The C. difficile strains were cultured in BHIS, TY, or TYC (TY with 5 mM cysteine) broth in an anaerobic chamber (ShelLab, USA). The following antibiotics were added to the medium as necessary, at the indicated concentrations: d-cycloserine, cefoxitin, lincomycin, or thiamphenicol (Heap et al., 2007). All materials used for C. difficile culture were pre-reduced in the anaerobic chamber for more than 2 h to remove oxygen. The E. coli strains were grown aerobically at 37°C in Luria–Bertani medium supplemented with the appropriate antibiotics. The bacterial strains and plasmids used in this study are presented in Table 1.

The C. difficile mutants were generated with the method of the ClosTron gene knock-out system, as described previously (Heap et al., 2007), combined with the TargeTron® Gene Knockout System Kit (Sigma-Aldrich). The intron DNA fragments were amplified with overlap PCR using specific retargeted primers (shown in Table S1) and then cloned into the HindIII and BsrGI restriction sites of pMTL007, using E. coli TOP10 as the host cells. After confirmation with PCR and DNA sequencing, the derived pMTL007 plasmids were each used to transform conjugative E. coli CA434, and then transferred into C. difficile 630Δerm by conjugation. The C. difficile transconjugants were selected in the presence of thiamphenicol, erythromycin, d-cycloserine, and cefoxitin. Once colonies appeared on the plates, the mutants were verified with PCR screening using the primers shown in Table S1. Southern blotting was performed with a DIG-High Prime Labeling and Detection Kit (Roche), according to the manufacturer's instructions. For the complementation experiments, each target gene, together with its native constitutive promoter, was cloned into pMTL84151. Using E. coli CA434 as the donor, the complementary plasmids were transferred individually into the C. difficile 630Δerm mutant strains to generate the strains used in this study (Table 1).

To isolate RNA, 3 ml aliquots of cultures grown in TY or TYC broth were harvested by centrifugation (4,000 × g for 10 min at 4°C). The total RNA was extracted from the cell pellets with the RNAprep pure Kit (for Cell/Bacteria) (TIANGEN). The PrimeScript RT Reagent Kit with gDNA Eraser (TaKaRa) was used to synthesize the cDNA, according to the manufacturer's instructions. The synthesized cDNA was stored at −20°C.

The relative expression levels of the target transcripts were determined with SYBR Premix Ex Taq (TaKaRa), according to the manufacturer's protocol. The primers used for real-time PCR in this study are presented in Table S1. The data were analyzed with the ΔC_T method as previously described (Schmittgen and Livak, 2008). The expression levels of the genes were normalized using the amplification efficiencies and the expression levels of the reference gene rpsJ. At least three biological replicates were assayed. The statistical analysis was performed with two-way analysis of variance (ANOVA) and a P-value ≤ 0.05 was considered significant.

Zymography was used to detect C-S-lyase activities as described previously (Auger et al., 2005) with some modifications. Crude extracts of the native proteins harvested from TY or TYC broth were run on a nondenaturing gel (12% polyacrylamide in Tris–glycine buffer). After electrophoresis, the gel was washed twice with 50 ml of Tris/HCl (50 mM, pH 7.4) and then incubated at 37°C for 1–4 h under anaerobic conditions in the following solution: 50 mM Tris/HCl (pH 7.4), 10 mM MgCl₂, 0.5 mM Pb(NO₃)₂, 0.4 mM PLP, 5 mM dithiothreitol and 10 mM cysteine. The H₂S formed during the enzymatic reaction precipitated as insoluble PbS, so the C-S-lyase activity was measured as the amount of precipitated PbS.

Cells were grown in TY or TYC medium for 12 h and H₂S production was then detected with Hydrogen Sulfide Test Strips (Sigma), which turned black in the presence of H₂S.

