Regulatory Effects of CsrA in Vibrio cholerae

Vibrio cholerae, a Gram-negative bacterium, is a natural inhabitant of the aqueous environment. However, once ingested, this bacterium can colonize the human host and cause the disease cholera.

and CsrD (Csr sRNAs). In E. coli, the gene named csrD encodes a protein that controls the degradation of CsrA (18), and MshH plays this role in V. cholerae (19). The three Csr sRNAs contain multiple CsrA-binding sites, characterized by GGA motifs located in the loop region of stem-loop structures in the RNA. When expressed, these sRNAs act to sequester CsrA, thus preventing CsrA from interacting with its target mRNAs. Therefore, CsrA regulation uses a feedback loop, whereby CsrA positively induces the expression of its antagonistic sRNAs. It also indicates that the availability of CsrA, or activity of CsrA, is tightly regulated within the cell.
In this study, we show that CsrA in V. cholerae is a global regulator that affects a large number of gene products. Direct targets for CsrA include mRNAs for regulators that in turn control expression of genes encoding a number of pathways, including those responsible for motility, rugosity, and virulence.

RESULTS
To better understand the regulatory role of CsrA in V. cholerae, we used RNA-seq transcriptome sequencing to compare the transcriptome of a csrA mutant strain to that of its wild-type parental strain, N16961. It should be noted that N16961 does not have an intact quorum sensing (QS) system, and this could lead to differences in the CsrA regulon compared with a QS1 strain. CsrA is essential for the growth of V. cholerae, and therefore a complete deletion of the csrA gene is not possible. We have isolated a csrA point mutant, NcsrA.R6H, which has wild-type growth (7) but does not exhibit the CsrA-dependent increase in ToxR levels in response to NRES, indicating that the CsrA.R6H protein is less functional than wild-type CsrA. Significantly, this mutant is avirulent in the infant mouse model of V. cholerae infection (7). The NcsrA.R6H mutant and the wild-type strain were grown in defined medium in the presence of NRES, which we have shown previously to increase the level of active CsrA (7,17). To give a more complete picture of the role of CsrA in regulating gene expression at different stages of V. cholerae growth, RNA was harvested from wild type and the csrA mutant grown to early exponential, mid-exponential, and stationary phase. Because the NcsrA. R6H mutant grows as well as its wild-type parental strain in defined medium supplemented with NRES (7), the observed differences in gene expression between the two strains are not due to differences in growth rate or yield.
CsrA is a global regulator in V. cholerae. In total, 832 genes were significantly differentially expressed in the NcsrA.R6H strain compared to the wild type (Fig. 1A). Only genes exhibiting a greater than 2-fold change in expression between the csrA mutant and the wild-type strain, with a false discovery rate (FDR) P value of less than 0.05, were included in the analysis (Table S1 in the supplemental material). CsrA has been shown to play a major role in regulating stationary-phase gene expression in other bacteria (8,9,20) and, not surprisingly, the majority of differentially regulated genes were observed at stationary phase. A total of 712 stationary phase CsrA-regulated genes were identified and, of these, 621 were regulated exclusively in stationary phase, showing that CsrA in V. cholerae, like in E. coli, is a major stationary phase regulator. Each growth phase was associated with the unique expression pattern of at least 35 genes. Additionally, 19 genes were found to be significantly differentially regulated at all phases of growth (Table S1). The genes regulated in all growth phases included the csrB and csrD genes, 4 transport genes, 5 metabolism genes, the hlyA gene encoding hemolysin, and 7 genes encoding proteins with uncharacterized or hypothetical functions.
Gene Ontology (GO) groupings were used to characterize the global effects of CsrA regulation on cellular processes (Fig. 1B) (Table S2). The PANTHER (Protein Analysis Through Evolutionary Relationships) bioinformatics web server (http://www.pantherdb .org) (21,22) was used to determine GO groupings that are overrepresented among significantly regulated genes in the csrA mutant compared with the wild-type strain within the three different growth phases. Based on the number of genes in a group, the analysis determined whether the number of genes that are regulated by CsrA is The V. cholerae CsrA Regulon ® significantly more or less than would be expected by chance. The ratio of observed to expected genes is expressed as fold enrichment (Fig. 1B). From this GO analysis, we found that CsrA affects the expression of multiple categories of processes, and some of the GO groupings are regulated in a growth phase-dependent manner. For example, at early exponential phase, there was a 19.8-fold enrichment of significantly more highly expressed genes in the tricarboxylic acid (TCA) cycle GO group in the csrA mutant than in the wild type. Genes in this GO group were also significantly overrepresented in stationary phase but, in contrast to early exponential phase, these genes were expressed at a lower level in the mutant compared to wild type in stationary phase. Genes in the glycolytic process GO group were also overrepresented at stationary phase, but these genes were more highly expressed in the csrA mutant than in the wild type. This pattern of expression is consistent with CsrA repressing the TCA cycle early in the growth cycle but allowing TCA cycle genes to be expressed in stationary phase, while at the same time repressing glycolysis in stationary phase. No significant that displayed a greater than 2-fold difference in gene expression and had an FDR P value of ,0.05 in NcsrA.R6H compared to N16961. (B) GO ontology analysis performed via PANTHER. The genes included in this analysis had a greater than 2-fold change in expression values. The bars that extend to the left of zero (blue shading) have an overrepresentation of downregulated genes, whereas the bars that extend to the right of zero (orange shading) have an overrepresentation of upregulated genes in the csrA mutant compared to the wild type at the indicated growth phase. Fold enrichment, as determined by PANTHER analysis, is the comparison of the regulation of genes in that category to the basal regulation for all genes in comparing the mutant to the wild type RNA-seq data. The Fisher's exact test was applied to determine significantly regulated categories (FDR P values: *, ,0.05; **, ,0.01; ***, ,0.001). enrichment of glycolysis genes was observed in the csrA mutant at either early-or midexponential phase.
