The Vibrio cholerae virulence regulatory cascade controls glucose uptake through activation of TarA, a small regulatory RNA

Summary

Vibrio cholerae causes the severe diarrheal disease cholera. A cascade of regulators controls expression of virulence determinants in V. cholerae at both transcriptional and post-transcriptional levels. ToxT is the direct transcription activator of the major virulence genes in V. cholerae. Here we describe TarA, a highly conserved, small regulatory RNA, whose transcription is activated by ToxT from toxboxes present upstream of the ToxT-activated gene tcpI. TarA regulates ptsG, encoding a major glucose transporter in V. cholerae. Cells overexpressing TarA exhibit decreased steady-state levels of ptsG mRNA and grow poorly in glucose-minimal media. A mutant lacking the ubiquitous regulatory protein Hfq expresses diminished TarA levels, indicating that TarA likely interacts with Hfq to regulate gene expression. RNAhybrid analysis of TarA and the putative ptsG mRNA leader suggests potential productive base-pairing between these two RNA molecules. A V. cholerae mutant lacking TarA is compromised for infant mouse colonization in competition with wild type, suggesting a role in the in vivo fitness of V. cholerae. Although somewhat functionally analogous to SgrS of E. coli, TarA does not encode a regulatory peptide, and its expression is activated by the virulence gene pathway in V. cholerae and not by glycolytic intermediates.

Introduction

Asiatic cholera, a severe diarrheal disease, remains a significant cause of morbidity and mortality in many regions of the globe (Sack et al., 2006). Cholera is caused by ingestion of the gram-negative, highly motile curved bacillus Vibrio cholerae in contaminated food or water. V. cholerae are present in many coastal areas worldwide; however, of the hundreds of known V cholerae serogroups, only the O1 and O139 serogroups can cause cholera (Sack et al., 2004, Reidl & Klose, 2002).

The two major V. cholerae virulence factors are the cholera toxin (CT) and the toxin-coregulated pilus (TCP). Secretion of CT into the intestinal lumen by V. cholerae directly causes the secretory diarrhea characteristic of cholera (Gill, 1976, Lonnroth & Holmgren, 1973). The genes encoding the two subunits of CT, ctxAB, are located within the genome of a filamentous bacteriophage, CTXΦ (Waldor & Mekalanos, 1996). The TCP, a type IV bundle-forming pilus, is required for intestinal colonization by V. cholerae and acts to initiate microcolony formation (Attridge et al., 1996, Herrington et al., 1988, Thelin & Taylor, 1996, Taylor et al., 1987). The genes encoding TCP are located within the Vibrio pathogenicity island (VPI), which may also be a mobile element but is not a bacteriophage (Manning, 1997, Karaolis et al., 1998, Rajanna et al., 2003, Faruque & Mekalanos, 2003). Other putative virulence genes located within the VPI are coordinately regulated with ctxAB and tcp, but their roles in pathogenesis are unclear.

Expression of genes encoding V. cholerae virulence factors is regulated by a cascade involving transcription activators, repressors and proteases (Matson et al., 2007); however, the direct transcription activator of virulence genes is ToxT (Matson et al., 2007). ToxT is a member of the AraC/XylS family of transcription factors (Higgins et al., 1992, Ogierman & Manning, 1992). ToxT binds to a 13 bp degenerate DNA sequence, the toxbox, to activate transcription (Withey & DiRita, 2006), using toxboxes configured as direct repeats, inverted repeats, and single toxboxes to activate transcription of different operons (Withey & DiRita, 2006, Withey & Dirita, 2005b).

Binding sites for ToxT have been characterized at most of the promoters it is known to activate (Withey & DiRita, 2006, Withey & DiRita, 2005a, Withey & Dirita, 2005b). An exception is the tcpI promoter, located on the VPI and encoding a putative methyl-accepting chemotaxis protein with an unknown role in pathogenesis (Harkey et al., 1994). Our sequence analysis identified two pairs of putative toxboxes upstream of tcpI; one pair, in an inverted configuration, is positioned proximal to the tcpI open reading frame (Withey & DiRita, 2006). The other pair, in a direct repeat configuration, is positioned at a site upstream of the proximal toxboxes (Figure 1a), and thus we refer to them as the distal toxboxes. Primer extension analysis performed previously by others identified two different start sites of tcpI transcription that would potentially correspond to transcription activated by ToxT from each set of toxboxes; though when those studies were carried out, the toxboxes had not yet been identified (Thomas et al., 1995, Murley et al., 2000). We performed experiments to characterize the role of ToxT in transcription from both of these putative sites and have found that there are indeed two separate ToxT-dependent transcripts produced upstream of tcpI. However, the tcpI-proximal promoter alone is responsible for transcription of tcpI itself (JHW and VJD, unpublished).

