CspC and CspA form a complex that does not include CspB

Previously, we showed that CspC levels in spores are greatly reduced in the absence of CspA [22,24] and that CspC and CspA levels are also greatly reduced in the absence of CspB [24]. Therefore, we hypothesized that CspC and CspA directly interact and that this interaction may require CspB. To test these hypotheses, we co-expressed untagged CspA and CspB with His₆-tagged CspC in E. coli and tested whether the untagged proteins co-purify with CspC-His₆. To recapitulate the CspB and CspA variants generated by YabG-mediated processing of the CspBA fusion protein [18,22], we produced the individual CspB_1-582 and CspA_583-1132 domains [18]. Since CspC and CspA have almost identical molecular weights (MWs) of 60 kDa, we distinguished the two proteins from each other by tagging CspC with a Vibrio cholerae MARTX toxin cysteine protease domain (CPD) [28] fused to a C-terminal hexahistidine tag (CPD-His₆, 24 kDa) [29,30]. CspB is readily distinguished from CspC and CspA because CspB undergoes autoprocessing of its N-terminal prodomain, and the processed CspB has a MW of ~55 kDa [19]. The pull-down analyses revealed that untagged CspA, but not untagged CspB, efficiently co-purifies with CspC-CPD-His₆, suggesting that CspC and CspA form a complex (Fig 1A). CspA failed to co-purify with a GFP fusion protein (GFP-CPD-His₆), used as a negative control, indicating that the CspC:CspA interaction is specific (Fig 1A). Interestingly, we found that CspC-CPD-His₆ pulled down untagged CspBA fusion protein at considerably lower levels than untagged CspA (S1 Fig), suggesting that YabG-mediated proteolytic processing of CspBA promotes CspC binding to CspA.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Interactions between Clostridioides difficile Csp proteins.

(A) Coomassie stain of co-affinity purifications of CspC-CPD-His₆ or GFP-CPD-His₆ as the bait and untagged CspB and CspA as the prey. + , induced fraction; CL, cleared lysate; E, elution. (B) Size exclusion chromatography (SEC) analysis of the elution fraction of a CspC-CPD-His₆:CspA co-affinity purification. The dashed line indicates the separation between the two peaks. 12.9 mL corresponds to an apparent MW of 121 kDa; 13.6 mL corresponds to an apparent MW of 84 kDa; and 13.7 mL corresponds to an apparent MW of 82 kDa. The SEC elution fractions were resolved using SDS-PAGE and stained with Coomassie. (Inset) Analysis of the stability of SEC-purified CspC-CPD-His₆:CspA complex. The complex was purified from the 12.6–13.0 mL fraction and then re-analyzed using SEC. (C) SEC analyses of individual Ni²⁺-affinity-purified Csp-His₆ proteins. 13.1 mL corresponds to an apparent MW of 106 kDa, 13.7 mL corresponds to an apparent MW of 83 kDa, and 14.9 mL corresponds to an apparent MW of 46 kDa. (bottom) Coomassie stain of SEC elution fractions resolved by SDS-PAGE. The CspB prodomain (~9 kDa) was run off the gel [19]. (D) Size exclusion chromatography (SEC) of separately purified CspC_VDAL and CspA-His₆ variants. Separately purified proteins were mixed and incubated for 1 hr at a 1:1 stoichiometry and then resolved using SEC. (bottom) western blot analysis of the SEC fractions. All data shown are representative of three replicates. The raw gel images in this figure can be found in S1 Raw Images. The data underlying panels B, C, and D can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003610.g001

To assess the stability of the CspC-CPD-His₆:CspA complex, we used size exclusion chromatography (SEC) (Fig 1B). Two distinct peaks were observed by SEC. Peak 1 eluted with an apparent MW of 121 kDa and contained CspA and CspC-CPD-His₆ at an apparent ratio of 1:1 (Fig 1B, SDS-PAGE), consistent with a CspC-CPD-His₆:CspA heterodimer. Peak 2 eluted with an apparent MW of 84 kDa. This peak contained mostly monomeric CspC-CPD-His₆, consistent with its predicted MW of 86 kDa (Fig 1B). Despite CspA having a predicted MW of 60 kDa, a small amount of untagged CspA was detected in this smaller peak, suggesting that CspA may form a dimer when not in complex with CspC. Still, nearly all the untagged CspA was found in the CspC-CPD-His₆:CspA peak, and when the complex was re-run on SEC, it eluted as a single peak at the same volume as the original CspC-CPD-His₆:CspA complex, indicating that the complex was stable (Fig 1B, inset). SEC analyses of individually purified CspC-His₆ and CspA-His₆ confirmed that CspA-His₆ formed an apparent homodimer, whereas CspC-His₆ was monomeric, consistent with our previous study [27] (Fig 1C).

To rule out any potential contributions of the CPD tag to this interaction, we individually purified CspC and CspA, mixed the two proteins at a 1:1 ratio, and analyzed the CspC:CspA heterodimer formation by SEC. Untagged CspC was first purified by inducing the CspC-CPD-His₆ fusion protein to undergo autoprocessing using inositol hexakisphosphate (InsP₆) [29,30], which adds a Val-Asp-Ala-Leu (VDAL) sequence to the C terminus of CspC. When CspC_VDAL and CspA-His₆ were mixed, a larger portion of the mixture eluted in peak 1, where the CspA homodimer typically elutes, consistent with CspC:CspA heterodimer formation (Fig 1D). These data indicate that CspA has a higher affinity for CspC than for itself and that the CspC:CspA heterodimer can form even when CspC and CspA are produced individually.

CspB and CspA form homodimers, but CspC does not

Since CspA forms an apparent homodimer detected by SEC, we assessed the ability of C. difficile Csp proteins to interact in pairwise analyses. Untagged CspA, CspB, or CspC were co-produced with CPD-His₆-tagged CspA, CspB, and CspC variants, respectively, in E. coli, and the resulting pairwise interactions were analyzed. As expected, untagged CspC co-purified with CspA-CPD-His₆ (S2 Fig), indicating that this interaction occurs when the tags are reversed. Untagged CspA co-purified with CspA-CPD-His₆ (S2 Fig), consistent with the ability of CspA to homodimerize. Untagged CspB also co-purified with CspB-CPD-His₆ (S2 Fig), suggesting that CspB forms a homodimer. Consistent with these data, SEC analyses of the individually purified Csp proteins (Fig 1C) confirmed that CspB and CspA each form homodimers. In contrast, untagged CspC was not pulled down by either CspB-CPD-His₆ or CspC-CPD-His₆ (S2 Fig), and it eluted as a monomer when individually purified and analyzed via SEC (Fig 1C). None of the untagged Csps interacted with the GFP-CPD-His₆ control (S2 Fig). These results are consistent with the known crystal structures of the C. perfringens CspB dimer¹⁶ and C. difficile CspC monomer²⁰ and provide the first evidence that C. difficile CspA forms a homodimer.

