Clostridium septicum manifests a bile salt germinant response mediated by Clostridioides difficile csp gene orthologs

Abstract

Clostridium septicum infections are highly predictive of certain malignancies in human patients. To initiate infections, C. septicum spores must first germinate and regain vegetative growth. Yet, what triggers the germination of C. septicum spores is still unknown. Here, we observe that C. septicum germinates in response to specific bile salts. Putative bile salt recognition genes are identified in C. septicum based on their similarity in sequence and organization to bile salt-responsive csp genes in Clostridioides difficile. Inactivating two of these csp orthologs (cspC-82 and cspC-1718) results in mutant spores that no longer germinate in the presence of their respective cognate bile salts. Additionally, inactivating the putative cspBA or sleC genes in C. septicum abrogates the germination response to all bile salt germinants, suggesting that both act at a convergent point downstream of cspC-82 and cspC-1718. Molecular dynamics simulations show that both CspC-82 and CspC-1718 bear a strong structural congruence with C. difficile’s CspC. The existence of functional bile salt germination sensors in C. septicum may be relevant to the association between infection and malignancy.

Two new bile salt germinant-responsive genes, which exhibit sequence and structural similarity to Clostridioides difficile cspC, have been identified in the cancer-associated bacteria Clostridium septicum.

Introduction

Clostridium septicum infections are infrequent in human patients, accounting for only 1.3% of clostridial sepsis cases¹. Yet, in a seminal review of 162 cases¹, the vast majority (80.8%) of C. septicum-infected patients were found to have an accompanying malignancy. Subsequent studies have also reported comparable figures (78-80%) for the strong association between C. septicum infections and cancer^2,3. In contrast, only 4% of Clostridium perfringens infections are linked to an existing cancer. C. septicum-linked cancers are usually of colorectal (34%) or hematological (40.1%) origin¹. In many cases (and especially for colorectal cancer), these cancers were unknown and only discovered after the infection. In short, C. septicum infections may be exceedingly rare, but its presence in a patient is highly predictive of a co-existing malignancy.

Besides its strong association with cancer, C. septicum also stands apart from the other pathogenic clostridia in how infections present and progress. C. septicum is the leading cause of spontaneous gas gangrene, which is highly lethal and arises without prior injury³. This lethality is due in large part to C. septicum’s α-toxin, which forms pores on plasma membranes, leading to death by pyroptosis^4,5. Left untreated, infections progress swiftly into fulminant gas gangrene and myonecrosis, with mortality approaching 100%¹. This is distinct from other clostridial infections, where injury is commonly the precursor and cause of infection⁶. Interestingly, when C. septicum is isolated from a patient, it tends to be the sole microorganism isolated, whereas most other clostridial infections result in mixed cultures⁶. This suggests that C. septicum’s host tropism may differ from other pathogenic clostridia.

There has been no explanation for why C. septicum infections are so different from the other pathogenic clostridia, and especially why they are so closely linked with malignancy. Since a clostridial spore can only colonize an environment that supports its germination, we reasoned that studying the germination triggers of C. septicum could provide clues as to why C. septicum preferentially infects cancer patients.

Many of the pathogenic clostridial species, which include C. septicum, C. perfringens and C. botulinum, share a common ancestry and form part of the 16 S rRNA-based phylogenetic group called cluster I⁷. Certain species (e.g. Clostridioides difficile and Paraclostridium bifermentans) which were formerly identified as Clostridium species were later renamed because they were excluded from cluster I by phylogenetic distance⁸. Despite this, germination studies of C. difficile and P. bifermentans are as highly relevant as their cluster I neighbors for understanding the diversity of germination machinery among the sporulating anaerobic bacteria^{7–12,11–14} From these studies, the clostridial germinant vocabulary spans a variety of L-amino acids^11–13,15, inorganic minerals^12,16–18 and organic molecules^11,15,19,20. Cognate receptors that sense these germinants respond to their presence by activating the downstream germination signaling cascade^21,22. For example, in C. perfringens, this germinant sensing function is played by GerKC, a germinant receptor which senses amino acid germinants and consequently activates the Csp (subtilisin-like pseudoprotease) gene products: CspA, CspB and CspC. These activated Csp proteases then proceed to activate pro-SleC by cleaving its N-terminal inhibitory pro-peptide, causing the release of catalytically active SleC^12,23–25. Next, SleC degrades the cortex, leading to calcium-DPA release, core rehydration and eventual spore outgrowth^21,26.

Recently, bile salts have been implicated in shaping the composition of the gut microbiome^27–29. Despite the evidence that the presence of specific bile acids can increase the germination frequency of many sporulating bacteria³⁰, C. difficile is currently the only one with a characterized bile salt-responsive germination sensor. This gene, cspC, was first implicated in bile salt signaling through a mutagenesis screen and further identified as a bile salt germinant receptor through an altered germinant specificity screen³¹. Guided by CspC’s crystal structure, further functional studies identified mutations that diminished CspC’s sensitivity for taurocholate and other co-germinants, hence establishing the role of CspC in taurocholate-mediated germination³². A subsequent study proposed that CspC and CspA inhibited CspB when associated with it, and that these inhibitory effects were released when germinants and co-germinants interacted with CspC and CspA respectively³³. Unhindered CspB would then process pro-SleC to its active SleC form in a manner similar to C. perfringens, leading to spore outgrowth^22,33–35.

In this current study, an unbiased screen of single factor germinants for C. septicum revealed that sodium deoxycholate, sodium chenodeoxycholate and sodium glycocholate are potent germinants. To identify the genes responsible for C. septicum’s bile salt-germination, genes with high similarity to C. difficile’s cspC were first identified as candidate bile salt-sensors. Mutational inactivation of two of these cspC orthologs abolished C. septicum’s bile salt germination response to distinct sets of bile salts. Further, cspBA and sleC mutant spores were unresponsive to all three bile salt germinants, which is consistent with both proteins acting downstream of the cspC orthologs. Finally, structural modeling and molecular dynamics simulations showed high congruence between C. difficile’s CspC and both CspC orthologs in C. septicum. The striking similarity between the bile salt-germinant sensing mechanisms of C. septicum and C. difficile raises the possibility that other gut bacteria may use the same pathway.

