A programmed reset mechanism primes Clostridioides difficile virulence at the onset of infection

Abstract

Clostridioides difficile is a major cause of antibiotic-associated diarrhea. This pathogen produces toxins essential for disease and flagella that promote intestinal colonization. Flagella and toxins are coregulated in an ON/OFF manner leading to a heterogeneous population of toxigenic, flagellated cells (flg-ON) and aflagellate cells with attenuated toxin production (flg-OFF). How selective pressures affect ON/OFF switching dynamics and overall population composition is unknown. Here, we use a mouse model of disease to evaluate the ON/OFF makeup of the C. difficile population in vivo. Two key results emerged: (i) Directional inversion of the flg switch to the “ON” orientation occurs during spore germination, and (ii) toxin-induced inflammation enriches for a flg-OFF population during infection. These findings reveal a cycle in which predominantly flg-OFF spores are disseminated into the environment, and spore germination in the next host resets C. difficile to a virulent population of motile, toxigenic cells at infection onset.

INTRODUCTION

The incidence of community-acquired Clostridioides difficile infection (CDI) has risen in recent years, accounting for almost 48% of cases in the United States and ranging from 10 to 43% of global cases annually (1). Around 20 to 30% of patients with CDI experience disease recurrence after initial antibiotic treatment, and a patient’s risk of subsequent episodes increases up to 65% (2).

An obligate anaerobe, C. difficile relies on the development of metabolically dormant spores for transmission between hosts (3). Once in the host, spores germinate in response to certain bile acids and amino acids found in the small intestine (4, 5). The gut microbiota contains bacterial species that affect bile acid metabolism through the conversion of primary bile acids to secondary bile acids (6, 7). Antibiotic use disrupts the microbiota, decreasing the abundance of bacterial species capable of this conversion and skewing the bile acid landscape away from secondary bile acids (inhibitory to C. difficile spore germination and vegetative cell growth) to mostly primary bile acids (conducive to C. difficile spore germination) (5, 8–10). In addition, antibiotic use suppresses colonization resistance by reducing competition for nutrients and the abundance of antimicrobial peptides in the gut, exposing niches for opportunistic pathogens like C. difficile to occupy (11, 12).

After germination, vegetative C. difficile cells produce virulence factors that promote bacterial survival in certain environments but may present a fitness cost in others. Individually and in combination, the toxins TcdA and TcdB disrupt the colonic epithelial barrier resulting in inflammation, neutrophil influx, and diarrheal disease (13–15). Toxin-mediated inflammation results in a reservoir of amino acids that C. difficile can exploit for nutrition and suppresses the protective host microbiota (16). C. difficile also produces flagella that confer swimming motility, promote adherence to mucosal surfaces and intestinal epithelial cells in vitro, and aid intestinal colonization in animal models (17–20). However, flagella are immunostimulatory and may pose a fitness cost, as flagellin is recognized by host Toll-like receptor 5 (TLR5) (21). Activation of TLR5 results in the release of inflammatory cytokines such as interleukin-8 and chemokine (C-C motif) ligand 20 (CCL20), enhancing inflammation at the site of infection and promoting pathogen clearance. Moreover, Batah et al. (22) showed that the activities of toxins and flagella induce a synergistic inflammatory response during CDI. Their findings indicate that toxin-mediated disruption of the colonic epithelial barrier allows flagellin to interact with TLR5 on the basolateral face of epithelial cells, resulting in more severe inflammation when both toxins and flagella are present.

One way that C. difficile may balance the advantages and disadvantages of flagellum and toxin production is by creating a phenotypically diverse population through phase variation (23, 24). Phase variation is a phenomenon that generates a heterogeneous population as a “bet-hedging” strategy, increasing the likelihood that a subpopulation is equipped to survive an environmental stress. The phenotypic composition of the population is typically dictated by selective pressures in the surrounding environment, with the most abundant phase-variable phenotype arising as a result of higher relative fitness in that environment (25–27).

In C. difficile, phase variation of flagella and toxins is mediated by the reversible inversion of the flagellar (flg) switch that controls transcription of a large flagellar operon in an ON/OFF manner (Fig. 1A) (28, 29). Within this operon is sigD, which encodes a sigma factor that activates tcdA and tcdB gene expression, thus linking flagellum and toxin biosynthesis (28, 30, 31). One orientation of the flg switch (corresponding to the sequence in the published R20291 genomes) is permissive for transcription of the flgB operon and SigD-dependent genes, resulting in toxigenic, flagellated cells (the flg-ON state) (29, 32–34). When the flg switch is in the inverse orientation, the resulting mRNA presents a binding site for the transcription termination factor Rho, suppressing expression of the flgB operon and resulting in aflagellate cells with attenuated toxin production (the flg-OFF state) (28). Because of stochastic inversion of the flg switch, C. difficile exists as a heterogeneous population of flg-ON and flg-OFF cells. We previously showed that growth of wild-type C. difficile in soft agar motility medium in vitro strongly selects for flg-ON cells, whereas prolonged incubation in standard agar medium in vitro enriches for flg-OFF cells (28). In addition, preventing flg switch inversion leads to phase-locked mutants with altered C. difficile colonization and virulence in a mouse model of CDI (35). Although we identified in vitro conditions that enrich for each variant, the host intestinal environment imposes complex selective pressures, including immune responses and interactions with the gut microbiome. The impact of flagellum and toxin phase variation on C. difficile fitness during infection, as well as the selective pressures that influence the population in vivo, are unknown.

Fig. 1. flg-ON and flg-OFF cells dominate at different phases of infection in a mouse model of CDI.

(A) Diagram of phase variation and the regulatory link between flagellum and toxin gene expression in C. difficile. The flg switch is a 154-bp sequence flanked by inverted repeats (denoted by gray triangles) and maps to the 5′ untranslated region of the flgB operon. Inversion of the flg switch is mediated by the tyrosine recombinase RecV (not shown). One orientation of the flg switch, but not the other, permits downstream gene expression. Within the flgB gene operon is sigD (σ^D), which encodes a sigma factor that activates late-stage flagellar genes as well as genes in the pathogenicity locus containing tcdA and tcdB. (B and C) Antibiotic-treated male and female C57BL/6 mice were inoculated with 100,000 spores of wild-type R20291 isolates with (B) flg-ON or (C) flg-OFF biases. Genomic DNA was isolated from the inoculums, and fecal samples were collected every 48 hours for quantification of the percentages of flg-ON and flg-OFF cells by qPCR. Blue, % flg-ON; orange, % flg-OFF. Data from two independent experiments are combined. Circles represent values from individual animals. Bars indicate the medians. **P < 0.01 and ***P < 0.001, one-way analysis of variance (ANOVA) with Dunnett’s posttest compared to the spore inoculums (sp).