CspA forms an antiparallel homodimer

To gain insight into the functional importance of the CspA homodimer uncovered by our co-affinity purification analyses, we determined the crystal structure of CspA. The structure was phased by molecular replacement using the known structure of the C. difficile CspC [27] and refined to 3.2 Å resolution (S1 Table). The C. difficile CspA structure is very similar to those of C. difficile CspC and C. perfringens CspB [19,27] (Fig 2A) and can be superimposed with RMSDs of 1.22 and 1.44 Å, respectively. All three proteins are composed of an N-terminal prodomain followed by the subtilase domain that contains the central jellyroll domain insertion unique to the Csp subtilisin-like serine protease family [19,27] (Figs 2A and S3).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Crystal structures of the Clostridioides difficile CspA homodimer and CspC:CspA heterodimer.

Ribbon and space-fill models of (A) C. perfringens CspB (PDB RI0W [19]), C. difficile CspC (PDB 6MW4 [27]), and C. difficile CspA (chain B) (this work—PDB 9PR9). The subtilase domains are shown in gray, jellyroll domains in green, and prodomains in purple. Ribbon models of the: (B) Crystal structure of the CspA homodimer solved to 3.2 Å (PDB 9PR9). Inset shows a close-up of one of two tripartite hydrogen bond networks found at the interface of the CspA homodimer. (C) Crystal structure of the CspC:CspA_F944E/Y1092E (CspC:CspA_EE) heterodimer solved to 3.4 Å (PDB 9PR8). CspC is shown in cyan, and CspA is shown in magenta. Inset (left): tripartite hydrogen bond network between residues Q516_CspC, T520_CspC, and R896_CspA with distances of 2.7 Å and 3.1 Å, respectively. Inset (center): hydrogen bond between the Y361_CspC main chain and the Q1090_CspA residue with a distance of 2.8 Å. Inset (right): two salt bridges between CspC and CspA. The D429_CspC, R1036_CspA salt bridge interaction is a distance of 3.6 Å, while the R456_CspC, D1008_CspA is 2.9 Å. (D) Crystal structure of CspC monomer (PDB 6MW4 [27]). Inset shows unstructured residues in the CspC monomer that form salt-bridges in the CspC:CspA heterodimer.

https://doi.org/10.1371/journal.pbio.3003610.g002

The asymmetric unit of the CspA crystals consists of two antiparallel, asymmetric homodimers, CspA_AB and CspA_CD (S4 Fig). The four copies of CspA are very similar and can be superimposed with RMSDs ranging between 0.30 and 0.42 Å. Chains A and C are similar to each other and the most complete, with 534 residues resolved out of 559 total residues. Chains B and D are also similar to each other but less complete, with 489 and 487 residues, respectively, resolved out of 559 total residues. The two CspA homodimers can be superimposed with an RMSD of 0.32 Å. The N-terminal prodomains in chains A and C are better resolved than their counterparts in chains B and D in two regions (S4 Fig). The A and C protomers are better resolved likely because of crystal contacts with neighboring protomers (S5 Fig).

To map residues at the CspA homodimer interface, we analyzed the buried surface area and interactions (hydrogen bonds and salt bridges) using PiSA interface analysis [31] (S2 Table). In the CspA homodimer, the interface features two symmetrical tripartite hydrogen-bond networks formed between the jellyroll and subtilase domains, termed twin H-bond networks (Fig 2B). Each H-bond network is formed by three residues, Arg896, Gln1094, and Thr1098 (for consistency, the numbering for CspA is based on the CspBA fusion protein) (Fig 2B, inset). Notably, there were no interactions at the center of the homodimer (Fig 2B), so the twin H-bond networks make up the majority of the CspA homodimer interface.

The CspC:CspA heterodimer has an extensive interface

To investigate the functional importance of the CspC:CspA heterodimer, we determined its crystal structure. Initial attempts to crystallize CspC_VDAL:CspA failed to yield diffraction-quality crystals, so we hypothesized that the CspA homodimer in the CspC_VDAL:CspA heterodimer preparations was interfering with the crystallization, based on a prior report that showed that homodimerization of Munc13-1 hindered crystallization of the Munc13-1:RIM2α heterodimer [32]. To mitigate this issue, we used a CspA_F944E/Y1092E (CspA_EE) mutant that increases CspC_VDAL:CspA_EE heterodimer formation (S6 Fig). By working with a CspA mutant favoring the formation of the CspC:CspA heterodimer, we were able to crystallize it. The crystal structure of the CspC_VDAL:CspA_EE heterodimer was determined to 3.35 Å resolution (Fig 2C and S1 Table).

There are many similarities between the CspC:CspA heterodimer and the CspA homodimer. Like the CspA homodimer (Figs 2B and S4), the CspC:CspA heterodimer has an antiparallel orientation (Fig 2C and S7 Fig). Likewise, there are two crystallographically independent CspC:CspA heterodimers in the asymmetric unit, with small variations between their structures (RMSD of 0.40 Å between the two CspC:CspA heterodimers, 0.41 Å between CspC_B and CspC_D, and 0.45 Å between CspA_A and CspA_C) (S7 Fig). Remarkably, the H-bond network formed by Arg896, Gln1094, and Thr1098 at the interface of the CspA homodimer is conserved at the CspC:CspA heterodimeric interface. Specifically, the same residue, Arg896_CspA, forms hydrogen bonds with Gln516_CspC and Thr520_CspC, which are homologous to Gln1094_CspA and Thr1098_CspA (Fig 2B, inset and 2C, left inset).

However, there are also notable differences. First, the CspC:CspA heterodimer has a more extensive interface than the CspA homodimer (Fig 2B and 2C). Indeed, the heterodimeric interface of the CspC:CspA buries a significantly larger area, 2,024 Ų on average, compared to the homodimeric interface of CspA, which buries 1,610 Ų on average (S2 Table). Second, in the CspC:CspA heterodimer, there is only one H-bond network at the interface (Fig 2C). Instead of the second H-bond network, a series of interactions tether the CspC jellyroll domain to the CspA subtilase domain, including a hydrogen bond between the carbonyl group of CspC residue Tyr361 and the amide of CspA residue Gln1090 (Fig 2C, center inset). Lastly, there are two salt bridges at the center of the interfaces of both heterodimers, between Asp429_CspC and Arg1036_CspA and between Arg456_CspC and Asp1008_CspA (Fig 2C, right inset). Interestingly, the CspC residues forming these salt bridges are unstructured in the CspC monomer structure [27] (Fig 2C and 2D, insets). Notably, the D chain Arg456_CspC is unstructured in the CspC_D:CspA_C heterodimer, so one of these salt bridges (Arg456_CspC and Asp1008_CspA) is absent (S7C Fig, inset). Previously, we showed that mutations of residues Asp429_CspC, Arg456_CspC, and Gly457_CspC enhance the sensitivity of spores to germinant and/or co-germinant signals [27]. These mutations would be expected to disrupt the salt bridges at the binding interface, potentially destabilizing the CspC:CspA heterodimer. Thus, the CspC:CspA heterodimer may be important for sensing both germinant and co-germinant signals.