Nucleotide biosynthesis processes were overrepresented at stationary phase. We observed upregulation of genes of the nucleotide biosynthetic process (3.5-fold overrepresentation) and the de novo inosine monophosphate (IMP) biosynthetic process GO groups in the csrA mutant at stationary phase. Nitrogen compound biosynthesis was similarly expressed at a higher level in the csrA mutant at stationary phase. These observations point to an overall repressive effect of CsrA on nucleotide biosynthesis during stationary-phase growth. CsrA may also act to repress gene expression and protein translation generally in stationary phase, as these two GO groups were both significantly overrepresented in stationary phase, showing a relatively higher expression in the csrA mutant. At the same time, cellular catabolic processes, including lipid catabolism, were expressed at lower levels in the mutant. These results point to a role of CsrA in decreasing gene expression and de novo protein biosynthesis, while increasing catabolic processes, in response to nutrient depletion in stationary phase. We found that genes encoding several inorganic ferric and ferrous iron uptake systems, including FbpABC (VC0608 to VC0610), FeoA (VC2078), and VctPDGC (VCA0227-to VCA0230) (reviewed in reference 23), were upregulated in the csrA mutant in stationary phase, while some genes encoding iron-storage, iron-binding, and iron-containing proteins, such as ferritin (VC0078), ferredoxin (VC0311), and iron-sulfur cluster-containing proteins (VC1512, VC2088, and VCA0985), were repressed. This suggests that CsrA may be critical for restricting the influx of free, unchelated iron during stationary phase growth, while sequestering any excess intracellular iron.
Another significantly overrepresented group in the GO analysis was flagellum-dependent cell motility. Genes whose expression was relatively lower in the csrA mutant at stationary phase were overrepresented in this category, suggesting that CsrA is an activator of motility. Of the 25 genes included in this category, 11 were regulated in response to CsrA, suggesting that CsrA affects flagellar gene expression early in the cascade or at multiple points in the flagellar gene expression hierarchy.
Among other notable targets of CsrA regulation observed in the RNA-seq analysis were several sigma factor genes, including rpoS (sigma-38, down 7-fold in the csrA mutant), rpoN (sigma-54, down 2.1-fold in the csrA mutant), rpoE (sigma-24, down 2-fold in the csrA mutant), and rpoD (sigma-70, up 2.6-fold in the csrA mutant). This pattern of sigma factor regulation suggests a larger picture involving CsrA-mediated upregulation of the stationary-phase stress response (rpoS, rpoE), with an overall dampening of gene expression through downregulation of the housekeeping sigma factor (rpoD). RpoS and RpoN are known to control many processes linked to the virulence of V. cholerae, including motility, quorum sensing, and biofilm formation, and their positive regulation by CsrA further highlight the importance of CsrA for the pathogenesis of V. cholerae. Additional CsrA-regulated genes with known or predicted roles in virulence identified in the RNA-seq analysis include varA (VC1213), which we have demonstrated previously to be a direct target of CsrA regulation (17), hlyA, encoding the major V. cholerae hemolysin (24), rstA and rstB, encoded on the CTX-phage (25), and aphA (VC2647), which is discussed in detail below.
Expression of the Csr sRNAs changes over time. In our previous work, we demonstrated an autoregulatory feedback loop in which CsrA positively regulates the expression of the three Csr sRNAs (CsrB, CsrC, and CsrD) through its posttranscriptional effects on the expression of varA, encoding an activator of transcription of all three Csr sRNAs (17). As the Csr sRNAs function to antagonize the activity of CsrA, this feedback loop likely prevents drastic oscillations of CsrA activity. In support of our original observations, the expression of each of the Csr sRNAs was significantly reduced in the csrA mutant ( Fig. 2A); however, the extent to which each sRNA level decreased varied in the csrA mutant. CsrC showed the most dramatic reduction in response to the loss of CsrA, particularly in early and mid-exponential phase, whereas CsrD was the least affected, despite being the most abundant in the wild type at stationary phase (Fig. 2B). This could suggest intrinsic differences between the sRNAs in their expression and dependence upon csrA. Our previous work indicated that the Csr sRNAs are not completely functionally redundant (17), leading us to hypothesize that the expression of the Csr sRNAs may vary in response to different conditions, including growth phase. We examined the expression of the Csr sRNAs in the wild-type strain at each growth phase. We found that the level of CsrB did not significantly differ from one growth phase to the next (Fig. 2B). CsrC expression was highest at early exponential phase, and then decreased over time and had the lowest abundance of the three sRNAs at mid-exponential and stationary phase. In stationary phase, the expression of CsrC was quite low, representing a 41-fold decrease compared to the level in early exponential phase. The levels of CsrD did not change from early to mid-exponential phase, but did increase substantially from mid-exponential to stationary phase. Notably, at stationary phase, CsrD was the most highly expressed RNA in the entire wild type transcriptome and accounted for nearly 10% of the total reads, with 93,842.2 transcripts out of a total of 1,000,000 transcripts. The differential expression of the Csr sRNAs supports the hypothesis that the Csr sRNAs have intrinsically different properties, and that this system is finely tuned to regulate the activity level of CsrA in response to changing environments.
