The conserved noncoding RNA ModT coordinates growth and virulence in Clostridioides difficile

ModT transcript abundance and stability resembles housekeeping RNAs

We previously demonstrated expression of ModT in C. difficile through genome-wide RNA sequencing-based mapping of transcriptional start sites (TSS) and transcript ends (annotated CDIF630nc_001 in [21,26]). ModT caught our attention due to its high sequence and secondary structure conservation across 9 taxonomic classes and 2 bacterial phyla [22]. RNA-seq-based transcript annotation revealed that expression of ModT is driven from a single intergenic TSS that is associated with a canonical housekeeping σ^A (SigA) promoter (Figs 1A and S1A [21]). Transcription of modT gives rise to a long (ModT (L), 260 nt) and a short (ModT(S), 215 nt) isoform that differ in their 3′ ends (Figs 1B and S1B). Again, our previous RNA-seq-based transcript annotation indicates 2 independent transcription termination events at sites that correspond to transcript lengths detected by northern blot (S1A Fig). Interestingly, analysis of existing RNA-seq data [26] shows that ModT is among the 30 most abundant transcripts in the cell, and only surpassed by some, but not all, ribosomal RNAs and tRNAs, as well as 6S RNA and the signal recognition particle RNA (Fig 1C and S1 Dataset).

Fig 1. ModT is a highly abundant and stable noncoding RNA.

(A) Schematic of the modT locus in C. difficile 630. Promoter region including the SigA -35 box and -10 box are shown on top, TSS and TTS are indicated with an arrow or a pin symbol, respectively. In silico secondary structure prediction (RNAfold) of the modT RNA is displayed in pink, with the 3′-end extension of the long isoform highlighted in purple. (B) ModT expression profile in C. difficile 630 across different growth stages. Total RNA was extracted from C. difficile 630 grown in TY medium supplemented with indicated carbon sources at indicated time points (for reference, see growth curves shown in Fig 3B). ModT transcript was detected by northern blot using a radioactively labeled DNA probe (FFO-67) that captured both ModT isoforms, as indicated with an arrow. 5S rRNA served as loading control. (C) Normalized read distribution of all expressed and annotated transcripts (n = 3,472) in C. difficile 630 sorted by transcript abundance [26]. Inset displays the 50 most abundant transcripts, with ModT labeled in pink. The underlying data can be found in S1 Dataset. (D) In vivo RNA half-lives for different noncoding RNA transcripts. C. difficile 630 was grown to mid-exponential phase and treated with Rifampicin. Total RNA was extracted at the indicated time points and analyzed by northern blotting using radioactively labeled DNA probes. The representative of 3 independent experiments is shown. TSS, transcriptional start site; TTS, transcription termination site.

https://doi.org/10.1371/journal.pbio.3002948.g001

To develop a better understanding of ModT function, we initially performed northern blot analysis to characterize its expression across growth stages and in the presence of different carbon and energy sources (Figs 1B and S1B). We observed prominent expression of both isoforms in TY (Tryptone-Yeast) medium throughout exponential growth, but it decreased to the limit of detection after transition into the stationary phase. In contrast, during growth on monosaccharides, such as glucose or fructose, steady-state levels of ModT(L), but not ModT(S) remained high even in late stationary phase. Since both isoforms are transcribed from the same promoter, we sought to explore whether their different fates in the stationary phase are determined by different RNA half-lives. Therefore, we monitored RNA levels over time, after Rifampicin-mediated inhibition of de novo modT transcription (Figs 1D and S1C). Intriguingly, the long isoform was substantially more stable (t_1/2 = >180 min) than the short isoform (t_1/2 = 45.94 min). This suggests that the accumulation of the long isoform in the stationary phase is, at least in part, due to a longer half-life relative to the short isoform. Moreover, the ModT(L) half-life was similar to the housekeeping 5S rRNA (t_1/2 = >180 min) and exceeded by far that of the canonical Hfq-dependent small RNAs CDIF630nc_008 (t_1/2 = 22.95 min) (Figs 1D and S1C). In addition, we also observed a higher GC content of ModT (45.6% for ModT(S) and 42.6% for ModT(L)) compared to the average GC content of the C. difficile genome (29.1%) or previously annotated sRNAs (27.5%) (S1D Fig). In summary, the ModT transcript exhibits several features that are reminiscent of housekeeping RNAs and suggests that this noncoding RNA fulfills a fundamental physiological function.

ModT isoforms adopt identical structures in vitro and in vivo

The function of many RNA elements is determined by their unique structure [27]. Given that both ModT isoforms exhibit very distinct in vivo half-lives, we wanted to explore whether they fold into distinct structures. First, we compared the C. difficile ModT sequence with 3 structure models for ModT, including the original structure predicted for ModT [15,22], the structure available from the Rfam database (RF03072), and a predicted structure optimized with R-scape (Fig 2A–2C). Of note, all 3 models represent the short isoform, likely because sequence conservation is poor in the extended 3′-end [15], and we appended our predicted terminator sequence to them. The recently updated consensus secondary structure model for ModT [22] has 2 pseudoknots (denoted pk1 and pk2) and 8 predicted base-paired regions (denoted P1 through P8), and except for P2, all of them are present in both isoforms of C. difficile ModT (Fig 2A). We also mapped the C. difficile ModT sequence onto the current Rfam structure (Fig 2B) as well as the R-scape optimized structure (Fig 2C) [28,29]. The 3 models show prominent structural differences in the regions comprising P3 and P4, as well as in the residues involved in potential pseudoknot formations. Therefore, we decided to include all of them in our analysis.

Fig 2. DMS probing uncovers structural dynamics of ModT during different growth phases and test conditions.