Mutation of the two salt bridges at the CspC:CspA heterodimer interface increases the sensitivity of C. difficile spores to TA germinant but has no effect on heterodimer stability

To dissect the functional roles of the aforementioned salt bridges, we mutated Asp429_CspC, Arg456_CspC, Arg1036_CspA, and Asp1008_CspA (Figs 2C and 3A) and tested the effect of the point mutations on C. difficile germination using an optical density (OD₆₀₀)-based germination assay, which measures the characteristic OD₆₀₀ drop of germinating spores that occurs due to spore cortex degradation and core rehydration [17,18,27] (S8 Fig). The point mutations were introduced into cspC or cspBA complementation constructs expressed from an ectopic locus [33], and the resulting mutants were tested for their ability to rescue the germination of ∆cspC and ∆cspBA mutants.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Mutation of two salt bridges at the CspC:CspA heterodimer interface increases the sensitivity of Clostridioides difficile spores to TA germinant.

(A) Two salt bridges at the center of the CspC:CspA heterodimer interface. (B, C) Gemination levels based on the change in optical density (OD₆₀₀) of purified spores suspended in rich medium over time following the addition of 0.6 mM taurocholate (TA). Germination in arbitrary units (AU) was calculated using the area below inverted OD₆₀₀ curves (S8 Fig). cspC complementation mutants were constructed in a ∆cspC background (B). cspBA complementation mutants were constructed in a ∆cspBA background (C). CspA residue numbers are based on the full-length CspBA fusion protein. Statistical significance relative to WT was determined using a one-way ANOVA and Dunnett’s multiple comparisons test. *** p < 0.001, ** p < 0.01. (D–F) Size exclusion chromatography analysis of CspC-CPD-His₆ and CspA co-affinity purifications with the indicated mutants. The dashed line indicates the separation between the two peaks. All C. difficile data shown (B, C) are representative of a minimum of three independent replicates. The SEC data shown were obtained once. The data underlying panels B–F can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003610.g003

We found that cspC_D429A spores were only slightly more sensitive to TA germinant, whereas cspC_R456A spores were significantly more sensitive to TA germinant (Fig 3B, p < 0.01). These results are consistent with our prior report showing that substitution at residue R456_CspC (cspC_R456G) resulted in spores that were more sensitive to TA germinant than a substitution at residue D429_CspC (cspC_D429K) [27]. The double mutant cspC_D429A/R456A also produced spores with increased TA germinant sensitivity, albeit slightly less than spores carrying the single cspC_R456A mutation (Fig 3B). cspA_R1036A and cspA_D1008K spores were significantly more sensitive to TA germinant (Fig 3C, p < 0.001), whereas cspA_D1008A spores exhibited WT TA germinant sensitivity (Fig 3C). cspA_{D1008A/R1036A} double mutant spores germinated similarly to the single cspA_R1036A mutant spores (Fig 3C).

Since we previously showed that mutations in this region affected co-germinant sensitivity [27], we tested the response of each of these mutants to the co-germinant calcium. To do this, we performed the OD-based germination assay in a buffer rather than rich medium, keeping the TA concentration constant as we varied the concentration of CaCl₂ [27]. We did not use glycine for this assay since endogenous calcium released from the spore during germination acts as a co-germinant; therefore, glycine-enhanced germination cannot be uncoupled from the contribution of calcium [12]. These analyses revealed that substitutions in the central dynamic region of the heterodimer also increased the sensitivity of spores to calcium, with the mutant spores exhibiting similar phenotypic trends to those observed in the TA-based assay (Figs 3B, 3C, and S9).

Western blot analyses confirmed that none of the salt-bridge substitutions altered Csp levels in spores relative to WT, although slight differences in mobility were observed with the CspC and CspA substitutions (S10 Fig). These differences in mobility are likely due to the altered charge of the Csp variants. Of note, increasing the resolution of the SDS-PAGE revealed multiple isoforms of CspA (S10 Fig). The three distinct bands observed may reflect the ability of YabG to target multiple sites during its interdomain processing of CspBA [18,22].

The location of these residues at the interface of the CspC:CspA heterodimer suggested that their substitution may destabilize the complex. To test the effect of these mutations on heterodimer stability in vitro, we used co-affinity purification followed by SEC analysis. We found that substituting both the CspC and CspA residues involved in the two salt bridges for alanine (CspC_D429A/R456A or CspA_{D1008A/R1036A}) had no effect on heterodimer stability in vitro, with both mutant heterodimers eluting like WT (Fig 3D–3F). Taken together, these data show that mutations in the central salt bridge of the CspC:CspA heterodimer increase TA germinant sensitivity but have no effect on heterodimer stability in vitro.

Mutations in the CspC:CspA heterodimer peripheral H-bonds reduce the sensitivity of C. difficile spores to TA germinant and have a minimal effect on heterodimer stability

To further probe the role of the CspC:CspA heterodimeric interface in C. difficile germination, we next targeted the H-bond network formed by the residues Gln516_CspC, Thr520_CspC, and Arg896_CspA, using mutagenesis (Figs 2C and 4A). Previously, we showed that cspC_Q516E and cspC_Q516R spores had WT sensitivity to TA germinant [27]. Here, we found that single alanine substitutions cspC_Q516A, cspC_T520A, and cspA_R896A also had no effect on sensitivity to TA germinant at high concentrations (18.6 mM) (Fig 4B and 4C). However, double cspC_Q516A/T520A and triple cspBA_R896A:cspC_Q516A/T520A mutant spores had decreased germinant sensitivity (Fig 4B and 4D). Moreover, substituting two or three H-bond network residues with glutamates (CspC_Q516E/T520E and CspBA_R896E:CspC_Q516E/T520E), which should disrupt the H-bond network by charge repulsion, completely abrogated germinant sensing (Fig 4B and 4D); even the single glutamate substitution CspBA_R896E partially decreased germinant sensitivity (Fig 4C).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Mutations in the CspC:CspA heterodimer peripheral H-bonds impair germination and impact heterodimer formation.