Figure 1. Schematic and diagram of tcpI promoters and the surrounding region.

A.) Schematic. Scale bar: 0.5 centimeters represents 200 bp of DNA. Bent arrows indicate putative transcriptional start sites, and long black arrows represent coding sequences of the known genes tcpP and tcpI. tcpI has been shortened from 1.6 kb for clarity. Toxboxes, or ToxT-recognition sequences, are indicated by paired boxes. B.) Diagram of tcpI distal promoter, tarA, and surrounding sequences. Bent arrows indicate transcriptional start sites for tcpP and tarA. Numbers indicate position relative to tarA transcriptional start site. ToxT binding sites (toxboxes) for the tcpI distal promoter are underlined with horizontal arrows. The −35 and −10 elements for tarA are indicated by thick boxes, and thin boxes represent potential base-pairing regions in putative hairpin structure.

Here we describe a small RNA (~91 nt), which we named TarA for ToxT-activated RNA, whose transcription from a gene located between tcpI and tcpP is activated by ToxT at the promoter distal to tcpI. The TarA nucleotide sequence is highly conserved among both epidemic and environmental V. cholerae strains that carry the VPI. We demonstrate that TarA influences glucose uptake through its effect on the transcript encoding the glucose transporter PtsG.

Results

ToxT activates transcription of a small non-coding RNA

To characterize ToxT-dependent activation of tcpI, we constructed lacZ fusions and assayed β–galactosidase activity in V. cholerae having either wild type toxT or a toxT deletion. Our results indicated that the minimal ToxT-responsive DNA sequence for activating a tcpI-lacZ fusion at this tcpI-proximal promoter extended only to −94 relative to the proximal tcpI promoter. Because a second separate start site for tcpI transcription had been previously described at −243 relative to the AUG (Murley et al., 2000, Harkey et al., 1994), we also constructed tcpI::lacZ fusions extending to −431 relative to the start site of transcription from the tcpI-proximal promoter to include both putative promoters (Figure 1A). However, additional upstream DNA, even if it contained the putative tcpI-distal promoter, did not affect the level of β–galactosidase produced. This result suggested that the distal promoter, if functional, does not read into tcpI under the conditions tested.

To determine whether the tcpI-distal promoter is functional, we created a set of lacZ fusions whose 3′ endpoints were located upstream of the tcpI-proximal promoter (at −173 relative to its start site and +36 relative to the distal promoter start site) and with 5′ endpoints that extended to −59, −81, or −223 relative to the start of transcription from the distal promoter (Figure 1B). Results of β–galactosidase assays using these constructs indicated that the tcpI-distal promoter was indeed functional and highly activated by ToxT (Table 1). Sequence analysis identified two consensus toxboxes configured as a direct repeat upstream of the tcpI-distal promoter −35 box that had the same spacing as the toxboxes we observed at the tcpA promoter (Withey & DiRita, 2006) (Figure 1B). Deletion of one of the two toxboxes resulted in a complete loss of ToxT-directed transcription from the tcpI-distal promoter (Table 1, Figure 1B (−59 construct)).

Table 1. Nested tarA::lacZ fusion analysis.

5′ and 3′ endpoints indicate region of V. cholerae chromosome ligated into pTL61T plasmid. Fusion plasmids were electroporated into WT O395 and an isogenic ΔtoxT strain. Cultures were grown under inducing conditions for 3 hours and then tested for β-galactosidase activity, with data reported in Miller units. Fold difference was determined by taking the ratio of wild type activity to ΔtoxT activity.

5′ Endpoint	3′ Endpoint	Miller Units (WT toxT)	Miller Units (ΔtoxT)	Fold Difference (toxT/ΔtoxT)
−223	+36	11,800 ± 500	804 ± 137	15x
−81	+36	18,000 ± 551	436 ± 44.8	41x
−59	+36	94.6 ± 9.35	93.1 ± 16.5	1x

Downstream of the start site of transcription is a G/C rich region that could potentially form a stem-loop structure (boxes in Figure 1B), followed by a tract of T nucleotides on the non-template strand, suggesting the presence of a transcription terminator. To determine if a small RNA product initiating at the tcpI-distal promoter and ending at the putative terminator is produced, we performed Northern blots on whole cell RNA purified from V. cholerae having a wild type or a deleted toxT gene, probing with an oligonucleotide complementary to the putative sRNA sequence. A single RNA product of the expected size, <100 nt, was detected in the Northern blotting experiments (Figure 2). Production of the sRNA required functional ToxT, was detectable beginning at three hours of growth (data not shown), continued through logarithmic phase, and peaked at seven hours of growth under ToxT-inducing conditions; no TarA was detected in non-inducing conditions (Figure 2). These findings are consistent with previous observations of ToxT activity (Yu & DiRita, 1999). Based on this evidence, we named this previously unknown sRNA “TarA” for ToxT-Activated RNA. TarA contains a putative binding site for the ubiquitous regulatory protein Hfq and lacks an open reading frame; thus, its function is likely to be regulatory.