(A–C) Manually curated secondary structure models of C. difficile 630 ModT based on available Rfam models (RF3072). The stem loop structure of the appended 3′ end is based on predictions by RNAfold. (A) Consensus secondary structure as described in [15,22], (B) current Rfam structure, and (C) the R-scape optimized structure of C. difficile 630 ModT. Black triangles indicate the most prominent scission events determined by In-line probing (see S2A–S2D Fig). Average DMS-reactivity for each residue, derived from in vitro structure probing of ModT(S) and ModT(L), is coded with a color-gradient (filled circles). (D) DMS reactivity profiles of adenine (A) and cytosine (C) nucleotides for residues 19–210 of ModT in following conditions: in vitro, in vivo native (Nat), and in deproteinated cell lysates (Dep). The Nat and Dep samples were analyzed in EP (3 h), TP (9 h), and stationary phase (24 h) of growth. Under “Nat” and “Dep” conditions, L displays reactivity specific for the long isoform, while S represents combined reactivities of the short and the long RNA isoforms. The conditions are sorted according to sample clustering in S2E Fig. The dendrogram has been omitted for clarity. The underlying data can be found in S5 Dataset. (E) Comparisons of DMS reactivity in different conditions or isoforms. The Δ reactivity for each nucleotide position was calculated by subtraction of normalized DMS reactivities obtained for the indicated conditions (see top of each graph). For “in vivo Native → Δ(ModT(L)—ModT(S))” and “ModT(L) → Δ(Deproteinated–Native),” the mean subtracted reactivity and its standard deviation was calculated from all growth stages (3 h, 9 h, and 24 h). For “In vivo Native → Δ(3h – 9h),” the profiles of ModT(L) and ModT(S) were combined to obtain the mean reactivity and standard deviation at these 2 time points. The underlying data can be found in S5 Dataset. EP, exponential phase; TP, transition phase.

https://doi.org/10.1371/journal.pbio.3002948.g002

In-line probing assays (ILP) on in vitro transcribed ModT(L) and ModT(S) (Figs 2A–2C and S2A–S2D) provided general support for all 3 in silico predicted structure models. More specifically, ILP indicated prominent cleavage events at nucleotides mostly predicted to be unpaired in each model. Notably, ILP of ModT(L) resulted in a cleavage pattern identical to ModT(S), suggesting that the extended 3′-end adopts a fold that does not alter the core structure of ModT(L). However, given that the structural information resolved by the ILP is limited for larger RNAs, we decided to obtain better insight into the ModT structure both in vitro and in vivo using dimethyl sulfate mutational profiling (DMS-MaP) [30,31]. DMS-MaP is a well-established method to infer secondary structure by exploiting DMS reactivity at single stranded, or exposed, A and C nucleotides. We generated DMS-reactivity profiles of ModT(S) and ModT(L) under various conditions: (1) in vitro transcribed ModT isoforms; (2) in vivo, from cultures grown to exponential, transition, and stationary phase; and (3) from deproteinated cell lysates obtained at the same growth phases (Fig 2D). Of note, we specifically determined the reactivity values of the long isoform (L) using a primer binding to the extended 3′-end for reverse transcription in vivo and in deproteinated cell lysates. As this is not possible for the short isoform, reactivities of these samples (S) represent a combined contribution of ModT(S) and ModT(L), which might mask minor structural differences.

Importantly, DMS reactivities for both ModT isoforms showed very high correlations (≥0.98) in each analyzed condition (S2E Fig), with no prominent differences between ModT(L) and ModT(S) within the common region (Fig 2D). This is particularly apparent when subtracting their normalized reactivities (Fig 2E). Hence, the DMS probing confirmed the ILP data, indicating that ModT(L) and ModT(S) adopt the same core structures in vitro and in vivo. Generally, we noticed a strong positive correlation across all analyzed samples (≥0.86), which indicates overall similar RNA structures for ModT in different growth conditions (S2E Fig). Nevertheless, hierarchical clustering using the single linkage algorithm showed that the samples from the late time points display the lowest reactivity across several regions, while those from the exponential phase (EP) have an overall increased reactivity (Fig 2D). Interestingly, higher overall reactivities were also observed for the Proteinase K digested lysates and for the in vitro transcribed RNA samples (Fig 2D). Arguably, the key difference between the native in vivo conditions, and the in vitro/deproteinated conditions, respectively, is the absence of proteins in the latter two, which indicates the interaction of ModT with proteins in vivo (Figs 2D and S2E).

Specifically comparing reactivity changes of single residues between conditions (e.g., in vivo deproteinated versus in vivo native, (Figs 2E and S2F) identified strong changes in base-paired regions P7 and P8, and in the regions with the lowest structural confidence across the 3 ModT structure models (P3-P5) (Fig 2A–2C). Therefore, we wanted to explore whether our experimental DMS-MaP data of ModT could identify a model with best fit to the ModT structural data. For that, we calculated the ROC-AUC score (receiver operator characteristic–area under the curve), a summary statistics, which serves as a measure of the agreement between the predicted base pairing in a model and the experimentally determined DMS reactivity [31]. Indeed, this analysis clearly favored the updated consensus secondary structure model (Fig 2A), as it displayed the highest ROC-AUC score across all tested conditions (S2G Fig).

In summary, our data suggest overall identical structures for both isoforms in vivo and in vitro. Furthermore, the identification of a best-fitting structure model provides a crucial step in delineating the structural constraints for ModT function as it will guide the design of mutations in critical base-paired regions in future experiments.

ModT acts in trans and regulates transition into the stationary phase of growth

As a first step towards identifying cellular functions of ModT, we deleted the first 215 nt of modT (ΔmodT) corresponding to the conserved ModT(S) isoform (Fig 3A). This abolished production of both isoforms without causing any polar effects on neighboring genes (CDIF630_00250 and CDIF630_00251) (S3A and S3B Fig). Due to the nutrient-dependent expression profile of ModT, we initially compared growth of wild-type (WT) and ΔmodT in TY medium supplemented with carbon and energy sources that can be utilized by C. difficile during gut colonization (Fig 3B) [32–34]. The deletion of modT was associated with an increased maximum optical density (OD) (S3C Fig), which appeared to be the result of extended exponential growth (Fig 3B). This overgrowth phenotype was particularly pronounced in media containing hexose monosaccharides (glucose, fructose, and mannose) or ethanolamine. Curiously, the overgrowth did not correlate with the carbon source-dependent ModT expression in late stationary phase, which does not occur during growth on ethanolamine (Fig 1C). Nevertheless, it indicates that the presence of ModT in exponential phase of growth is needed for a timely transition into the stationary phase. Importantly, in trans chromosomal complementation of ΔmodT with either the short or the long isoform resulted in an expression level and pattern similar to wild-type ModT (S3B Fig) and restored growth to wild-type level for both complementation strains (Fig 3B). This demonstrates that the short isoform is sufficient for ModT function. Furthermore, and in agreement with a recent study in C. acetobutylicum [22], this experiment rules out a cis-acting riboswitch function of ModT, which was originally proposed based on its genetic location and complex secondary structure [15].

Fig 3. Growth analysis and transcriptomic profiling of modT mutants reveal a trans-acting function important for the transition into the stationary phase.