Here, we investigate the dynamics of phase variation in the complex host environment over the course of infection in a murine model of CDI. We identified two major shifts in the phenotypic makeup of the C. difficile population during infection: Flagellated, toxigenic cells dominate early during infection, while aflagellate and toxigenic cells are more fit at later time points. The early predominance of flg-ON cells is driven by a regulated mechanism that preferentially inverts the flg switch to the ON orientation during spore germination. Subsequently, flg-OFF cells accumulate due to lower relative fitness of flg-ON cells under toxin-mediated inflammation. Moreover, we show that other invertible elements in C. difficile are subject to preferential, directional switching during germination in vitro and in murine infection. Collectively, these data support a model in which C. difficile uses a regulated “reset” mechanism during germination, preferentially inverting multiple switches to achieve a genotype poised to establish infection.

RESULTS

Motile, toxigenic C. difficile blooms early during infection, while nonmotile, toxin-attenuated C. difficile dominates over time

We hypothesized that the phase-variable phenotypes associated with flg-ON and flg-OFF cells confer distinct benefits and costs in the dynamically changing intestinal environment, affecting the flg-ON/OFF makeup of the population during infection. To test this hypothesis, we monitored the proportions of flg-ON and flg-OFF cells during infection using a mouse model of CDI (36, 37). Antibiotic-treated male and female C57BL/6 mice were inoculated with 10⁵ spores of a wild-type R20291 isolate (Fig. 1B). DNA was isolated from fecal samples collected every 48 hours, and quantitative polymerase chain reaction (qPCR) was used to quantify the percentage of each switch orientation as a proxy for the proportion of flg-ON and flg-OFF cells in the sample. When starting with a spore inoculum ∼92.0% flg-ON, the proportion of flg-ON and flg-OFF cells remained stable through day 3 postinoculation (PI) (Fig. 1B). By day 5 PI, we observed an increase in the heterogeneity of flg-ON and flg-OFF cells, and flg-OFF cells became dominant in the population on days 7 and 9 PI (90.1 and 94.4% flg-OFF, respectively) (Fig. 1B). This increase in the proportion of flg-OFF cells suggests that nonmotile, toxin-attenuated C. difficile is more fit at later time points during infection.

What remained unclear was whether the flg-ON bias seen on days 1 and 3 PI was due to a fitness advantage of motile and toxigenic cells in early infection or if the population simply remained transiently stable. To distinguish between these possibilities, we inoculated mice with spores of a wild-type isolate with a flg-OFF (89.5%) bias (Fig. 1C). By day 1 PI, the C. difficile populations in fecal samples showed a notable shift to 93.1% flg-ON, suggesting that motile and toxigenic cells are more advantageous when establishing infection. Thereafter, the proportion of flg-OFF cells followed a dynamic similar to that observed with the flg-ON inoculum, with >95% flg-OFF populations present by day 9 PI (Fig. 1C). In these experiments, the mice were also monitored for weight loss, and fecal samples were collected daily to determine bacterial burden. Mice from each group showed comparable colony-forming units (CFU), weight loss, and toxin production regardless of the inoculum (fig. S1).

We next considered that the C. difficile populations shed in feces may differ from those retained in the intestinal tract. Accordingly, cecums, cecal contents, and fecal samples were collected from mice inoculated with wild-type R20291 (flg-ON) on days 2, 4, and 7 PI for qPCR analysis. For most of the animals (90%), we found that the proportion of flg-ON and flg-OFF cells associated with the cecum and in cecal contents was consistent with that in feces from the same animal on average by ±5%. These results indicate that the data collected from fecal pellets is representative of the C. difficile population associated with cecal tissue and cecal contents (fig. S2). Hereafter, we used data collected from fecal samples as a surrogate measure of both fecal and tissue-associated C. difficile. Together, these data suggest that phase variation occurs during infection, with a bloom of motile and toxigenic cells followed by a shift to the flg-OFF state later in infection.

Nonmotile, toxin-attenuated C. difficile is more fit during infection

The results above suggest that motile, toxigenic cells are more fit in early infection, whereas the nonmotile, toxin-attenuated phenotype becomes advantageous in late infection. We directly assessed the relative fitness of flg-ON and flg-OFF populations by coinfecting mice with C. difficile mutants containing the flg switch locked in either the ON (flg-Δ3-ON) or the OFF (flg-Δ3-OFF) orientation and consequently unable to phase vary flagella and toxins (35). The flg-Δ3-ON mutant was shown to be motile and to produce TcdA and TcdB in vitro and in mouse and hamster models of CDI (35). In contrast, the flg-Δ3-OFF mutant is nonmotile and attenuated for toxin production in vitro and in hamsters but produces comparable levels of toxins to wild type in fecal samples from mice (35). To distinguish the mutants during coinfection, the flg-Δ3-ON strain was marked with a spectinomycin resistance cassette integrated on the chromosome at a neutral site (38, 39). Incorporation of the spectinomycin resistance cassette was previously shown not to affect fitness of wild-type C. difficile (39). The flg-Δ3-ON^R and flg-Δ3-OFF^S mutants were first competed in vitro by coculturing the two strains in rich medium and subculturing and enumerating CFU daily for 10 days. The competition indices (CIs) remained around 1 at each time point (medians ranged from 0.4 to 1.2 per day), indicating no difference in general fitness between flg-ON and flg-OFF cells (Fig. 2A).

Fig. 2. C. difficile phase-locked flg-OFF cells outcompete phase-locked flg-ON cells in mice.

(A) In vitro competition of flg-Δ3-ON^R and flg-Δ3-OFF^S mutants. The differentially marked mutants were mixed 1:1, diluted into fresh TY broth every 24 hours, and plated daily on selective media. Each symbol represents one biological replicate; lines indicate the means. Input, starting culture. (B and C) Antibiotic-treated male and female C57BL/6 mice were inoculated with 100,000 spores of flg-Δ3-ON^R and flg-Δ3-OFF^S mutants combined 1:1. Fecal samples were collected daily and plated on selective media to enumerate CFU per gram feces over time for each individual strain (B) and to calculate the CI (C). Data are combined from two independent experiments. (B) Symbols represent the medians; error bars represent the range. (C) Dotted line indicates equal fitness (CI = 1). In vitro: As a control, 1:1 mixtures of flg-Δ3-ON^R and flg-Δ3-OFF^S spores were incubated in 0.1% taurocholic acid (TA) to induce germination and plated on selective media after 16 hours for calculation of the CI. Circles represent values from individual animals/replicates; open circles indicate approximate CI where flg-Δ3-ON^R was not recovered. Lines indicate the medians. [(A) to (C)] *P < 0.05, **P < 0.01, ***P < 0.001, and #P < 0.0001, Wilcoxon signed rank test, hypothetical = 1. ns, not significant.