CsrA binds directly to multiple transcripts encoding regulatory proteins. Because CsrA had such a substantial effect on the V. cholerae transcriptome, we hypothesized that CsrA might be directly regulating multiple regulatory proteins. To  test this model, we performed a directed RNA-CsrA coimmunoprecipitation (co-IP) (Fig. 3) using CsrA-overproducing cells grown to mid-exponential phase, and real-time quantitative PCR (RT-qPCR) to determine relative enrichment of specific RNA targets. The targets chosen were mRNAs identified as CsrA-regulated in the RNA-seq analysis and encode known or putative regulatory proteins. The amount of the RNA target was quantified relative to an internal standard in the RNA-CsrA coimmunoprecipitated sample compared with the initial RNA population. As a positive control for the enrichment of direct CsrA targets, CsrB was included in the panel of RNAs investigated. CsrB was enriched over 100-fold in the immunoprecipitated sample, indicating this method can detect direct targets of CsrA. Targets less abundant than CsrB can also be detected through co-IP analysis; our previous work demonstrated that varA mRNA, described above as encoding the transcriptional activator of Csr sRNA expression, was enriched 5-fold in the CsrA-coimmunoprecipitated RNA pool (17). We found that the selected mRNAs that encode regulatory proteins coimmunoprecipitated with CsrA. For example, transcripts encoding the sigma factors RpoS and RpoE, the flagellar regulator FlrC, and the QS master regulator AphA were significantly enriched with CsrA. This finding suggests that CsrA may exert large regulatory effects on the V. cholerae transcriptome by controlling the expression of other regulators. Consistent with our observation that CsrA is required for virulence within the host, we identified multiple CsrA-interacting mRNAs associated with pathogenesis, including the following: aphA (26); flrC (27); epsC, which encodes a component of the type two secretion system that is responsible for the secretion of the cholera toxin (28); cobB, which encodes an NAD1 dependent deacetylase (29); and VCA0965, encoding a functional cyclic di-GMP synthase (30). Selected regulators are discussed in more detail below.
CsrA regulates the expression of the stationary phase regulator RpoS. A large proportion of the CsrA-regulated genes identified in the RNA-seq analysis were exclusive to stationary-phase cells, suggesting that CsrA may be a major regulator of stationary phase gene expression. Expression of rpoS, which encodes the stationary phase sigma factor, sigma-38, was downregulated ;7-fold at stationary phase in the csrA mutant (Table S1), suggesting CsrA is critical for induction of rpoS in stationary phase. In addition, we found that rpoS mRNA coimmunoprecipitated with CsrA ( Fig. 3), consistent with rpoS being a direct target of CsrA regulation. This result suggests that at least part of the observed regulation of stationary phase gene expression by CsrA may occur via RpoS. To investigate this idea further, we determined whether there was an overlap between the CsrA and RpoS regulons by comparing genes regulated by RpoS at stationary phase, identified by Nielsen et al. (31), to genes regulated by CsrA at stationary phase. It is important to note that these two RNA-seq analyses were performed in different media, which may affect the expression of some of the genes. We found that, at stationary phase, 195 genes exhibited regulation by both CsrA and RpoS. Interestingly, of these genes, 63% are regulated in a similar way by CsrA and RpoS, exhibiting either upregulation or downregulation by both regulators. Taken together, these data suggest that regulation of stationary phase gene expression by CsrA occurs, at least in part, through its control of RpoS levels.
CsrA positively regulates motility in V. cholerae. In our RNA-seq analysis, we found that over half of all annotated flagellum-dependent motility genes had significantly lower expression in the csrA mutant relative to the wild-type strain at stationary phase. This difference suggests a major role for CsrA in the regulation of flagellum-dependent motility in stationary phase. Figure 4A shows the flagellar proteins encoded by the genes that were significantly regulated in the RNA-seq analysis. In addition to flagellar structural genes, we found that several regulatory genes involved in flagellar gene expression were differentially expressed in the csrA mutant, including rpoN, rpoS, and flrC, suggesting the effects of CsrA on motility gene expression could be due to CsrA-mediated regulation of these central regulators. In support of this, we found that CsrA bound to both rpoS and flrC mRNA in vivo, as determined by the CsrA-RNA coimmunoprecipitation experiments (Fig. 3).
To determine whether the observed regulation of flagellar gene expression by CsrA is relevant for motility in V. cholerae, the ability of the csrA mutant to swim through a semisolid agar medium was assessed. The csrA mutant strain displayed a reduced zone of spread compared to the wild type, indicating the csrA mutant is less motile than the wild type (Fig. 4B). The motility defect was restored by ectopically expressing the wildtype csrA allele in the NcsrA.R6H strain, demonstrating that the motility defect is due to the csrA mutation. Because the csrA mutant is less motile than the wild type, we conclude that CsrA positively regulates motility in V. cholerae, consistent with the observed downregulation of motility gene expression in the csrA mutant.
CsrA regulates rugose switching in V. cholerae. In order to cope with extracellular stressors, V. cholerae may alter its outer surface through the excretion of extracellular polysaccharides, resulting in a rugose colony morphology. The rugose morphology arises through phase variation, yielding colonies with a distinctive rough, wrinkled surface (32). The rugose phenotype is associated with an overproduction of biofilms (33) and increased resistance to osmotic and oxidative stressors (33,34) and antimicrobial compounds (35). The rugose variant has been linked to outbreaks of cholera (35) and has been proposed to confer increased persistence in the environment (36). Many of the regulatory and structural genes known to play a role in the production of the extracellular polysaccharide matrix of V. cholerae (reviewed in reference 37) were identified in the RNA-seq and CsrA-RNA coimmunoprecipitation analyses as being potential targets of CsrA regulation. These include genes encoding the sigma factors RpoS, RpoE, and RpoN, as well as the biofilm-specific transcriptional regulator VpsT (VCA0952), and several genes in the vpsI, vpsII, and rbm biofilm gene clusters (VC0916 to VC0939). While many regulatory genes, including vpsT, were expressed at lower levels in the absence of CsrA, a number of the structural genes were differentially regulated through the growth phases, being generally downregulated or neutral in mid-exponential phase but significantly more abundant at stationary phase in the csrA mutant. This pattern of expression suggests that CsrA may be required for induction of exopolysaccharide gene expression during rapid growth, but may have a repressive effect in stationary phase, indicating temporal regulation of this process by CsrA. Interestingly, at least one transcript involved in exopolysaccharide production, rbmA (VC0928), was identified as an RNA target of CsrA binding (Fig. 3), suggesting direct regulation by CsrA.