(A—top) Tripartite hydrogen bond network in the CspC:CspA heterodimer, and (A—bottom) hydrogen bond between the Y361_CspC main chain and the Q1090_CspA residue. (B, C) Germination levels based on the change in optical density (OD₆₀₀) of purified spores suspended in rich medium over time following the addition of the 18.6 mM taurocholate (TA). Germination in arbitrary units (AU) was calculated using the area below inverted OD₆₀₀ curves (S8 Fig). cspC complementation mutants were constructed in a ∆cspC background (B). cspBA complementation mutants were constructed in a ∆cspBA background (C). cspBAC complementation mutants were constructed in a ∆cspBAC background (D). CspA residue numbers are based on the full-length CspBA fusion protein. ∆cspBAC/3xAla = triple-Ala substitution cspBA_R896A-cspC_Q516A/T520A, ∆cspBAC/3xGlu = triple-Glu substitution cspBA_R896E-cspC_Q516E/T520E. Statistical significance relative to WT was determined using a one-way ANOVA and Dunnett’s multiple comparisons test. **** p < 0.0001. (E–L) Size exclusion chromatography (SEC) analyses of CspC-CPD-His₆:CspA co-affinity purifications. The dashed line indicates the separation between the two peaks. The trace for CspC_WT-CPD-His₆:CspA_WT is reproduced from Fig 3D. All Clostridioides difficile data shown (B–D) are representative of a minimum of three independent replicates. E–L are representative of a minimum of two independent replicates, except for CspC_WT-CPD-His₆:CspA_Q1090A, which was resolved via SEC once. The data underlying panels B–L can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003610.g004

We next probed the role of a single hydrogen bond between the main chain carbonyl group of residue Tyr361_CspC and residue Gln1090_CspA on the reciprocal side of the heterodimer interface from the H-bond network (Figs 2C and 4A) by substituting residue Gln1090_CspA for alanine. Similar to the heterodimer H-bond network mutations, the cspA_Q1090A mutation greatly reduced TA germinant sensing even at high concentrations (18.6 mM) (Fig 4C, p < 0.0001). None of the mutations affected Csp levels in spores (S11 Fig), indicating that the observed germination defects are not due to reduced levels of germination proteins in spores. Taken together, these data imply that disrupting the H-bonds at the periphery of the CspC:CspA heterodimer interface impairs TA germinant sensing.

To test the effects of peripheral H-bond substitutions on heterodimer stability in vitro, we used co-affinity purification followed by SEC (Fig 4E–4L). Most mutations had no effect on the stability of the CspC:CspA heterodimer. However, the double glutamate mutant CspC_Q516E/T520E had a reduced CspC:CspA heterodimer peak by SEC (Fig 4G), and the amount of untagged CspA in the co-affinity pull-down was reduced compared to WT CspC (S12A Fig). Notably, similar to the WT heterodimer (Fig 1B), the CspC_Q516E/T520E-CPD-His₆:CspA heterodimer remained stable upon being re-purified using SEC (S13A Fig). These data demonstrate that the CspC_Q516E/T520E-CPD-His₆:CspA heterodimer is stable, indicating that the SEC analyses primarily reflect the efficiency of the co-affinity pull-down. Consistent with this conclusion, variable amounts of CspC_Q516E/T520E-CPD-His₆:CspA heterodimer in the co-affinity purifications resulted in different proportions of CspC_Q516E/T520E-CPD-His₆:CspA heterodimer to free CspC-CPD-His₆ and CspA homodimer observed in the SEC analyses (S14 Fig). This variability may result from the sensitivity of competitive interactions, caused by the change in equilibria between the CspC:CspA heterodimer and CspA:CspA homodimer mutants, which can occur due to slight variations in the E. coli growth conditions or incubation times during co-affinity purification. Nonetheless, our data support the conclusion that the Q516/T520_CspC residues promote CspC:CspA heterodimerization. Moreover, since the cspC_Q516E/T520E mutant is one of two germination-null mutants (Fig 4B), disruption of this H-bond network reduces both heterodimer formation in vitro and TA germinant sensing.

All other mutants tested eluted similarly to the WT heterodimer, apart from the CspA_R896A and CspA_R896E variants, which eluted exclusively in complex with CspC (Fig 4H and 4I). Notably, Arg896_CspA also participates in H-bond networks at the CspA homodimer interface (Fig 2B, inset). Since the second peak in the WT CspC-CPD-His₆:CspA SEC trace contains both the CspC monomer and some CspA homodimer (Figs 1B and 4E), disrupting the interactions formed by Arg896_CspA likely impairs CspA homodimerization, promoting CspC:CspA interaction.

Like the CspC_Q516E/T520E mutant, the triple glutamate substitution in the H-bond network of the CspC:CspA heterodimer (CspC_Q516E/T520E:CspA_R896E) abrogated germination at high concentrations of TA germinant (Fig 4D). However, unlike the double glutamate mutant CspC_Q516E/T520E (Fig 4G), the triple mutant CspC_Q516E/T520E:CspA_R896E eluted like the WT heterodimer (Fig 4L). Since CspA_R896E may impair CspA homodimerization, promoting CspC:CspA interaction, it is possible that combining this mutant with CspC_Q516E/T520E counterbalances the destabilizing effect of the CspC mutations, resulting in an SEC trace similar to the WT heterodimer. Regardless, these data reveal that there is no direct correlation between heterodimer stability in vitro and TA germinant sensitivity in spores and that the H-bond network at the interface of CspC:CspA is important for heterodimerization and critical for germinant sensing.

Mutations in the CspA H-bond networks disrupt CspA homodimerization but have variable effects on germination

While our analyses pointed to the importance of the CspC:CspA heterodimer in C. difficile germinant signaling, the role of the CspA homodimer detected in vitro (Figs 1 and 2) in germination remained unclear. To address this, we designed mutations targeting the twin H-bond networks in the CspA homodimer, formed by Arg896, Gln1094, and Thr1098 (Figs 2B and 5A). Arg896_CspA is also involved in the analogous H-bond network in the CspC:CspA heterodimer (Fig 2B, inset and 2C, left inset), and Arg896_CspA mutants reduced C. difficile TA germinant sensing and favored CspC:CspA heterodimer formation in vitro (Fig 4). To further interrogate the role of the twin H-bond networks in the CspA homodimer, we generated cspA mutants encoding double and triple substitutions targeting H-bond network residues. Similar to cspBA_R896E mutant spores, cspBA_{Q1094E/T1098E} mutant spores were less sensitive to TA germinant at both low and high germinant concentrations (Fig 5B and 5C). While the cspBA_{Q1094E/T1098E} mutant spores were less sensitive to TA germinant at high concentrations of germinant (18.6 mM, Fig 5C), these phenotypes were relatively subtle compared to the analogous mutations in the CspC:CspA heterodimer H-bond network, which completely abrogated TA germinant sensing (cspC_Q516E/T520E, Fig 4B). By contrast, cspBA_R896E/Q1094E and cspBA_{R896A/Q1094A/T1098A} (3xAla) mutant spores were more sensitive to low concentrations of TA germinant (0.6 mM) (Fig 5B). Since none of the CspA mutations impacted Csp levels in spores in western blot analyses (S15 Fig), these data indicate that mutation of the CspA homodimer H-bond networks had relatively minor and variable effects on germinant sensitivity compared to the analogous mutations of the H-bond network in the CspC:CspA heterodimer (Fig 4).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Mutations in the CspA twin H-bond networks disrupt CspA homodimerization but have variable effects on germination.

