scraped off with water and then incubated for 7 days at 4°C to allow the release of spores from the cells. Spores were purified by centrifugation using a HistoDenz (Sigma-Aldrich) gradient as described previously (58). For the germination efficiency tests, we monitored the OD₆₀₀ of purified spores incubated under anaerobiosis in BHI supplemented with 0.5% taurocholate with bryostatin or muropeptides.

To measure motility, plates containing 25 ml of BHI agar (0.3%, wt/vol), Tm, and ATc were inoculated with 5 μl of an exponential-phase culture. Plates were incubated for 48 h at room temperature, and the zone of motility was then measured. For the biofilm assay, 1 ml of BHIS medium (BHI, 5 g/liter yeast extract, cysteine 0.1%) containing 0.1 M glucose and polymyxin B (20 μg·ml⁻¹) or DOC (0.01%) was inoculated and deposited in a well of a 24-well microplate. Microplates were incubated at 37°C. The biofilm was washed with PBS (phosphate-buffered saline), stained with 1 ml of crystal violet (0.2%), and washed twice with PBS. The OD₅₇₀ was measured after resuspension of the cells in 80% methanol–20% acetone using noninoculated medium as a negative control. For dispersion of preformed biofilms, 24-h and 48-h biofilms were treated with DNase I (100 μg/ml in 150 mM NaCl, 1 mM CaCl₂) under anaerobic conditions at 37°C for 1 h (35). Control wells were treated with buffer without DNase. Biofilms were then washed, stained, and quantified as described above.

Phase-contrast and transmission electron microscopy.

C. difficile strains were grown for 5 h in TY medium. For phase-contrast microscopy, cells were analyzed using an Axioskop microscope (Zeiss). For TEM, cultures were mixed with 1 volume of a fixative solution containing 4% paraformaldehyde (PFA) and 4% glutaraldehyde (GA) in 1× PHEM buffer, pH 7.3 [60 mM piperazine-N,N′-bis(2-ethanesulfonic acid), 25 mM HEPES, 10 mM EGTA, 2 mM MgCl₂], and then incubated for 30 min at room temperature. After centrifugation, pellets were resuspended in a second fixative solution (2% PFA and 2% GA in PHEM buffer, pH 7.3) and incubated for 1 h. Samples were then washed twice in PBS prior to being subjected to a high-pressure (>2.0 × 10⁸ Pa) freezing in 1-hexadecane using a BAL-TEC HPM 010 (Leica). Freeze substitution was done with 2% OsO₄ in acetone, followed by several steps: −90°C for 42 h, warmed to −30°C (5°C/h), incubated for 12 h, warmed to 0°C (10°C/h), and incubated for 1 h. Samples were then washed with acetone on ice and incubated in Epon-acetone at different volume ratios (1:3 for 3 h, 1:1 for 3 h, 2:1 overnight, 3:1 for 4 h), followed by an incubation in pure Epon (2 h, overnight, 6 h). Samples were then incubated overnight in Epon and the hardener benzyldimethylamine (BDMA) prior to polymerization at 60°C for 48 h. Sections (60 to 70 nM) were obtained on an FC6/UC6 ultramicrotome (Leica) and transferred onto 200-mesh square copper grids coated with Formvar and carbon (CF-200-Cu50; Delta Microscopy). Samples were stained with 4% uranyl acetate and counterstained with lead citrate. Images were recorded with a Tecnai Spirit 120 kV (with a bottom-mounted Eagle 4K×4K camera).

SNAP labeling, fluorescence microscopy, and image analysis.