To directly test whether CsrA affects the ability of V. cholerae to undergo phase variation and induce exopolysaccharide production, we compared the frequency of transition from the smooth to the rough, or rugose, colony morphology in the wild type and csrA mutant at 24 and 48 h of growth without shaking in alkaline peptone water (APW) number 3 growth medium, which promotes switching to the rugose morphology (36,38). We found that after 24 h of growth, NcsrA.R6H had not undergone phase variation at all, as none of the cells formed rugose colonies on solid medium. In contrast, 74% of the wild-type colonies displayed the rugose colony morphology at this time point (Fig. 5). However, at 48 h, both the wild-type and the csrA mutant colonies displayed the rugose variant phenotype at a frequency of approximately 80%. Additionally, this temporal regulation correlates with the RNA-seq data. Taken together, these observations suggest temporal regulation of rugose switching by CsrA; CsrA may act as a positive regulator of exopolysaccharide production early in the growth cycle, whereas later in the growth cycle the interplay between CsrA and other stationary phase regulators may serve to repress production of the extracellular matrix.
CsrA binds to the aphA mRNA and positively regulates AphA protein production. AphA is a transcriptional regulator essential for the induction of virulence gene expression in V. cholerae (26). AphA also activates biofilm gene expression by inducing expression of vpsT (39). We found that the aphA (VC2647) transcript was enriched 15-fold in the CsrA-RNA coimmunoprecipitation (Fig. 3), suggesting that aphA is a direct regulatory target of CsrA. CsrA plays a key role in both virulence (7) and biofilm (vps) gene expression (Table S1), and some of this regulation could be mediated by aphA. To confirm that CsrA binds directly to the aphA mRNA, we performed an in vitro RNA electrophoretic mobility shift assay (EMSA) with purified CsrA and a 39 biotinylated aphA mRNA probe. The aphA probe begins at the transcriptional start site, as determined by Papenfort et al. (40), and includes, in addition to the 59 untranslated region (59 UTR), 228 nucleotides of the aphA coding sequence (11 to 1432 relative to the start of transcription). The sequence of the aphA probe contains one GGA motif in the 59 UTR and four additional GGA motifs in the coding sequence. The RNA EMSA The V. cholerae CsrA Regulon ® revealed an initial shifted aphA species with the addition of 375 nM of purified CsrA (Fig. 6A). As the concentration of CsrA increased, the extent to which the aphA species was shifted also increased. To compete with the biotinylated aphA mRNA (aphA*), unlabeled aphA mRNA was added in 10-fold (lane 5) or 50-fold (lane 6) excess of aphA*. The addition of unlabeled aphA competed for CsrA binding, resulting in a reduction in the extent of aphA* species shifting. This result shows that CsrA is capable of binding to the aphA mRNA both in vitro and in vivo, indicating that aphA is a direct target of CsrA regulation.
Because both CsrA and AphA are positive regulators of virulence gene expression and aphA expression is itself modestly, but positively, influenced by CsrA in exponential growth phase (17) (Fig. S1), we hypothesized that CsrA positively regulates AphA protein production by binding to the aphA transcript and increasing the efficiency of translation of the aphA mRNA. To determine whether CsrA regulates AphA protein production, we compared the amount of AphA protein produced in the wild type versus the csrA mutant. In order to visualize the AphA protein, a V5 epitope tag sequence was fused in-frame to the 39 end of the aphA gene in the chromosome of both the wild type and the csrA mutant, giving rise to strains NaphA-V5 and NcsrA.R6H.aphA-V5. Strains were grown in defined medium supplemented with NRES, and the level of AphA-V5 was determined by Western blotting using anti-V5 antisera (Fig. 6B). There was less AphA-V5 produced in the csrA mutant than in the wild type, and the AphA-V5 level was restored by complementing the csrA mutant with the wild type csrA allele on a plasmid. This result shows that the observed decrease in AphA-V5 protein production was due to the csrA mutation, confirming that CsrA positively regulates AphA protein production.
The regulation of AphA protein synthesis by CsrA could occur through an increase in translational efficiency when CsrA binds to the aphA transcript. To test whether CsrA affects the translation of aphA, a translational fusion of aphA to the lacZ gene was constructed for testing in the wild type and the csrA mutant. For comparison, a transcriptional aphA-lacZ promoter fusion was generated as well. CsrA is a posttranscriptional regulator with no known direct effects on transcription; however, we wanted to rule out any indirect effects of CsrA on the expression of aphA in the reporter assays. In order to ensure all b-galactosidase is produced from the reporter constructs, the strains Nlacz::kan and NcsrA.lacZ::kan were used as the wild type and csrA mutant (17). The b-galactosidase activity from the transcriptional reporter was the same in both the wild type and the csrA mutant (Fig. 6C), consistent with CsrA having no effect on the expression of the aphA gene. In contrast, the translational reporter, containing the 59UTR and partial coding sequence of aphA fused to lacZ, had significantly reduced b-galactosidase activity in the csrA mutant compared with the wild type, suggesting CsrA is required for optimal translation of the aphA mRNA. Taken together, these results point to a mechanism of regulation whereby CsrA binds directly to the aphA mRNA and enhances its translation. The exact nature of the translational regulation of aphA by CsrA remains to be determined.