Figure 2. TarA is produced under ToxT-inducing conditions.

Wild type and ΔtoxT strains were grown under inducing and non-inducing conditions (LB pH 6.5, 30°C and LB pH 8.5, 37°C, respectively) and total RNA was collected at indicated timepoints. Equal amounts of RNA from each sample (as determined by OD₂₆₀) were subjected to Northern blot analysis, using an oligonucleotide complementary to the putative sRNA sequence as a probe. Methylene blue staining was used to confirm load amounts (data not shown).

TarA is highly conserved among V. cholerae strains

To determine whether TarA has homology to any other sRNA species, we performed searches with its sequence from the classical V. cholerae strain O395 using both the RNAfam database and BLAST (Touzet & Perriquet, 2004). The former search indicated that TarA has no homology to any known sRNA family. However, the latter search indicated that TarA has a very high degree of sequence conservation among V. cholerae (Table 2, Figure 3). TarA sequences from eight different V. cholerae strains were obtained from the BLAST search; two of these are classical biotype V. cholerae, two are El Tor biotype V. cholerae, and four are non-O1 V. cholerae. The two classical strains have identical TarA sequences and the sequences of the two El Tor strains are also identical (Table 2 and Figure 3). However, the El Tor sequences have a single nucleotide insertion and two substitutions that differ from the classical sequences. The non-O1 V. cholerae strains, commonly known as environmental V. cholerae to distinguish them from the O1 epidemic strains, carry the VPI but are not capable of causing cholera. These environmental V. cholerae strains have TarA sequences apparently derived from the classical biotype or vice versa. Three of these four non-O1 strains have TarA sequences identical to the classical TarA sequences; the fourth non-O1 strain has a single nucleotide change from the classical TarA sequences at a position that differs from the changes observed between classical and the El Tor biotype strains.

Table 2. Conservation of TarA and upstream sequences.

Asterisks indicate toxigenic strains.

V. cholerae strain	Biotype	O Antigen	TarA Conservation	Upstream DNA Conservation
O395*	Classical	O1	99/99 nt	99/99 bp
Z17561*	Classical	O1	99/99 nt	99/99 bp
N16961*	El Tor	O1	96/99 nt	62/99 bp
H1*	El Tor	O1	96/99 nt	75/99 bp
SCE4	Environmental	O8	99/99 nt	75/99 bp
SCE200*	Environmental	O44	99/99 nt	91/99 bp
SCE256	Environmental	O42	99/99 nt	68/99 bp
SCE226	Environmental	O35	98/99 nt	85/99 bp

^*Toxigenic strain

Figure 3. Comparison of tarA sequences among V. cholerae strains.

Sequences from strains indicated in Table 2 are aligned, with bold letters indicating mutations. The box indicates a conserved potential Hfq binding site.

To determine whether the degree of TarA sequence conservation among different V. cholerae strains is a general feature of the VPI, we compared the DNA sequences 99 bp directly upstream of the tarA start site in the above strains to the sequence in O395. This analysis indicated that the tarA sequence was preferentially conserved over the surrounding intergenic sequence (Table 2). The degree of conservation of the 99 bp directly upstream of tarA varies to a wide degree, even between strains that have identical tarA sequences. For example, classical O395 and environmental SCE 256 have 99/99 identical base pairs within the tarA gene but only 68/99 identical base-pairs within the upstream sequence. These upstream 99 bp include both the promoter and the two toxboxes that are required for expression of tarA; notably, the toxbox sequences are 100% conserved in every strain. We have also observed TarA expression in the El Tor strain N16961 under ToxT-inducing conditions (data not shown). These results strongly suggest that TarA conveys an evolutionary advantage to both environmental and epidemic V. cholerae and that its regulation by ToxT is highly conserved.

TarA regulates ptsG mRNA levels in V. cholerae

To identify potential regulatory targets of TarA, we probed microarrays of the V. cholerae genome with RNA isolated from classical strain O395 and its ΔtarA derivative. The ΔtarA strain features a markerless deletion of the 91 nucleotide RNA sequence with the promoters left intact. The whole-genome expression data were analyzed using the Significance Analysis of Microarrays (SAM) program (Tusher et al., 2001) using ≥2.0 fold differences in gene expression and a ≤1% false discovery rate (FDR) as cutoff values. A gene annotated as ptsG (VC2013 in the N16961 genome sequence) was upregulated by greater than two-fold in the ΔtarA strain compared with wild type. This is consistent with a previous microarray study (Bina et al., 2003), in which VC2013 RNA was shown to be upregulated in ΔtoxT V. cholerae relative to wild type. This pattern of regulation – with VC2013 expression being negatively controlled by ToxT – is in contrast to that of nearly all other known ToxT-regulated genes, such as the toxin co-regulated pilus (tcp) and the cholera toxin (ctxAB), which are positively controlled by ToxT. That ptsG may be the target of sRNA regulation in V. cholerae is consistent with the fact that ptsG in E. coli is also regulated by a small RNA, SgrS (Vanderpool & Gottesman, 2004).