(A) Schematic of chromosomal modT deletion, and in trans complementations at the pyrE locus of the short and the long isoforms, respectively. (B) Growth curves of C. difficile 630 WT, ΔmodT and complemented strains in TY medium supplemented with indicated carbon sources. The underlying data can be found in S5 Dataset. (C, D) Volcano plots showing transcriptomic comparison between C. difficile 630 WT and ΔmodT grown in TYG to EP (3.5 h) and TP (9 h). FDR cutoff was set to 35] are indicated in red, genes induced during sporulation in a Spo0A-dependent manner [39] are labeled blue. Up-regulated tRNA genes are colored in purple. The underlying data can be found in S2 and S5 Datasets. EP, exponential phase; FDR, false discovery rate; TP, transition phase; WT, wild type.

https://doi.org/10.1371/journal.pbio.3002948.g003

To identify cellular pathways that are regulated by ModT, we performed comparative transcriptome analysis. Wild-type and ΔmodT strains were grown in TY medium supplemented with glucose (TYG). We collected samples in the EP (3, 5 h) and the TP (9 h), when the knockout strain begins to outgrow the wild type (S3D Fig). This analysis identified 145 differentially regulated genes in the exponential phase and 344 differentially regulated genes in the TP (each with a false discovery rate (FDR) ≤ 0.1 and a log₂ FC ≥ 1 or ≤ -1) (Figs 3C, 3D, and S3E and S2 Dataset). The list of regulated genes was dominated at both timepoints by metabolic genes and ABC-type sugar and amino acid transporters, which might be linked to the observed carbon-source dependence of ModT expression and the growth phenotype of the mutant.

Given the outgrowth phenotype of ΔmodT, we focused on differentially expressed genes in the TP. We noticed that deletion of modT was associated with broad down-regulation of the σ^B-dependent regulon. Sigma B regulates the general stress response in C. difficile, mediating tolerance towards many stress conditions, but also adaptive responses during intestinal colonization [35,36]. In our dataset, approximately 25% (62/248) of the genes that are under positive control of σ^B [35], were significantly down-regulated in ΔmodT during the TP (S3F Fig and S3 Dataset). In accordance with a low σ^B activity during early exponential phase [36], only 7% of the regulon (16 genes) was down-regulated at this time point.

We further observed down-regulation of genes under the control of σ^H, a central regulator of the TP [37]. Its regulon is less defined than the σ^B regulon, but of the 40 genes known to harbor a σ^H promoter [37], 18 (45%) were down-regulated in ΔmodT during the TP (S3 Dataset). Furthermore, we found globally increased tRNA abundance in the mutant, with 75 out of 90 tRNA loci being up-regulated. Given the technical challenges associated with tRNA isolation and sequencing [38], we sought to validate this observation by northern blotting. To that end, we generated a high-resolution time curve covering all growth stages of C. difficile and probed for 2 different tRNAs (S3G Fig). This validated the increased abundance of these tRNAs in the knockout during the TP, in comparison to the wild type. Interestingly, the difference increases at late stationary phase time points due to a reduction of the tRNA fraction in the wild type, while levels are stably maintained in the knockout. Intriguingly, probing for 5S indicated that this trend also occurs with ribosomal RNA.

In addition to a broad down-regulation of the σ^B and σ^H regulons, we identified pathways that directly regulate toxin production and sporulation. Specifically, we observed up-regulation of the tcdA gene, encoding for the large clostridial toxin A (log₂FC = 1.1 and 0.9 in exponential and transition phase, respectively), and of the tcdR gene (log₂FC = 2.3 in TP), encoding for the alternative sigma factor TcdR that positively controls synthesis of the toxin genes tcdA and tcdB. This indicated that ModT activity may limit toxin production. To test whether the observed transcriptome changes translate into higher toxin levels, we compared these between wild type and mutant. This revealed that the modT mutant produced more toxin than the wild type (Fig 4A, left panel), which ultimately resulted in higher levels of toxin released into the culture supernatant (Fig 4A, right panel). Importantly, as seen for the growth phenotype, complementation of either the short or the long isoform of ModT restored toxin production to wild-type levels.

Fig 4. ModT regulates stationary phase associated spore formation and toxin production.

(A) ELISA-based toxin quantification (TcdA and TcdB combined) in cell lysates (“intracellular toxin”) and cell free supernatant (“secreted toxin”) of C. difficile 630 WT, ΔmodT and complemented strains grown in TYG medium. Relative toxin amount is displayed as absorbance at 450 nm (A₄₅₀) normalized by the absorbance at 620 nm (A₆₂₀). n = 3 biological replicates. The underlying data can be found in S5 Dataset. (B) Representative phase-contrast micrographs (n = 3 biological replicates) of C. difficile 630 WT, and modT mutants in 70:30 medium. Time points are indicated relative to the onset of the TP. Scale bar = 5 μm. (C) Sporulation frequency calculated from the phase-contrast microscopy displayed in (B). n = 3 biological replicates. Time points are indicated relative to the onset of the TP. The underlying data can be found in S5 Dataset. Significance for data shown in (A) and (C) was determined by two-way ANOVA with Tukey’s multiple comparisons test. Only significant differences with P P P P

https://doi.org/10.1371/journal.pbio.3002948.g004

Furthermore, 23 sporulation genes [39] were down-regulated in TP (Fig 3D and S3 Dataset), including those encoding for the sporulation sigma factors σ^E (log₂FC = −1.1), σ^F (log₂FC = −0.92), and σ^G (log₂FC = −1.05), all of which coordinate spore developmental steps downstream of the master regulator of sporulation Spo0A. Therefore, we quantified sporulation frequencies in wild-type, ΔmodT and complemented strains over an extended period of 5 days. Spore enumeration by phase contrast microscopy revealed a significant decrease in sporulation frequency in the knockout (Fig 4B and 4C), which was independently validated by CFU-based calculation of vegetative cells and ethanol-resistant spores (S4A and S4B Fig). This sporulation defect remained throughout the experiment, indicating that sporulation is not just delayed in ΔmodT, but permanently repressed. Surprisingly, while chromosomal complementation of ΔmodT with the long isoform (ΔmodT C-long) reconstituted sporulation to wild-type levels, complementation with the short isoform (ΔmodT C-short) resulted in a pronounced hyper-sporulation phenotype (Figs 4B, 4C, S4A, and S4B). Northern blot validation of ModT revealed both a mild increase in ModT(S) transcript levels and a prolonged expression of ModT(S) at later time points in the short complementation strain, relative to the WT or ΔmodT C-long (S4C Fig). This either suggests that small alterations in ModT abundance and/or presence of ModT in stationary phase can have a strong impact on sporulation, or that the 2 isoforms differ in their capacity to regulate sporulation.