DISCUSSION
In this study, we have begun the characterization of the CsrA regulon in V. cholerae. By comparing differences in gene expression between a csrA mutant and its wild-type parental strain at multiple phases of growth, we found that CsrA affects the expression of ;22% of the V. cholerae transcriptome. Further, by analyzing the effects of CsrA at multiple phases of growth, we were able to identify genes that showed growth-phasedependent expression patterns, revealing significant changes during the V. cholerae growth cycle that would otherwise have been missed. The regulatory roles of CsrA identified in this analysis are summarized in Fig. 7. It is possible that the number of regulated genes is even greater, because the NcsrA.R6H mutant strain produces a CsrA protein with partial activity.  The most plausible explanation for the substantial effect of CsrA on the V. cholerae transcriptome is that CsrA regulates other regulators. We found that CsrA directly bound multiple mRNAs that encode transcriptional regulators, two of which are sigma factors, RpoS and RpoE, each controlling a large subset of genes. Regulation of RpoS by CsrA is likely responsible for the large proportion of the genes that were regulated exclusively during stationary phase, and there is substantial overlap between genes regulated by RpoS (31) and CsrA at stationary phase. The importance of CsrA in the adaptation of V. cholerae to the nutrient limitations and other stresses of stationary phase growth was also evident in the categories of genes regulated by CsrA. At stationary phase, CsrA had a repressive effect on genes involved in nucleotide biosynthesis, gene expression, and protein translation, suggesting that CsrA is critical for slowing down metabolism when nutrients are scarce. At the same time, cellular catabolic processes were enriched by CsrA, which would increase the supply of readily metabolizable macromolecules. During stationary phase growth, CsrA repressed genes involved in iron uptake, while increasing expression of genes encoding iron storage and other ironcontaining proteins, pointing to a major role for CsrA in both reducing influx of iron and increasing iron storage. By limiting the import of free iron, increasing iron storage, and increasing the expression of iron-binding proteins, CsrA may help to mitigate the toxic effects of free iron in stationary phase (41). Fewer iron-related genes were affected by CsrA in exponential phase, and these included genes encoding iron sulfur cluster-binding proteins. In contrast to the observed regulation in stationary phase, these genes were repressed by CsrA during mid-exponential phase, suggesting that CsrA may serve to redirect the available iron away from iron-intense metabolic pathways and toward essential biosynthetic pathways when iron is less abundant (42). In E. coli, CsrA was found to repress synthesis of intracellular iron storage proteins during growth in exponential phase, thus freeing up iron when the requirement for iron in metabolic and respiratory pathways is high (9,43). Together with our findings, this suggests that CsrA regulates iron levels to increase access to iron in the rapidly growing cell, but acts to mitigate the toxic effects of free iron when the iron requirements are less.
Central carbon metabolism was also a target of CsrA regulation. This was expected, as CsrA was originally named for its role in carbon storage regulation in E. coli (44). It was shown in E. coli that CsrA activates glycolysis both at mid-exponential phase and stationary phase through positive regulation of the expression and protein production of pfkA, which encodes the glycolytic enzyme phosphofructokinase I (9,(45)(46)(47)(48). Unlike the findings in E. coli, we did not detect a strong effect of V. cholerae CsrA on glycolysis at mid-exponential phase; the genes within this pathway were mostly unaffected by the csrA mutation. CsrA did, however, have an effect on glycolytic gene expression at stationary phase; expression of pfkA (VC2689) was significantly more highly expressed in the csrA mutant, suggesting that CsrA is a repressor of pfkA in V. cholerae, rather than an activator as in E. coli. Additional differences between E. coli and V. cholerae CsrA-mediated regulation of carbon metabolism were observed in the TCA cycle. While E. coli CsrA represses TCA cycle genes under multiple growth conditions (9,46), V. cholerae CsrA appears to repress expression in early exponential phase, and it clearly activates gene expression in stationary phase. Differences in CsrA-mediated regulation of TCA and glycolysis genes indicate that CsrA can regulate these processes independently in V. cholerae, whereas the regulation of these two pathways appears to be linked in E. coli (45,46,49).
Results from the RNA-seq experiments (Table S1) suggest that CsrA may activate the acetate switch in stationary phase. The acetate switch promotes the transition from acetate excretion during rapid growth to acetate assimilation when other, preferred, carbon sources become depleted. Expression of genes involved in the acetate switch, including genes encoding the two-component system CrbRS (VC0303/VC2702), a putative cation-acetate symporter system (VC2704/VC2705), acetyl-CoA synthase (VC0298) (50), and CobB (VC1509), which activates the acetyl-CoA synthase through deacetylation (29), was reduced in the csrA mutant at stationary phase, consistent with positive regulation of these genes by CsrA. Although the statistical threshold for inclusion in the data set (P , 0.05) was not met in every case for these genes, the overall trend in gene expression suggests that V. cholerae is no longer excreting acetate via glycolysis and mixed acid fermentation in stationary phase, but is instead taking up acetate from the medium and using it in the TCA cycle in a CsrA-dependent process. In addition, the cobB (VC1509) transcript was a direct target of CsrA binding at mid-exponential phase (Fig. 3), suggesting CsrA may also regulate acetate metabolism during rapid growth.
We observed that genes within the de novo inosine monophosphate (IMP) pathway were significantly enriched in the PANTHER analysis, with 7 out of 12 genes in this category expressed at a higher level in the csrA mutant compared to the wild type at stationary phase (Fig. 1B). Activation of the IMP pathway is a crucial step in the stringent response, as it is required for the synthesis of the alarmone (p)ppGpp (51). In E. coli, a csrA mutant had an increase in the expression of the (p)ppGpp synthase, relA, as well as increased levels of (p)ppGpp, indicating that CsrA acts to repress aspects of the stringent response (52). Interestingly, relA (VC2451) was not significantly differentially regulated in the RNA-seq; however, V. cholerae has another (p)ppGpp synthase, relV (VC1124) (53), which was upregulated 2.8-fold at stationary phase in NcsrA.R6H. CsrAdependent repression of the stringent response in E. coli, together with the overrepresentation of upregulated genes within the de novo IMP pathway, and upregulation of relV in NcsrA.R6H, suggest that in the wild type CsrA may repress the stringent response in V. cholerae as well.