Spores of the flg-Δ3-ON^R and flg-Δ3-OFF^S mutants were combined 1:1 and inoculated into antibiotic-treated mice (10⁵ spores total). Fecal samples were collected daily to enumerate each strain and to calculate the CI. Over time, flg-Δ3-OFF^S was recovered in increasingly higher numbers than flg-Δ3-ON^R until flg-Δ3-ON^R fell below the limit of detection in most mice (Fig. 2, B and C), indicating that aflagellate and toxin-attenuated cells are more fit later in CDI. Although wild-type flg-ON cells reached 93.1% of the population early during infection (Fig. 1B), the flg-Δ3-ON^R mutant did not outcompete flg-Δ3-OFF^S at any point (Fig. 2C), which suggests that the bloom of flg-ON cells on day 1 PI is not due to a fitness advantage of motile, toxigenic cells over the aflagellate, toxin-attenuated variant (Fig. 1C).

Accumulation of the flg-OFF variant during infection is dependent on toxin-mediated inflammation

Given our data demonstrating that flg-OFF cells are more fit during CDI, we sought to identify the bacterial factors driving enrichment of the OFF variant over time. Are the virulence factors associated with the flg-ON state—toxins, flagella, or both—important for the accumulation of the flg-OFF variant during infection? We reasoned that by removing the virulence factors associated with the flg-ON state, we would alleviate potential selective pressures acting on the population and disrupt the population shift to flg-OFF over time. We tested a ΔtcdBEA mutant that does not produce TcdA or TcdB but displays wild-type motility, a ΔfliC mutant that is aflagellate and nonmotile but produces TcdA and TcdB, and a sigD::erm mutant lacking flagella and attenuated for toxin production (fig. S3) (28, 40). Spores of each mutant or the wild-type were used to inoculate antibiotic-treated mice, which were then assessed daily for weight loss and bacterial burden in feces. No differences in CFU were observed among the infected groups, indicating that these factors do not affect colonization (fig. S4A). Mice inoculated with wild-type R20291 and ΔfliC lost the most weight, but this did not reach statistical significance (fig. S4B). Consistent with attenuated toxin production in vitro, sigD::erm-infected mice did not lose weight. However, mice infected with sigD::erm had comparable fecal toxin levels to mice infected with wild-type R20291 (fig. S3A and S4, B and C), suggesting that the mutants may occupy distinct niches within the intestinal tract resulting in differences in toxin localization and, consequently, disease symptoms.

From these mice, fecal samples were collected every 48 hours for DNA extraction and qPCR analysis of flg switch orientation. For all infected groups, there was a shift from a predominantly flg-OFF spore inoculum to a flg-ON bias (>80%) 1 day PI (Fig. 3, A to C). Throughout days 3 to 9 PI, the sigD::erm populations maintained a flg-ON bias, with only 19.2% flg-OFF cells on day 9 PI (Fig. 3A), indicating that flagella and/or toxins are necessary for the shift to a flg-OFF population over time. The ΔtcdBEA populations similarly remained primarily flg-ON from days 3 to 9 PI, although we observed a significant increase in the proportion of flg-OFF cells on days 7 and 9 PI (23.8 and 34.9%, respectively, versus 7.7% flg-OFF cells on day 1 PI), suggesting a delayed shift compared to wild-type infection (Figs. 1, B and C, and 3B). In contrast, after the shift to 81.4% flg-ON on day 1 PI, the ΔfliC populations showed an accelerated shift to flg-OFF, reaching 93.3% on day 3 PI compared to 15.5% for the wild-type on the same day (Figs. 1, B and C, and 3C).

Fig. 3. Selection against flg switch variants is dependent on toxin-induced inflammation in mice.

Antibiotic-treated male and female C57BL/6 mice were inoculated with 100,000 spores of (A) sigD::erm, (B) ΔtcdBEA, or (C) ΔfliC mutants. Fecal samples were collected every other day for genomic DNA isolation and qPCR analysis to quantify the percentages of flg-ON and flg-OFF cells. Circles represent values from individual animals; bars indicate the medians. Data are combined from two independent experiments. **P < 0.01 and #P < 0.0001, one-way ANOVA with Dunnett’s posttest compared to day 1 PI (D to I) Cecums were collected from C57BL/6 mice on day 2 PI, formalin fixed, and routine processed to paraffin for histology. Representative images at 200× of mice inoculated with 100,000 spores of (D) sigD::erm, (E) ΔtcdBEA, (F) ΔfliC, (G) wild-type (WT) C. difficile R20291, or (H) healthy uninfected mock animals. Magnification bars = 50 μm. (I) Histological scores for neutrophilic inflammation, submucosal edema, and epithelial damage for individual mice in each group. The horizontal lines represent the median scores for each group of animals. *P < 0.05, ANOVA with Kruskal-Wallis posttest compared to mock controls.

These findings indicate that the shift to a predominantly flg-OFF population requires the presence of TcdA and TcdB and likely toxin-induced inflammation. To assess this association, we evaluated epithelial damage in cecal tissue on day 2 PI Histopathology was most severe in mice infected with wild-type C. difficile R20291 (Fig. 3, G and I). Mice infected with either ΔfliC or sigD::erm exhibited less tissue damage compared to the mice inoculated with wild-type C. difficile (Fig. 3, D, F, and I), whereas mice infected with ΔtcdBEA were indistinguishable from mock-inoculated mice, with no obvious tissue damage in either group (Fig. 3, E, H, and I). Together, these results support a model in which toxin-mediated inflammation drives the greater relative fitness of the flg-OFF variant during infection.

The flg switch preferentially inverts from OFF to ON during germination

We next sought to identify the signals and/or conditions that promote the shift to a flg-ON population observed in wild-type, sigD::erm, ΔtcdBEA, and ΔfliC populations on day 1 PI. In the process of preparing spore inoculums of wild-type flg-ON or flg-OFF C. difficile, we assessed the flg switch orientation in the starting broth cultures and the purified spores derived from those cultures. We found that the proportion of flg-ON and flg-OFF cells remained stable after sporulation, indicating that phase variation does not occur during the sporulation process (fig. S5). Given the rapidity of the shift from a predominantly flg-OFF spore inoculum to a flg-ON population on day 1 PI (Fig. 1C), we postulated that germination could promote flg switch inversion. To examine this possibility, we determined the flg switch orientation after incubation of the purified spores with a bile acid germinant, taurocholic acid (TA). When beginning with spore populations skewed 89.0% flg-OFF, the proportion of flg-ON cells increased to 96.8% after 24 hours with TA (fig. S5A). The reverse inversion was not observed when starting with flg-ON spores (fig. S5B).