Results

Characterization of C. septicum germination

To identify potential germinants, forty-eight compounds known to affect the growth, sporulation or germination of various clostridia (Supplementary Table T1) were individually tested for germinant activity as measured by a decrease in optical density of the spores over time. Of the screened compounds, only the bile salt sodium deoxycholate (DCA) triggered a significant OD drop (Supplementary Fig. S1a). The other two bile salts in the screen, sodium cholate (CA) and sodium taurocholate (TCA), did not elicit germination, suggesting that DCA’s putative germinant activity was specific. In a follow-up experiment, dormant phase bright spores incubated with DCA turned phase dark, confirming that DCA was indeed a germinant (Supplementary Fig. S1b). Spores incubated with CA remained phase bright, supporting the initial screen results. Intriguingly, none of the L-amino acids in the screen were identified as germinants, even though L-amino acids are the dominant class of compounds acting either as germinants or co-germinants in other Bacillus and Clostridium spores^9,36. Beyond single amino acids, the combination of L-asparagine, D-glucose, D-fructose, and K⁺ (AGFK), which is a well-known potent germinant for B. subtilis and C. perfringens spores^12,37, also failed to germinate C. septicum spores (Supplementary Fig. S2).

Since the physical properties of bile salts are affected by pH, DCA’s germinant activity was profiled from pH 6 to 8 (Supplementary Fig. S3a). C. septicum spores were observed to germinate most rapidly in the range of pH 6 to 7. However, a time-dependent OD increase after 100 min was also observed at pH 6 and 6.5. This artifact, possibly caused by complexation between spore-released Ca²⁺ and DCA³⁸, was unrelated to germination since all germinated spores remained phase dark after 100 min. This artifact was investigated by incubating Ca²⁺, DPA and DCA in sodium phosphate buffer at various pHs (Supplementary Fig. S3b). A significant increase in OD was observed at pH 6.0 but not at pH 6.5 or pH 7.0. This pH-dependent effect was specific to DCA and not observed with either CA or GCA. To avoid this artifact in future experiments, all absorbance assays from this point were performed at pH 7.0.

We wondered if C. septicum responded to other bile salts besides DCA. When tested against a panel of bile salts, C. septicum spores germinated not just to DCA, but also to sodium chenodeoxycholate (CDCA) and sodium glycocholate (GCA) (Fig. 1a, Supplementary Fig. S4). GCA induced germination most rapidly, followed by DCA then CDCA. Sodium glycodeoxycholate (GDCA) induced weak partial germination at the highest concentrations, and hence we considered it a borderline germinant (Fig. 1a, Supplementary Fig. S4). No germination was observed with the other bile salts tested - sodium cholate (CA), sodium taurocholate (TCA) and sodium taurodeoxycholate (TDCA). Spore germination in these experiments was confirmed with phase contrast microscopy (Fig. 1b).

Fig. 1. C. septicum germinates robustly in the presence of bile salts.

a Germination measured by optical density (OD) is shown. OD values were normalized to the starting OD. Purified wild-type spores were incubated with 25 mM of each bile salt. Germination, as indicated by decreasing OD values. was observed with CDCA, DCA and GCA. b Representative phase contrast images of the spores at the end of the OD assay in a are shown. Spores treated with CDCA, DCA and GCA were phase dark, while spores treated with CA, GDCA, TCA and TDCA remained phase bright. Scale bar, 2 μm. c Germination response curves were plotted for the three bile salt germinants. Germination was measured at 900 min post-exposure to increasing bile salt concentrations. CDCA was the most potent germinant, followed by DCA and GCA. d Maximal germination velocity curves were plotted for the three bile salt germinants. For (a, c, d) graphed data represent the average at least 3 biologically independent experiments using at least two different spore preparations. For all bile salts, n = 3 except for GCA, where n = 5. Error envelopes (dashed lines) represent the standard error of the mean. For (a), statistical significance relative to water was determined by one-way ANOVA and Dunnett’s test. For (c, d), statistical significance relative to CDCA was determined by unpaired t test between groups at each time point. ****p < 0.0001, ***p < 0.001, **p < 0.01 *p < 0.05.

When spores were incubated with the various bile salts in the panel and plated on LB agar, the top germinators in liquid media (GCA, CDCA, DCA, GDCA) also produced the most colonies by rank order (Supplementary Fig. S5). Only GCA, CDCA and GDCA however were significantly higher than the negative control. Although CA, TCA and TDCA did not produce an OD decrease with the germination assay, they generated slightly higher numbers of colonies on LB agar than the control without bile salt, albeit without statistical significance. In this study, we focused on GCA, CDCA and DCA for subsequent experiments but we also note that solid growth agar may modulate the germinant effects of bile salts on C. septicum. All three germinants behaved somewhat differently in terms of germination extent and velocity as bile salt concentration was varied. Notably, the EC₅₀ values for these germinants are within the physiological concentrations found in the intestinal lumen, and are also in the ballpark of taurocholate concentrations found to germinate C. difficile spores (Fig. 1c)^31,39. Interestingly, germination extent was saturated at 80% for CDCA, compared to 100% for both DCA and GCA. All spores were observed to be either phase bright or phase dark, reflecting that CDCA germinated ~80% of spores in that experiment as opposed to a partial germination of all spores. Next, the maximal germination (V_max) velocity was plotted against concentration for each bile salt (Fig. 1d). All three V_max curves plateaued at different points in the concentration range. CDCA saturated early, with its V_max maxing out at a low value after 6.25 mM. In contrast, the V_max of GCA exceeded both DCA and CDCA after 12.5 mM, and had not plateaued by 50 mM, which was the highest concentration tested. DCA’s V_max response straddled both GCA and CDCA and plateaued at around 12.5 to 50 mM. What is clear is that all three germinants have distinguishable germination kinetics. Notably, in comparison to GCA and DCA, CDCA exhibited slow kinetics, yet non-inferior potency.

Although we had previously demonstrated that single amino acids were insufficient to trigger germination of C. septicum spores, amino acids can act as potent co-germinants for other sporulating bacteria¹¹. To investigate this possibility for C. septicum spores, we screened an admixture of 20 amino acids for their ability to potentiate germination with suboptimal concentrations of CDCA, DCA or GCA. At these concentrations where germination was minimally induced, significant potentiation was observed with GCA, but not with CDCA or DCA (Supplementary Fig 6a). We further screened each of the 20 amino acids with the same suboptimal concentration of GCA (3.13 mM) and discovered that arginine produced the strongest potentiation while glutamic acid, asparagine and aspartic acid had relatively weaker effects (Supplementary Fig 6b).

The effect of heat activation on bile salt-induced germination was also investigated (Supplementary Fig. S7). Two trends were noted. Firstly, the kinetics of germination increased slightly when spores were heat activated at 60 °C or 68 °C as compared to no heat activation. This was reflected most noticeably by the steeper slopes for DCA and CDCA, and the increased ability of the spore to germinate to the weak germinant GDCA. Secondly, as activation temperature increased, the extent of germination induced by GCA decreased as spores went from no heat activation to heat activation at 68 °C. No germination was observed at ≥ 78 °C for all bile salts. With the exception of the weak germinant GDCA, heat activation below 78 °C did not enhance bile salt germination. Hence, heat activation was not incorporated into downstream germination assays in this study.