For membrane and chromosome staining, 500 μl of exponential-phase cells was centrifuged and resuspended in 100 μl of PBS containing the fluorescent membrane dye FM4-64 (1 μg/ml; Molecular Probes, Invitrogen) and the DNA stain DAPI (2 μg/ml; Sigma). Samples were incubated for 2 min in the dark and mounted on a 1.2% agarose pad. Strain CDIP1357 was grown for 2 h in TY, and expression of the SNAP^Cd-prkC fusion was induced with 50 ng/ml of ATc for 2 h. For SNAP labeling, the TMR-Star substrate (New England Biolabs) was added at 250 nM, and the mixture was incubated for 30 min in the dark under anaerobiosis. Cells were then collected by centrifugation, washed, and resuspended in PBS. Cell suspension (3 μl) was mounted on 1.7% agarose-coated glass slides. The images were taken with 300-ms exposure times for the autofluorescence and 900-ms exposure times for the SNAP using a Nikon Eclipse TI-E microscope 100× lens objective and captured with a CoolSNAP HQ2 camera. The images were analyzed using ImageJ (62).

Detection of phosphothreonine by Western blotting on soluble and insoluble fractions.

Cells were grown for 6 h in TY and then harvested by centrifugation. The pellets were resuspended in PBS containing protease and phosphatase inhibitor cocktails (Sigma-Aldrich) and 0.12 μg/ml of DNase. Cells were lysed for 45 min at 37°C and then centrifuged. The soluble fraction was diluted twice with 2× sample buffer (150 mM Tris-HCl, pH 6.8, 30% glycerol, 1.5% SDS, 15% β-mercaptoethanol, 2 μg/ml of bromophenol blue). After 2 washes with PBS, the insoluble fraction was resuspended in 2× PBS containing 1% SDS and mixed with 2× sample buffer. Western blots were performed using an anti-P-Thr primary antibody (Cell Signaling), followed by a goat anti-rabbit horseradish peroxidase-conjugated secondary antibody (Sigma-Aldrich), and developed using the SuperSignalWest Femto chemiluminescent kit (Thermo Scientific).

PG and PSII analysis.

Peptidoglycan (PG) samples were prepared from 1.5 liters of C. difficile grown in BHI (OD₆₀₀ of 1) as previously described (36). Purified PG was then digested with mutanolysin (Sigma-Aldrich), and the soluble muropeptides were separated by reverse-phase high-performance liquid chromatography (RP-HPLC) (36). PSII was extracted from C. difficile grown in BHI (OD₆₀₀ of 1) as previously described (40). PSII was lyophilized and its structure was checked by ¹H and ¹³C NMR. Spectra were acquired on a 400-MHz Bruker spectrometer equipped with a Prodigy probe. PSII was lyophilized and its structure was determined by ¹H NMR. For the quantification of PG and PSII, 500 ml of culture (OD₆₀₀ of 1) was used to purify PSII covalently linked to PG (PG-PSII). PG-PSII was purified as previously described (63), without the acetone treatment. Linkage between PG and PSII was disrupted after an incubation for 48 h at 4°C in hydrofluoric acid (48%). The pellet containing PG was washed three times in H₂O, and the hydrofluoric acid supernatant that contains PSII subunits was evaporated and resuspended in H₂O. Both PG and PSII were lyophilized and then weighed.

For the PSII dot blot, exponential-phase cultures were harvested by centrifugation. Supernatant and total crude cell extracts were kept as separate fractions. The supernatant fraction was recovered and precipitated with 10% trichloroacetic acid (TCA) for 30 min. The supernatant and the total crude cell fractions were treated with 100 μg/ml of proteinase K (Sigma) for 1 h at 37°C. Samples were then serially diluted, and 5 μl of each dilution was spotted onto an activated polyvinylidene difluoride membrane. The membrane was washed in H₂O, blocked for 15 min in TBST (20 mM Tris-HCl, 150 mM NaCl, 0.05% Tween 20, pH 7.5) containing 10% milk, and then washed in 5% milk in TBST for 2 min. After overnight incubation in PSII-LTB rabbit antiserum (1:8,000) (38), the membrane was washed once in TBST with 5% milk, twice in TBST for 5 min, and once in TBST with 5% milk for 10 min. Following incubation with goat anti-rabbit horseradish peroxidase-conjugated secondary antibody at 1:10,000 dilution for 1 h, the membrane was washed 5 times in TBST for 5 min and revealed using the SuperSignal West Femto chemiluminescent substrate.