Taken together, the extended exponential growth and the global down-regulation of the σ^B and σ^H regulons in ΔmodT indicates that ModT regulates the transition from exponential to stationary phase of growth in C. difficile. Concomitantly, we observe changes in spore formation and toxin production, 2 virulence-determining processes that are tightly regulated in response to the physiological state of the cell.

ModT regulates sporulation by modulating cellular cyclic di-GMP levels

We sought to better understand the mechanistic basis for the observed changes in toxicity and sporulation in the modT mutant. Our transcriptome analysis highlighted several differentially expressed genes shown to affect sporulation (highlighted in Fig 5A). Among them, the dccA gene caught our attention, because it was the top up-regulated transcript at both time points in ΔmodT (Fig 3C and 3D and S2 Dataset). DccA is a diguanylate kinase containing a GGDEF domain that catalyzes the synthesis of the second messenger c-di-GMP [40]. Overexpression of DccA was shown to suppress early sporulation events, downstream of Spo0A, by increasing cellular c-di-GMP levels [41] through a yet unknown mechanism. Increased dccA transcription, and consequently, increased c-di-GMP levels, have been associated with a sinRR’ deletion mutant which is asporogenous and displays global gene expression changes in pathways regulating sporulation [42]. Aligning with that, the bicistronic operon encoding SinR and its protein antagonist SinR’ [42], was down-regulated in ΔmodT at both time points (Fig 5A). Hence, our transcriptome analysis of the modT mutant (S2 Dataset) suggested that the pronounced effects on sporulation could be linked to changes in cellular levels of c-di-GMP.

Fig 5. ModT modulates sporulation and toxin production via changes in cellular c-di-GMP levels.

(A) Schematic overview of differentially expressed genes in ΔmodT that were shown to be involved in the regulation of spore formation in C. difficile [40–42,56,65]. (B) Changes in c-di-GMP levels in ΔmodT and in complemented strains grown in TYG medium. Measurements were done using a fluorescent c-di-GMP type II riboswitch reporter. Values in the ΔmodT and in complemented strains were calculated as fold-change relative to the C. difficile 630 WT signal. Data represent the mean of 3 independent experiments performed with n = 3 biological replicates per experiment (total n = 9). The underlying data can be found in S5 Dataset. (C) c-di-GMP levels in WT, ΔmodT and ΔmodTΔdccA-01583. Measurements were done as in (B) Values in the ΔmodT and ΔmodTΔdccA-01583 strains were calculated as fold-change relative to the C. difficile 630 WT signal. The underlying data can be found in S5 Dataset. (D) Assessment of sporulation frequency in 70:30 medium by phase-contrast microscopy. The underlying data can be found in S5 Dataset. (E) Relative quantification of toxin (TcdA and TcdB combined) as determined by ELISA in cell free supernatants of C. difficile 630 WT, ΔmodT and ΔmodTΔdccA-01583 grown in TYG medium. Each experiment was performed in 3 (n = 3) biological replicates. The underlying data can be found in S5 Dataset. For the experiments shown in (B) and (C), the fluorescence values were normalized to the control (pRS^A>G) as described in S5B Fig. For data shown in (B) and (C), significance was determined by two-way ANOVA with Tukey’s multiple comparisons test; for data shown in (D) and (E), one way ANOVA with Tukey’s multiple comparisons test was applied. Only significant differences with P P P P

https://doi.org/10.1371/journal.pbio.3002948.g005

To investigate this link, we assessed whether deletion of ModT was associated with increased cellular c-di-GMP levels. For this, we adopted a reporter system in which a c-di-GMP sensitive riboswitch represses transcription of a downstream encoded mCherry protein at low c-di-GMP levels [43–45] (S5A Fig). Hence, increasing mCherry fluorescence serves as a proxy for increased c-di-GMP levels. To control for background fluorescence, we employed a reporter with an A70G mutation (pRS^A70G) that renders the riboswitch unresponsive to c-di-GMP. Wild-type, ΔmodT and the complemented strains, each harboring the different reporter constructs and an empty vector control, were grown in TYG medium to mid-exponential (6 h) and stationary phase (24 h), and mean fluorescence intensities were assessed by flow cytometry (S5B Fig). Comparing the mean fluorescence in each strain background carrying the functional reporter (pRS), the mutated reporter (pRS^A70G), and the empty vector control (pEV), revealed some residual fluorescence activity for the mutated reporter (S5B Fig). Therefore, all fluorescence data were normalized to the background fluorescence of the same strain carrying the mutated reporter construct (Fig 5B). There were no overall significant differences between the WT and the ΔmodT strain in the early stages of growth. However, at 24 h, we consistently detected increased fluorescence in the modT mutant. Importantly, this increase in reporter activity could be reverted to WT levels in both complemented ModT strains (Fig 5B). This confirmed our initial hypothesis that deletion of modT leads to increased cellular c-di-GMP levels, presumably through increased levels of DccA.

The dccA gene is co-transcribed with CDIF630_01583, which is a putative diguanylate kinase signaling protein. However, it also contains several potential sensory domains of the Per-Arnt-Sim (PAS) type, in addition to an EAL-domain which may degrade c-di-GMP [46]. The regulatory impact of CDIF630_01583 has not been studied so far. Since the entire operon was up-regulated in our dataset, we wanted to test if and how CDIF630_01583-dccA up-regulation was contributing to the changes in sporulation in the modT mutant. For that, we deleted the dccA-01583 operon in ΔmodT and compared the fluorescence of the resulting double knockout to ΔmodT and the wild type. This revealed that c-di-GMP levels did not increase in a ΔmodT ΔdccA-01583 double knockout, and even dropped below WT levels (Fig 5C). Consequently, deletion of the dccA-01583 operon in the ΔmodT strain background did not just restore sporulation frequency to WT levels, but actually increased the frequency (Fig 5D). Taken together, these data show that modT deletion leads to sporulation inhibition via increased DccA-mediated c-di-GMP synthesis. In contrast to sporulation, we observed only partial restoration of toxin production in the ΔmodT ΔdccA-01583 double knockout (Fig 5E), suggesting additional, c-di-GMP-independent, regulatory mechanisms contributing to the observed increases in toxin production in ΔmodT.

In summary, the effects of ModT on spore formation are mediated through the ubiquitous second messenger c-di-GMP.