A correlation of bile acid structure with germinant activity revealed an intriguing observation. Taking GCA as the reference, the loss of either the R2 hydroxyl (as in GDCA) or the R4 conjugated glycine (as in CA) destroyed germination activity, implying that the presence of both features were important for the recognition of GCA (Supplementary Fig. S8). However, DCA lacked both these features but was still a germinant. This raised the possibility that there are at least two distinct proteins which separately respond to DCA and GCA. We next set out to identify the relevant sensors for the three bile salt germinants.

General features of the C. septicum ATCC 11424 genome

To identify the bile salt-responsive elements in C. septicum, we sequenced the C. septicum genome in search of genes with high similarity to C. difficile’s bile-salt germination genes, which are the only characterized bile salt germinant-responsive genes to date. De novo sequencing was necessary because the strain used in this paper (C. septicum ATCC 11424) was not among any of the strains sequenced before⁴⁰. Genome assembly yielded a complete chromosome scaffold with three contigs and two plasmids (Supplementary Figs. S9–S11). The assembled chromosomal scaffold reveals features consistent with other sequenced C. septicum strains: a similar genome size of 3.41 million bp and a low G/C content of 27.79%. PGAP annotation uncovered a total of 3107 CDS, 115 RNA genes and one CRISPR array (Supplementary Table T2). Other features of the genome are described in Supplementary Tables T3-T4, Supplementary Data 2, and Supplementary Figs. S9–S11.

Identification of putative csp operons

Since C. difficile’s subtilisin-like pseudo-protease (CspC) is currently the only known bile salt germinant sensor, we searched for orthologs of CspC in the C. septicum genome. Five cspC orthologs were identified and these were divided among three operons (Supplementary Figs. S12) as predicted by Operon-mapper⁴¹. For ease of reference, these orthologs are referred to in this study by their imputed orthologous gene name (cspC) hyphenated with a number denoting their gene order within the C. septicum genome. All orthologs contained the subtilase domain, which is the hallmark of Csp serine proteases⁹. The Asp-His-Ser (or DHS) catalytic triad within this domain was preserved in all but one ortholog (cspC-1718) where Ser was replaced by Cys, suggesting that cspC-1718 may function as a cysteine protease instead of a serine protease (Supplementary Figs. S13, S14). One of the three identified operons was structured similarly to C. perfringens S40 and C. difficile 630 (Fig. 2a). Specifically, this operon contained the genes cspBA-81, cspC-82 and sleC-83, which had similar counterparts in C. difficile and C. perfringens arranged in the same order (Fig. 2a). Two minor differences inconsequential to this work are however noted. Firstly, cspB and cspA are separate genes in C. perfringens whereas both genes are fused as cspBA in both C. septicum and C. difficile. Secondly, in C. septicum and C. perfringens, sleC follows immediately after cspC, whereas sleC is located further downstream (1.9 Mb) of cspC in C. difficile.

Fig. 2. Identification of putative cspC genes and scheme for genetic inactivation.

a Genetic layout of the csp operons for C. septicum, C. difficile and C. perfringens. cspBA-81, cspC-82 and sleC-83 is part of a pentacistronic operon as predicted by OperonMapper. The other two upstream genes were not predicted to be part of the germination mechanism and were hence omitted to avoid clutter. b Schematic of insertion sites of the TargeTron vector for generation of mutants in the csp/sleC operon and cspC-1718.

Targeted insertional mutagenesis using a group II intron vector^42,43 was performed to probe the function of all seven orthologs (five cspCs, one cspBA and one sleC). For technical reasons, we were unable to inactivate three of the orthologs. The remaining four orthologs shown in Fig. 2b (cspBA-81, cspC-82, sleC-83 and cspC-1718) were successfully inactivated and verified by Southern blot analysis to have single intron insertions (Supplementary Fig. S15). These four mutants were sufficient to provide an insight into the bile salt germination mechanism of C. septicum.

Inactivation of cspC orthologs abrogates germination to specific bile salts

Spores with inactivated cspC-82 or cspC-1718 were evaluated for germination in the presence of bile salts in three ways: (1) drop in optical density, (2) release of calcium-dipicolinate (Ca-DPA) from the spore core^44,45 and (3) phase contrast microscopy.

Inactivation of cspC-1718 attenuated the germination response to GCA and inactivation of cspC-82 did the same for CDCA and DCA (Fig. 3, Supplementary Figs. S16). While intron insertional methods can lead to polar effects on the downstream genes, this was not observed in our mutants. Complementation of the mutants with their respective inactivated gene products restored the original germination response. Similar results were also observed with an independent spore batch, further supporting the previous observations (Supplementary Figs. S17). These results, demonstrating that two gene products were responsible for sensing non-coincidental bile salt germinants, was consistent with our earlier hypothesis that there were at least two bile salt germination sensors in C. septicum. As expected, there was no appreciable change in germination phenotype with the weakly-germinant and non-germinant bile salts - CA, GDCA, TCA and TDCA (Fig. 3, Supplementary Figs. S16, S17). Interestingly, the OD decrease and Ca-DPA release triggered by GDCA was attenuated by cspC-82 inactivation but restored by complementation. This suggests that GDCA, which is structurally similar to GCA, may be weakly recognized by the cognate receptor for GCA.

Fig. 3. Both cspC-82 and cspC-1718 mutants show a differential response to bile salt germinants.

Purified spores of wild-type (WT), cspC mutants (cspC-82::ermB and cspC-1718::ermB) and complemented cspC mutants (cspC-82::ermB pcspC-82 and cspC-1718::ermB pcspC-1718) were incubated with 10 mM bile salts in the presence of oxyrase. a Heat map depicting %OD drop values at the end of the assay. Colors indicate extent of germination (red: germination, yellow: no germination). The cspC-82 mutant was able to germinate only in the presence of GCA while the cspC-1718 mutant was able to germinate in the presence of CDCA and DCA. Complementation of the mutant with their respective gene rescued the mutant phenotype. Data represent the average of n = 2 biologically independent experiments. b Representative phase contrast images of WT, cspC mutants and their respective complemented cspC mutants after 900 minutes of incubation with various bile-salts. Scale bar, 2 μm.

Interestingly, cspC-1718 mutant had approximately one log lower Ca-DPA content compared to its WT counterpart (Supplementary Fig. S18). Although complementation of the cspC-1718 mutant did not reverse this phenotype, germination of the cspC-1718 mutant and its complement still released Ca-DPA, accompanied by the usual OD drop and phase dark spores (Fig. 3, Supplementary Figs. S16, S17).