(A) One of the two tripartite hydrogen bond networks at the CspA homodimer interface. (B, C) Germination levels based on the change in optical density (OD₆₀₀) of purified spores suspended in rich medium over time following the addition of the indicated concentration of taurocholate (TA). Germination in arbitrary units (AU) was calculated using the area below inverted OD₆₀₀ curves (S8 Fig). cspBA complementation mutants were constructed in a ∆cspBA background. CspA residue numbers are based on the full-length CspBA fusion protein. ∆cspBA/3xAla = cspBA_{R896A/Q1094A/T1098A}. The cspBA_R896A and cspBA_R896E measurements presented in (C) are reproduced from Fig 4C. Statistical significance relative to WT was determined using a one-way ANOVA and Dunnett’s multiple comparisons test. **** p < 0.0001, *** p < 0.001, ** p < 0.01. (D–I) Size exclusion chromatography (SEC) analyses of CspA-His₆ variants following affinity purification. The dashed line indicates the separation between the two peaks. All Clostridioides difficile data shown (B, C) are representative of a minimum of three independent replicates. The data shown in D–I were obtained once, except for CspA_R896A-His₆, which was resolved via SEC using three independent replicates. The data underlying panels B–I can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003610.g005

To determine how the H-bond network substitutions affected CspA homodimerization in vitro, we analyzed the SEC profiles of purified CspA-His₆ variants. While alanine substitutions in H-bond network residues (CspA_R896A-His₆ and CspA_{R896A/Q1094A/T1098A}-His₆) did not affect CspA homodimerization (Fig 5D, 5E, and 5I), even single glutamate substitutions were sufficient to disrupt the CspA homodimer formation (CspA_R896E-His₆, CspA_{Q1094E/T1098E}-His₆, and CspA_R896E/Q1094E-His₆, Fig 5F–5H). Importantly, we found no correlation between the effect of the mutations on CspA homodimer stability and C. difficile spore sensitivity to TA germinant. The double glutamate substitution cspBA_R896E/Q1094E disrupted CspA homodimerization in vitro (Fig 5H) and slightly increased spore sensitivity to TA germinant (Fig 5B), whereas other glutamate substitutions (CspA_R896E and CspA_{Q1094E/T1098E}) disrupted CspA homodimer formation (Fig 5F and 5G) and decreased spore sensitivity to TA germinant (Fig 5B and 5C). Further, CspA substitutions in the other residues of the homodimer H-bond network stabilized the CspC:CspA heterodimer in vitro in SEC analyses (S16 Fig), further supporting the hypothesis that disruption of the CspA homodimer promotes the CspC:CspA heterodimer formation. Since no notable correlation was observed between CspA homodimer formation in vitro and C. difficile spore TA germinant sensitivity, our data suggest that CspA homodimerization, which occurs under in vitro conditions, has unclear physiological significance.

The CspC:CspA heterodimer is extremely stable in the presence of germinants and co-germinants

Since our mutagenesis revealed that residues at the CspC:CspA heterodimer interface are important for germinant sensing (Figs 3 and 4), we considered the possibility that disruption of the CspC:CspA heterodimer is a key step in TA germinant and co-germinant sensing. However, despite extensive mutagenesis across the CspC:CspA heterodimer interface, we identified only a single CspC variant that partially disrupted the heterodimer in vitro (CspC_Q516E/T520E, Fig 4G). This finding strongly suggests that the CspC:CspA heterodimer is extremely stable. To assess its stability, we performed limited proteolysis on the CspC-CPD-His₆:CspA heterodimer, the CspC-CPD-His₆ monomer, and the CspA-His₆ homodimer. Consistent with prior work [27], the CspC monomer was sensitive to high levels of chymotrypsin (Fig 6A). The CspA homodimer was similarly sensitive to chymotrypsin (Fig 6A). By contrast, the CspC:CspA heterodimer was highly resistant to proteolysis by chymotrypsin, with the most notable degradation occurring with the top ~82 kDa CspC-CPD-His₆ band at 40 µg/mL of chymotrypsin (Fig 6A). By analyzing the limited proteolysis reactions with CspC and CPD antibodies using western blotting, we determined that the CPD tag was susceptible to proteolysis while CspC remained largely intact (S17 Fig). Notably, substitutions in the central dynamic region (CspC-CPD-His₆:CspA_{D1008A-R1036A}) and in the H-bond network (CspC_Q516E-T520E-CPD-His₆:CspA) of the heterodimer did not change the susceptibility of the complex to proteolysis (S18A and S18B Fig). This is consistent with our findings that both the WT and CspC_Q516E-T520E-CPD-His₆:CspA heterodimer complexes remain stable upon being re-run on the SEC column (Figs 1B and S13A). Taken together, these data show that the heterodimer is a stable complex resistant to proteolysis even with the amino-acid substitutions at its interface.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. The CspC:CspA heterodimer is extremely stable in the presence and absence of germinants and co-germinants.

(A, B) Limited proteolysis analyses of purified CspC, CspA, and the CspC-CPD-His₆:CspA heterodimer using the indicated concentrations of chymotrypsin. CspA-His₆ and CspC-CPD-His₆:CspA heterodimer were affinity-purified. The CspC monomer was affinity-purified as a CspC-CPD-His₆ fusion. Cleavage of the CspC-CPD-His₆:CspA heterodimer at 40 μg/mL chymotrypsin (*) is due to digestion of the CPD-His₆ tag (S17 Fig). (B) Limited proteolysis of purified CspC-CPD-His₆:CspA heterodimer in the presence and absence of the indicated concentrations of taurocholate (TA) and glycine (Gly). Ca²⁺ co-germinant was not tested in this assay because high concentrations of CaCl₂ inhibit chymotrypsin activity. (C) Size exclusion chromatography analyses of CspC-CPD-His₆:CspA co-affinity purifications in the presence or absence of the indicated concentrations of TA and CaCl₂. All data shown are representative of at least two independent replicates. The raw gel images in panels A and B can be found in S1 Raw Images. The data underlying panel C can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003610.g006

We next considered the possibility that germinants and co-germinants might disrupt the heterodimer or induce conformational changes detectable by limited proteolysis. However, dual exposure to TA germinant and the co-germinant glycine did not alter the protease sensitivity of the WT or mutant CspC:CspA heterodimers (Figs 6B, S18A, and S18B). We did not test the influence of calcium on the digestion of the heterodimer, as calcium can alter the activity of chymotrypsin [34]. We next assessed whether the CspC-CPD-His₆:CspA heterodimer might be destabilized during SEC in the presence of TA germinant and calcium co-germinant, since this method does not analyze proteins at equilibrium. No difference in the SEC profiles of the WT or mutant heterodimers was observed in the presence or absence of germinant and co-germinant (Figs 6C, S19A, and S19B). Thus, the presence of the TA germinant and the co-germinants does not affect the stability or overall conformation of the CspC:CspA heterodimer in vitro.