ModT orthologs from distant phylogenetic families are functionally conserved

Given the high sequence conservation of ModT family members, we wanted to explore whether ModT orthologs from diverse species are capable of complementing ModT function in C. difficile. To this end, we exchanged the sequence of the consensus ModT motif (first 215 nt) in C. difficile (S6A and S6B Fig) with the consensus motif of 3 ModT orthologs from the phylum Bacillota, including Clostridium perfringens and unclassified Clostridium sp. CAG:138 (genus Clostridium, family Clostridiaceae), as well as Paeniclostridium sordellii (genus Paeniclostridium, family Clostridiaceae). In addition, we chose 1 ortholog from the phylum Actinomycetota, Acetohalobium arabaticum (genus Acetohalobium, family Halobacteroidaceae). All 4 orthologs possess core structural features of the ModT family, but vary in size and sequence, particularly in the regions of the variable P5, P7, and P8 stems (Figs 6A and S6A). For example, while the P7 and P8 stem regions comprise only ~51 nt in C. difficile, they are 73 nt and 118 nt long in CAG:138 and A. arabaticum, respectively. Furthermore, the primary sequences of the P5 loop vary in all strains and contain covarying mutations in various places. Northern blot validations of ModT transcripts in complemented strains showed that the orthologs from C. perfringens and CAG:138 are expressed at levels similar to WT ModT, although the ratios between short and long isoform vary (S6C Fig). In contrast, strains expressing orthologs from P. sordellii and A. arabaticum showed low ModT levels (S6C Fig). Despite expression level differences, we assessed toxin production and sporulation frequency in all complemented strains to determine whether these orthologs can replace ModT function in C. difficile. In comparison to the ΔmodT mutant, the orthologs of C. perfringens and CAG:138 were able to fully restore both toxin production and sporulation frequency to WT levels (Figs 6B, 6C, and S6D–S6F). P. sordellii ModT was able to restore sporulation frequency, whereas toxin levels remained comparable to ΔmodT. Similarly, complementation with ModT from A. arabaticum was able to partially rescue sporulation but did not restore toxin production. The fact that P. sordellii and A. arabaticum orthologs can only partially rescue the ΔmodT associated phenotypes might simply be due to their weaker expression in C. difficile, but it could also be an indication for a functional divergence.

Fig 6. Orthologs from distant phylogenetic families can replace ModT function in C. difficile.

(A) Consensus secondary structure models of ModT orthologs used for complementation. Regions varying in primary sequence and secondary structure across all orthologs are highlighted in yellow. (B) Relative quantification of toxin (TcdA and TcdB combined) determined by ELISA in cell free supernatant of WT, ΔmodT and C. difficile with ModT ortholog replacements grown in TYG medium. Each experiment was performed in 3 (n = 3) biological replicates. The underlying data can be found in S5 Dataset. (C) Assessment of sporulation frequency in 70:30 medium based on phase-contrast microscopy images (n = 3 biological replicates) shown in S6D Fig. The underlying data can be found in S5 Dataset. (D) Northern blot-based expression analysis of ModT in C. perfringens and P. sordellii across different growth stages in various rich media, including BHI, BHI + yeast extract (BHIS), Tryptone Yeast (TY), and TY supplemented with 0.5% glucose (TYG). 5S rRNA served as loading control. Depicted is a representative of 3 biological replicates. Significance for data shown in (B) and (C) was determined using two-way ANOVA with Dunnett’s multiple comparisons test. Only significant differences with P P P P

https://doi.org/10.1371/journal.pbio.3002948.g006

To develop a better understanding of whether ModT orthologs carry out conserved functions, we sought to explore their expression and isoform profiles within their native hosts. For that purpose, we chose the 2 human pathogens Clostridium perfringens and Paeniclostridium sordellii that belong to the family Clostridiaceae. Although they are phylogenetically distant from C. difficile (family Peptostreptococcaceae), they are both spore formers and encode homologs of the large clostridial toxins TcdA and TcdB [47,48]. Both bacteria were grown in media with varying levels of glucose to collect time-resolved expression profiles across the different growth stages. This expression analysis revealed several interesting parallels to C. difficile ModT (Fig 6D, compare to Fig 1C). Firstly, both orthologs occur in 2 isoforms, one matching the conserved ModT motif (~210 nt) and a larger isoform of ~240 nt. Secondly, modT is also strongly expressed during exponential growth, but depleted in the stationary phase. Third, medium supplementation with glucose extends the expression of the RNAs into the stationary phase. Although a recent expression analysis of ModT in C. acetobutylicum did not analyze the effect of different carbon sources on ModT expression, it also showed that ModT transcript levels decrease after entry into stationary phase [22]. In addition, the promoter regions of modT in all 4 clostridial species contain a highly conserved sigA consensus motif (S6G Fig), strongly suggesting that SigA-mediated modT expression in the exponential phase is widespread and may be a key aspect of its regulation. Despite these similarities, there is one major difference in that ModT(S) isoforms appear more stable than the ModT(L) isoforms in C. perfringens and P. sordellii, while the long isoform is lacking entirely in C. acetobutylicum.

In summary, the conserved expression profiles in various bacteria, along with the capacity of orthologs to complement ModT activity in C. difficile suggests that ModT members from diverse phylogenetic backgrounds are functionally conserved.

Bacterial strains and growth conditions

A list of bacterial strains used in this study is provided in S4 Dataset. C. difficile strain 630 (DSM 27543) cultures were routinely grown at 37°C inside a Coy chamber (85% N₂, 10% H₂, and 5% CO₂) on brain heart infusion (BHI) agar plates (1.5% agar) or in autoclaved BHI broth (37 g/l). Petri dishes were anaerobically incubated for at least 4 h prior to inoculation, while liquids were kept in the Coy chamber overnight to allow diffusion of entrapped oxygen. For selection of plasmid-harboring C. difficile 630, antibiotics were added at the following concentrations: thiamphenicol 15 μg/ml, cefoxitin 8 μg/ml, cycloserine 250 μg/ml. Cryostocks were stored at −80°C in BHI broth containing 15% glycerol.

To cultivate C. difficile with different carbon and energy sources autoclaved TY medium (30 g/l tryptone (Roth), 20 g/l yeast extract (Roth)) was supplemented with the respective source from 20% (w/V) sterile filtered stock solutions in ddH₂O to a final concentration of 0.5%. For growth curves, TY day cultures were inoculated to an OD₆₀₀ of 0.05 from TY overnight cultures, grown to exponential phase (OD₆₀₀ ~0.5) and subsequently diluted to an OD₆₀₀ of 0.01 in TY containing the respective carbon source. For growth curves in BHI and BHIS medium (BHI with 5 g/l of yeast extract (Roth)), cultures were inoculated from exponential phase BHI cultures. Optical density at 600 nm was recorded in a transparent 96-well plate in a Biotek H1 microplate reader.