These findings suggest a directional switch from flg-OFF to flg-ON with exposure to TA, potentially during germination. We thus assessed flg switch orientation in spore and vegetative cell populations with and without TA (Fig. 4A). When 87.3% flg-OFF spores were germinated with TA, a shift to a 97.6% flg-ON population was observed (Fig. 4B). A significant increase in the proportion of flg-ON cells (from 10.4 to 26.0%) was observed for flg-OFF vegetative cells after incubation with TA, although this effect was not nearly as strong as with germinating spores (Fig. 4B). In contrast, 92.1% flg-ON spores and 98.1% flg-ON vegetative cells remained 97.4 and 99.1% flg-ON, respectively, both with and without TA (Fig. 4C). That flg switch inversion only occurred during germination of flg-OFF spores suggests that C. difficile phase varies flagella and toxins from OFF to ON in medium with TA. We observed that the same phenomenon occurs in other C. difficile strains representing clades 1, 2, and 4, indicating that the directional switch is conserved among different phylogenetic backgrounds (fig. S6).

Fig. 4. Preferential flg switch inversion occurs before vegetative cell outgrowth, independent of the germination stimulus.

Genomic DNA was isolated from the starting populations (T0) and the indicated samples for qPCR to quantify percentages of flg-ON and flg-OFF cells. (A) Inversion of the flg switch in the presence of TA. Wild-type spores and vegetative (veg) cells with (B) flg-OFF or (C) flg-ON biases were incubated with and without 0.1% TA for 24 hours. (D) Inversion of the flg switch with other bile acids. Wild-type spores and vegetative (veg) cells with (E) flg-OFF or (F) flg-ON biases were incubated with 0.1% TA, 0.1% glycocholic acid (GCA), 0.1% cholic acid (CA), 0.01% deoxycholic acid (DCA), or no bile acid (BA) for 24 hours. (G and H) flg switch inversion with artificial germination. Wild-type spore populations with flg-OFF or flg-ON biases were treated with lysozyme (Lz) to induce germination and incubated in TY broth for 24 hours. (I and J) Timing of flg switch inversion during germination versus vegetative cell outgrowth. Wild-type spore populations with a flg-OFF bias were incubated with 0.1% TA, and samples were collected at the indicated intervals for (I) qPCR analysis and (J) plating for CFU. [(B), (C), (E), (F), (H), (I), and (J)] each symbol represents one biological replicate. Bars indicate the means. [(B) and (C)] **P < 0.01 and #P < 0.0001, spore populations analyzed via unpaired t test, T0 compared to +TA. [(E), (F), and (I)] *P < 0.05, **P < 0.01, ***P < 0.001, and #P < 0.0001, one-way ANOVA with Dunnett’s posttest compared to the T0 population. nd, not determined due to lack of spore germination without TA. (H) #P < 0.0001, unpaired t test, T0 compared to +Lz. (J) **P < 0.01, two-way ANOVA with Tukey’s posttest compared to 30 min.

Preferential inversion of the flg switch from OFF to ON occurs independently of the signal used to induce germination

We next considered whether the shift from flg-OFF to flg-ON during spore germination is dependent on the bile acid stimulus. Spore and vegetative cell populations with a flg-ON or flg-OFF bias were incubated with primary and secondary bile acids capable of inducing germination: taurocholate (TA), glycocholate (GCA), cholate (CA), or deoxycholate (DCA) (Fig. 4D). In all bile acids tested, spore populations starting with a flg-OFF bias shifted to 97.8 to 99.0% flg-ON, whereas the proportions in vegetative cell populations remained stable regardless of bile acid added (Fig. 4E). Spore and vegetative cell populations with a flg-ON bias remained flg-ON regardless of the bile acid used (Fig. 4F).

We subsequently assessed whether the shift from flg-OFF to flg-ON during spore germination was possible without a bile acid signal. After artificial germination of flg-OFF and flg-ON spores with lysozyme (41–43), the germinated populations shifted to primarily flg-ON in both cases (Fig. 4, G and H). Collectively, these results indicate that flg switch inversion from OFF to ON occurs independently of the stimulus for germination.

It remained unclear whether the directional switch from flg-OFF to flg-ON occurred during the germination process, which begins within minutes of exposure to germinant (5), or whether flg-ON cells were enriched during vegetative cell outgrowth, which resumes 90 to 120 min postexposure (44, 45). We repeated the germination experiment starting with spores that were 26.5% flg-ON and then collected samples at intervals after the addition of TA. The percentage of flg-ON cells significantly increased to 42.1% after 30 min, with the complete OFF-to-ON shift (reaching 94.4% flg-ON) occurring by 90 min post-TA addition (Fig. 4I). From the same samples, vegetative cell outgrowth was assessed by quantifying CFU at each time point; a change in CFU was not detectable until after 2 hours exposure to TA. (Fig. 4J). Together, these results indicate that flg switch inversion from OFF to ON occurs during germination specifically, before vegetative cell outgrowth. Hereafter, we will denote the directional flg switch inversion from OFF to ON during germination as the “OFF-to-ON switch.”

Bypassing germination in a mouse model of CDI bypasses the OFF-to-ON switch

Since the OFF-to-ON switch occurred during germination in vitro, we posited that the bloom in flg-ON bacteria on day 1 PI in mice also occurred during germination. To evaluate this possibility, we bypassed the in vivo germination step by inoculating mice with wild-type vegetative cell populations with a 70.5% flg-OFF bias and collected fecal samples every 48 hours for qPCR analysis of flg switch orientation (Fig. 5A). In contrast to spore inoculums that exhibited the OFF-to-ON switch, with vegetative cell inoculums, the proportions of flg-ON and flg-OFF cells in the population remained stable (75.3% flg-OFF) on day 1 PI (Fig. 5B). The populations then shifted to 97.9% flg-OFF by day 3 PI and remained flg-OFF through day 9 PI. The mice did not lose any weight during infection, despite colonization levels (CFU) similar to those observed with wild-type spore inoculums (Fig. 5, C and D, and fig. S1). These results indicate that the bloom of flg-ON bacteria on day 1 PI in mice is a consequence of the OFF-to-ON switch during germination at the initiation of infection, and abolishing this switch results in attenuated disease symptoms.

Fig. 5. Bypassing germination with a vegetative cell inoculum abolishes the OFF-to-ON switch in mice.

(A) Antibiotic-treated male and female C57BL/6 mice were inoculated with 10⁸ wild-type R20291 vegetative cell populations with a flg-OFF bias, and fecal samples were collected daily. (B) Genomic DNA was isolated from the inoculums (veg), fecal samples were collected every other day PI, and the percentages of each flg switch orientation were determined by qPCR. Each symbol represents one animal. Bars indicate the means. **P < 0.01, ANOVA with Dunnett’s posttest compared to veg. (C) CFU were enumerated from fecal samples collected daily to determine bacterial burden. (D) Mice were weighed daily to assess weight loss. [(C) and (D)] Symbols represent the medians, and error bars represent the ranges. LOD, limit of detection.