Two regions of the CspC:CspA heterodimer function independently to regulate germinant and co-germinant signal integration

Our mutational analyses revealed that disrupting interactions across two distinct regions of the heterodimer interface results in markedly distinct phenotypes: the salt bridge substitutions enhance germinant sensitivity [27] (Fig 3), while the peripheral H-bond substitutions impair germination (Fig 4). However, it remained unclear whether each region plays a functionally distinct, independent role in signaling or whether the regions signal through a linear, i.e., hierarchical, pathway. To distinguish between these possibilities, we combined mutations with the strongest germination phenotypes in a quadruple mutant ∆cspBAC/cspBA_{D1008A/R1036A}:cspC_Q516E/T520E (4x mut) (Figs 2C and 7A). We then used this mutant to assess whether one region functionally constrains the other—in other words, is epistatically dominant—or if the two regions function independently, in parallel [35,36].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Two regions of the CspC:CspA heterodimer function independently to regulate germinant and co-germinant signal integration.

(A—top) Tripartite hydrogen bond network in the CspC:CspA heterodimer, and (A—bottom) salt bridges at the center of the CspC:CspA heterodimer interface. (B, C) Gemination levels based on the change in optical density (OD₆₀₀) of purified spores suspended in rich medium over time following the addition of 0.6 mM taurocholate (TA) (B) or 4.7 mM TA (C). Germination in arbitrary units (AU) was calculated using the area below inverted OD₆₀₀ curves (S8 Fig). Germination phenotypes of double and quadruple mutants were best resolved at 4.7 mM TA (S8 Fig). cspBAC complementation mutants were constructed in a ∆cspBAC background; cspBA complementation mutant was constructed in a ∆cspBA background; and cspC complementation mutants were constructed in a ∆cspC background. CspA residue numbers are based on the full-length CspBA fusion protein. ∆cspBAC/4x mut = cspBA_{D1008A/R1036A}-cspC_Q516E/T520E. Statistical significance relative to WT was determined using a one-way ANOVA and Dunnett’s multiple comparisons test. **** p < 0.0001, ** p < 0.01. (D) Size exclusion chromatography (SEC) analyses of CspC-CPD-His₆:CspA co-affinity purifications. The dashed line indicates the separation between the two peaks. The traces for CspC_WT-CPD-His₆:CspA_WT, CspC_WT-CPD-His₆:CspA_{D1008A/R1036A}, and CspC_Q516E/T520E-CPD-His₆:CspA_WT are reproduced from Figs 3D, 3F, and 4G, respectively. All Clostridioides difficile data shown (B, C) are representative of a minimum of three independent replicates, apart from ∆cspC/cspC_Q516E/T520E at 0.03% TA, which was performed with n = 2. The data shown in D were obtained twice, except for CspC-CPD-His₆:CspA_{D1008A/R1036A}, which was resolved via SEC once. The data underlying panels B–D can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003610.g007

If one region exhibits epistasis over the other, then a mutant containing substitutions in both regions would have the same phenotype as the mutant containing substitutions only in the dominant region [35]. Instead, we found that 4× mutant spores exhibited an intermediate germinant phenotype: they germinated better than the germination-null cspC_Q516E/T520E mutant spores but worse than the hypersensitive cspA_{D1008A/R1036A} mutant spores (Fig 7B and 7C). Notably, the 4x mutations resulted in slightly decreased levels of Csps in dormant spores (~30%, S20 Fig). While this modest decrease in Csp levels could partially contribute to reduced germination, the cspC_Q516E/T520E and cspA_{D1008A/R1036A} mutations had no impact on Csp levels in spores (S10B and S11A Figs), and their phenotypes reflect changes in CspC:CspA function rather than abundance. Therefore, the 4× mutant phenotype likely reflects a partial contribution from both regions, suggesting that neither region is epistatically dominant over the other. Thus, signal transduction through these regions of the CspC:CspA heterodimer does not occur in a hierarchical fashion and, instead, the two regions likely signal through independent parallel pathways.

We next analyzed the effect of the 4x substitutions on the stability of the CspC:CspA heterodimer using recombinant proteins. We found that combining the 4× substitutions decreased CspC:CspA heterodimer stability even more than the CspC_Q516E/T520E variant, with less untagged CspA_{D1008A/R1036A} being pulled down by the CspC_Q516E/T520E-CPD-His₆ variant during affinity purification compared with WT and the constituent CspA or CspC mutants alone (S12B Fig). Furthermore, the 4× mutant heterodimer was less stable in SEC analyses than the CspC_Q516E/T520E mutant (Fig 7D). Despite targeting two regions of the CspC:CspA heterodimer, the 4× mutant did not fully disrupt heterodimer formation in vitro (Fig 7D). This finding suggests that the Y361_CspC:Q1090_CspA main chain H-bond contributes to the stability of the CspC:CspA heterodimer (Figs 2C, middle inset and 4A, lower inset). Like the CspC_Q516E/T520E mutant, the 4× mutant displayed variability across replicates in the amount of untagged CspA pulled down in the co-affinity purifications and the relative amount of CspC_Q516E/T520E-CPD-His₆:CspA_{D1008A/R1036A} heterodimer observed in SEC analyses (S14B, S14D, and S14E Fig). Nevertheless, re-running the SEC-purified 4× mutant (CspC_Q516E/T520E-CPD-His₆:CspA_{D1008A/R1036A}) complex on SEC resulted in a stable heterodimer (S13B Fig), similar to the CspC_Q516E/T520E mutant (Fig 13A), highlighting the stability of the mutant complexes. Accordingly, the 4× mutant exhibited the same chymotrypsin digestion pattern with or without TA and glycine as WT and the previously tested mutants (Figs 6A, 6B, and S18). TA and calcium also had no effect on the elution of the 4× mutant by SEC (S19C Fig).

Collectively, we found no direct correlation between germinant sensitivity and CspC:CspA heterodimer stability in vitro (Fig 7B–7D). Thus, our data suggest that, while the CspC:CspA heterodimer plays a central role in germinant sensing, its stability in vitro does not determine germinant sensitivity. Indeed, the behavior of this complex in vitro may not accurately reflect what occurs in the C. difficile spore. These data also support the idea that different regions at the heterodimer interface function independently to mediate germinant signal transduction.