C. perfringens ATCC 133124 and P. sordellii ATCC 9714 were cultured under the conditions indicated for C. difficile 630, with the following changes: For growth of P. sordellii in BHIS medium, 10% w/V sterile filtered cysteine solution was added to a final 0.1% (w/V) after autoclaving. For C. perfringens, the overnight culture was replaced with a day culture. Escherichia coli (E. coli) strains were aerobically cultivated in autoclaved Luria–Bertani (LB) broth (10 g/l tryptone, 5 g/l yeast extract, 10 g/l NaCl) at 37°C with shaking at 220 revolutions per minute (RPM), or on LB-agar (1,5% agar). Plasmids were maintained by adding 20 μg/ml chloramphenicol as a selection antibiotic. Cryostocks were stored at −80°C in LB broth with 25% glycerol.

Toxin ELISA

For C. difficile toxin quantification, 2.5 OD₆₀₀ units of bacterial culture were centrifuged at 10,000 × g, 4°C for 10 min. The supernatant was transferred to a new tube and centrifuged again under the same conditions to remove any residual cells. Approximately 250 μl cell free supernatant was stored at −80°C until quantification of the “secreted toxin.”

The cell pellet (used for quantification of “intracellular toxin”) was washed with 1 ml PBS (phosphate-buffered saline), then centrifuged again to remove the remaining liquid. Samples were kept at −80°C until further processing.

Cell pellets were suspended in 500 μl of ddH₂O and transferred to a new tube containing 500 μl 0.1 mm glass beads. For lysis, samples were ground in pre-cooled adapters (−20°C) in a Retsch Mixer Mill at 30 Hertz for 10 min, and subsequently centrifuged at 10,000 × g, 4°C for 15 min to remove debris and beads; 200 μl of the clear supernatant was transferred to a fresh tube and the protein concentration was estimated with a Bradford assay using RotiNano Quant solution (Roth) according to the manufacturer’s instructions. For relative toxin quantification, 4 μg of protein from the “intracellular toxin” sample were used as input, while “secreted toxin” was assessed from 50 μl of the cleared supernatant. The relative amount was determined by toxin ELISA (TGC-E001-1, TGC Biomics) performed and evaluated according to the manufacturer’s instructions. ELISA signal was quantified with a Tecan infinite 200 Pro plate reader assessing absorption at 450 nm and 620 nm. As a final readout, the signal at 620 nm was subtracted from the 450 nm signal.

Sporulation assays

Sporulation assays were performed in 70:30 sporulation broth medium according to Edwards and colleagues with minor alterations [56]. Overnight TY cultures were used to inoculate day cultures to an OD₆₀₀ of 0.05 in BHIS containing 0.5% fructose and 0.1% taurocholate. After growth to mid-exponential phase, 70:30 medium was inoculated to an OD₆₀₀ of 0.01.

At the end of transition phase (approximately 8 to 10 h after inoculation), the total CFUs from untreated cultures were assessed by serial dilution and spotting in BHIS, and in addition, an aliquot treated with ethanol for 30 min was serially diluted on BHIS agar with 0.1% taurocholate to check for spore carryover. This time point is referred to as 0 hours. At the subsequent time points, vegetative CFUs were assessed by serial dilution and plating on BHIS agar and spores were quantified from culture treated with ethanol (EtOH) (final concentration 48% v/V) for 30 min, followed by serial dilution in PBS and plating on BHIS containing 0.1% taurocholate.

For a second independent analysis at all time points, 500 μl culture aliquots were concentrated by brief centrifugation and approximately 3 μl of the concentrated cells were spotted onto an agarose pad on a glass slide for microscopy. Spores were visualized by phase contrast microscopy using a HC PLAN FLUOTAR 100×/1.32 PH3 oil immersion objective on a LEICA DM2500 microscope. Two fields of view were acquired per biological replicate from which at least 300 cells per field were counted to determine the sporulation frequency. In both methods, sporulation frequency was calculated as [spore count] / [spore count + vegetative cell count] = sporulation frequency.

Transformations

Plasmids were transformed to E. coli Top10 by applying a heat shock for 1 min at 42°C. After chilling on ice for 2 min, bacteria were incubated at 37°C, 350 RPM for 1 to 4 h in LB medium without antibiotics and were subsequently plated on selective LB-agar.

For transformations into E. coli CA434, 80 μl of a bacterial suspension was thawed on ice and mixed with 1 to 2 μl plasmid in a pre-chilled electroporation cuvette. Plasmid was delivered with an electric pulse (1.8 kV, 200 Ω, 4–5 s). Bacteria were recovered at 37°C in LB broth at 350 RPM for 1 h and subsequently streaked on selective LB agar.

Plasmid conjugation to C. difficile 630

For conjugations, 1 ml of an E. coli CA434 overnight culture harboring the respective plasmid was centrifuged at 4,000 × g, RT for 2 min. The pellet was transferred into the Coy chamber, and resuspended in 200 μl overnight culture of the respective C. difficile strain. After spotting onto BHI agar and 8 h of incubation at 37°C, cells were scraped from the plate with approximately 700 μl BHI broth and restreaked on BHI agar containing antibiotics to select for bacteria harboring the plasmid. Colonies appeared after 24 to 72 h and were restreaked to purity on selective BHI agar plates.

Plasmid assemblies

All used and generated plasmids are listed in S4 Dataset.

Chromosomal manipulation of C. difficile 630

For chromosomal manipulation, the two-step allelic exchange method was applied as described previously [21,57].

After plasmid conjugation, C. difficile 630 was restreaked at least twice and subsequently screened for single crossover events. Integrations into the modT locus (pFF-173/314/320/321/377/378) were identified by PCR using FFO-360/498, FFO-361/499 and FFO-498/499 and FFO-360/613. Integrations into the pyrE locus (pFF-221/222) were identified using FFO-361/612 and FFO-612/613. Single crossover events for generating Δ01583-dccA (pFF-451) were identified using FFO-1746/361 and FFO-1743/360.