In addition to regulation of metabolic pathways, V. cholerae CsrA plays a significant role in control of motility and virulence, and the pattern of CsrA as a regulator of regulators emerges here, as well as in stationary phase regulation. One of the mRNAs enriched in the CsrA-RNA co-IP was VCA0965, which encodes a cyclic di-guanylate (c-di-GMP) synthase. C-di-GMP is a secondary metabolite that has been shown to inversely regulate motility and biofilm formation (30,54,55), as well as regulate the expression of V. cholerae virulence genes (56). It has also been shown to regulate the expression of multiple transcriptional regulators such as vpsT and aphA (57,58). A decrease in c-di-GMP in the csrA mutant could explain the downregulation of vpsT expression, as cdi-GMP is a known inducer of vpsT expression (34). VpsT is the major activator of genes within the vibrio polysaccharide synthesis (VPS) gene cluster (VC0916 to VC0939), and some of the vps genes (VC0931 to VC0938) had reduced expression in the csrA mutant at mid-exponential phase (59). This would suggest that CsrA has a positive effect on VPS production in V. cholerae; however, many vps genes, including vpsA (VC0917) and rbmA (VC0928), were significantly upregulated in the csrA mutant in stationary phase, indicating the effect of CsrA is to repress these genes in stationary phase. Thus, the effect of CsrA on exopolysaccharide production depends on growth phase, and likely other factors. The timing of expression of different genes involved in exopolysaccharide production could help explain the temporal regulation of rugose switching by CsrA. It is not known if temporal regulation of exopolysaccharide production through CsrA is a general feature of V. cholerae, or whether it is unique to QS-deficient strains.
CsrA positively regulates motility in V. cholerae, as has been described in E. coli. A large proportion of the flagellar assembly genes were downregulated in the csrA mutant, and we demonstrated that CsrA coimmunoprecipitated with the transcripts for two regulators of motility genes: (i) flrC transcript, encoding the class III transcriptional regulator of motility genes (27), and (ii) rpoN mRNA, encoding a sigma factor involved in activating both class II and class III flagellar genes (60). This suggests that CsrA regulates motility at multiple points, and that regulation could be occurring directly, on flrC, and both directly and indirectly through rpoN. This is of particular interest because it has been demonstrated in V. cholerae that these regulators are also required for proper expression of virulence genes (27). Further, it has been demonstrated that DflaA or DmotAB mutants, which are nonmotile, do not colonize the infant mouse small intestine, and either fail to express (DflaA), or misregulate (DmotAB), the ctxA gene in vitro and in vivo (61). Thus, motility and virulence are linked in V. cholerae, and CsrA could affect pathogenesis by regulating motility.
One of the central transcriptional regulators coordinating quorum sensing and other environmental stimuli, such as oxygen levels and pH, with virulence gene expression in V. cholerae is AphA (26). When activated, AphA induces tcpPH transcription, and TcpP, together with ToxR, induces toxT expression. ToxT, in turn, activates expression of the toxin coregulated pilus, TCP, and cholera toxin, CTX (reviewed in reference 62). The aphA transcript was a target of CsrA binding in the CsrA-RNA co-IP, and we demonstrated that CsrA can bind directly to an RNA sequence representing the 59UTR and initial sequence of the aphA mRNA in vitro. Validating the relevance of the CsrA-aphA interaction, we showed that AphA protein levels were significantly decreased in the csrA mutant, and CsrA was required for optimal AphA translation, but not aphA transcription. The precise mechanism of enhanced translation by CsrA in V. cholerae is unknown and under investigation.
We previously demonstrated that CsrA is required for colonization of the infant mouse model, and that CsrA positively regulates ToxR protein production. Here, we demonstrate that CsrA positively regulates other processes that are also important for pathogenesis, such as motility and AphA protein production. Virulence gene regulation in V. cholerae comprises an exceedingly complex network of sensory two-component systems, regulatory RNAs and proteins, and small signaling molecules. From our RNAseq and CsrA-RNA co-IP data, we show that CsrA plays a central role in how signals are sensed and relayed, and in how responses are coordinated. It is likely the colonization defect of the csrA mutant is due to effects on several, not just one, pathway leading to virulence gene expression.

MATERIALS AND METHODS
Bacterial strains and growth conditions. All strains (Table S3) were maintained at 280°C in tryptic soy broth plus 20% (vol/vol) glycerol. Cultures were grown at 37°C with shaking in Luria-Bertani (LB) broth (1% [wt/vol] tryptone, 0.5% [wt/vol] yeast extract, and 1% [wt/vol] NaCl) or on LB agar plates. LB overnight cultures inoculated from single colonies were subcultured 1:100 into the minimal medium and grown at 37°C with shaking. The minimal medium, a modified T medium, was supplemented with a mixture of the amino acids asparagine, arginine, glutamate, and serine (NRES), each dissolved in water and added to a final concentration of 3.125 mM. The T medium was modified to contain 0.2% (wt/vol) sucrose, 20 mM FeSO 4 , and VA vitamin solution (https://www.genome.wisc.edu/resources/protocols/ ezmedium.htm). Unless stated, all experiments were performed with at least three biological replicates. For the RNA-sequencing experiment, LB cultures from single colonies were grown overnight, then diluted 1:100 in T medium with NRES and grown to early exponential phase (optical density at 650 nM [OD 650 ] ;0.1), mid-exponential phase (OD 650 ;0.5), and stationary phase (OD 650 ;2.0). For the CsrA-RNA coimmunoprecipitation experiment, LB cultures from single colonies were grown overnight, then diluted 1:100 in T medium with NRES and grown to mid-exponential phase (OD 650 ;0.5). The following antibiotics were used at the indicated concentrations: for E. coli, 50mg/ml kanamycin, 50 mg/ml ampicillin, 30 mg/ml chloramphenicol, and 250 mg/ml carbenicillin; for V. cholerae, 50mg/ml kanamycin, 25 mg/ ml ampicillin, 6 mg/ml chloramphenicol, 125 mg/ml carbenicillin, and 20 mg/ml polymyxin B.