Inactivation of cspBA-81 and sleC-83 abrogates germination to all bile salts

Both cspBA-81 and sleC-83 are potentially important for bile salt germination because they reside in the same operon as cspC-82, the gene shown by previous experiments to mediate CDCA/DCA-triggered germination. The cspBA, cspC, and sleC genes in C. difficile play a pivotal role in its bile salt germination pathway. Specifically, in C. difficile, CspBA is cleaved by the YabG protease into CspB and CspA⁴⁶. CspB then undergoes autoprocessing to remove its own inhibitory prodomain, resulting in an active form which converts pro-SleC to SleC. The activated SleC proceeds to hydrolyze the spore peptidoglycan to initiate spore germination^34,47,48.

To study if cspBA-81 and sleC-83 in C. septicum played a similar role to their counterparts in C. difficile, spores with these genes inactivated were studied in the same way as the cspC mutants. Here, we observed that inactivation of either gene negated the germination response of C. septicum spores to all three bile salt germinants, and complementation with their respective gene product restored germinant sensitivity, suggesting that no polar effects were induced by Targetron insertional mutagenesis (Fig. 4, Supplementary Fig. S19, S20). Not surprisingly, and similar to the cspC mutants, there was no appreciable change in germination phenotype with the borderline or non-germinant bile salts - CA, GDCA, TCA and TDCA (Fig. 4, Supplementary Figs. S19). These results were also replicated using an independent spore batch (Supplementary Figs. S20).

Fig. 4. Inactivating cspBA or sleC removes the germination response to all bile salt germinants.

Purified spores of cspBA or sleC mutants and their complemented counterparts were incubated with 10 mM bile salts in the presence of oxyrase. a Heat map depicting %OD drop values at the end of the assay. Colors indicate extent of germination (red: germination, yellow: no germination). cspBA-81 and sleC-83 mutants were unable to germinate with any bile salts. Complementation of the mutant with their respective gene rescued the mutant phenotype. Data represent the average of n = 2 biologically independent experiments. b Representative phase contrast images of WT, cspBA and sleC mutants and their respective complements after 900 minutes of incubation with various bile-salts. Scale bar, 2μm.

FInally, we tested the viability of the mutant and complemented spores on BHI-S plates (Supplementary Fig. S21). All cspC mutant spores were viable, but cspC-1718 spores exhibited approximately 10-fold fewer colonies compared to WT. This lower germination efficiency seems to be consistent with the one log lower Ca-DPA content previously observed for cspC-1718 (Supplementary Fig. S18). Complementing cspC-1718 restored viability to WT levels, suggesting that the cspC-1718 may somehow influence spore Ca-DPA content. Compared to the cspC mutants, cspBA and sleC mutant spores were not viable, further corroborating our observations in the previous germination assays. As expected, complementing cspBA and sleC restored colony counts to WT levels.

Structural comparisons of CspC-82 and CspC-1718 with C. difficile CspC

Having established that CspC-82 and CspC-1718 were similar to C. difficile’s CspC in biological function, we wondered what structural similarities existed between the three proteins. A Molecular Dynamics (MD) simulation approach was used to study this question. These MD experiments were not intended to provide functional insight, just to cross-compare the overarching physical attributes of all three proteins.

Homology models for CspC-82 and CspC-1718 were generated using i-TASSER and AlphaFold since crystal structures for both of these proteins were unavailable (Supplementary Data 1). The structural analog identified with the closest similarity to both proteins was C. difficile’s CspC, with a high TM score of 0.899 and 0.936 respectively when both proteins were aligned to it. MD simulations of both proteins were then performed to test and refine the accuracy of their homology-modeled structures (Supplementary Table T5). The same simulation was also performed for C. difficile’s CspC using its known crystal structure³².

First the structural stability of the three proteins were verified through the time evolution of the structure over the simulation runs. Specifically, the RMSD with respect to the initial energy minimized structure was calculated. These RMSD values provide microscopic insight on local dynamics of different segments of the protein structure. The RMSD values were calculated by considering only the non-hydrogen atoms of the protein backbone and amino acid side chains. For all three proteins, three independent simulations were performed using three different initial structures to check the observed simulation results are independent of initial protein configuration (Supplementary Fig. S22). Figure 5a shows the change in RMSD with respect to time for three independent protein simulations of C. difficile’s CspC. For all three simulations, RMSD showed similar time evolution, ensuring that the relative protein dynamics were independent of the initial structures. Similar protein dynamics were also observed for CspC-82 and CspC-1718. Figure 5b shows the time evolution of the RMSD values for all three proteins. For all three proteins, RMSD quickly increased with time and reached a plateau within 20 ns, demonstrating that for all cases, the structures of the proteins stabilized rapidly. The RMSD profiles exhibited stable fluctuations around mean values of 4.07 Å, 4.22 Å and 1.92 Å for CspC-82, CspC-1718 and C. difficile’s CspC respectively. The lower RMSD value of C. difficile’s CspC indicates higher structural rigidity compared to the other two proteins. Based on the observed RMSDs, the first 20 ns was used as the equilibration run and the next 30 ns as the production run.

Fig. 5. Molecular Dynamics Simulation.

The structure and stability of C. septicum’s (CS) CspC-82 and CspC-1718 were compared with C. difficile’s (CD) CspC. a The change in RMSD with respect to time for three independent protein simulations of C. difficile’s CspC is shown. Only non-hydrogen atoms were considered for the calculation of RMSDs. b The time evolution of RMSD from the initial energy minimized structure is shown for the three proteins. c Space-filled representations of CD_CspC, CS_CspC-82 and CS_CspC-1718, each with four domains, are shown. d The time evolution of RMSD measured in (a), but resolved for the individual domains shown in b, were calculated to understand the local structural fluctuations of the three studied proteins.

The RMSD of the different segments of the protein were further analyzed at the domain level to explore the microscopic details of the local configurational fluctuations within the protein. For the purpose of this analysis, we defined four domains within each protein (Supplementary Table T6) and their relative position within the protein structures are shown in Fig. 5c. The domain-level analyses show that the RMSD values for domains one and two were significantly larger for CspC-82 and CspC-1718 in comparison to C. difficile’s CspC, indicating that both domains had higher flexibility (Fig. 5d). On the other hand, the similar RMSDs for domains three and four suggest that all three proteins exhibit similar rigidity in these regions. In summary, the previous observation of higher RMSD values for CspC-82 and CspC-1718 were due to the higher flexibility of domains one and two.