To allow the second recombination event, single crossover integrants were incubated on BHI-agar for approximately 72 h, harvested in 900 μl PBS, and plated in serial dilutions onto CDMM agar plates containing 50 μg/ml 5′-Fluorcytosine. Emerging colonies were restreaked onto BHI and BHI containing thiamphenicol. Cultures growing on BHI, but not on selective BHI were screened for second recombination events.

Successful manipulation at the modT locus was determined by colony PCR using FFO-498/499 and integrity of homology and the knockout region was verified by Sanger sequencing (FFO-498, 499, 482, 483). Similarly, double crossover recombinants at the pyrE locus were identified using FFO-612/613 and verified by Sanger sequencing (FFO-610, 611, 612, and 613). For sequencing of the 01583-dccA operon knockout, FFO-1743, 1746, and FFO-1749 were used.

Hot phenol extraction of total RNA and DNase I digestion

For collection, samples were mixed with ice-cold Stop Mix (95% EtOH, 5% acidic phenol) equivalent to 1/5 of the culture volume and subsequently frozen in liquid N₂. For RNA extraction, samples were thawed on ice, centrifuged for 20 min at 4,000 × g, 4°C. Pellets were suspended in 600 μl of 10 mg/ml lysozyme in TE buffer (pH 8) and digested for 10 min at 37°C, and 60 μl of 10% (w/V) SDS was added, samples were mixed by inversion and incubated at 64°C for 1 to 2 min. Subsequently, 66 μl of 3 M NaOAc, pH 5.2 with 1 mM EDTA were added and mixed by inversion, followed by addition of 750 μl of acidic phenol for RNA extraction (Roth). Samples were mixed by shaking and incubated at 64°C in a water bath for 6 min with intermediate mixing. Phases were separated by centrifuging at 4°C, 16,000 × g for 15 min and the aqueous phase was transferred to a Phase Lock Gel tube (Eppendorf), where an equal volume of chloroform was added. Phases were thoroughly mixed and subsequently separated by centrifugation at 16,000 × g, 15°C for 12 min. Again, the aqueous phase was transferred to a new tube and 2 volumes of a 30:1 Ethanol to 3 M NaOAc (pH 6.5) mix was added and mixed by inversion. RNA was precipitated overnight, centrifuged at 16,000 × g, 4°C for 30 min and washed twice with 500 μl ice cold 75% Ethanol, with 10-min centrifugation steps in between. After complete removal of liquid, the pellet was air dried at RT for approximately 5 min and eluted in 50 μl of RNase free water by shaking at 1,000 RPM, 65°C, for 10 min. RNA concentration and purity was assessed with a Nanodrop 2000 spectrophotometer (Thermo Fisher).

Approximately 5 μg of RNA were digested with 5 U DNase I and 10 U RNase Inhibitor (Takara) at 37°C for 1 h. For RNA purification, the sample was adjusted to 100 μl with DEPC-treated ddH₂O and transferred to a Phase Lock Gel tube, where it was mixed with 100 μl P:C:I for RNA (Roth) by inversion. Phases were separated by centrifugation at 16,000 × g, 4°C for 15 min. The aqueous phase was transferred to a new tube and precipitated overnight after adding 2.5 volumes of 30:1 Ethanol to NaOAc (pH 6.5) and mixing. Precipitated RNA was centrifuged for 30 min at 16,000 × g, 4°C, washed twice with 200 μl 75% EtOH with 10-min centrifugation steps at 16,000 × g, 4°C, dried for 5 min after complete liquid removal and eluted in 20 μl DEPC-treated water by shaking for 5 min at 1,000 RPM, 65°C.

Rifampicin assay

To inhibit transcription of growing cultures, Rifampicin was added to C. difficile WT cells grown in TYG medium to mid-exponential phase, to a final concentration of 200 μg/ml per culture and mixed by swirling. RNA samples were aliquoted at the indicated time points and samples were fixed by adding ice-cold stop mix, inversion and immediate freezing in liquid nitrogen. RNA was extracted as indicated in the hot phenol protocol.

Northern blotting

For northern blot analysis, 3 to 6 μg of RNA were suspended in 2× Gel loading buffer II (95% formamide, 18 mM EDTA, 0.025% SDS, 2% bromophenol blue), boiled for 5 min at 95°C and subsequently cooled for 5 min on ice. After separation on a 6% polyacrylamide (PAA) gel containing 7 M Urea for 2 h at 300 V, RNA was transferred to a Hybond N+ membrane (GE Healthcare) in 1× TBE using a wet blotting chamber for 1 h at 50 V, 4°C and subsequently cross-linked at 254 nm (0.12 J/cm²).

Membranes were pre-hybridized in Roti Hybri-Quick buffer (Roth) for 1 h at 42°C. RNA of interest was detected with [γ-³²P]-labeled oligonucleotides by overnight hybridization at 42°C. A list of all oligonucleotides used in this study can be found in supplementary S4 Dataset. The membrane was washed sequentially in 5×, 1×, and 0.5× SSC containing 0.1% SDS, dried and finally exposed on a phosphor screen for 1 to 7 days. Signals were visualized using a Typhoon FLA7000 phosphor imager. Band intensities were assessed using ImageJ v1.52u [58].

Blots were stripped with 0.1% SDS in distilled, boiled water for 15 min and subsequently washed in distilled water. Blots were hybridized in Roti Hybri-Quick buffer before adding a new oligonucleotide probe.

Oligonucleotide labeling

For [γ-³²P] 5′-labeling, 10 pmol of DNA-oligonucleotide probes were incubated with 5 U T4 Polynucleotide Kinase (NEB) together with 1 μl [γ-³²P]-ATP (10 μCi/μl) in a volume of 10 μl. After 1 h incubation at 37°C, 10 μl of sterile, DNase- and RNase free water was added and radiolabeled oligonucleotides were separated from free [γ-³²P]-ATP using a sephadex G-25 column (Cytiva) according to the manufacturer’s instructions.

In vitro transcription of RNA, RNA labeling, and in-line probing (ILP)

ILP assays were performed as described previously [53]. The short and the long isoform of ModT were PCR amplified using FFO-1235/1089 and FFO-1235/1088 to add a T7 promoter. RNA was in vitro transcribed using the T7 Megascript RT Kit (Thermo Scientific), resolved on a 6% PAA, 7 M Urea gel, excised and eluted overnight in 750 μl RNA elution buffer (0.1 M NaOAc, 0.1% SDS, 10 mM EDTA) at 1,000 RPM, 8°C. RNA was purified by adding 750 μl P:C:I for RNA (Roth), thoroughly mixed and phases were separated by centrifugation. The aqueous phase was transferred to a new tube and RNA was precipitated by adding 1 ml 30:1 EtOH:NaOAc (pH 6.5).