Isolation of V. cholerae RNA. Approximately 10 9 bacterial cells, grown in minimal medium supplemented with NRES, were pelleted by centrifugation and resuspended in 100 ml of the growth medium. One ml of RNA-BEE (Tel-Test, Inc., Friendswood, TX), a solution containing phenol and quinidine thiocyanate, was added to the cells. Two hundred microliters of RNase-free chloroform was added, and the mixture was centrifuged at 4°C to separate the organic and aquatic phases. The aquatic phase was removed by pipetting and added to an equal volume of RNase-free isopropanol, and placed at 280°C overnight. The precipitated nucleic acids were pelleted by centrifuging the samples at 21,100 Â g for 10 min at 4°C. The nucleic acids were washed with 500 ml of 75% RNase-free ethanol. The dried RNA/DNA pellet was resuspended in 20 to 50 ml of RNase-free H 2 O. The nucleic acids were treated with DNase I (New England BioLabs [NEB], Ipswich, MA) as per the manufacturer's directions and subsequently stored at 280°C overnight. The RNA was washed with 500 ml of 75% RNase-free ethanol and the dried RNA pellet was resuspended in 20 to 50 ml of RNase-free H 2 O. The isolated RNA was either used immediately or placed at 280°C until further use.
Whole-genome RNA-sequencing. The purified RNA was sent to the Genomic Sequencing and Analysis Facility at The University of Texas at Austin, where the RNA samples were assembled into a single-ended RNA-seq library and sequenced by Illumina-based next-generation sequencing. The RNA-seq samples were sequenced on the HiSeq 4000, resulting in an average of 26.9 million reads per sample. The Illumina reads were then imported into the CLC Genomics Workbench 11.0.1 (Qiagen, Valencia, CA) and mapped to the N16961 reference genome (NC_002505 and NC_002506), with a slight modification.
The experimentally determined 59 UTRs were added into the gene track to permit the RNA reads to properly align to the 59 UTR regions within the genome (40). The standard CLC Genomic Workbench parameters were then used to map each sample's reads individually to the genome. The expression value for each gene was reported as a transcripts per million (TPM). The TPM expression value was selected because it normalizes the number of reads for a gene to the length of that gene. The samples were normalized by quantile, grouped, and comparisons between the csrA mutant and the wild type were made. The Baggerley's test was used to determine statistical significance (63). From this analysis, a gene was considered to be significantly differentially expressed between N16961 and NcsrA.R6H if that gene had a false discovery rate (FDR) P value of ,0.05.
GO ontology analysis. Genes with a statistically significant (FDR P value of ,0.05) and a weighted proportions fold change of .2-fold were selected for analysis via PANTHER Gene Ontology (GO). The statistical overrepresentation test with the GO biological process complete annotation data set was used. To determine processes that were overrepresented, Fisher's exact test was applied, and categories with an FDR P value of ,0.05 were considered significant.
CsrA-RNA coimmunoprecipitation. A csrA allele encoding a CsrA protein tagged with a single V5 tag sequence on the C terminus, inserted into the inducible vector pBAD18-cm, was generously provided by Bryan Davies (The University of Texas at Austin). The NcsrA.R6H strain harboring an arabinoseinducible V5 epitope-tagged csrA allele was grown to mid-exponential phase and csrA-V5 expression was induced. RNA was harvested from 5% of the lysate and the remainder was incubated with magnetic beads coated with anti-V5 antibody. The beads were collected and washed, and then CsrA-bound RNA was isolated from the beads. CsrA-V5 expression was induced in mid-exponential phase culture by adding 0.1% arabinose to the medium for 15 min. The cells were centrifuged and resuspended in 2.5 ml of binding/wash buffer (100 mM MOPS [pH 7.0] [KOH], 10 mM MgCl 2 , 100 mM KCl). The cells were lysed in a French pressure cell with a final volume of ;2 ml. One hundred microliters of the cell lysate was added directly to RNA-Bee, and the RNA was further purified. The remaining lysate was incubated with protein G Dynabeads (Thermo Fisher Scientific, Waltham, MA) coated with anti-V5 antibodies (Sigma-Aldrich, St. Louis, MO; number V8012) for an hour. The beads were washed 4 times with binding/wash buffer, resuspended in 100 ml of buffer, and added to RNA-Bee for further RNA purification.
Generating cDNA and quantitative PCR. SuperScript III (Thermo Fisher Scientific) was used to reverse transcribe 2 mg of RNA into cDNA using the random primers supplied with the SuperScript III kit. Fresh cDNA was diluted 1:10 with ddH 2 O and was used immediately. Primers were designed by using Primer3 (http://bioinfo.ut.ee/primer3-0.4.0/) and were selected to be located within the first half of the target gene, and to produce a product between 50 and 150 nucleotides in length. Real-time qPCR was performed with Power SYBR Green (Thermo Fisher Scientific). The Applied Biosystems ViiA 7 instrument was used with the following parameters for qPCR: (i) holding stage, 50°C for 2 min and 95°C for 10 min; (ii) PCR stage, 95°C for 15 sec and 60°C for 1 min, repeated 40 times with fluorescence recorded at 60°C; and (iii) melting curve stage, 90°C for 15 sec, 60°C for 1 min, then 95°C for 15 sec with the fluorescence recorded every 0.05 sec. Each reaction produced only one melting curve, indicating only one target was amplified during the qPCR. The threshold cycle (DDC T ) method was used to determine the relative amount of RNA present in the sample tested. The atpI gene was used as an internal reference to normalize the CsrA-RNA coimmunoprecipitation because we have previously shown that CsrA does not bind to the mRNA transcript of this gene (17).