The interface solvent accessible surface area (interSASA) for each protein domain was also analyzed at the domain level. A larger interSASA value implies a greater chance for interaction between that domain and the rest of the protein. Conversely, a smaller interSASA value indicates a lower surface area for interactions and potential for more flexibility. Here, we observed that all three proteins internally showed the same relative pattern of interSASA values across their four domains, with domain 2 having the highest interSASA and domains 1 and 3 having the lowest interSASA (Supplementary Table T7). Within each domain, C. difficile’s CspC showed the highest interSASA while the C. septicum CspC orthologs had similar and smaller interSASA values. This result supports the RMSD analysis showing the comparatively greater structural flexibility of C. septicum’s CspC orthologs at both the protein and domain levels.

To summarize, there was a high degree of general structural congruence between the 3 proteins, providing further corroboration that C. septicum CspC orthologs play a similar bile salt-sensing role to C. difficile’s CspC. Yet, there were also microscopic differences in domain flexibility and interSASA which remain to be explored and could shed light on differences in bile salt specificity or signaling.

Discussion

In this study, we investigated the germinant vocabulary of C. septicum. The most noteworthy finding was the strong germination response to bile salts germinants, and the identification of two cspC genes mediating this effect. While C. difficile has hitherto provided the only example of a cspC gene playing a bile salt germinant-sensing role, the two new examples now identified in C. septicum create more opportunities for understanding the nature of bile salt-sensing by CspC. Further, we speculate that bile salt dysregulation may plausibly link C. septicum with cancer.

Up till now, C. difficile’s cspC has been the only known gene known so far to mediate bacterial spore germination through bile salt sensing^11,31,32. Hence, C. septicum was sequenced with the aim of identifying cspC orthologs which could play the same germinant-sensing function. While cspC orthologs were our best lead, the mere presence of cspC genes does not guarantee bile salt germination. For example, the well-characterized Csp proteases in C. perfringens are not germinant sensors, but instead act downstream of germinant receptor GerKC to activate SleC for cortex hydrolysis^23,24. In another example, P. bifermentans does not germinate in response to bile salts despite the presence of cspC⁴⁹. On the flip side, many microbes (besides C. difficile) which are known to exhibit bile salt-enhanced germination are also predicted in a Blastp search to possess at least one cspC ortholog. These microbes include Clostridium perfringens, Paeniclostridium sordellii (previously Clostridium. sordellii), Clostridium baratii, Clostridium innocuum, Flavonifractor plautii (previously Eubacterium plautii) and Hungatella hathewayi (previously Clostridium hathewayi)^30,50,51. Now that C. septicum has presented us with a second example of cspC’s role in bile salt-sensing, it would be relevant to investigate if the cspC orthologs in the abovementioned microbes play a similar bile salt-sensing role. All four deposited C. septicum genomes in the NCBI database submitted prior to our study contain paralogs identical in sequence to CspC-82 and CspC-1718. The bile salt-sensing function of C. septicum CspCs demonstrated in this study is hence unlikely to be strain-specific. We postulate that CspC’s bile sensing properties extend far beyond C. difficile and C. septicum. Recent work with C. perfringens has reported that taurocholate, glycocholate, taurochenodeoxycholate and taurodeoxycholate can potently induce spore germination when combined with L-alanine⁵¹. If Ger proteins are the canonical sensors for amino acid germinants, it is conceivable that Csp proteins may likewise form the molecular basis for bile acid germinant sensing.

All proteolytically functional Csp proteins belong to the PA clan (Proteases of mixed nucleophile, superfamily A) and have a catalytic triad which is either DHS (aspartate-histidine-serine) or DHC (aspartate-histidine-cysteine). In contrast, C. difficile’s CspC triad has been rendered non-functional through mutation to DTG. Recently, it was speculated that this loss of its catalytic triad was required for C. difficile to evolve a bile salt sensing function independent of its original proteolytic one⁵². The evidence in this study seems to suggest otherwise. Both of C. septicum’s bile salt-sensing orthologs (CspC-1718 and CspC-82) possess catalytic triads, showing that a catalytic triad is not a barrier to gaining a new bile salt-sensing function. Further, this appears to be true regardless of whether the catalytic triad is DHC (as for CspC-1718) or DHS (as for CspC-82). It remains to be seen whether these triads in C. septicum are truly functional. If so, it would imply that CspC’s bile salt sensing and proteolytic activity are independent of each other.

While the gene inactivation results suggest that CspC-1718 and CspC-82 are bona fide bile salt germination sensors, the structural modeling experiments show that there is a strong similarity between these proteins and C. difficile’s CspC (Fig. 5). A cross-comparison between CspC-82, CspC-1718 and C. difficile’s CspC showed a high level of structural as well as dynamical congruence between the three proteins. The major structural domains of C. difficile’s CspC were conserved in C. septicum’s orthologs — the jellyroll domain (domain 3), which imparts structural rigidity to CspC, as well as the prodomain (domain 1), which functions as an intramolecular chaperone (IMC) for protein folding. The jellyroll domain is thought to play an important role in bile salt germinant sensing. In a recent study, a single amino acid substitution (R358A) in the jellyroll domain of C. difficile’s CspC dramatically reduced C. difficile’s germination response to taurocholate³². One major limitation of these simulations is that they are limited only to bile salt-responsive CspCs. A full molecular dynamical comparison of bile salt responsive CspC’s from C. septicum and C. difficile, against bile salt non-responsive CspC’s from other clostridia is warranted. This may reveal structural features within CspC which are essential for bile salt sensing.

Altogether, this study provides sufficient data to begin piecing together a rudimentary germination signaling framework for C. septicum. The four bile salt-responsive elements identified in this study corresponded to similar elements in C. difficile. Among them, CspC-82 and CspC-1718 responded to non-overlapping subsets of the bile salt germinants while CspBA-81 and SleC-83 responded to all of them (Fig. 6a). This suggests that CspC-82 and CspC-1718 may be individual bile sensing elements, both signaling through downstream CspBA-81 and SleC-83 to cause germination. Here, there is a strong parallel with C. difficile, where CspC is the bile salt element which signals through CspB (cleaved from CspBA) to activate SleC’s cortex degrading activity (Fig. 6b)³⁴. Since CspB also signals downstream to SleC in C. perfringens²⁴, we hypothesize that information flow may be similar in C. septicum, proceeding from both CspC elements to CspBA-82 and then SleC-83.

Fig. 6. Summary of bile-salt response and proposed germination pathway.

a Summary of bile-salts and their germinant effects on the various mutants. b The proposed germination pathway of C. septicum is compared with the known pathway elements for C. difficile and C. perfringens. CspC-1718 and CspC-82 are bile salt sensors (similar to C. difficile’s CspC) which recognize their cognate bile-salt germinants, leading to downstream activation events which trigger germination. While C. septicum does not appear to have a germinant response to amino acids (unlike C. difficile and C. perfringens), several amino acids can act as co-germinants with sub-optimal levels of GCA (Supplementary Fig. S6A). We have not included potential mediators of amino acid sensing in this model. A dotted arrow joins CspBA-81 and SleC-83 because it is unclear if SleC is activated solely by CspB (like in C. difficile), redundantly by multiple Csp proteins (like in C. perfringens), or a novel alternative configuration.