Approximately 50 pmol of in vitro transcribed RNAs were dephosphorylated by CIP (NEB) treatment, purified and precipitated as described before and 20 pmol of eluted RNAs were radioactively 5′-labeled using 10 U T4 Polynucleotide kinase (Thermo Scientific) and 2 μl [γ-³²P]-ATP (10 μCi/μl) in 1× PNK Buffer A (Thermo Scientific) in a total volume of 20 μl for 30 min at 37°C. Labeled RNA was purified using a sephadex G-50 column (Cytiva) and a denaturing PAA gel, isolated by overnight elution from the gel and purified by organic phase separation and precipitation of the aqueous phase as described above.

For ILP reactions, labeled ModT was denatured at 95°C for 1 min and subsequently chilled on ice for 5 min, and 0.04 pmol/μl ModT was incubated in a buffer containing final concentration of 100 mM KCl, 20 mM MgCl₂, 50 mM Tris-HCl for approximately 40 h at RT in a volume of 15 μl. Control reactions were set up in RNase free water. For a T1 ladder, 0.01 U/μl of RNaseT1 (Thermo Scientific) was incubated with 0.02 pmol/μl of labeled ModT RNA for 5 min at 37°C in a volume of 15 μl. For the alkaline hydrolysis control reaction, 0.02 pmol/μl ModT was heated to 95°C for 5 min in sodium bicarbonate (alkaline hydrolysis) buffer, pH 9 (Thermo Scientific). All reactions were stopped by adding an equal volume of 2× colorless loading dye (10 M Urea, 1.5 mM EDTA). Finally, 10 μl of the quenched reactions was separated on 60-cm long denaturing PAA gels at 40 W, which were dried onto whatman paper for 2 h at 80°C. The radioactive signal was exposed to a phosphor screen, which was read out with a Typhoon FLA 7000 imager (Fuji series).

DMS-structure probing and sequencing

Reverse transcription and quantitative PCR

For cDNA strand synthesis, 1 μg of DNase I digested RNA was reverse transcribed using the M-MLV reverse transcriptase Kit (Invitrogen) with random hexamer primers (Invitrogen) according to the manufacturer’s instructions.

qPCR reactions were performed using 1 μl of 20-fold diluted cDNA and with 2x Takyon NO ROX SYBR Master Mix blue TTP (Eurogentec) with 20 nM primers in a total volume of 10 μl. Each reaction was carried out at least in technical duplicates per biological replicate. Fold change was calculated with the 2^-ΔΔCT method [62] using 16S rRNA as a housekeeping reference.

RNA sequencing

Samples for RNA sequencing were collected from C. difficile WT and ΔmodT cultures grown in TYG medium to EP (3, 5h) and to TP (9 h). RNA was extracted using the hot phenol method, DNase I digested and purified as indicated above. cDNA library preparation and sequencing were performed by Vertis Biotechnologie AG. RNA samples were examined by capillary electrophoresis. Ribosomal RNA was depleted using an in-house developed protocol. The ribodepleted RNA samples were fragmented using ultrasound (1 pulse of 30 s at 4°C) and an oligonucleotide adapter was ligated to the 3′ end of the RNA molecules. First-strand cDNA synthesis was performed using M-MLV reverse transcriptase and the 3′ adapter as primer. The first-strand cDNA was purified and the 5′ Illumina TruSeq sequencing adapter was ligated to the 3′ end of the antisense cDNA and the resulting cDNA was PCR-amplified (15 cycles) to about 10 to 20 ng/μl using a high fidelity DNA polymerase. The cDNA was purified using the Agencourt AMPure XP kit (Beckman Coulter Genomics) and analyzed by capillary electrophoresis to estimate library quality and concentration. For Illumina NextSeq sequencing, the samples were pooled in approximately equimolar amounts. The cDNA pool was size fractionated in the size range of 200 to 600 bp using a preparative agarose gel. An aliquot of the size fractionated pool was analyzed by capillary electrophoresis for quality control. The cDNA pool was single-read sequenced on an Illumina NextSeq 500 system using 75 bp read length.

Bioinformatic analysis of RNA sequencing

Raw reads of RNA-seq libraries were trimmed and filtered using BBDuk (ktrim = r, k = 23, mink = 11, hdist = 1, tpe, tbo, qtrim = r, trimq = 20, minlength = 25). Mapping and read counting was conducted with READemption 0.4.3 (READemption align:—poly_a_clipping,—realign, READemption gene_quanti:—skip_antisense,—no_count_splitting_by_gene_no) [63]. Reads were mapped to C. difficile 630 CP010905.2 with additional annotations of UTRs and ncRNAs. Differential expression analysis was performed with edgeR [64]. Normalization factors were calculated using the trimmed means of M values method (TMM). All features with less than 10 reads in at least 2 libraries were excluded from the analysis.

Fluorescence quantification by flow cytometry

At the respective time point, 200 μl of each sample was transferred to a 96-well plate and centrifuged at 4,500 × g, 4°C for 10 min. Supernatant was removed using a multichannel pipette and bacterial pellets were suspended in 150 μl of 4% PFA and fixed at RT in a dark plate for 30 min. Bacteria were washed 3 times in 200 μl PBS with centrifuging steps as indicated above. The final cell pellet was suspended in 30 μl PBS containing 1 μg/ml DAPI and kept overnight in the dark at 4°C to allow full fluorophore maturation. The next day, samples were centrifuged, supernatant was removed and pellets were thoroughly suspended in 200 μl of PBS. Cells were diluted to an appropriate concentration and red fluorescence (λ_ex = 561 nm, signal detection with the custom “PE-Texas Red” bandwidth filter detecting 615 +/−20 nm) was determined using a Novocyte Quanteon flow cytometer (Agilent) from 100,000 events, selecting exclusively DAPI-positive cells (λ_ex = 405 nm, applying the “Pacific-Blue” filter (detecting 445 +/−45 nm) and a FSC-H (>5,000 AU) cutoff to discriminate cells from false positive debris events. The mean fluorescence of the selected events was taken as readout. To normalize fluorescence, the mean fluorescence from the inactivated riboswitch (pRS^A70G) was subtracted from the functional, c-di-GMP sensitive reporter (pRS) in each respective background. Fluorescence of all strains carrying the nonfluorescent control (pEV) was measured, but was not included in the calculation.