Evidence that the OFF-to-ON switch during germination involves a regulated mechanism of flg switch inversion

Several lines of evidence indicate that the OFF-to-ON switch is not due to greater fitness and selection of flg-ON cells. First, the flg-Δ3-ON^R mutant did not outcompete the flg-Δ3-OFF^S counterpart at any point during murine infection (Fig. 2C), suggesting that the OFF-to-ON switch is not due to higher relative fitness of flg-ON cells over the flg-OFF variant during germination. Second, we also tested flg-Δ3-ON^R and flg-Δ3-OFF^S spore populations in competition during in vitro germination, which revealed a slight advantage for flg-OFF cells (CI = 0.4, P = 0.02) (Fig. 6A). Third, ΔtcdBEA, ΔfliC, and sigD::erm spores exhibited the OFF-to-ON switch in mice (Fig. 3), indicating that flagella and toxins, presumed targets of selection, do not contribute. We directly tested the ability of ΔtcdBEA, ΔfliC, and sigD::erm spore and vegetative cell populations to undergo the OFF-to-ON switch during germination in vitro. All three mutant spore populations, which started with a flg-OFF bias, shifted to ≥99% when germinated with TA (Fig. 6B). The mutant vegetative cell populations retained the flg-OFF bias. Last, the OFF-to-ON switch occurs specifically during germination (Fig. 4, I and J); selection of one variant over the other is unlikely to occur without vegetative cell outgrowth. Collectively, these findings support the conclusion that the OFF-to-ON switch is not due to selection of motile and toxigenic cells and instead suggest that directional phase variation of C. difficile during germination is a regulated process.

Fig. 6. The OFF-to-ON flg switch inversion during germination is not due to selection of motile and toxigenic cells.

(A) Spores of flg-Δ3-ON^R and flg-Δ3-OFF^S were combined 1:1 and plated on medium containing germinant (0.1% TA), with and without Spec1200 to differentiate the mutants. Each symbol represents one biological replicate, and the line indicates the mean. P = 0.02, one sample t test, hypothetical = 1. (B) Spore and vegetative cell populations of sigD::erm, ΔtcdBEA, or ΔfliC were incubated with and without 0.1% TA for 24 hours. Genomic DNA was isolated from the starting populations (T0) and each condition, and the percentages of the flg switch orientation in each sample were determined by qPCR. Each symbol represents one biological replicate. Bars indicate the means. *P < 0.05 and #P < 0.0001, unpaired t test, T0 compared to +TA.

Other invertible elements are subject to switching during germination but remain at the same frequency during infection

Previous work from our group identified RecV as the recombinase responsible for catalyzing flg switch inversion in C. difficile (28). We wondered whether the OFF-to-ON switch is dependent on RecV or due to the activity of another recombinase. We found that a recV-null mutant (recV::erm) with the flg switch locked in the OFF orientation remained flg-OFF after germination (Fig. 7A), and a recV-null mutant with the flg switch locked in the ON orientation remained flg-ON (Fig. 7B). RecV is thus required for flg switch inversion during germination, including for the OFF-to-ON switch.

Fig. 7. Other invertible elements are subject to directional inversion during germination.

Spore populations of (A) recV::erm flg-OFF and (B) recV::erm flg-ON were incubated with and without 0.1% TA for 24 hours. No significant differences by unpaired t test, T0 compared to +TA. (C to I) Left: Switch inversion during germination in vitro. Spores of wild-type R20291 (T0) with ON/published (PUB) and OFF/inverse (INV) orientations of the (C) flg, (D) cwpV, (E) cmr, (F) pdcB, (G) pdcC, (H) algI, and (I) Cdi7 switches were incubated with 0.1% TA for 24 hours before genomic DNA isolation and qPCR analysis. Each symbol represents one biological replicate. Bars indicate the means. Lines connect the same biological replicate before and after germination. Right: Switch inversion during murine infection. Antibiotic-treated male and female C57BL/6 mice were inoculated with 100,000 spores of wild-type R20291 with ON/published and OFF/inverse orientations of the (C) flg, (D) cwpV, (E) cmr, (F) pdcB, (G) pdcC, (H) algI, and (I) Cdi7 switches. Fecal samples were collected on days 1, 5, and 9 PI for genomic DNA isolation and qPCR analysis. Sp, spore inoculums. Blue, % flg-ON; orange, % flg-OFF; green, % ON/published; yellow, % OFF/inverse. Switches measured by “% published/inverse” refer to the switch sequence in the published genome of wild-type R20291, as it is unknown if these switches have regulatory function. Circles represent values from individual mice. Bars indicate the medians. Data are combined from two independent experiments.

Eight C. difficile inversion (Cdi) sites have been identified, and seven of which are present in strain R20291 (46, 47). Five of these invertible sites have been shown to control expression of adjacent genes in a phase-variable manner. In addition to the flg switch (Cdi4), Cdi1 controls expression of cwpV encoding a cell wall protein, Cdi2 and Cdi3 modulate expression of cyclic diguanylate (c-di-GMP) phosphodiesterase genes pdcB and pdcC, respectively, and Cdi6 modulates expression of the cmrRST operon encoding a signal transduction system (47–50). It is unknown whether Cdi5, upstream of an alginate O-acetyltransferase gene (algI), or Cdi7, which spans an open reading frame encoding a hypothetical protein (CDR20291_3417) of unknown function, have regulatory function. RecV is required for Cdi1, Cdi4, Cdi5, and Cdi6 inversion, influences the inversion of Cdi2 and Cdi3 when overexpressed, and does not affect Cdi7 (46).

We postulated that other switches, especially those that require RecV for inversion, may also exhibit directional inversion during germination. To serve as starting populations, ON- and OFF-biased spore populations of wild-type R20291 were obtained for most of the switches: Cdi1 (cwpV switch), Cdi2 (pdcB switch), Cdi3 (pdcC switch), Cdi4 (flg switch), and Cdi6 (cmr switch). We note that apart from the flg and cmr switches, the laboratory culturing conditions that select for ON- and OFF-biased populations for the other switches are still poorly understood. Consequently, we were limited in our ability to obtain independent isolates with some of the genotypes. Genomic DNA was isolated from spores and cells germinated with TA in vitro to assess the percentage of each switch orientation as a proxy for the proportion of ON and OFF cells for each switch. Convergence to one orientation during germination was observed for the flg, cwpV, cmr, and pdcB switches (toward ON, ON, OFF, and OFF respectively) (Fig. 7, C to F, left). The pdcC switch stabilized at 30 to 40% ON (Fig. 7G, left). The algI and Cdi7 switches remained in the same orientation as in the starting spore populations (Fig. 7, H and I, left), although we were unable to obtain isolates with the inverse orientation of Cdi5 (algI switch) or Cdi7 for comparison.