We do not know if CspBA-81 is processed to yield CspB and CspA, but we speculate this is likely because C. difficile’s orthologous CspBA fusion gene product is non-functional until proteolytically cleaved into functional CspB and CspA by the YabG protease⁴⁶. This suggests that cleavage is necessary for function. In comparison, the need for processing is removed in C. perfringens because both CspB and CspA are already expressed as distinct functional proteins.

Although the 20 canonical amino acids have been shown to trigger spore germination in most clostridia^12,13,53,54, it turns out that C. septicum is one of the rare cases which only uses these amino acids as co-germinants (Supplementary Figs. S1a, S6). Arginine exhibited the strongest co-germinant effect, triggering spore germination in the presence of sub-germinant concentrations of GCA. This co-germination effect was not observed with the other bile salt germinants (DCA and CDCA), suggesting that amino acid co-germination is specific for the GCA sensing pathway. Based on our current understanding, there are two likely mechanisms through which arginine sensing could occur. In one mechanism, C. septicum’s Csp proteins sense arginine similarly to how C. difficile’s CspA³³ and CspC³¹ are involved in sensing glycine as a co-germinant. Since there is already a strong parallel in how both bacteria use CspC to sense bile salt germinants, C. septicum might also use the same co-germinant sensing paradigm as C. difficile. On the other hand, there is one big difference between both bacteria; Ger family orthologs are present in the genome of C. septicum (Supplementary Table T3) but not C. difficile. Since Ger orthologs are the canonical sensor for amino acid germinants in other sporulating bacteria (e.g. C. perfringens¹²), it is conceivable that Ger orthologs mediate co-germinant sensing in C. septicum. Only further genetic studies with CspA and GerK can shed further light on this and hence we have not speculated on the signaling framework for co-germinant sensing in Fig. 6.

While there are contrasting reports debating the presence of C. septicum in humans^55–58, our results now suggest that C. septicum is arguably a native inhabitant of the vertebrate gut, especially since bile acids are found exclusively in the vertebrate gut²⁷. Physiological concentrations of bile salts within the intestinal lumen³⁹ fall within the range of EC₅₀ values of the bile germinants in this study (Fig. 1c).

Since the bile salt germinants for C. septicum (CDCA, GCA, DCA) are found to varying degrees in healthy human individuals, it is perhaps surprising that C. septicum is not more commonly detected in the gut microbiome. However, we speculate that its presence in a healthy colon may be suppressed by growth-inhibitory bile salts as well as mucosal immunity. A shift to a germination and growth-friendly bile salt profile, accompanied by a suitable necrotic or ischemic environment for colonization, are likely necessary conditions for C. septicum to replicate efficiently within the colon. This is consistent with the current thought that C. septicum infections begin in the gut and enter the bloodstream through gastric lesions, which are often but not exclusively caused or facilitated by malignancies. Once in the bloodstream, C. septicum can establish itself at distant anatomical sites, leading to spontaneous myonecrosis⁵⁹. Compromised mucosal immunity, a common feature of certain malignancies⁶⁰ could further release the brakes on such infections which could explain the high association with malignancy.

In light of our study, it is timely to revisit C. septicum’s unique association with colorectal and hematological malignancies¹. Recent studies have now established a link between certain bile acids and cancer. For example, up to 90% of colorectal cancer cases are influenced by high lipid diets that upregulate bile acid secretion in the gut, with DCA being one such bile acid⁶¹. Further, DCA and CDCA are known carcinogens implicated in colorectal carcinogenesis, while GCA has been proposed to be a biomarker for hepatocellular carcinoma^62–67. If the same bile acids which promote carcinogenesis also trigger C. septicum spore germination, then bile acid dysbiosis could very well be the common cause linking C. septicum infection with malignancy.

Materials and methods

Bacterial strains and growth conditions

C. septicum strains were grown on BHIS (Brain Heart Infusion supplemented with 0.5 g/l L-cysteine) in an anaerobic chamber (Plas lab) at 37 °C (85% N₂, 10% H₂, and 5% CO₂). Antibiotics were added as needed (10 μg/mL thiamphenicol, polymyxin B 60 U/ml or 2.5 μg/mL erythromycin). The E. coli strains, XL1-Blue and S17-1 (Biomedal S.L) were cultured aerobically at 37 ^oC on either LB plates or 2xYT broth and supplemented with 10 μg/ml chloramphenicol. XL1-Blue was used as the host for plasmid construction while S17-1 was used for conjugal transfer of plasmids to C. septicum.

Generation of csp and sleC mutants

Oligonucleotides and synthesized DNA fragments were listed in Supplementary Materials 1, 2. The Targetron plasmid pJIR750ai (Sigma) was used as the starting plasmid for generation of mutants and was modified in several ways. First, pRJ1 was created by cloning the oriT site into pJIR750ai. Next, BtsMutI restriction sites were knocked out and re-assembled with NEBuilder HiFi DNA assembly (NEB) to yield the plasmid pRJ2. The original promoter for the transcription of the group II intron machinery, beta-2p, was replaced with the promoter for the alpha toxin gene in C. septicum (Pcsa) to yield pRJ3. The ErmB retrotransposition activated selectable marker (RAM), which confers erythromycin resistance upon insertion of the intron into the genome, was inserted into the MluI site of PFGv2-Pcsa to finally yield the base plasmid pRJ4.

Insertion sites in each gene were scored and predicted per gene using an algorithm⁶⁸. Our choices of insertion sites were chosen according to two criteria, 1) high scores 2) position of insertion site relative to the N-terminus of the protein so that the protein is truncated early and abolishment of the function can be determined. The identified retargeting regions were synthesized by Gene Universal and ligated into pRJ4 at the HindIII and BsrGI restriction site. These plasmids were then sequenced and used for electroporation of the transfer strain S17-1

csp and sleC mutants were generated in C. septicum ATCC 11424 with the TargeTron mutagenesis system. Conjugal transfer of the Targetron plasmid (Supplementary Material 3) to the recipient C. septicum strains was performed in the anaerobic chamber by dripping 200 μl of the conjugation mixture on BHIS plates for six hours. The mix was then harvested with 1 ml of PBS and 100 μl of the slurry was spreaded on a BHIS plate supplemented with 10 μg/ml thiamphenicol and 60U/ml polymyxin B for 16–24 h. Replica plating was then performed on erythromycin containing BHIS plates (2.5 μg/ml) and grown for a further 16–24 h before screening with PCR to confirm insertion of introns. Mutants were then further streaked on BHIS plates to shed the TargeTron plasmid.