Bacterial strains and growth conditions

Escherichia coli DH5α strains are listed in S1 Table and were grown in Luria-Bertani (LB) broth at 37°C with 225 rpm shaking. For protein production, E. coli BL21(DE3) strains were grown in autoinduction broth (Terrific broth [Thermo Fisher] supplemented with 0.5% glycerol, 0.05% glucose, and 0.1% α-lactose monohydrate). Media were supplemented with 20 μg/mL chloramphenicol, 100 μg/mL ampicillin, or 30 μg/mL kanamycin as needed.

C. difficile strains are listed in S2 Table and were grown on brain heart infusion medium supplemented with 0.5% w/v yeast extract and 0.1% w/v L-cysteine (BHIS) with TA (0.1% w/v; 1.9 mM), thiamphenicol (10–15 μg/mL), kanamycin (50 μg/mL), or cefoxitin (8 μg/mL) as needed. C. difficile defined medium (CDDM) was used for constructing complementation strains [54]. Cultures were grown swirling at 37°C under anaerobic conditions using a gas mixture containing 85% N₂, 5% CO₂, and 10% H₂.

E. coli strain construction

All plasmids used in this study are compiled in S1 Table, containing external links to plasmid maps with primer sequences used during cloning. C. difficile genes were codon-optimized (co) for E. coli, and plasmids were cloned via Gibson assembly and transformed into E. coli (DH5α). All plasmids were confirmed with whole plasmid sequencing by Plasmidsaurus using Oxford Nanopore Technology. Confirmed plasmids were transformed into E. coli HB101 for conjugation with C. difficile or E. coli BL21(DE3) for protein expression.

BL21(DE3) E. coli strains were designed to encode Csps under the control of a pT7_lac promoter (Novagen). CspC-CPD-His₆ and CspA-His₆ were expressed in pET22b plasmids containing an ampicillin resistance cassette (Novagen), while untagged CspA was expressed in a pRSFDuet1 plasmid containing a kanamycin resistance cassette (Novagen).

Recombinant protein affinity purification

BL21(DE3) E. coli strains were grown in 20 mL LB with antibiotics as needed, back-diluted 1:1,000 into 1 L autoinduction broth with antibiotics as needed and grown for ~60 hours at 20°C with 225 rpm shaking. Cultures were pelleted, resuspended in 50 mL low imidazole buffer (LIB; 500 mM NaCl, 50 mM Tris-HCl pH 7.5, 15 mM imidazole, 10% glycerol, 2 mM β-mercaptoethanol), and flash frozen in liquid nitrogen. Once thawed, cells were probe sonicated [Branson] in 3 × 45-s rounds at 40% amplitude with 5 min on ice between. The insoluble material was pelleted, and tagged proteins were affinity-purified from cleared lysates using Ni-NTA agarose beads [Qiagen] with gentle rocking at 4°C for 2 hours. The beads were washed three times with LIB before elution with either high imidazole buffer (HIB; 500 mM NaCl, 50 mM Tris-HCl pH 7.5, 200 mM imidazole, 10% glycerol, 2 mM β-mercaptoethanol) or by inducing cleavage of the CPD tag [29,30] with 200 μM inositol hexakisphosphate (InsP₆, Sigma) in LIB. Samples of whole cells before purification (induced), after lysis and removal of insoluble material (cleared lysate), and elution were analyzed by SDS-PAGE with Coomassie staining.

Size exclusion chromatography

Affinity (or co-affinity) purified protein was buffer-exchanged into SEC buffer (200 mM NaCl, 10 mM Tris-HCl pH 7.5, 5% glycerol, 1 mM dithiothreitol) and concentrated using an Amicon Ultra-15 30 kDa cutoff centrifugal filter [Millipore Sigma]. Proteins were separated using a Superdex 200 Increase 10/300 GL column [GE] on an AKTA Pure high-pressure liquid chromatography instrument [GE] with the following parameters: 500 μL loop volume, empty loop with 1 mL; 0.2 mL/min flow rate for 1 column volume. Apparent MWs were calculated using a BioRad gel filtration standard. Fractions were collected every 0.5 mL for SDS-PAGE analyses using Coomassie staining or Western blotting.

CspA and CspC mixing experiments

CspA and CspC were purified as described above, mixed in a 1:1 ratio (w:w) in SEC buffer, and incubated for 2 hours at room temperature. The mixture was analyzed by SEC as described above. SEC fractions were analyzed by SEC followed by Western blotting. Western blots were imaged using LiCOR Odyssey instrument. Blots were quantified in LiCOR Image Studio software and each band was normalized to the sum of all bands in the respective channel.

Protein purification for crystallography

BL21(DE3) strains #3330 (S1 Table) containing pET22b-CspC-CPD-His₆ and pRSFDuet1-CspA_F944E/_Y1092E plasmids, and strain #3127 (S1 Table) containing pET22b-CspA-His₆ were used for crystallography. For the case of the CspA homodimer, Ni-NTA agarose beads were eluted using high imidazole buffer. For the CspA:CspC heterodimer, instead of eluting the proteins using high imidazole, InsP₆ was used to cleave the C-terminally CPD-His₆ tag [29,30] from CspC to purify the untagged version of the CspC-CspA heterodimer. Ni-NTA agarose beads were incubated with 200 µM InsP₆ in LIB buffer with gentle shaking at 4°C overnight. For both protein complexes, after the incubation with high imidazole or InsP₆, Ni-NTA agarose beads were pelleted, and the supernatant collected. Pooled proteins were concentrated and injected into the Superdex 200 Increase 10/300 GL column. Fractions from the SEC peaks containing the pure CspA homodimer and CspA:CspC heterodimer, respectively, were collected and re-concentrated using Amicon Ultra-15 30 kDa cutoff centrifugal filter [Millipore Sigma].

Crystallization and structure determination

C. difficile strain construction

Complementation strains were constructed as previously described [27,29,30]. Briefly, mutant csp genes were assembled in pMTL-YN1C plasmids containing the truncated pyrE gene [33,58], cloned into DH5α cells, and sequence-confirmed using Plasmidsaurus. Sequence-confirmed plasmids were transformed into HB101 E. coli strains containing the necessary genes for conjugation. HB101 strains containing mutant csp complementation plasmids were combined with the associated ∆csp C. difficile strain to construct the C. difficile complementation strains. CDDM media was used to select for recombination of the construct into the pyrE locus, which restores the uracil prototrophy. Two independent clones from each conjugation were phenotypically characterized.