We next examined the ON/OFF proportions for the other switches in mice inoculated with wild-type populations, comparing the spore inoculums with fecal populations collected on days 1, 5, and 9 PI. Notably, the proportions of each switch orientation on day 1 PI generally matched that in populations germinated in vitro (Fig. 7, C to I, right compared to left). The exception is the pdcB switch, which converged to the OFF orientation during in vitro germination but showed a skew toward pdcB-ON in mice albeit with greater variability between replicates (Fig. 7F). These results are consistent with a model in which C. difficile germinates to a specific “phasotype,” or combination of switch orientations and associated phenotypes, at the onset of infection (24). In contrast to the flg switch, which changes from predominantly ON to OFF between 1 and 9 days PI (Figs. 1 and 7C, right), the other switches maintained the proportions of ON/OFF during this time frame (Fig. 7, D to I, right). Together, these results support a model in which a regulated reset mechanism during germination preferentially inverts multiple invertible elements, yielding a vegetative cell population with a uniform genotype and phenotypic profile.

DISCUSSION

This study has revealed how the host intestinal environment affects phase variation at different points throughout infection and shapes the phenotypic makeup of C. difficile populations. Our results support a model in which spore germination spawns a flg-ON population of flagellated, toxigenic vegetative cells that can establish infection (Fig. 8). Under toxin-mediated inflammation, selective pressures favor the flg-OFF state over the flg-ON state until aflagellate, toxin-attenuated cells predominate. Upon transmission to a new host, flg-OFF spores again switch to the flg-ON virulent state upon germination. This cycle may extend to other phase variable traits, most of which also reset to a specific ON/OFF state during germination, presumably yielding a phasotype with greater fitness at infection onset. Broadly, the heterogeneity generated by phase variation is widely understood to improve overall survival outcomes by increasing the likelihood that one subpopulation can survive abrupt environmental stressors. Our results challenge this paradigm through the identification of the existence of a mechanism that regulates phase variable factors in response to spore germination, indicating that a homogeneous, phenotypically fixed population is optimal under specific stimuli, stressors, or environmental conditions.

Fig. 8. Model for phase variation of flagella and toxins during infection.

An individual ingests C. difficile spores. Regardless of the starting orientation of the flg switch, spore germination leads to phase variation (PV) and spawns a population in a predominantly flg-ON state, yielding flagellated, toxigenic cells that can establish infection. Under toxin-mediated inflammation, selective pressures in the intestinal environment favor the flg-OFF state until aflagellate, toxin-attenuated cells become dominant. During infection, C. difficile is shed into the environment, potentially with an increasing proportion of flg-OFF spores. Upon transmission to a new host, and possibly during disease recurrence, flg-OFF spores will germinate to the flg-ON state, restoring virulence.

The preferential OFF-to-ON switch during germination occurs independent of the stimulus. The shift to a primarily flg-ON population occurred in the presence of primary and secondary bile acids, as well as after lysozyme-induced germination, indicating that this process does not occur in response to a specific signal (41). In addition, the populations shifted toward flg-ON as early as 30 min after exposure to TA and reached ∼95% flg-ON after 90 min, before any detectable increase in vegetative cells. Previous work determined that the transition from an ungerminated spore to a fully developed vegetative cell takes 180 min (44, 45), although some vegetative cells have been visualized exiting the spore as early as 90 min postexposure to TA (45). Our results thus indicate that the OFF-to-ON switch occurs specifically during germination, before vegetative cell outgrowth. Bypassing germination at the start of infection by inoculating mice with C. difficile vegetative cells eliminated the OFF-to-ON switch, averting the weight loss that is characteristically observed during CDI in this mouse model.

Multiple lines of evidence indicate that the germination-specific OFF-to-ON switch is not driven by selection of flg-ON cells and instead support a regulated mechanism of directional switch inversion. One potential mechanism involves an alternate recombinase (51). However, we demonstrated that the OFF-to-ON switch requires RecV and found no evidence for another recombinase during germination. Phase variation via site-specific recombination can also be regulated posttranscriptionally, such as by an additional cofactor or recombination directionality factor (RDF) working in tandem with the recombinase (52, 53). RecV was shown to catalyze bidirectional flg switch inversion in C. difficile, but rates of inversion in each direction were not assessed (28). We posit that a germination-specific cofactor interacts with RecV and/or its DNA substrates to modulate the specificity or efficiency of recombination. RecV lacks an N-terminal high-affinity DNA binding domain common to tyrosine recombinases (54). An RDF may substitute for this missing domain, although it is possible that the domain is dispensable for RecV-mediated switch inversion. In addition, the inverted repeats flanking the flg and other invertible switches (most of which also exhibited directional switching during germination in vitro and in mice) share similarity but are not identical (46, 55), raising the possibility that multiple, alternative cofactors interact with RecV to modulate recognition of different inverted repeat sequences. C. difficile does not encode orthologs of characterized cofactors such as Xis, Fis, H-NS, etc., and the lack of conserved domains or synteny with the recombinase gene makes these factors difficult to identify bioinformatically. Identifying the cofactor/RDF involved in directional switch inversion may reveal ways to inhibit the OFF-to-ON switch and consequently prevent C. difficile from germinating to a virulent state, ameliorating disease symptoms. Defining the mechanism of directional switch inversion is a priority for future studies.

Within the first few days of infection, flg-OFF cells gain an increasing fitness advantage over flg-ON cells, leading to an enrichment of aflagellate, toxin-attenuated cells in the population. Our findings indicate that in the presence of TcdA- and TcdB-dependent inflammation, flagellum biosynthesis contributes to the lower relative fitness of flg-ON cells during infection. Consistent with this idea, the absence of both toxins and flagella in the sigD::erm mutant allowed the flg-ON population appearing on day 1 PI to be maintained for the duration of the 10-day experiment. However, that the shift to a flg-OFF population is accelerated during infection by the ΔfliC mutant suggests that selection is flagellin independent. During infection by wild-type C. difficile, toxin-induced inflammation may promote immune-mediated clearance of flg-ON cells through disruption of the epithelium and exposure of flagellin to host immune receptors and recruitment of macrophages and neutrophils to the site of infection (21, 22, 56, 57). Flagellated C. difficile cells may be preferentially targeted through recognition of surface structures, as shown in Pseudomonas aeruginosa infection, where neutrophils identify motile, flagellated bacteria and release neutrophil extracellular traps to entrap and immobilize cells before macrophage phagocytosis (58). It is similarly plausible that flagellated C. difficile cells are preferentially phagocytized by neutrophils, leaving aflagellate and less-toxigenic cells.