Complementation of C.septicum mutants

The modular plasmid pMTL83153⁶⁹ was modified by replacing the promoter P_fdx with P_csa at the SacI and AscI restriction site to yield pRJ5. Finally, the base constitutive expression plasmid pRJ6 was constructed by inverse PCR with primers complementary to the FLAG tag and the end of the alpha toxin promoter sequence. The respective genes for complementation were amplified by the Q5 polymerase (NEB) with their respective primers and assembled into pRJ6 with NEBuilder HiFi DNA assembly (NEB). These plasmids (Supplementary Material 3) were then sequenced and used for electroporation of the transfer strain S17-1. Conjugal transfer of the respective plasmids to the respective mutant recipient strains was performed as described above. The recipient strains carrying their respective plasmids were verified by PCR.

Genomic DNA isolation

Genomic DNA (gDNA) was isolated with a few modifications from the original protocol described in ref. ⁷⁰. RNAseA (Sigma) and Proteinase K (Roche) treatment was only done following crude DNA extraction for 2 h each at 37 ^oC. Lastly, gDNA was purified by phenol/chloroform extraction and quantified with Picogreen (Life Technologies) using lambda DNA (NEB) as standards.

Southern blot analysis

0.5 μg of gDNA was digested with 10 U of AseI, separated by gel electrophoresis in a 0.8% agarose gel and transferred to a nylon membrane (Amersham Hybond-N + ) with a vacuum blotting system (GE, VacuGeneXL). Hybridization of the blot with the intron probe was done following manufacturer instructions (Roche) The blot was hybridized with a 378 bp DIG-labeled intron probe (Roche) generated from the plasmid pRJ3.

DNA sequencing, assembly and annotation

High molecular genomic DNA was sheared, and the 10kB fragments were isolated for end-repair and universal hairpin adapters ligation. Subsequent steps were followed as per PacBio protocol to prepare the SMRTbell library. The library was sequenced in PacBio Sequel and the resulting sequencing reads were assembled with Flye v2.8⁷¹. Genome annotation was carried out using NCBI’s prokaryotic genome annotation pipeline (PGAP) build 6021⁷². Schematic representations of the contigs and plasmids were generated with Proksee (https://proksee.ca/). Prophage analysis was predicted with the PHASTER server⁷³. Analysis of Insertion elements were carried out with ISEscan⁷⁴. Evaluation of genome completeness was executed with CheckM v1.0.18⁷⁵. Genome assembly statistics was reported by Quast 4.4 and PGAP^72,76. Both Quast and CheckM were hosted on Kbase⁷⁷. The assembled genome can be found on NCBI’s GenBank through accession number JARRAV000000000 under Bioproject ID PRJNA872817.

Bioinformatic identification and comparison of Ger/Csp orthologs

To identify putative CspC orthologs in C. septicum, CspC from C. difficile was used to query the C. septicum CDS sequences with the Protein Blast (Blastp) algorithm. Protein Blast (Blastp) was performed over nucleotide blast (Blastn) to account for synonymous codons. GerK orthologs in C. septicum were identified by querying the sequences of C. perfringens GerAA, GerKA, GerKB and GerKC. Identification of putative cspC orthologs of C. baratii (taxid: 1561), C. innocuum (taxid: 1522), F. plautii (taxid: 292800) H. hathewayi (taxid: 154046) and P. bifermentans (taxid: 1490) was performed with Blastp using CspC (accession number: WP_003433821) from C. difficile. CspBA and CspC orthologs of C. perfringens, C. difficile and C. septicum were analyzed in MegAlign Pro v13 (DNASTAR) using the Clustal Omega algorithm.

Sporulation and purification of C. septicum spores

C. septicum spores were purified as described previously⁷⁸, with some modifications. Briefly, overnight cultures of C. septicum were diluted 50X into 100 ml of BHI-S and incubated until the OD600 is 1.5 to 3 before the entire culture was added into 900 ml of sporulation media (0.5% L-cysteine, 3% Bacto polypeptone, 5% dehydrated cooked meat medium and 10% fetal bovine serum). Thiamphenicol (5 μg/ml) or erythromycin (2.5 μg/ml) were added when necessary. This sporulation media was then incubated for a further 5 days. Spores were purified from vegetative cells on a 80% discontinuous Percoll gradient, washed twice, resuspended in water and stored at 4 ^oC until use. Purified spores were observed to be >99% phase bright.

Germination of C. septicum spores

All germination assays were conducted at 37 ^oC in the presence of 0.2x Oxyrase for broth (Oxyrase®). Single factor screening experiments of WT spores (Supplementary Table T1) were performed (Oxyrase®) in a 384 well plate (Greiner) and covered with a plastic film (Excel Scientific). Determination of the optimum pH for WT spore germination was performed as above except WT spores were germinated in a germination assay buffer containing 10 mM sodium phosphate buffer with a range of pH from 6 to 8, 20 mM DCA and bacterial spores adjusted to an initial OD₆₀₀ of 1.0士0.2. Data from the above experiments were collected from the infinite 200 Microplate reader (Tecan).

Subsequently all data was collected from the Spark Multimode Microplate reader (Tecan) To investigate whether Ca-DPA release from the spores was responsible for the high OD artifact observed with DCA induced spore germination, we performed germination assays as described above, except that Ca²⁺ and DPA was added instead of spores. Briefly, a germination solution consisting of 10 mM sodium phosphate buffer at the tested pH of 6, 6.5 or 7, 20 mM DCA, CA, or GCA, 50 uM Ca²⁺ and DPA was incubated at 37 ^oC and the OD was measured every 5 mins over 15 hours.

Germination assays of WT spores in serial dilutions of bile salts was performed in a germination assay media containing 10 mM sodium phosphate buffer pH 7.0, 0.2X Oxyrase, serial dilutions of bile salts (final concentrations 50 mM–0.78 mM) and lastly bacterial spores adjusted to an initial OD₆₀₀ of 1.0士0.2 in a 384 well UV transparentplate (Greiner UV Star plate) covered with an optically clear qPCR film (ABI technologies) to maintain anaerobicity. All measurements were taken in intervals of five minutes for 15 hours.

Amino acid co-germination assays were performed as above with sub-optimal concentrations of CDCA (1.56 mM), DCA (1.56 mM) and GCA (3.13 mM), except that a 20- amino acid mastermix was added to the germination buffer at a final concentration of 5 mM for each amino acid. Individual amino acids (5 mM) were also tested with GCA (3.13 mM) as described above. The final percentage of OD decrease relative to its initial OD (OD_final/OD_initial) of each individual amino acid with GCA is then subtracted by the GCA only control to establish the effect of the amino acid.