Isolation of cell wall proteins and proteomic analysis.

C. difficile strains were grown for 6 h in TY at 37°C, and 20 ml of each culture was then centrifuged. The cell pellets were washed with PBS, resuspended (OD₆₀₀ of 100) in 75 mM Tris-HCl, pH 6.8, 15% glycerol, 7.5% β-mercaptoethanol, 0.75% SDS, and boiled for 10 min at 100°C. Proteins were precipitated with 10% TCA, washed with 90% cold acetone, dried, resuspended in 8 M urea–100 mM NH₄HCO₃, and sonicated. Total protein extracts (50 μg) were reduced with Tris(2-carboxyethyl)phosphine hydrochloride (TCEP; 10 mM) (Sigma) for 30 min and alkylated with iodoacetamide 20 mM (Sigma) for 1 h. Proteins were digested with 1 μg rLys-C (Promega) for 4 h at 37°C and with 1 μg trypsin (Promega) overnight at 37°C. The digestion was stopped with 4% formic acid (FA). Peptides were desalted on C₁₈ Sep-Pak cartridges (WAT054955; Waters) and eluted with 50% acetonitrile (ACN)–0.1% FA and then 80% ACN–0.1% FA before being dried in a vacuum centrifuge. Peptides were resuspended with 2% ACN–0.1% FA. A nanochromatographic system (Proxeon EASY-nLC 1000; Thermo Fisher Scientific) was coupled on-line to a Q Exactive HF mass spectrometer (Thermo Fisher Scientific). Peptides (1 μg) were injected onto a 47-cm C₁₈ column (1.9 μm particles, 100-Å pore size; ReproSil-Pur Basic C₁₈; Dr. Maisch GmbH) and separated with a gradient from 2% to 45% ACN at a flow rate of 250 nl/min over 132 min. Column temperature was set to 60°C. Data were acquired as previously described (64). Raw data were analyzed using MaxQuant software, version 1.5.1.2 (65), using the Andromeda search engine (66). The MS/MS spectra were searched as previously described (64) against an internal C. difficile database containing 3,957 proteins and the contaminant file included in MaxQuant. The statistical analysis was performed with Perseus 1.5.2.6 (67) as previously described (64). Missing values for label-free quantification (LFQ) intensities were imputed and replaced by random LFQ intensities that are drawn from a normal distribution at the low detection level. Statistical significance was assessed with a two-sided t test of the log₂-transformed LFQ intensities with a permutation-based false discovery rate calculation at 5% and S0 = 2 (68). Differentially regulated proteins are visualized on a VolcanoPlot.

Golden Syrian hamster infections.

Golden Syrian hamsters were first treated with a single oral dose of 50 mg/kg of body weight of clindamycin. Five days after the antibiotic treatment, the hamsters were infected by gavage with 5,000 spores of either the 630Δerm strain or the ΔprkC mutant. Spore inocula were standardized before challenge. Eight animals per strain per experiment were used for the infection, and we performed two independent experiments. Colonization was followed by enumeration of C. difficile cells in feces samples and was started 2 days postinfection and each day until the death of the animal. Briefly, feces were resuspended in PBS at 10 mg·ml⁻¹ and serially diluted with PBS before plating on BHI supplemented with 3% defibrinated horse blood and the C. difficile selective supplement containing cycloserine (25 μg/ml) plus cefoxitin (8 μg/ml). All animal experiments were conducted according to the European Union guidelines for the handling of laboratory animals, and procedures for infection, euthanasia, and specimen collection were approved by the Central Animal Care Facilities and Use Committee of University Paris-Sud (agreement 92-019-01; protocol number 2012-107).

Data availability.

The mass spectrometry proteomics data have been deposited with the ProteomeXchange Consortium via the PRIDE (69) partner repository with the data set identifier PXD012241.