Another, nonmutually exclusive possibility is that the host microbiota imposes selective pressures that enrich for flg-OFF cells. Changes in microbiota-derived metabolites and antimicrobial compounds, which fluctuate during infection, could influence the relative fitness of C. difficile phase variants (5, 9, 10, 59–62). We recently showed that the opportunistic pathogen Enterococcus faecalis affects phase variation of the CmrRST signal transduction system in C. difficile (40). The mechanism has not yet been identified but may involve amino acid exchange between the two species (63). In addition, the presence of membrane-associated proteins, like flagella, may increase membrane permeability of flg-ON cells, increasing susceptibility to antimicrobial compounds produced in the host. Recent work showed that the presence of virulence factors, including flagella, embedded in the envelope of Salmonella enterica serotype Typhimurium, confer higher sensitivity to membrane stress (64).

Beyond flagella and toxins, SigD regulates genes involved in membrane transport, metabolism, and cell wall proteins that could influence fitness during infection (31). Accordingly, the flg-ON and flg-OFF variants may display different cell surface proteins and exhibit distinct immunogenic profiles. In addition, toxin activity liberates nutrients from the epithelium, and CDI-associated inflammation affects the integrity and architecture of the mucus layer, altering nutrient sources (16, 65, 66). Changes in nutrient availability may lead different phase variants to occupy different spatial or nutritional niches, as glucose availability has been shown to affect phase variation of colony morphology in C. difficile (67). A recent study reported no differences in the spatial localization of toxin-expressing cells in a mouse model of CDI (68); however, the C. difficile strain used in that work (630Δerm; ribotype 012) does not invert the flg switch and consequently does not phase vary flagella and toxins (69).

Selective pressures within the murine intestinal environment appear only to influence the proportion of flg switch variants. After germination as a specific phasotype, the population shifts from flg-ON to flg-OFF over time, while the cwpV, cmr, pdcB, pdcC, algI, and Cdi7 switches maintained stable ON/OFF ratios throughout infection. How phase variation of the products controlled by these switches affects bacterial fitness and infection dynamics remains to be determined. Different dynamics may occur in other animal models and in human patients with varying disease severity, microbiota composition, or asymptomatic carriage, each of which may impose distinct selective pressures. Extending the approaches here to human samples could potentially reveal phase variation signatures that can be used to develop diagnostic tools based on ON/OFF switch combinations, particularly for the flg switch.

A limitation of this work is the assumption that the switch orientation reflects an associated phenotype, as additional regulatory layers may override the genotype. For example, c-di-GMP riboswitches exist upstream of the flg and cmr switches, and consequently, the amount of c-di-GMP in the cell can regulate expression of these genes (70–73). Phase variation of the c-di-GMP phosphodiesterases PdcB and PdcC may influence transcription of flagellar genes and cmrRST, adding further complexity to the phenotypic heterogeneity exhibited by C. difficile (49). Despite these variables, the composition of the C. difficile population was altered when the virulence factors associated with the flg-ON state were removed, implying that the phenotype imparted by orientation of the flg switch matches its genotype. Another limitation of this work is the reliance on population level analyses of switch inversion, which can mask heterogeneity among individual cells. Single-cell approaches could reveal whether distinct phase variants occupy different intestinal niches, such as within the lumen, mucus, or epithelial surfaces, or in proximity to specific commensal microbes. Emerging tools, including dual-fluorescent reporters optimized for in vivo imaging (68, 74), single-cell RNA and DNA fluorescence in situ hybridization (75, 76), and computational approaches and single-cell sequencing to identify bacterial invertible elements (77–79) provide promising avenues to reveal phase variation dynamics during germination and infection. By revealing how phenotypic heterogeneity influences host-pathogen interactions, these findings may inform strategies to prevent recurrent CDI and mitigate disease severity.

MATERIALS AND METHODS

Bacterial strains and growth conditions

Strains and plasmids used in this study are listed in table S1. C. difficile was grown in an anaerobic chamber (Coy Laboratories) using a gas mix consisting of 85% N2, 5% CO₂, and 10% H₂. C. difficile was routinely cultured statically under anaerobic conditions at 37°C in tryptone yeast (TY) broth (Becton Dickinson) or in brain heart infusion medium (Becton Dickinson) supplemented with 5% yeast extract (Gibco) (BHIS). Unless otherwise specified, C. difficile experiments were performed using R20291, a ribotype 027 strain (33). Escherichia coli DH5α and HB101(pRK24) derivatives were cultured under aerobic conditions in Miller’s Luria-Bertani (LB) medium (Thermo Fisher Scientific) at 37°C. Plasmids were maintained with ampicillin (100 μg/ml), chloramphenicol (10 μg/ml), thiamphenicol (10 μg/ml), or spectinomycin (1200 μg/ml; Spec1200) as indicated.

Plasmid and strain construction

A full list of oligonucleotides used for plasmid and strain construction is provided in table S2. Allelic exchange in C. difficile was done using the pMSR0 plasmid (80). To create a marked strain of C. difficile R20291 with the flg switch locked in the ON orientation, the aad9 spectinomycin resistance cassette from E. faecalis was integrated on the chromosome of the previously made flg-Δ3-ON mutant between the convergently oriented genes CDR20291_2492 and CDR20291_2493 as described by Santos-Santiago et al. (35, 38, 39). Briefly, pRT3098 was introduced into heat-shocked C. difficile R20291 flg-Δ3-ON via conjugation with E. coli HB101(pRK24) (81). Allelic exchange was performed as previously described (80), and spectinomycin-resistant clones were screened by PCR using primers R3472 + R3473 and confirmed by sequencing.

To delete the tcdBEA region, the regions upstream of tcdB and downstream of tcdA were amplified from R20291 genomic DNA by PCR using R3312 + R3313 and R3314 + R3315, respectively. The fragments were cloned in SalI/XhoI-digested pMSR0 via Gibson assembly. Colonies were screened by PCR using primers R2743 + R2744 and confirmed by sequencing. The resulting plasmid, pRT3031 (pMSR0::ΔtcdBEA), was introduced into heat-shocked C. difficile R20291 with E. coli HB101(pRK24) for allelic exchange as described (35). Deletion mutants were identified by PCR using primers R3261 + R3262, and the integrity of the deletion was confirmed by Sanger sequencing and phenotyping.

Isolation of wild-type C. difficile flg-ON and flg-OFF populations

Isolation of flg-ON and flg-OFF populations of wild-type R20291 was achieved by PCR screening individual colonies using primers R857 + R1614 (flg-ON) or R857 + R1615 (flg-OFF). Once clones skewed mostly flg-ON or flg-OFF were identified, the percentage of flg-ON and flg-OFF cells in each population was quantified by qPCR using primers R2175 + R2177 (flg-ON) and R2176 + R2177 (flg-OFF).

qPCR analysis of invertible elements in C. difficile

The orientations of invertible elements were quantified using qPCR as described previously (46). Briefly, genomic DNA was purified from tissue, fecal, spore, and broth culture samples by treating with lysozyme (10 mg/ml) and bead beating for 2 × 90 s, followed by phenol/chloroform extraction (82). Genomic DNA (100 ng) was used as a template with SYBR Green Real-Time qPCR reagents (Bioline). For C. difficile broth cultures and spores, primers were used at a final concentration of 0.1 μM and an annealing temperature of 60°C. For fecal DNA samples, final primer concentrations of 1 μM and an annealing temperature of 55°C were used (35). The primer sequences used to detect the published and inverse orientations of the flg, cmr, pdcB, pdcC, cwpV, algI, and Cdi7 switches are listed in table S2. Quantification of the percentage of published and inverse amplicons was done as previously described using rpoC as the reference gene (46).

Spore purification

Broth cultures (100 μl) were spread on 70:30 agar and incubated for 72 hours at 37°C (83). Bacterial growth was collected in 10 ml of 1× DPBS (Dulbecco’s phosphate-buffered saline; Gibco) and incubated aerobically at room temperature for 24 hours. Spores were washed five times with 1× DPBS and purified using a 50% sucrose gradient as described (84). Spores were washed five more times with DPBS + 1% bovine serum albumin and stored at room temperature until use.

Animal experiments

Mouse experiments were performed under the guidance of veterinary staff within the University of North Carolina at Chapel Hill (UNC-CH) Division of Comparative Medicine. All animal studies were done with prior approval from the UNC-CH Institutional Animal Care and Use Committee under protocol 24-222. Animals that were considered moribund, as well as all animals at the experimental endpoint, were euthanized by CO₂ asphyxiation and thoracotomy in accordance with the Panel on Euthanasia of the American Veterinary Medical Association. Male and female C57BL/6 mice (Charles River Laboratories) aged 8 to 10 weeks were given a previously described antibiotic regimen to render them susceptible to C. difficile infection (36, 37). The mice were given a cocktail of antibiotics [kanamycin (400 μg/ml), gentamicin (35 μg/ml), colistin (850 U/ml), vancomycin (45 μg/ml), and metronidazole (215 μg/ml)] 7 days before inoculation for 3 days and a single intraperitoneal dose of clindamycin (10 mg/kg body weight) 2 days before inoculation. Mice were inoculated with 10⁵ spores by oral gavage. Mock-inoculated animals were included as controls. Animal weights were recorded daily, and fecal samples were collected every 24 hours for 10 days PI for enumeration of CFU on taurocholate-cycloserine-cefoxitin-fructose agar (TCCFA), determination of flg switch orientation, and quantification of toxin accumulation (35). Tissues (cecum and cecal contents) were collected on days 2, 4, and 7 PI for genomic DNA extraction. Cecums were cut open, cecal contents were collected, and the tissue was gently rinsed with 1× DPBS. For competition experiments, spores were combined 1:1 before inoculation by oral gavage, and fecal samples were collected daily for enumeration of CFU on TCCFA with and without Spec1200. The CIs were calculated by dividing the ratio of the two strains in the output [number of resistant CFU/(total CFU − resistant CFU)]_output by the ratio in the spore inoculum [number of resistant CFU/(total CFU − resistant CFU)]_input. Each experiment was independently repeated, and the data were combined for analysis.

Histopathologic examination of inflammation

Cecums were collected at the time of euthanasia from mice on day 2 PI for histopathological examination. Cecal tissue was fixed in 10% neutral formalin for 48 hours, transferred to 70% ethanol, and routinely processed to paraffin. Sections (5 μm) were stained with hematoxylin and eosin and analyzed in a masked fashion by a board-certified veterinary pathologist. Tissue sections were evaluated on the severity of epithelial damage, epithelial hyperplasia, submucosal edema, and neutrophilic inflammation based on the rubric described by Reeves et al. (85). Each metric was scored on a scale of 0 to 4, and values were summed for each animal to obtain a histopathology score.

Toxin quantification

Toxin activity in bacterial cultures and in feces from mice inoculated with wild-type C. difficile was quantified using a Vero cell rounding assay as previously described (86). An enzyme-linked immunosorbent assay (ELISA) kit (tgcBIOMICS, #TGC-E001-1) was used for simultaneous detection of C. difficile toxins A and B in feces from mice inoculated with wild-type C. difficile and ΔtcdBEA, ΔfliC, and sigD::erm mutants. Fecal samples were thawed on ice, and 50 mg of feces was homogenized in the provided dilution buffer and centrifuged for 5 min at 2500g. The clarified fecal supernatants were added to each well. The ELISA was conducted according to the manufacturer’s instructions, and the absorbance was measured at 450 nm with an Agilent BioTek Synergy H1 plate reader.

In vitro competition experiments

Broth cultures from each strain were combined at a 1:1 ratio, diluted, and plated on BHIS agar ± Spec1200 to determine the starting ratio. The mixture was subcultured 1:60 in 3 ml of TY, and serial dilutions were plated every 24 hours for 10 days on BHIS agar ± Spec1200. To assess fitness during germination, an equal ratio of spores was mixed and plated immediately on BHIS agar + 0.1% TA ± Spec1200. CIs were calculated as described above.

Swimming motility assay

Individual colonies from freshly streaked BHIS agar were inoculated into 0.5× BHIS–0.3% agar using sterile pipette tips. The diameter of motile growth was measured after 24- and 48-hour incubation at 37°C (72). Plates were imaged using a G:BOX Chemi imaging system with the upper white light illuminator.

Quantifying flg switch orientation after germination

Approximately 1 × 10⁴ spores were first heat shocked at 55°C for 20 min. Spores and vegetative cell controls were inoculated into 3 ml of TY broth with 0.1% TA, GCA, or CA; medium without germinant was included as a control. Because DCA inhibits C. difficile vegetative cell growth (5), 0.01% DCA was used to germinate spores while mitigating inhibition of outgrowth. After incubation for 24 hours, cells were collected for genomic DNA isolation and qPCR analysis of flg switch orientation. For germination timeline experiments, the input was increased to approximately 5 × 10⁷ spores to increase the yield of genomic DNA at early time points. After 30, 60, 90, 120, and 360 min of exposure to TA, serial dilutions were plated on BHIS agar ± 0.1% TA for CFU enumeration. The remaining culture was used for DNA isolation and qPCR analysis as above. For experiments using artificial germination, approximately 1 × 10⁴ spores were pelleted, suspended in 250 mM sodium thioglycolate, and incubated at 50°C for 30 min (41–43). Spores were subsequently washed with 1× DPBS, suspended in lysozyme (4 mg/ml), and incubated at 37°C for 15 min. As a control, spores were suspended in DPBS without lysozyme. Spores were washed with 1× DPBS and suspended 3 ml of TY broth. After 24 hours of incubation, cultures were harvested for genomic DNA isolation and qPCR analysis of flg switch orientation.

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S6

Tables S1 and S2

References