For heat activation of WT spores, WT spores were incubated in a thermocycler (AIT Biotech) in their respective temperatures for 15 minutes before being cooled on ice for five minutes prior to performing the germination assay as described above with a few modifications. The respective bile salts and TbCl₃ were added to a final concentration of 10 mM and 0.1 mM respectively. Released DPA complexes with terbium to form a fluorescence complex and is measured at an excitation of 272 nm with emission of 545 nm at z = 18216 μm and a time lag of 20 μs before measurement. All measurements were taken in kinetic mode in 5 minute steps for 15 hours. Germination assays of WT, csp, sleC mutants and complemented mutants were conducted in the same way as heat activation assays except spores were not heat activated. All values below zero were considered to be zero. Due to low Ca-DPA content in cspC-1718 mutants and its complemented mutant spores, spores were added to an initial OD600 of 1.3 士 0.2 and the acquisition parameters were optimized as follows: number of flashes (50), lag and integration time (2000 us).

Total DPA release assay

To assess the total DPA content in the spore preparation, spores suspensions (OD 0.5 in 100 ul) were heated at 100 ^oC for an hour. The spores were then cooled on ice for five minutes. Next, the spore suspension is spun at 17,000 g for 5 minutes and 20ul of the supernatant was assayed for DPA content in the same buffer as described above for the spore germination assay for the mutant, except that bile salts are excluded.

Recovery of colonies from bile salt exposed C. septicum spores on agar

2 × 10⁷ C. septicum spores were incubated in 0.1% solutions of various bile acids for 10 min in the anaerobic chamber before they were plated on LB overnight. The colonies were counted the next day and compared to GCA exposed spores.

Comparison of the viability of C. septicum mutant and complemented spores

Working concentrations of C. septicum WT, mutant and complemented spores (2 × 10⁷ spores/ml) were serially diluted and 5ul was plated on BHI-S plates and left to dry for 30 minutes. After drying, plates were incubated in a GasPak jar (BD, 260628) overnight.

Protein structure prediction and simulation details

The all-atom molecular dynamics (MD) simulation was performed for three systems: (i) C. difficile CspC, (ii) C. septicum CspC-82, and (iii) C. septicum CspC-1718 to explore the structural stability and similarities of the three proteins. NAMD simulation program v2.14 with CHARMM36 force field was used to perform all simulations^79,80. The initial configuration for the simulation of C. difficile CspC was obtained from Protein Data Bank (PDB entry 6MW4; 1.1 Å resolution), whereas the initial configurations of C. septicum CspC-82 and C. septicum CspC-1718 were modeled using i-TASSER⁸¹ and AlphaFold⁸². The end residues for all proteins were properly capped (Acetyl group at the N-terminal and N-methyl group at the C-terminal). The side-chain protonation states of asparagine, glutamine, and histidine were considered and resolved through optimization of local hydrogen bonding. The three proteins were solvated by cubic TIP3 water boxes, and the overall charge of each system was neutralized by adding counter ions. Each of the three systems were energy minimized by performing 50,000 steps of Steepest Descent to eliminate the close van der Waals contacts. After that, the temperature of the systems was gradually increased to room temperature (300 K), followed by 50 ps equilibration using the NPT algorithm, where temperature was controlled by the Langevin algorithm. The three systems were then run for 50 ns under a constant pressure of 1 atm, and a constant temperature of 303 K (NPT ensemble). The time steps for each simulation was 2 fs, and trajectories were stored every 2 ps for analysis. The long-range electrostatic forces were calculated using Particle Mesh Ewald method. Periodic boundary condition and a 10 Å cutoff were applied for non-bonded short range interactions.The interface Solvent-Accessible Surface Area (interSASA) for a specific domain i among the set of all domains P is defined as: interSASA_{i} = SASA_{i} + SASA_P-{i} - SASA_P where SASA_{i}, SASA_P-{i} and SASA_P are the SASA values for all domains, domain i and all domains excluding i respectively.

Statistics and Reproducibility

Spores of the order of 10⁹ CFU/ml were used for all germination assays. All experiments were independently replicated at least twice with two embedded technical replicates. All attempts at replicating our findings were successful. Where applicable, spores from a different batch were also assayed. For viability and recovery of spores on agar, at least 2 × 10⁷ spores/ml were used and this amount produced colonies which were serially diluted for accurate counting. Figures were generated on Graphpad Prism 9 for Windows (Graphpad software). Error bars represent the standard error of the mean. Statistical analysis of germination curves, CFU recovery, amino acid co-germination, heat activation and total DPA content was done with one way ANOVA and Dunnett with the final endpoint data. Comparison of OD increase of different bile acids in varying pH was performed with two way ANOVA and Dunnett with the final endpoint data.

To screen for potential germinants, OD drop over the moving average of 6 timepoints (25 min) was calculated and the maximum OD drop velocity value over the duration of the experiment was used. For bile salt concentration response, 100% germination response was defined as the maximal germination response with DCA since it showed the highest OD drop while 0% germination response was defined as the germination response in water instead of bile salts. The standard curve of each bile salt germinant was generated using a four-parameter logistic curve and EC₅₀ values were interpolated from the generated curves. To fit V_max curves, V_max was calculated after the assay started for 30 min over a moving window of three time points (10 min) and the maximum velocity of each concentration was used. The curve of each bile salt was generated using the built-in allosteric sigmoidal curve fit. An unpaired t-test was conducted to assess if there were differences in EC₅₀ and V_max values between the germinants.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Conceptualization (R.S., A.S., I.C). Data curation (R.S., S.J.M.L, S.S., M.S., I.C). Formal analysis (R.S., S.J.M.L, S.S., M.S., I.C.). Funding acquisition (I.C.). Investigation (R.S., S.J.M.L, A.S., S.S. M.S. W.L., A.M., T.W.L., M.D.C., G.J.H). Methodology (R.S., S.S., M.S., I.C.). Project administration (R.S). Software (S.S., M.S). Supervision (R.S., I.C.). Validation (R.S). Visualization (R.S., S.J.M.L, S.S., M.S., I.C.). Writing – original draft (R.S., S.J.M.L, A.S., I.C.). Writing – review & editing (R.S., S.J.M.L, A.S., S.S., M.S., W.L., A.M., T.W.L., M.D.C., G.J.H., I.C.).

Peer review

Peer review information

Communications Biology thanks Ernesto Abel-Santos and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editor: Tobias Goris. A peer review file is available.

Data availability

All data is available in the main text or the Supplementary Information. The assembled genome can be found on NCBI’s GenBank through accession number JARRAV000000000. The source data behind the graphs in the paper can be found in Supplementary Data 3.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-024-06617-4.