Frequency of rugosity. To determine the percentage of rugose conversion, the strains were grown in alkaline peptone water (APW) number 3 medium, which has been previously shown to promote high rates of conversion (36). APW number 3 (3 ml) was inoculated with a single colony and placed at 37°C without shaking. At 24 or 48 h, the cultures were vigorously pipetted up and down through a P1000, then a P100 Gilson pipetman tip, to break up the cellular aggregates. The cells were serially diluted in saline, and the appropriate dilution to obtain 20 to 40 colonies was plated onto LB plates without antibiotics and incubated for 24 h at 37°C. After an additional 24 to 48 h of growth at room temperature, the number of smooth and rugose colonies was enumerated. To calculate the frequency of conversion, the number of rugose colonies was divided by the total number of colonies on the plate. At least 100 colonies were observed for each strain.
Motility. A 0.1 to 10-ml pipette tip was used to stab a single colony into the motility agar. The motility medium was T medium supplemented with NRES, and contained 0.3% agar. The inoculated plates were placed upside-down at 37°C for 24 h. Relative rates of motility were then determined by measuring the diameter of growth.
Construction of V. cholerae strains and plasmids. All strains and plasmids constructed during this study are listed in Table S3. Constructing the aphA-V5 chromosomal fusion proteins was performed by allelic exchange as previously described (17). The primers used to generate the suicide vector pSaphA-V5 are listed in Table S4. To create the aphA transcriptional fusion, an AT-nucleotide-rich (65% AT) region directly upstream of the aphA coding sequence was selected as the likely promoter. This region is 228 nucleotides in length (2821121 to 2821348) and goes from 2208 to 120 relative to the transcriptional start site determined by Papenfort et al. (40). The predicted promoter region was PCR amplified using primers listed in Table S4. This PCR fragment was digested with NcoI and BamHI and cloned into the NcoI and BamHI sites located upstream of a promoterless lacZ gene in the plasmid pQF50, generating the full-length reporter construct pQFaphA.TS. The translational reporter pQElacZ (17) was used to generate the aphA translational reporter. The aphA translational reporter construct begins at 1100 and goes to 1424 relative to the start of transcription (2821428 to 2821752). The primers used to generate this fragment are listed in Table S4. The promoterless aphA construct was cloned into the MfeI and NotI The V. cholerae CsrA Regulon ® sites downstream of the T5 promoter in pQELacZ to generate the translational reporter plasmid pQEaphA.TL.
SDS-PAGE and immunoblotting. Cultures were grown to mid-exponential phase (OD 650 ;0.5) and samples containing an equivalent number of cells were resuspended in Laemmli SDS sample buffer (64). Samples were resolved by electrophoresis through 10% SDS-polyacrylamide gels and the proteins were visualized by Coomassie brilliant blue staining or immunoblotting. The proteins for immunoblotting were transferred to a 0.45-mm-pore-size nitrocellulose membrane (GE Healthcare, Chicago, IL). The AphA-V5 fusion protein was immunodetected by mouse monoclonal anti-V5 antibodies (diluted 1:10,000) (Sigma-Aldrich, number V8012). To visualize the protein, horseradish peroxidase (HRP)-conjugated goat antimouse IgG (diluted 1:10,000) (Bio-Rad Laboratories, Hercules, CA) was used. To ensure equal loading, the relative densities of the immunoblotted samples were assessed by Coomassie staining of a duplicate gel.
Generating RNA probes. The MEGAshortscript kit (Thermo Fisher Scientific) was used to in vitro transcribe RNA. For this method, the PCR product of the aphA mRNA was generated using a forward primer containing the T7 promoter (Table S4). The region PCR amplified includes 204 nucleotides upstream from the ATG and 228 nucleotides into the coding region (2821329 to 2821760). The purified T7 PCR product was used as the template at a concentration of between 25 and 125 nM. After the in vitro transcription reaction, the RNA in the 20-ml reaction was precipitated by adding 5 ml RNase-free 3 M sodium acetate, followed by 125 ml of 100% RNase-free isopropanol. Once the precipitated RNA pelleted dried completely, the RNA was resuspended in RNase-free H 2 O. The Pierce RNA 39 end biotinylation kit (Thermo Fisher Scientific) was then used as per the manufacturer's directions to biotinylate the RNA. The biotinylated RNA was precipitated and washed, as described above, and resuspended in RNase-free H 2 O. The biotinylated RNA was either used immediately or placed at 280°C until further use.
RNA electrophoretic mobility shift assays. RNA oligonucleotides were made as described above. CsrA was purified as described in reference 17. The CsrA-RNA electrophoretic mobility shift assay (EMSA) samples were prepared according to the LightShift Chemiluminescent RNA EMSA kit (Thermo Fisher Scientific) directions, with the slight modification of a 10-ml final volume instead of a 25-ml final volume. CsrA-RNA complexes were resolved by electrophoresis through a 5% nondenaturing polyacrylamide gel, transferred to the BrightStar-Plus (Thermo Fisher Scientific) positively charged nylon membrane, and UV cross-linked with 150 mJ. The LightShift Chemiluminescent RNA EMSA kit was used to visualize the biotinylated RNA.
lacZ-based reporter assays. Strains harboring either a transcriptional or translational lacZ-based reporter were grown to mid-exponential phase in minimal medium with or without NRES, and the optical density at OD 650 was recorded. To ensure equivalent expression from the translational reporter, 100 mM IPTG (isopropyl-b-D-thiogalactopyranoside) was added to the growth medium. Cell suspension (100 ml) was placed into 900 ml of Z-Buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl, 1 mM MgSO 4 , and 2.5 mM b-ME) (65). The cells were permeabilized by adding 20 ml of chloroform, and then 10 ml of 0.1% SDS. After 15 min, 200 ml of freshly made ONPG solution (4 mg/ml ONPG dissolved in phosphate buffer [60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 ]) was added to the cell suspension in Z-Buffer solution. The duration of time between adding the ONPG and the solution turning light yellow was recorded, and once color change was observed, 500 ml of stop solution (1 M Na 2 CO 3 ) was added. The OD 420 and OD 550 were recorded to determine the b-galactosidase activity. Then, by using the following calculation, the Miller units were determined: