Table 1 lists the strains and plasmids used in this study. Clostridioides difficile strains were cultured in BHIS media (brain heart infusion broth supplemented with 0.5% yeast extract and 0.1% L-cysteine, and 1.5% agar for agar plates) at 37°C in an anaerobic chamber (90% N₂, 5% H₂, and 5% CO₂). For spores preparation, C. difficile strains were cultured in Clospore media and purified as described earlier (Perez et al., 2011). Escherichia coli DH5α and E. coli HB101/pRK24 were grown aerobically at 37°C in LB media (1% tryptone, 0.5% yeast extract, and 1% NaCl). Escherichia coli DH5α was used as a cloning host, and E. coli HB101/pRK24 was used as a conjugation donor host. Antibiotics were added when needed for E. coli, 15μg/ml chloramphenicol; for C. difficile, 15μg/ml thiamphenicol, 250μg/ml D-cycloserine, 50μg/ml kanamycin, 8μg/ml cefoxitin, and 500ng/ml anhydrotetracycline.

DNA manipulations were carried out according to standard techniques (Chong, 2001). Plasmids were conjugated into C. difficile as described earlier (Heap et al., 2010). The DNA markers, protein markers, PCR product purification kit, DNA gel extraction kit, restriction enzymes, cDNA synthesis kit, and SYBR Green RT-qPCR kit were purchased from ThermoFisher Scientific (Waltham, United States). PCRs were performed with the high-fidelity DNA polymerase NEB Q5 Master Mix, and PCR products were assembled into target plasmids with NEBuilder HIFI DNA Assembly Master Mix (New England, United Kingdom). Primers (Supplementary Table 1) were purchased from IDT (Coralville, United States). All chemicals were purchased from Sigma-Aldrich (St. Louis, United States) unless those stated otherwise.

Gene edit plasmid pDL-1 containing Cas12a (AsCpfI) under the control of tetracycline-inducing promoter was constructed and used for C. difficile gene deletion according to a previous report (Hong et al., 2018). The target sgRNA was designed with an available website tool,¹ and the off-target prediction was analyzed on the Cas-OFFinder website.² The sgRNA, up- and down-homologous arms, were assembled into pDL-1. Two target sgRNAs for one gene deletion were selected and used for gene deletion plasmid construction in C. difficile, respectively. Briefly, the gene deletion plasmid was constructed in the cloning host E. coli DH5α and was transformed into the donor host E. coli HB101/pRK24, and subsequently was conjugated into R20291. Potential successful transconjugants were selected with selective antibiotic BHIS-TKC plates (15μg/ml thiamphenicol, 50μg/ml kanamycin, and 8μg/ml cefoxitin). The transconjugants were cultured in BHIS-Tm broth (15μg/ml thiamphenicol) to log phase, then the subsequent cultures were diluted with PBS serially and plated on the inducing plates (BHIS-Tm-ATc: 15μg/ml thiamphenicol and 500ng/ml anhydrotetracycline). The plates were incubated at 37°C in the anaerobic chamber for 24–48h, then 20–40 colonies were used as templates for colony PCR test with check primers for correct gene deletion colony isolation. The correct gene deletion colony was sub-cultured into BHIS broth without antibiotics and was passaged several times to cure the deletion plasmid, and then the cultures were plated on BHIS plates and subsequent colonies were replica plated on BHIS-Tm plates to isolate pure gene deletion mutants. The genome of R20291ΔfliW (referred hereafter as R20291ΔW) and R20291ΔfliW-csrA (referred hereafter as R20291ΔWA) were isolated and used as templates for the PCR test with check primers, and the PCR products were sequenced to confirm the correct gene deletion.

The fliW (396bp; primers 3-F/R), csrA (213bp; primers 4-F/R), and fliW-csrA (599bp; primers 5-F/R) genes were amplified and assembled into SacI-BamHI digested pMTL84153 plasmid, yielding the complementation plasmid pMTL84153-fliW, pMTL84153-csrA, and pMTL84153-fliW-csrA, and were subsequently conjugated into R20291ΔWA, R20291ΔW, and R20291 yielding complementation strain R20291ΔWA/pMTL84153-fliW (referred as R20291ΔWA-W), R20291ΔWA/pMTL84153-csrA (R20291ΔWA-A), R20291ΔWA/pMTL84153-fliW-csrA (R20291ΔWA-WA), and R20291ΔW/pMTL84153-fliW (R20291ΔW-W), and overexpression strain R20291/pMTL84153-fliW (R20291-W), R20291/pMTL84153-csrA (R20291-A), and R20291/pMTL84153- fliW-csrA (R20291-WA).

Clostridioides difficile strains were incubated to an optical density of OD₆₀₀ of 0.8 in BHIS media and were diluted to an OD₆₀₀ of 0.2. Then, 1% of the culture was inoculated into fresh BHIS, followed by measuring OD₆₀₀ for 32h.

To examine the effect of fliW and fliW-csrA deletion on C. difficile motility, R20291, R20291ΔWA, and R20291ΔW were cultured to an OD₆₀₀ of 0.8. For swimming analysis, 2μl of C. difficile culture was penetrated into soft BHIS agar (0.175%) plates, meanwhile, 2μl of culture was dropped onto 0.3% BHIS agar plates for swarming analysis. The swimming assay plates were incubated for 24h, and the swarming plates were incubated for 48h, respectively.

For biofilm formation analysis, wild-type and mutant strains were cultured to an OD₆₀₀ of 0.8, and 1% of C. difficile cultures were inoculated into reinforced clostridial medium (RCM) with eight-well repeats in a 96-well plate and incubated in the anaerobic chamber at 37°C for 48h. Biofilm formation was analyzed by crystal violet dye. Briefly, C. difficile cultures were removed by pipette carefully. Then, 100μl of 2.5% glutaraldehyde was added into the well to fix the bottom biofilm, and the plate was kept at room temperature for 30min. Next, the wells were washed with PBS three times and dyed with 0.25% (w/v) crystal violet for 10min. The crystal violet solution was removed, and the wells were washed five times with PBS, followed by the addition of acetone into wells to dissolve the crystal violet of the cells. The dissolved solution was further diluted with ethanol 2–4 times, and biomass was determined at OD₅₇₀.

Clostridioides difficile adhesion ability was evaluated with HCT-8 cells (ATCC CCL-244; Janvilisri et al., 2010). Briefly, HCT-8 cells were grown to 95% confluence (2×10⁵/well) in a 24-well plate and then moved into the anaerobic chamber, followed by infecting with 6×10⁶ of log phase of C. difficile vegetative cells at a multiplicity of infection (MOI) of 30:1. The plate was cultured at 37°C for 30min. After incubation, the infected cells were washed with 300μl of PBS three times, and then suspended in RPMI media with trypsin and plated on BHIS agar plates to enumerate the adhered C. difficile cells. The adhesion ability of C. difficile to HCT-8 cells was calculated as follows: CFU of adhered bacteria/total cell numbers.

To visualize the adherence of C. difficile to HCT-8 cells, C. difficile vegetative cells were labeled with the chemical 5(6)-CFDA (5- and−6)-Carboxyfluorescein diacetate (Fuller et al., 2000). Briefly, C. difficile strains were cultured to an OD₆₀₀ of 0.8, then washed with PBS 3 times and resuspended in fresh BHIS supplemented with 50-mM 5(6)-CFDA, followed by incubation at 37°C for 30min in the anaerobic chamber. After post-incubation, the labeled C. difficile cells were collected and washed with PBS three times, and then resuspended in RPMI medium. Afterward, the labeled C. difficile cells were used for the infection experiment as described above. After 30-min post-infection, the fluorescence of each well was scanned by the multi-mode reader (excitation, 485nm; emission, 528nm), the relative fluorescence unit (RFU) was recorded as F0. Following, the plates were washed with PBS three times to remove unbound C. difficile cells, then the plates were scanned, and the RFU was recorded as F1. The adhesion ratio was calculated as follows: F1/F0. After scanning, the infected cell plates were further detected by the fluorescence microscope.

For fliC transcription analysis, 2ml of 24-h post-inoculated C. difficile cultures were centrifuged at 4°C, 12,000×g for 5min, respectively. Then, the total RNA of different strains was extracted with TRIzol reagent. The transcription of fliC was measured by RT-qPCR with primers Q-fliC-F/R. All RT-qPCRs were repeated in triplicate, independently. Data were analyzed by the comparative CT (2^-∆∆CT) method with 16s rRNA as a control.

To analyze the FliC protein level, C. difficile cell lysates from overnight cultures were used for Western blot analysis. Briefly, overnight C. difficile cultures were collected and washed three times with PBS and then resuspended in 5ml of distilled water. The suspensions were lysed with TissueLyser LT (Qiagen), followed centrifuged at 4°C, 25,000×g for 1h. The final pellets were resuspended in 30μl of PBS, and the total protein concentration was measured by using a BCA protein assay (Thermo Scientific, Suwanee, GA, United States). Protein extracts were subjected to 10% SDS-PAGE. Sigma A protein (SigA) was used as a loading control protein in SDS-PAGE (Mukherjee et al., 2013). FliC and SigA proteins on the gel were detected with anti-FliC and anti-SigA primary antibody (1:1,000, a generous gift from Dr. Daniel Kearns at Indiana University) and horseradish peroxidase-conjugated secondary antibody goat anti-mouse (Cat: ab97023, IgG, 1:3,000, Abcam, Cambridge, MA, United States) by Western blot, respectively. Anti-FliC antibody used in the Western blot analysis is an anti-FliCD serum, generated in the laboratory. FliCD is a fusion protein containing C. difficile FliC and FliD (Wang et al., 2018). The relative intensity of blot bands was analyzed by ImageJ software, and FliC relative intensity was normalized to SigA control.

To evaluate toxin expression in C. difficile strains, one single colony from each strain was inoculated into 25ml of BHIS and incubated in an anaerobic chamber at 37°C, and 10ml of C. difficile cultures from different strains were collected at 24- and 48-h post-incubation. The cultures were adjusted to the same OD₆₀₀ value with fresh BHIS. Then, the collected C. difficile cultures were centrifuged at 4°C, 8,000×g for 15min, filtered with 0.22μm filters, and used for ELISA. Anti-TcdA (PCG4.1, Novus Biologicals, United States) and anti-TcdB (AI, Gene Tex, United States) were used as coating antibodies for ELISA, and HRP-Chicken anti-TcdA and HRP-Chicken anti-TcdB (Gallus Immunotech, United States) were used as detection antibodies.

For toxin transcription analysis, 2ml of 24- and 48-h post-inoculated C. difficile cultures were centrifuged at 4°C, 12,000×g for 5min, respectively. Next, the total RNA of different strains was extracted with TRIzol reagent. The transcription of tcdA and tcdB was measured by RT-qPCR with primers Q-tcdA-F/R and Q-tcdB-F/R, respectively. All RT-qPCRs were repeated in triplicate, independently. Data were analyzed by using the comparative CT (2^-∆∆CT) method with 16s rRNA as a control.

Clostridioides difficile germination and sporulation analysis were conducted as reported earlier (Zhu et al., 2019). Briefly, for C. difficile sporulation analysis, C. difficile strains were cultured in Clospore media for 4days. Afterward, the CFU of cultures from 48 and 96h were counted on BHIS plates with 0.1% TA to detect sporulation ratio, respectively. The sporulation ratio was calculated as CFU (65°C heated, 20min)/CFU (no heated). For C. difficile germination analysis, C. difficile spores were collected from 2-week Clospore media-cultured bacteria and purified with sucrose gradient layer (50, 45, 35, 25, and 10%). The heated purified spores were diluted to an OD₆₀₀ of 1.0 in the germination buffer [10mM Tris (pH 7.5), 150mM NaCl, 100mM glycine, and 10mM taurocholic acid (TA)] to detect the germination ratio. The value of OD₆₀₀ was monitored immediately (0min, t₀), and was detected once every 2min (t_x) for 20min at 37°C. The germination ratio was calculated as OD₆₀₀ (tx)/OD₆₀₀ (T₀). Spores in germination buffer without TA were used as the negative control.

C57BL/6 female mice (6weeks old) were ordered from Charles River Laboratories, Cambridge, MA. All studies were approved by the Institutional Animal Care and Use Committee of University of South Florida. The experimental design and antibiotic administration were conducted as described earlier (Sun et al., 2011). Briefly, 30 mice were divided into three groups in six cages. Group 1 mice were challenged with R20291 spores, group 2 mice with R20291ΔWA spores, and group 3 mice with R20291ΔW spores, respectively. Mice were given an orally administered antibiotic cocktail (kanamycin 0.4mg/ml, gentamicin 0.035mg/ml, colistin 0.042mg/ml, metronidazole 0.215mg/ml, and vancomycin 0.045mg/ml) in drinking water for 4days. After 4days of antibiotic treatment, all mice were given autoclaved water for 2days, followed by one dose of clindamycin (10mg/kg, intraperitoneal route) 24h before spores challenge (Day 0). After that mice were orally gavaged with 10⁶ spores and monitored daily for a week for changes in weight, diarrhea, and mortality. If body weight loss was equal to or greater than 20%, the mouse was euthanized and counted as a dead one. Mortality also included mice that were succumbed to disease. Diarrhea was defined as soft or watery feces. All survived mice were humanely euthanized on day 7 of post-C. difficile challenge.

Fecal pellets from post-infection day 0 to day 7 were collected from each mouse and stored at −80°C. To enumerate C. difficile spores, feces were diluted with PBS at a final concentration of 0.1g/ml, followed by adding 900μl of absolute ethanol into 100μl of the fecal solution, and kept at room temperature for 1h to inactivate vegetative cells. Afterward, 200μl of vegetative cells inactivated fecal solution from the same group and the same day was mixed. Then, fecal samples were serially diluted and plated on BHIS-CCT plates (250μg/ml D-cycloserine, 8μg/ml cefoxitin, and 0.1% TA). After 48-h incubation, colonies were counted and expressed as CFU/g feces. To evaluate toxin tilter in feces, 0.1g/ml of the fecal solution was diluted two times with PBS, followed by examining TcdA and TcdB ELISA.

The reported experiments were conducted in independent biological triplicates, and each sample was additionally taken in technical triplicates. Animal survivals were analyzed by Kaplan–Meier survival analysis and compared by the log-rank test. One-way ANOVA with post hoc Tukey test was used for more than two groups’ comparison. Results were expressed as mean±SEM. Differences were considered statistically significant if p<0.05 (^*).

DNA and protein sequences of fliW and csrA from 10 C. difficile strains belonging to different ribotypes (RTs), including RT106, RT027, RT001, RT078, RT009, RT012, RT046, and RT017 were selected and aligned to those of R20291 (Table 2). We found that fliW and csrA genes are broadly found in C. difficile genomes, and both DNA and protein sequences of fliW and csrA are conserved across different C. difficile strains. These results motivated us to investigate the functions of fliW and csrA in C. difficile.

The C. difficile R20291 flagellar gene operon was analyzed through the IMG/M website,³ and the late-stage flagellar genes (F1) are drawn as Figure 1A (Stevenson et al., 2015). Among them, fliW and csrA genes have a 10bp overlap and were demonstrated as cotranscription by RT-PCR (Supplementary Figure 1).

To analyze the role of fliW and csrA in R20291 (NC_013316.1), CRISPR-AsCpfI-based plasmid pDL1 (pMTL82151-Ptet-AscpfI) was constructed for gene deletion in C. difficile (Zhu et al., 2021). pDL1-fliW and pDL1-csrA gene deletion plasmids were constructed, and the fliW gene (288bp deletion; R20291ΔW) was deleted successfully. However, after several trials, we could not get the csrA gene deletion mutant possibly due to its small size (213bp) or particularly unknown roles for R20291. We also tried to use Clostron and pyrE gene edit system to delete csrA gene, but failed to get the correct mutant. Therefore, we constructed fliW-csrA codeletion plasmid pDL1-fliW-csrA. Part of fliW-csrA (445bp deletion) gene was codeleted, and the plasmid curing mutant R20291ΔWA was obtained (Figure 1B,C). To study the role of csrA in R20291, the single gene complementation strain R20291ΔWA-W and R20291ΔWA-A were constructed. R20291, R20291-pMTL84153 (R20291-E), R20291ΔW-pMTL84153 (R20291ΔW-E), and R20291ΔWA-pMTL84153 (R20291ΔWA-E) were used as control strains when needed.

The effects of fliW and fliW-csrA deletion on R20291 growth were evaluated. Figure 1D shows that there was no significant difference in bacterial growth between parent strain and mutants in BHIS media.

To characterize the effects of fliW and fliW-csrA deletions on C. difficile motility, swimming, and swarming motilities of R20291, R20291ΔWA, and R20291ΔW were first analyzed at 24 and 48-h post-inoculation (Figure 2A; Supplementary Figure 2), respectively. The diameter of the swimming halo of R20291ΔWA increased by 27.2% (p<0.05), while that of R20291ΔW decreased by 58.4% (p<0.05) compared to that of R20291. Next, we examined the motility of the complementation strains (Figure 2B; Supplementary Figure 2), and similar results were obtained among R20291-E, R20291ΔWA-E (with the swimming halo increased by 74.8%, p<0.05), and R20291ΔW-E (with the swimming halo decreased by 59.2%, p<0.05; Figure 2B). No significant difference was detected between complementation strain R20291ΔWA-WA, R20291ΔWA-W, R20291ΔW-W, and the parent strain R20291-E except R20291ΔWA-A which decreased by 52.0% (p<0.05) in swimming halo (Figure 2B). The swarming (48h) and swimming (24h) motilities analyzed on agar plates are shown in Supplementary Figure 2.

The effects of fliW and fliW-csrA deletions on C. difficile biofilm formation were also analyzed. In comparison with R20291, the biofilm formation of R20291ΔW increased by 49.5% (p<0.01), and no significant difference in biofilm formation was detected in R20291ΔWA (Figure 2C). The biofilm formation of R20291ΔW-E increased 112.3% (p<0.001) and R20291ΔWA-A increased by 79.9% (p<0.001) compared to R20291-E (Figure 2D). Meanwhile, the biofilm formation of R20291ΔWA-WA and R20291ΔWA-W decreased by 42.8% (p<0.01) and 25.2% (p<0.05), respectively.

Together, these data indicate that loss of FliW impairs C. difficile motility, and increases biofilm production. The decrease of motility and increase in biofilm production were also detected in R20291ΔWA-A, which was largely restored by coexpressing fliW with csrA in R20291ΔWA (Figures 2B,D), indicating that FliW could antagonize CsrA to regulate bacterial motility and biofilm production.

The ability of C. difficile vegetative cells to adhere to HCT-8 cells in vitro was analyzed. Figure 3A shows that the mean adhesion number of R20291 was 2.40±0.70 bacteria/cell, while that of R20291ΔW was 7.17±0.61, which was 3.0-fold (p < 0.0001) of R20291. No significant difference was detected between R20291ΔWA and R20291. In the complementation strains, we detected a similar result which showed that the mean adhesion number of R20291ΔW-E (6.17±0.64) was 3.20-fold (p < 0.0001) of R20291-E (1.93±0.25; Figure 3B). The adhesion ability of complementation strains nearly recovered to that of wild-type strain except for R20291ΔWA-A (7.13±0.66, p < 0.0001) which was 3.69-fold of R20291-E in the mean adhesion number (Figure 3B).

To visualize the adhesion of C. difficile to HCT-8 cells, the C. difficile vegetative cells were labeled with the chemical 5(6)-CFDA. Figures 3C,D shows that the fluorescence intensity of R20291ΔW was 3.50-fold (p < 0.0001) of that in R20291, and the fluorescence intensity of R20291ΔW-E was 2.36-fold (p < 0.001), and R20291ΔWA-A was 4.08-fold (p < 0.0001) of that in R20291-E, respectively, which is consistent with the results shown in Figures 3A,B. Meanwhile, the adherence of C. difficile to HCT-8 cells was also visualized by fluorescence microscopy (Supplementary Figure 3).

Our data showed that FliW negatively affects bacterial adherence. CsrA complementation in R20291ΔWA increased adherence, while the phenotype change can be recovered partially when fliW was coexpressed with csrA in R20291ΔWA, suggesting that FliW could antagonize CsrA to regulate bacterial adherence. The results from bacterial adherence analysis were consistent with biofilm production analysis indicating the close relation between biofilm production and adherence in C. difficile.

In B. subtilis, FliW interacts with CsrA to regulate hag (a homolog of fliC) translation. We reasoned that FliW and CsrA would also regulate fliC expression in C. difficile. As shown in Figure 4A, the transcription of fliC in R20291ΔWA increased 1.12-fold (p<0.05), while the fliW deletion impaired the fliC transcription slightly while no significant difference. Figure 4B shows the production of FliC in R20291ΔW dramatically decreased (10.4-fold reduction, p<0.001), while that of R20291ΔWA increased significantly (increased by 27.5%, p<0.05). To further determine the role of the single-gene csrA on FliC synthesis, csrA and fliW were complemented into R20291ΔWA or overexpressed in R20291, respectively. Results showed that the significant difference of fliC transcription could only be detected in R20291ΔWA-E (increased by 32.3%, p<0.05; Figure 4C) and R20291-W (increased by 69.8%) compared to R20291-E (Figure 4E). Interestingly, the FliC production of R20291ΔWA-A decreased 3.2-fold (p<0.001) compared to that of R20291-E, while that of R20291ΔWA-WA only decreased by 14.3% (p<0.05), and no significant difference of FliC production in R20291ΔWA-W was detected (Figure 4D). As shown in Figures 4E,F, the fliC transcription of R20291-A was not affected compared to R20291-E, but the FliC production in R20291-A decreased 5.3-fold (p<0.0001). The decrease in FliC production in R20291-A can be partially recovered when fliW was coexpressed with csrA (R20291-WA decreased by 16.2%, p<0.05).

Collectively, our data indicate that CsrA negatively modulates fliC expression post-transcriptionally and FliW antagonizes CsrA to regulate fliC expression possibly through inhibiting CsrA-mediated negative post-transcriptional regulation.

It has been reported that the expression of csrA could affect toxin expression in C. difficile (Gu et al., 2018). To evaluate the effects of fliW and fliW-csrA deletions on toxin production, the supernatants of C. difficile cultures were collected at 24- and 48-h post-inoculation, and the toxin concentration was determined by ELISA. Figure 5A shows that the TcdA concentration of R20291ΔWA decreased by 28.6% (p < 0.05), while R20291ΔW increased by 65.1% (p < 0.01) compared to R20291 at 24-h post-inoculation. However, after 48-h incubation, no significant difference was detected. In Figure 5B, TcdB concentration of R20291ΔWA decreased by 26.4% (p < 0.05) at 24-h post-inoculation, while that of R20291ΔW increased by 93.6% (p < 0.01) at 24h and 33.0% (p < 0.05) at 48h. Similar results were also detected in the complementation strains group (Figures 5C,D). As shown in Figures 5C,D, after 24-h post-inoculation, TcdA (Figure 5C) concentration of R20291ΔWA-E and R20291ΔWA-W decreased by 33.0% (*p < 0.05) and 47.7% (p < 0.01), and TcdB (Figure 5D) concentration of R20291ΔWA-E and R20291ΔWA-W decreased by 37.9% (p < 0.05) and 31.3% (p < 0.05), respectively, while TcdA concentration of R20291ΔW-E, R20291ΔWA-A, and R20291ΔW-W increased by 83.1% (p < 0.01), 64.7% (p < 0.05), and 56.5% (p < 0.05), respectively. Meanwhile, TcdB concentration of R20291ΔW-E increased by 100.2% (p < 0.01). At 48-h post-inoculation, though no significant difference in TcdA production was detected among different C. difficile strains, TcdB concentration of R20291ΔWA-A increased by 28.5% (p < 0.05) compared to R20291-E.

To analyze the transcription of tcdA and tcdB in the complementation strains, RT-qPCR was performed. As shown in Figures 5E,F, the transcription of tcdA and tcdB of R20291ΔWA-E and R20291ΔWA-W decreased significantly (p < 0.05), while that of R20291ΔW-E increased significantly (p < 0.05). Interestingly, the tcdA transcription of R20291ΔWA-A also showed a significant increase (p < 0.05) compared to the wild-type strain. Our data indicate that FliW negatively regulates toxin expression, while CsrA plays a positive regulation role in toxin expression.

To assay the sporulation ratio of C. difficile strains, R20291, R20291ΔWA, and R20291ΔW were cultured in Clospore media for 48 and 96h, respectively. Results (Supplementary Figure 4A) showed that no significant difference in the sporulation ratio was detected between the wild-type strain and the mutants. The germination ratio of C. difficile spores was evaluated as well. Purified spores of R20291, R20291ΔWA, and R20291ΔW were incubated in the germination buffer supplemented with taurocholic acid (TA). As shown in Supplementary Figure 4B, there was no significant difference in the germination ratio between the wild-type strain and the mutants.

To evaluate the effects of fliW and fliW-csrA deletions on C. difficile virulence in vivo, the mouse model of CDI was used. Thirty mice (n=10 per group) were orally challenged with R20291, R20291ΔWA, or R20291ΔW spores (1×10⁶ spores/mouse) after antibiotic treatment. As shown in Figure 6A, the R20291ΔW infection group lost more weight at post-challenge days 1 (p < 0.05), and the R20291ΔWA infection group lost less weight at post-challenge days 3 (p < 0.05) compared to the R20291 infection group. Figure 6B shows that 60% of mice succumbed to severe disease within 4days in the R20291ΔW infection group and 20% in the R20291ΔWA infection group compared to 50% mortality in the R20291 infection group (no significant difference with log-rank analysis, p=0.1629). Meanwhile, 100% of mice developed diarrhea in both the R20291ΔW and R20291 infection groups vs. 80% in the R20291ΔWA infection group at post-challenge days 2 (Figure 6C). As shown in Figure 6D, the spores CFU of the R20291ΔW infection group increased in the fecal shedding samples at post-challenge days 1 and 2 (p < 0.05), while the spores CFU of the R20291ΔWA infection group decreased at post-challenge days 1, 5, and 6 (p < 0.05) compared to the R20291 infection group. Interestingly, while we did not detect significant differences in bacterial growth, germination, and sporulation between the wild-type strain and mutants, the spore numbers from different infection groups were different (Figure 6D). This kind of difference implied that the culture media we used in vitro cannot simulate the complicated intestine environment well, which can lead to the different outcomes in bacterial physiology between in vitro and in vivo analysis. CsrA, as a carbon storage regulator, its regulation on carbon metabolization, and other potential roles in the complicated gut environment in vivo remain to be further studied.

To evaluate the toxin level in the gut, the concentrations of TcdA and TcdB in the feces were measured by ELISA. In comparison with the R20291 infection group, the TcdA of the R20291ΔW infection group increased significantly at post-challenge days 1 (p < 0.05), 2 (p < 0.05), 3 (p < 0.01), and 5 (p < 0.05; Figure 6E), while the TcdA of the R20291ΔWA infection group decreased significantly at post-challenge days 1 (p < 0.05) and 4 (p < 0.05; Figure 6E). As shown in Figure 6F, the TcdB concentration of the R20291ΔWA infection group decreased significantly at post-challenge days 1 (p < 0.05), 2 (p < 0.05), and 3 (p < 0.05), and that of the R20291ΔW increased significantly at post-challenge days 1 (p < 0.05), 2 (p < 0.01), and 3 (p < 0.01). Taken together, our results indicate that the FliW defect increases R20291 pathogenicity in vivo, while the fliW-csrA codeletion impairs R20291 pathogenicity.