Sporulation and spore purification

Sporulation and spore purification were performed as previously described [27,44]. Briefly, C. difficile strains were streaked from frozen glycerol stocks onto BHIS plates and grown overnight. Liquid BHIS cultures were grown until they reached an OD₆₀₀ between 0.35 and 0.75. To induce sporulation, 100–120 μL of these cultures were spread onto 70:30 agar plates. Spores were collected from 70:30 plates a minimum of 60 hours later and resuspended in ice-cold sterile water. To confirm proper levels of sporulation, a small sample was removed and examined via phase-contrast microscopy. Spore samples were washed with ice-cold water 5–8 times and incubated on ice overnight. The next day, the samples were treated with DNAse I [New England Biolabs] for 1 hr at 37°C. After treatment, the spores were washed 1–2 times in ice-cold water and then purified on a 20%−50% Histodenz [Sigma Aldrich] gradient. After 2 more washes with water, spore purity was confirmed using phase-contrast microscopy. The samples were stored at 4°C until use.

OD₆₀₀ kinetics assay

The OD₆₀₀ kinetics assay protocol was adapted from previous works [11,27]. Spores were resuspended in BHIS media to an OD₆₀₀ of 0.89 (~7 × 10⁵ spore-forming units/μL), and 180 µL were added into wells of a 96-well flat-bottom tissue culture plate [Falcon] for each condition tested. The spores were exposed to different concentrations of TA (1%, 18.6 mM; 0.5%, 9.3 mM; 0.25%, 4.7 mM; 0.125%, 2.3 mM; 0.06%, 1.2 mM; 0.03%, 0.6 mM) or water (untreated) to a final volume of 200 µL per well (final spore OD₆₀₀ of 0.8). The OD₆₀₀ was measured every 3 min using a Synergy H1 microplate reader [Biotek] at 37°C with constant shaking between readings. Readings were subtracted by background OD₆₀₀ of spore-free media, and the values were normalized to the first measurements (time 0) and plotted as OD₆₀₀ versus time (S8 Fig). Germination (arbitrary units) for each strain was calculated by first subtracting each of OD₆₀₀ values from 1 to invert the curves (S8 Fig). The area under the inverted curves was plotted for each strain as a bar graph (S8 Fig). Area under the curve calculations were performed using Prism. All germination assays were initially performed using all 7 concentrations of TA to determine the concentration that best displays a given germinant sensitivity phenotype. This assay was performed for each strain at a minimum of three biological replicates, with three individually prepared spore isolates.

For OD₆₀₀ kinetics assays using TA and calcium, spores were resuspended in 50 mM HEPES buffer + 100 mM NaCl to an OD₆₀₀ of 0.89 and aliquoted into wells of a 96-well plate. The spores were exposed to 0.25% (4.7 mM) TA and different concentrations of CaCl₂ (0.6 mM or 1.25 mM). The rest of the assay was performed as described above.

Western blot analysis

Spore samples for immunoblotting were prepared as previously described [18,22,27]. Spores were diluted to an OD₆₀₀ of 1.4 in 100 μL of sterile water (1.5 × 10⁸ spore-forming units), pelleted, and resuspended in 50 μL EBB buffer (8 M urea, 2 M thiourea, 4% (w/v) SDS, 2% (v/v) β-mercaptoethanol). The samples were boiled for 20 min with vigorous vortexing during incubation, pelleted, and resuspended. Final sample buffer with bromophenol blue was added to 1X to stain the sample. To maximally solubilize spore proteins, the samples were boiled again for 5–10 min, vortexed vigorously, and pelleted. The samples were resolved by 12% SDS-PAGE gels and transferred to Millipore Immobilon-FL PVDF membranes. The membranes were blocked with Odyssey Blocking Buffer with 0.1% (v/v) Tween 20 and probed with mouse monoclonal anti-CspC [22] or anti-SleC [19] and/or rabbit polyclonal anti-CspB [19] or anti-CspA that was raised against SEC-purified CspA-His₆ in rabbits (CoCalico Biologicals). The anti-CspC and anti-CspA antibodies were used at 1:2,000 dilutions, the anti-CspB antibody was used at 1:3,000, and the anti-SleC antibody was used at 1:5,000. The membranes were probed with primary antibodies for 4 hours, washed 3× with PBS with 1.5× Tween 20, and probed with IRDye 680CW and 800CW infrared dye-conjugated secondary antibodies at 1:12,000. After incubation for 1hour in the dark, the membranes were washed twice with PBS with 1.5× Tween 20, and secondary antibody infrared fluorescence emissions were visualized using an Odyssey LiCOR CLx. The results shown are representative of three independently prepared spore isolates.

For western blot analysis of limited proteolysis samples, the samples were probed with mouse anti-CspC antibody and rabbit anti-CPD antibody [59] at a 1:2,000 dilution. The rest of the assay was performed as described above.

Limited proteolysis

Limited proteolysis of purified C. difficile proteins was performed as previously described in Rohlfing and colleagues [27]. Purified C. difficile Csp proteins were diluted to 15 µM in 10 mM Tris pH 7.5 buffer. Twenty-four µL of the protein solution was aliquoted into 0.2 mL PCR tubes. A 1 mg/mL chymotrypsin stock solution was diluted in 10 mM Tris pH 7.5 buffer to generate 10-fold dilutions. 1 µL of the appropriate chymotrypsin dilution was added to the protein solution to achieve the indicated concentration of chymotrypsin. 1 µL of 10 mM Tris pH 7.5 buffer was added to the protein samples as an untreated control. The protein and chymotrypsin solutions were incubated at 37°C for 1 hour. Chymotrypsin activity was then quenched with the addition of NuPAGE 4× LDS Sample Buffer [Invitrogen] and boiled for 3 min at 98°C. The samples were resolved on a 15% SDS-PAGE gel and visualized using Coomassie staining.

For limited proteolysis experiments with the addition of TA and glycine, chymotrypsin was diluted as above, while the proteins were diluted to a final concentration of 7.5 µM in 10 mM Tris buffer. Twenty-three µL of the protein solution was added to the 0.2 mL PCR tubes. A 10% (186 mM) TA stock solution and a 200 mM glycine stock solution were diluted to 25 mM in 10 mM Tris pH 7.5 buffer. 1 µL of the TA/glycine solution was added to the indicated protein samples for final concentrations of 1 mM TA and 1 mM glycine. 1 µL of the appropriate chymotrypsin dilution was added to the protein solution to achieve the indicated concentration of chymotrypsin. Incubation, protease quenching, resolution, and visualization were performed as described above.

Sequence alignment

CspA and CspC sequences were retrieved from Lewis and colleagues [37] and Knight and colleagues [38] (S5 Table) and aligned with default parameters using MUSCLE [60]. Alignments were manually inspected in Geneious Prime 2025.2.2.

Data visualization

All graphs were prepared using GraphPad Prism. Gels and blots were imaged using a LiCOR Odyssey instrument. Protein structure visualization was performed using Pymol.