To assay cytotoxicity, cultures of the C. difficile strains grown anaerobically in TY or TYC broth were centrifuged (4,000 × g for 10 min at 4°C), and filter sterilized. The filter-sterilized supernatants were then serially diluted twofold or 4-fold and added to monolayers of Vero cells preincubated in 96-well plates. Cytotoxicity was recorded after 24 h. The negative control was treated with fresh medium. The end-point titer was defined as the first dilution in the series in which the morphology of the Vero cells was the same as that of the negative control. For the neutralization assay, the diluted supernatants from 24 h cultures were incubated with appropriately diluted anti-TcdA or anti-TcdB serum for 1 h at 37°C and then added to Vero cell monolayers. Each experiment was performed three times in duplicate. The statistical analysis was performed with two-way ANOVA with multiple-comparisons test, and P ≤ 0.05 was considered statistically significant.

Vero cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Gibco) and 1% (v/v) penicillin–streptomycin at 37°C in a humidified atmosphere containing 5% CO₂. For the cytotoxicity assay, Vero cells were seeded into 96-well plates at densities of approximately 5 × 10⁴ cells/ml and incubated for 20–24 h before the assay.

The supernatants from C. difficile cultures grown anaerobically in TY broth were concentrated with 10-kDa Amicon® Ultra-4 Centrifugal Filter Units (Merck Millipore), and the cell densities were standardized. The samples were spotted onto a nitrocellulose membrane and air-dried. The membranes were blocked with 5% nonfat dry milk in phosphate-buffered saline (PBS) containing 0.05% Tween 20 for 1 h, and incubated with mouse anti-TcdA serum (maintained in our laboratory) and then with a horseradish-peroxidase conjugated anti-mouse secondary antibody. The immunological spots were visualized with the SuperSignal West Pico Chemiluminescent Substrate Kit (Thermo Scientific), according to the manufacturer's instructions.

The extraction and purification of total RNA were performed as described above. Transcriptional start site (TSS) of cdsB was determined using the SMARTer™ RACE cDNA Amplification Kit (Clontech) according to manufacturer instructions. Gene-specific primers and universal primer mix were used to amplify the 5′ end of cdsB mRNA (Table S1).

To generate the cdsB promoter-reporter gene fusions, regions of various lengths upstream of the cdsB promoters were PCR-amplified from strain 630Δerm genomic DNA with primers listed in Table S1. These products were independently ligated into the KpnI/SacI sites of pRPF185, which contains a C. difficile codon-optimized gusA gene (Fagan and Fairweather, 2011), to generate the plasmids listed in Table 1. The recombinant plasmids were confirmed by sequencing and then transferred into 630Δerm, the sigL::ermB or the cdsR::ermB strains, which were selected by resistance to thiamphenicol and verified by PCR.

The C. difficile strains containing the promoter–reporter gene fusions listed in Table 1 were grown to late exponential phase (OD₆₀₀, 1.5), and the cells were harvested and stored at −20°C. The samples were then lysed and analyzed for β-glucuronidase activity, as previously described (Fagan and Fairweather, 2011). Briefly, the pellets stored at −20°C were resuspended in 0.5 ml of PBS and incubated at 37°C for 40 min to generate whole-cell lysates. The enzymatic reaction was initiated by the addition of 100 μl of a 6 mM solution of p-nitrophenyl-β-d-glucuronide. After incubation at 37°C, the reactions were stopped by the addition of 240 μl of 1 M Na₂CO₃. The tubes were centrifuged at 12,000 × g for 5 min to remove the cell debris and OD₄₀₅ was measured with a MD SpectraMax M5 spectrophotometer. Specific activity was calculated with the formula: (OD₄₀₅ × 1,000)/(OD₆₀₀ × t [min] × 1.25 × volume [ml]). The results are presented as the means of the calculated activity and the standard errors of the means of at least three replicates. The data were analyzed with two-way ANOVA with Dunnett's multiple-comparisons test.

A cysteine desulfidase, CdsB, has been described as catalyzing the hydrolysis of cysteine to sulfide, ammonia, and pyruvate in M. jannaschii (Tchong et al., 2005). When we searched the genome of C. difficile 630, we identified the cdsB homologous gene CD630_32320. A phylogenetic analysis was performed with the MEGA software using the deduced amino acid sequences of the CdsB proteins derived from the genomes of C. difficile 630 and 14 other bacterial strains. This showed that the CdsB protein is conserved in many facultatively anaerobic and anaerobic bacteria, but not in aerobic bacteria, and that it has evolved into two clearly demarcated branches, the enbacterial branch and the archeal branch (Figure 1A). CdsB homologs were also found widely distributed in the genomes of the genus Clostridium. Tchong et al. identified four conserved cysteine residues (C25, C282, C322, and C329) in the CdsB protein in M. jannaschii. Three of them (C282, C322, and C329) are ligands for the [4Fe–4S] center, and the thiolate of Cys25 acts as a base to abstract the R-hydrogen in the first step of elimination (Tchong et al., 2005). Based on a multiple-sequence alignment of the CdsB proteins using ClustalW, the CdsB protein encoded in C. difficile contains all four cysteine residues that are conserved in specific positions of the l-cysteine desulfidase of M. jannaschii (Figure 1B). As well as the conserved cysteine residues, an additional 25 amino acids reported in the CdsB protein of M. jannaschii are conserved in C. difficile. Among them, nine amino acid residues (K49, H139, R255, S269, E292, S301, D362, K371, and D422) that are considered to be involved in the catalytic mechanism or substrate binding in M. jannaschii CdsB also occur in C. difficile CdsB (Figure 1B). On the basis of these conserved amino acids and the phylogenetic analysis results shown above, we speculated that CdsB has similar functions in C. difficile as in M. jannaschii and other organisms, and is probably a cysteine desulfidase involved in bacterial virulence.

To explore the role of CdsB in C. difficile growth and toxin production during supplementation with cysteine, the expression of cdsB was first analyzed with qRT-PCR during different growth phases of C. difficile 630Δerm in TY or TYC broth (Figure 2). In TY broth, the cdsB gene was mainly expressed during the late exponential phase (10 h) and its expression during the early and mid- exponential phases (3 and 6 h, respectively) was lower. The trend in the transcription of cdsB was quite different when the cells were cultured with cysteine. The cdsB expression was much higher than in cells grown without cysteine during both the early and mid-exponential phases, but decreased during the late exponential phases of growth. These results indicate that transcription of cdsB is induced early in growth by the presence of cysteine. The reduced expression of cdsB at the end of the exponential growth phase in TYC might be attributable to the degradation of cysteine and its consequent depletion.

To explore the role of CdsB in C. difficile, we inactivated the cdsB gene in C. difficile 630Δerm with the ClosTron technology. The insertion of a group II intron into the cdsB gene (insertion site at 815s) was confirmed with PCR amplification and Southern blotting (Figure S1). To generate the complemented strain, the vector pMTL84151-cdsB was transferred into strain C. difficile 630Δerm cdsB::ermB to produce strain cdsB::ermB/pMTL-cdsB. To exclude the effects of the heterologous plasmid on the strains, the pMTL84151 vector was transferred into strains C. difficile 630Δerm and C. difficile 630Δerm cdsB::ermB to generate the parental strain WT/pMTL84151 and strain cdsB::ermB/pMTL84151.

We performed zymography to compare the C-S-lyase activities in the C. difficile strains and to investigate the role of CdsB in cysteine degradation in C. difficile. The C-S-lyase activities in crude extracts of the parental WT/pMTL84151 strain, the cdsB::ermB/pMTL84151 strain, and the cdsB::ermB/ pMTL-cdsB strain cultured in TY or TYC broth were monitored directly on a native gel with incubation in the presence of cysteine and Pb(NO₃)₂. Three bands (α, β, and γ) were detected in the crude extracts of the WT/pMTL84151 strain, which is consistent with a previous study (Dubois et al., 2016). The intensity of the γ band was clearly increased in the presence of cysteine (Figure 3A, lane 1). However, unlike the parental WT/pMTL84151 strain, the γ band was absent from crude extracts of the cdsB::ermB/pMTL84151 strain grown in either TY or TYC broth, whereas the γ band reappeared in the complemented strain cdsB::ermB/pMTL-cdsB (Figure 3A, lanes 2 and 3). This result suggests that the enzyme in the γ band was a CdsB, and was a cysteine-induced enzyme, clearly induced in the presence of cysteine. Interestingly, the expression of the CdsB protein corresponding to the γ band was strongly reduced in the sigL::ermB/ pMTL84151 strain and restored in the sigL::ermB/pMTL-sigL strain, suggesting that the expression of CdsB is regulated by SigL. The expression of the enzyme in the β band was relatively steady, with or without the addition of cysteine, suggesting that it is a constitutively expressed cysteine-independent enzyme involved in cysteine degradation in C. difficile. A previous study indicated that the enzyme corresponding to α band is MalY-dependent, and that in the presence of cysteine, the expression of MalY was reduced in the sigL mutant (Dubois et al., 2016). In the present study, all the strains expressed a small amount of protein corresponding to α band when cultured without cysteine, which probably reflected the constitutive expression of MalY. However, when cultured with cysteine, the expression of MalY decreased in strain sigL::ermB/pMTL84151, which is consistent with a previous study (Dubois et al., 2016), suggesting that expression of malY is possibly regulated by SigL. The expression of MalY in strain cdsB::ermB/pMTL84151 increased, which is probably due to the disruption of a major cysteine-inducible cysteine desulfidase gene (cdsB) and the expression of other proteins that with C-S-lyase activities may be affected by the presence of cysteine. According to these results, the expression of MalY might be affected by cysteine, SigL, and CdsB, the regulatory mechanisms for this phenomenon remain to be identified.

We also showed that the production of H₂S as a result of cysteine degradation was strongly reduced in the cdsB::ermB/pMTL84151 strain, compared with that in the parental WT/pMTL84151 strain, but was restored in the complemented cdsB::ermB/pMTL-cdsB strain (Figure 3B). The H₂S production by the sigL::ermB/pMTL84151 strain and sigL::ermB/pMTL-sigL strain, detected with H₂S test strips, was consistent with the results of a previous study (Dubois et al., 2016), which showed that the inactivation of sigL strongly reduced the amount of H₂S generated in the presence of cysteine. It has been reported that cells of the strain 630Δerm turn black when grown in TYC broth (Dubois et al., 2016), which we also observed in the parental WT/pMTL84151 strain, the complemented cdsB::ermB/pMTL-cdsB strain, and the complemented sigL::ermB/pMTL-sigL strain, but not in the cdsB::ermB/pMTL84151 strain or the sigL::ermB/pMTL84151 strain (Figure 3C). The black deposit in the anaerobic waste collection system was probably an iron-sulfide precipitate resulting from the production of high levels of H₂S during cysteine degradation (Nielsen et al., 2008). Therefore, the inactivation of the cdsB gene in C. difficile results not only in changes in the expression of the active C-S-lyases, but also in lower H₂S production in the presence of cysteine. This suggests that CdsB in C. difficile is an important cysteine-inducible cysteine-degrading enzyme.

The toxin levels in the supernatant of strain cdsB::ermB/pMTL84151 were compared with those in the parental WT/pMTL84151 with cell cytotoxicity assay. The results are presented in Figure 4A and Figure S2. In the parental WT/pMTL84151 strain, the toxin levels were significantly lower in TYC broth than in TY broth. However, in the cdsB::ermB/pMTL84151 strain, no obvious difference in toxin levels was detected in TY and TYC broth. The level of toxin production in the complemented cdsB::ermB/pMTL-cdsB strain was similar to that in the parental WT/pMTL84151 strain and was repressed by the addition of cysteine. TcdA in the culture supernatants of these strains were also analyzed with dot blotting. As shown in Figure 4B, TcdA in the supernatants of the parental WT/pMTL84151 strain and the complemented cdsB::ermB/pMTL-cdsB strain disappeared in TYC broth, whereas that in the supernatant of strain cdsB::ermB/pMTL84151 was similar regardless of whether it was grown in TYC or TY broth. The mRNA levels of tcdA and tcdB in the C. difficile strains cultured in the presence of cysteine were detected with qRT-PCR, as shown in Figure 4C. The transcription levels of both tcdA and tcdB were higher in strain cdsB::ermB/pMTL84151 than in the parental or complemented strain. These results clearly demonstrate that the repression of toxin synthesis by cysteine was abolished in strain cdsB::ermB/pMTL84151, indicating that the cysteine-dependent repression of toxin expression is closely related to the cysteine degradation.

As the cysteine-dependent repression of toxin expression is closely related to the cysteine degradation, which is similar to a previous study's view that it is the accumulation of cysteine by-products rather than the cysteine itself that involved in the repression of toxin expression (Dubois et al., 2016), we also analyzed the toxin expression in the strains grown in TY broth supplemented with either 10 mM cysteine or 10 mM cysteine by-product (Na₂S, pyruvate, or NH₄Cl). As shown in Figure 5, the toxin level with the addition of Na₂S, as well as with the addition of cysteine, was significantly reduced in the strain WT/pMTL84151. However, there were no significant differences on the toxin levels between the WT/pMTL84151 strain, the cdsB::ermB/pMTL84151 strain, and the cdsB::ermB/pMTL-cdsB strain in the presence of Na₂S, which suggested that the inactivation of cdsB couldn't abolish the Na₂S-dependent repression of toxin synthesis. A previous study showed that the transcripts of the tcdA and tcdB genes were downregulated by pyruvate in 630Δerm, and suggested that the effect of pyruvate on toxin gene expression was partly mediated by the two-component system CD2602–CD2601 (Dubois et al., 2016). In this study, the toxin levels detected at the protein level were also slightly reduced in the strain WT/pMTL84151 in response to pyruvate. The toxin levels in the presence of pyruvate showed no significant differences among the three strains, suggesting that cdsB gene is irrelevant to the repression of toxin production by pyruvate. The addition of NH₄Cl in the medium has no effects on toxin levels of the three strains. These results indicate that similar to cysteine, excess Na₂S can significantly repress toxin expression in C. difficile, but the repression cannot be abolished by the inactivation of cdsB, which provide a reasonable proof for the view that the cysteine-dependent toxin repression is probably related to the accumulation of cysteine by-products (Dubois et al., 2016).

To assess the effect of cdsB disruption on C. difficile growth, growth curves of these strains in TY and TYC broth were constructed based on their OD₆₀₀ values. Single colonies of the strains were picked and cultured in BHIS broth for 16 h, and then subcultured in TY broth or TYC broth. The growth of the cdsB::ermB/pMTL84151 strain was similar to that of the parental strain in TY broth. However, when supplemented with 5 mM cysteine, the cdsB::ermB/pMTL84151 strain showed a growth defect relative to the parental strain or the cdsB::ermB/pMTL-cdsB strain (Figure 6). No obvious difference in growth was observed between the parental WT/pMTL84151 strain and the cdsB::ermB/pMTL-cdsB strain when the cells were cultured in TY or TYC broth. These results indicate that the inactivation of cdsB increased the sensitivity of C. difficile to cysteine.

Based on the results described above and in a previous report (Dubois et al., 2016), we inferred that the cdsB gene is probably controlled by δ⁵⁴. We searched upstream from the cdsB gene and found a consensus sequence of promoters recognized by δ⁵⁴ (TTGGCACG-N4-TTGC) (Nie et al., 2015) in the region upstream from the start codon of the cdsB gene (Figure 7A). To confirm this, a recombinant plasmid consisting of the cdsB promoter fused to the gusA reporter gene was constructed and the β-glucuronidase activities were measured in C. difficile strains. The activities of the reporter fusions were assessed in the C. difficile 630Δerm strain and the sigL::ermB strain (Figure 7B). As predicted, this region contained the promoter elements necessary to support transcription, as indicated by the β-glucuronidase activity expressed. In C. difficile 630Δerm, higher β-glucuronidase activity was observed in the presence of cysteine than without cysteine, which is consistent with the zymographic results and qPCR described above, indicating that CdsB expression is induced by cysteine. The activity of the cdsB promoter was significantly reduced in the sigL::ermB strain cultured with TYC, which suggests that δ⁵⁴ is required for cysteine-dependent expression from the cdsB promoter. The effect of Na₂S addition on the β-glucuronidase activity was also compared between the strains (Figure 7B), and the results indicated that it was the cysteine but not cysteine by-products that induced the expression of cdsB. To determine the TSSs of cdsB, 5′-RACE analysis was further performed as described in the materials and methods. An A residue located 63 bp upstream from the cdsB start codon was identified (Figure 7A). Therefore, these results strongly suggested that cdsB is transcribed by δ⁵⁴ associated with the core enzyme of the RNA-polymerase.

Unlike δ⁷⁰, the δ⁵⁴-RNA polymerase forms a stable closed complex at the promoter site, and the conversion to an open complex absolutely requires the input of ATP from an associated activator (enhancer binding proteins, EBP⁵⁴s), which binds to the upstream activating sequence (UAS) remotely upstream from the transcription start site (Francke et al., 2011). These EBP⁵⁴s usually consist of an N-terminal signal-recognition domain, a central activator domain, which provides the ATPase activity, and a C-terminal DNA-binding domain, and are often encoded adjacent to the controlled genes (Francke et al., 2011). Interestingly, the CD630_32330 gene (renamed as cdsR in this study) upstream from the cdsB gene encodes a transcriptional regulator of 591 amino acids, which contains PAS, AAA-type ATPase, and DNA-binding Fis domains, and shares strong similarities with the EBP⁵⁴s. The highly conserved “GAFTGA” amino acid sequence, which is essential for the interaction between δ⁵⁴ and the EBP⁵⁴ activators (Rappas et al., 2005), is also found in the CdsR amino acid sequence. To test whether CdsR is the EBP⁵⁴ for cdsB expression, the activity of the cdsB promoter fused to the gusA reporter gene was assessed in the cdsR::ermB strain. Compared with the strain 630Δerm, the reporter expression in this strain was significantly reduced when cultured in TYC (Figure 7B), suggesting that the cdsR gene is also required for the expression of cdsB in C. difficile.

The EBP⁵⁴s usually bind UASs located between 100 and 200 bp upstream from the TSS. Sequence analyses of the region revealed two inverted repeat sequences spanning from nt −179 to −190 and nt −200 to −211, suggesting that these areas might be the UAS site for EBP⁵⁴ and involved in regulation (Figure 7A). We created several fusions to determine the necessary sequence required for the transcription of cdsB (Table 1). Compared with the control plasmid pRPF185, the plasmids containing the reporter fusions with the 153-bp upstream region from the TSS of cdsB gene did not generate significant β-glucuronidase activity in strain 630Δerm in the presence of cysteine (Figure 7C). However, the pRPF-P_cdsB−223 fusion was inducible and had strong activity in the presence of cysteine, demonstrating that the 223 bp upstream region from the TSS of cdsB gene is sufficient for its transcription. The region between 153 and 223 bp upstream from the cdsB TSS containing the putative UAS site for EBP⁵⁴ is the necessary element for the transcription of cdsB. Together, these results strongly suggested that the CdsR regulator which is probably the EBP⁵⁴ of cdsB is required in the transcription of cdsB, and that the region between 153 and 223 bp upstream from the TSS of cdsB gene containing the putative UAS site for EBP⁵⁴ is necessary for its transcription.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.