To confirm the microarray result, we used quantitative real time PCR (qRT-PCR) comparing ptsG mRNA levels in wild type and tarA mutant V. cholerae strain O395. Consistent with the microarray experiments, we observed an approximately three-fold increase in ptsG mRNA in mutant cells compared with wild type (Figure 4), supporting the hypothesis that TarA acts to decrease the levels of ptsG mRNA. To be sure that regulation was due to the specific absence of tarA and not due to a polar effect on tcpI, a ΔtcpI strain was also tested. ptsG transcript levels in the ΔtcpI strain closely resemble those seen in wild type O395, ruling out this possibility.

Figure 4. Confirmation of ptsG regulation by qRT-PCR analysis.

Cultures of the listed strains were grown under ToxT-inducing conditions for 7 hours and total RNA was isolated using Trizol. cDNA was prepared from equivalent amounts of RNA for each sample, and cDNA was used for SYBR Green qRT-PCR. Numbers indicate fold change in transcript level between strains, calculated using the ΔΔCT method and recA transcript levels as an internal control.

Growth of V. cholerae in minimal glucose medium is limited in cells overexpressing TarA

In E. coli, SgrS acts to reduce ptsG mRNA levels and to limit glucose uptake under conditions where flux through the glycolytic pathway may be disrupted, thereby allowing the accumulation of potentially lethal phospho-glucoside compounds (Vanderpool & Gottesman, 2004). When E. coli is grown with the glucose analogue alpha-methyl-glucoside (αMG), the compound is phosphorylated and taken up by PtsG, but phospho-αMG cannot be catabolized by the normal glucose metabolic pathway and therefore accumulates. Under such conditions, sgrS expression is induced and the resulting degradation of ptsG mRNA reduces further uptake of αMG. Mutants lacking sgrS are highly sensitive to αMG (Vanderpool & Gottesman, 2004). We tested whether ΔtarA V. cholerae exhibited enhanced sensitivity to αMG relative to wild type, which would be a physiological consequence of the fact that TarA regulates ptsG mRNA in V. cholerae. The two strains were grown in LB supplemented with αMG in concentrations ranging from 1-10% weight by volume; we did not see a consistent effect of this glucose analogue on ΔtarA V. cholerae at any of the concentrations tested.

Failing to identify an αMG effect, we tested another ptsG-related hypothesis. We reasoned that constitutive overexpression of TarA, which would remove the RNA from its normal regulation by ToxT, would affect the growth physiology of V. cholerae, leading to diminished growth in glucose-containing minimal media if indeed ptsG mRNA levels are regulated by TarA. Overexpression of TarA in LB-grown V. cholerae had a slight effect on culture growth rate, although the final optical density achieved by this culture was close to wild type. Growth of ΔtarA and wild type V. cholerae carrying the vector alone was unaffected compared to wild type (Figure 5A). However, after dilution from LB into minimal MOPS/0.5% glucose, the cells overexpressing TarA were severely deficient for continued growth and resembled the ΔptsG strains, while the other strains grew similarly after a lag phase of approximately three hours (Figure 5B). The ΔptsG strain complemented with a constitutively-expressed copy of ptsG exhibited partially restored growth in this medium. Analysis of glucose remaining in the supernatants of these cultures revealed similar levels of glucose for approximately four hours, after which the culture overexpressing TarA and that of the ΔptsG mutants nearly stopped removing glucose from the medium while the others continued to do so (Figure 5C). Thus, the poor growth of wild type cells overexpressing TarA correlated with decreased levels of glucose uptake, similar to cells lacking ptsG.

Figure 5. Overexpression of TarA results in severely reduced glucose uptake.

(A.) LB growth. Indicated strains were grown overnight in LB media, subcultured into fresh LB and grown to mid-exponential phase, then washed with PBS and diluted into fresh LB to OD = 0.04. Cultures were grown in a 96-well plate at 30°C with constant aeration, and the OD₆₀₀ was measured every 20 minutes. Growth assays were performed at least three times, and a representative growth curve is shown. (B.) MOPS/glucose growth. Strains were prepared as in (A), but diluted into MOPS/0.5% glucose instead of LB and then grown as in (A). Growth assays were performed at least three times, and a representative growth curve is shown. (C.) Strains were grown overnight in LB, subcultured into LB and grown to mid-exponential phase, then washed and diluted into MOPS/0.5% glucose, and grown in flasks. At specified timepoints, supernatants were collected and the glucose concentration was tested using a tetrazolium blue reducing sugar assay. Calculated glucose concentrations were divided by the starting concentration to determine the amount of glucose remaining in the media. This experiment was performed three times, and a representative assay is shown. Error bars indicate the standard deviation among triplicate samples in the representative experiment.

To more carefully examine association between TarA and ptsG mRNA levels, we placed tarA expression under control of an IPTG-inducible promoter in pHDB3 and followed steady state levels of ptsG mRNA message upon TarA induction. ΔtarA V. cholerae carrying this plasmid was grown in LB to mid-logarithmic levels before addition of IPTG to induce tarA expression. Thirty minutes after IPTG induction, cells were collected, washed, resuspended in fresh media and allowed to continue growing for another 30 minutes. RNA was harvested from cells at 0, 2, 10 and 30 minutes after addition of IPTG, and then at 10 and 30 minutes after the IPTG was washed out. RNA was analyzed by Northern blotting using a probe specific for TarA or ptsG mRNA.

Prior to induction (0 minutes) TarA was undetectable while ptsG mRNA was abundant. Within two minutes after induction by IPTG, TarA became detectable and the amount of ptsG mRNA diminished. This continued throughout the course of the experiment, with TarA expression remaining at high levels and ptsG mRNA detectable at levels considerably lower than those seen prior to induction (Figure 6). Cells carrying the cloning vector alone expressed no detectable TarA and very high levels of ptsG mRNA. The levels of ptsG mRNA were higher in the strain carrying only the cloning vector than in the cells carrying the tarA clone without induction, which could be due to undetectable yet functional levels of TarA being expressed even prior to addition of IPTG. There is a slight decrease in ptsG mRNA levels in the vector control after thirty minutes of induction and ten minutes of washout (t=40 timepoint), which we attribute to transcript turnover as the cells near stationary phase.

Figure 6. Induction of TarA reduces ptsG transcript levels.

Strains AR059 and AR060, containing an IPTG-inducible copy of TarA and an empty vector control, respectively, were grown overnight in LB. Cultures were diluted 1:200 into fresh LB, grown until OD ~ 0.4, and stimulated with IPTG. Thirty minutes after induction, cultures were washed and resuspended in fresh media lacking IPTG. RNA was collected at indicated timepoints (from the start of induction) using Trizol, and equal amounts of RNA for each sample were subjected to Northern blot analysis. Methylene blue staining was used to confirm load amounts (not shown). Blots were probed with oligonucleotides complementary to TarA (A) and to the ptsG transcript (B).

Base complementarity between TarA and the 5′ region of the ptsG mRNA

In E. coli, SgrS controls ptsG transcript levels in part by associating with the 5′ end of the ptsG mRNA and causing its degradation in a process that depends on Hfq and RNAse E. (Maki et al., 2008, Wadler & Vanderpool, 2007). We used the alignment prediction program RNAhybrid (http://bibiserv.techfak.uni-bielefeld.de/rnahybrid/submission.html) to identify potential areas of complementarity between TarA and the leader sequence of the ptsG transcript that would be expected for a similar mechanism to be operating in V. cholerae. This program produced a structure with a favorable free energy (−54.4 kcal/mol) and a predicted unpaired region, which includes the putative Hfq binding site (Figure 7). For purposes of comparison, we tested the potential complementarity of SgrS and the 100 bp directly upstream of ptsG in E. coli with the same program, which yielded a similarly-composed structure with a minimal free energy of −49.2 kcal/mol (data not shown).

Figure 7. Alignment of TarA and ptsG upstream sequences.

The program RNAhybrid (http://bibiserv.techfak.uni-bielefeld.de/rnahybrid/welcome.html) was used to predict potential base-pairing between TarA and the ptsG leader sequence (defined here as 100 bp upstream of the AUG, including the +1 A). The TarA sequence is shown in green and the ptsG leader sequence in red. The putative Hfq binding site is indicated. The minimum predicted free energy for this alignment is −54.4 kcal/mol.

TarA is unstable in the absence of Hfq

Given the putative Hfq binding site in the TarA sequence, we hypothesized that TarA interacts with Hfq. It is well-established that Hfq-interacting small RNAs are markedly less stable in the absence of Hfq (Morita et al., 2004, Kawamoto et al., 2005, Sledjeski et al., 2001, Moller et al., 2002, Masse et al., 2003, Moll et al., 2003). To explore a potential TarA/Hfq interaction, we assessed TarA levels by Northern blot analysis of RNA from wild type and hfq mutant V. cholerae. Cells were grown for seven hours in ToxT-inducing conditions prior to RNA isolation and blotting. As predicted, the hfq mutant had significantly reduced TarA RNA in comparison to the wild type strain O395 (Figure 8).

Figure 8. TarA is unstable in the absence of Hfq.

Wild type, ΔtarA, and Δhfq strains were grown under inducing conditions (LB pH 6.5, 30°C) and total RNA was collected after seven hours of growth. Equal amounts of RNA from each sample (as determined by OD₂₆₀) were subjected to Northern blot analysis, using an oligonucleotide complementary to TarA as a probe. Methylene blue staining was used to confirm load amounts (data not shown).

Discussion

Here we describe a small, non-coding regulatory RNA, TarA, whose transcription is directly activated by the major V. cholerae virulence activator protein ToxT. Activation of a regulatory RNA by ToxT adds another level to the complex V. cholerae virulence regulatory cascade and contributes to our understanding of ToxT-dependent gene activation. Based on our previous analyses of other ToxT-activated genes, the promoter architecture of tarA most closely resembles that of tcpA (Withey & DiRita, 2006). Furthermore, the degree of activation conferred by ToxT to the two genes is very similar – 41-fold for tarA and 44-fold for tcpA. The two tarA toxboxes have 7/7 and 6/7 of the toxbox core determinants conserved (Matson et al., 2007), the highest match to consensus we have observed for any operon having two toxboxes. However, while the degree of ToxT activation of tarA is very high, it is not higher than the ToxT activation of tcpA, which has a poorer match to consensus at one toxbox. This suggests that the toxbox sequences may have evolved to be maximally efficient at any given locus, such that surrounding sequence that is not conserved in the consensus alignment plays a more important role than previously thought.

TarA joins a growing list of regulatory sRNAs in Vibrio cholerae. Four sRNAs, Qrr1,2,3, and 4, are involved in regulating the quorum-sensing response, together with Hfq protein, by negatively affecting the stability of hapR mRNA (Lenz et al., 2004). Another three sRNAs, CsrB,C,D, are also involved in quorum sensing; these act to inhibit activity of CsrA protein, which subsequently leads to expression of the Qrr sRNAs (Lenz et al., 2005). Another sRNA is RyhB, which is involved in regulating iron-containing proteins (Davis et al., 2005). RyhB acts with Hfq to destabilize the mRNAs encoding these proteins. While many other sRNAs have been described in E. coli, very few of these are conserved in V. cholerae. The exceptions are SsrA, also known as tmRNA, RnpB, Spf, and Ffs (Livny & Waldor, 2007). Recently, large numbers of sRNAs have been identified in V. cholerae using bioinformatics and parallel sequencing methods (Livny et al., 2008, Vogel et al., 2003). One of these, IRG7, also contributes to regulation of carbon utilization by negatively regulating a mannitol transport gene, mtlA (Liu et al., 2009). TarA was not identified in either of these studies.

Our results suggest that a target of negative regulation by TarA in V. cholerae is ptsG, encoding a glucose transporter in the phosphotransferase system. In E. coli, ptsG is negatively regulated by SgrS, an sRNA whose mechanisms of action have been the subject of a significant amount of research. Expression of sgrS is activated by a conserved regulatory protein called SgrR, in response to glucose-phosphate stress (Vanderpool & Gottesman, 2004). SgrS is a bifunctional molecule, postulated to inhibit PtsG through two distinct mechanisms. One is by acting as a regulatory RNA directly by participating in a ternary complex with Hfq and the ptsG mRNA through specific base-pairing between the two RNA molecules. The site on ptsG mRNA that is base-paired to SgrS is in the 5′ end of the transcript at the Shine-Dalgarno sequence (Vanderpool & Gottesman, 2004), blocking translation and eventually leading to degradation of ptsG mRNA through a mechanism that requires RNAse E (Maki et al., 2008). We think it likely that TarA also acts in a similar fashion, given the base complementarity between TarA and ptsG RNA, overlapping the latter’s Shine-Dalgarno site. AU-rich sequences within a predicted unpaired region of TarA may also correspond to an Hfq binding domain (Figure 7) (Link et al., 2009). Our observation that TarA is less stable in an hfq mutant lends support to our hypothesis that Hfq and TarA interact. The second mechanism by which SgrS regulates PtsG is through a small protein, SgrT, encoded at the 5′ end of the same RNA that includes SgrS in E. coli and in a variety of other species (Wadler & Vanderpool, 2007), (Horler & Vanderpool, 2009). SgrT is hypothesized to act post-translationally on PtsG function (Wadler & Vanderpool, 2007). TarA does not have an open reading frame and is consequently much shorter than SgrS. Another distinction between SgrS and TarA is that, unlike functional SgrS homologues in many other Gram negative bacteria (Horler & Vanderpool, 2009), a homologue of the SgrR regulatory protein is not found encoded next to tarA in the VPI. As sequence homology to sgrR was used to locate SgrS functional homologues in other enteric bacteria, the fact that an SgrS homologue was not predicted in Vibrio cholerae is not surprising.

Unlike sgrS in E. coli, tarA expression is not controlled by the glucose-phosphate response regulator SgrR, but rather by ToxT, the activator responsible for expression of the major virulence factors of V. cholerae. The high degree of similarity between the tarA and tcpA toxboxes suggests that the two are expressed synchronously during infection. tarA makes a slight contribution to the fitness of V. cholerae during infection, as competition experiments between wild type and tarA mutant V. cholerae in the infant mouse model resulted in a small but consistent colonization defect of the tarA mutant (avg. competitive index = 0.45). Cells expressing elevated PtsG during infection could be at a competitive disadvantage if glucose is not the preferred in vivo carbon source. This hypothesis is under investigation. An alternative explanation for how the link between tarA expression and virulence gene activation may have evolved comes from work demonstrating that V. cholerae in stool samples of cholera patients store glucose in the form of glycogen granules, and that glycogen biosynthesis genes are expressed late during infection (Schild et al., 2007). Perhaps tarA expression by ToxT during infection, with a subsequent reduction in ptsG mRNA, prepares the cells for the apparent metabolic conversion to glycogen production by reducing the amount of glucose transported into the glycolytic pathway.

Experimental procedures

V. cholerae strains and plasmids

Strains used in this work were classical V. cholerae O395, El Tor strain N16961, and derivatives (listed in SI Table 1). The ΔtarA, ΔptsG, and Δhfq strains were constructed from O395 and N16961 by mating with E. coli SM10λpir using the pKAS32 suicide vector as previously described (Skorupski & Taylor, 1996). The deletion was confirmed by PCR using primers flanking the gene. Strains were grown in LB medium at 37° C (overnight cultures), in LB medium with a starting pH of 6.5 at 30° C (inducing conditions for classical strains), or under AKI conditions (inducing conditions for El Tor strains). Strains were maintained in LB/20% glycerol at −70°C. Antibiotics were used in the following concentrations: ampicillin: 100 μg/ml, streptomycin: 100 μg/ml and chloramphenicol: 30 μg/ml.

Strains used for β-galactosidase assays were either O395 or an O395 ΔtoxT derivative (VJ740,(Champion et al., 1997)) carrying the respective tarA::lacZ fusion in plasmid pTL61T (Linn & St Pierre, 1990). All constructs were cloned between the HindIII and XbaI sites of pTL61T. For constitutive and inducible expression of TarA, plasmids pKK177-3RI (Opdyke et al., 2004) and pHDB3 (Ulbrandt et al., 1997) were used, respectively. In pKK177-3RI, the tarA sequence alone was cloned between the EcoRI and HindIII sites, creating plasmid pJM292. The tarA sequence preceded by the P_LlacO operator sequence was cloned between the BamHI and HindIII sites of pHDB3 to generate pAR6. These plasmids, and the empty vector controls, were electroporated into wild type O395 and N16961 (pKK177-3RI and pJM292) and the corresponding ΔtarA strains (pHDB3 and pAR6). For ptsG complementation, the ptsG sequence preceded by a ribosome binding site was cloned between the EcoRI and XbaI sites of pUC19, generating plasmid pAR7. This plasmid and the empty vector control were then electroporated into N16961 ΔptsG. All plasmid sequences were confirmed by the University of Michigan DNA sequencing core.

β-galactosidase assays

Strains were cultured overnight in LB medium at 37°C, then sub-cultured at a 1/50 dilution into inducing medium and grown for 3 hours at 30°C with vigorous aeration. Bacteria were then placed on ice and chloramphenicol was added to 0.5 mg ml⁻¹. β-galactosidase assays were performed using the basic procedure of Miller (Miller, 1972) as previously described (Withey & Dirita, 2005b).

DNA manipulations

Plasmids were purified using the Promega Wizard Plus Miniprep kit. Polymerase chain reaction (PCR) was performed using Taq DNA polymerase (Roche) as specified by the manufacturer. Restriction enzymes were purchased from New England Biolabs and used as specified by the manufacturer.

Growth analysis and glucose quantification

Strains N16961, AR43, AR38, AR39, AR74, AR80, and AR81 were grown overnight in LB, then subcultured 1:100 into fresh LB and grown to mid-log phase (six hours). OD₆₀₀ was measured for each culture, and 1×10⁹ bacteria of each strain were pelleted, washed in PBS, and resuspended to a final OD₆₀₀ of 2 in PBS. These suspensions were then diluted 1:50 (final OD₆₀₀ = 0.04) into LB and MOPS minimal medium supplemented with 0.4% glucose (Neidhardt et al., 1974), and dispensed into a 96-well plate. The plate was incubated at 30°C with constant shaking for 8 hours in an Omega Polarstar plate reader, with OD₆₀₀ measured every 20 minutes. For glucose quantification, cultures were prepared as described and grown in 125 mL flasks on a ThermoFisher orbital shaker at 37°C. One milliliter samples were taken at 0, 4 and 8 hours of growth. Bacteria were pelleted, and 10-fold diluted supernatants were used in a tetrazolium blue reducing sugar assay as described (Jue & Lipke, 1985). Glucose concentration was determined for each sample and divided by the starting concentration to determine the percent glucose remaining. Assays were run in triplicate three times.

Northern blots

Whole cell RNA was purified from O395 and VJ740 (ΔtoxT) V. cholerae grown in parallel under inducing and non-inducing conditions (Figure 2), from strains AR59 and AR60 after addition of IPTG (Figure 6), and from strains O395, AR15, and AR24 grown under inducing conditions (Figure 8). At specific points (as detailed in figure legends for each experiment), the OD₆₀₀ of each culture was read and aliquots of each culture equal to 1 ml of a 0.7 OD₆₀₀ culture were removed. RNA purification was performed using the TRIzol reagent as specified by the manufacturer (Invitrogen). For TarA visualization, 7 Δg total RNA from each sample were separated by 8% polyacrylamide gel electrophoresis and transferred to an Amersham Hybond nylon membrane (GE Healthcare) (0.45μ) membrane with a Semi-Phor apparatus (Hoefer). An oligonucleotide probe complementary to nt 12-36 of TarA was radiolabeled with γ–³²P-ATP with T4 polynucleotide kinase (New England Biolabs). Probing conditions were as previously described (Sambrook, 1989). To visualize ptsG transcript, 12.5 μg of each sample were run on a 1.2% agarose/formaldehyde gel and transferred to an Amersham Hybond nylon membrane (GE Healthcare). Membranes were probed with an oligonucleotide complementary to nt 176-208 of the ptsG transcript, labeled as above.

Microarray and qRT-PCR analysis

Cultures of O395 and ΔtarA strains were grown overnight and then diluted 1:100 into LB pH 6.5 and grown at 30°C. Once cultures reached an OD₆₀₀ ~0.35, they were diluted 1:10 and grown until OD₆₀₀ ~0.35 and 2 ml aliquots of the samples were harvested. Cell pellets were resuspended in TRIzol (Invitrogen) and total RNA was isolated according to the manufacturer’s protocol as described previously (Yildiz et al., 2004). Whole genome expression analysis was performed using 70-mer oligos representing the open reading frames present in the V. cholerae N16961 genome. cDNA synthesis, microarray hybridization, and analysis of expression data were performed as described previously (Beyhan et al., 2006) using two biological and four technical replicates (for each biological replicate).

RNA samples for qRT-PCR were DNase treated, run on an agarose gel to check quality, and quantified by measuring the OD₂₆₀. 2.5 μg of each sample were treated with Moloney murine leukemia virus (M-MLV) reverse transcriptase (Invitrogen) according to the manufacturer’s specifications. For detection of transcripts, primers amplifying a 200-bp region in the middle of the mRNA were used with SYBR Green Master Mix (Stratagene) on a Stratagene MX3000P thermocycler. Primers were designed using the OligoPerfect tool (Invitrogen). Each test was performed in triplicate at least three times, and fold change in expression was calculated using the ΔΔCT method (Pfaffl, 2001), with recA transcript levels used as controls between cDNA samples.

Infant mouse colonization assays

Four-day-old CD1 mice (Charles River) were inoculated intragastrically with approximately 10⁶ bacteria. The inocula were 50:50 mixtures of the strain of interest (either tarA⁻/ lacZ⁺ V. cholerae or the tarA⁺/ lacZ⁺ control) and the control strain, which is a lacZ⁻ O395 derivative. Inoculated mice were incubated at 30° C for 16 hours, after which the mice were sacrificed and their intestines were removed and homogenized. Serial dilutions of the individual intestinal homogenates were plated on LB agar plates containing streptomycin and X-gal. Blue and white colonies were counted after overnight growth at 37°C. To determine the ratio of strains in the initial inoculum, the mixtures used to inoculate the mice were diluted into PBS and plated onto LB agar plates containing streptomycin and X-gal, and again the blue and white colonies were counted. The competitive index was calculated by dividing the blue:white ratio of the initial inoculum by the in vivo blue:white ratio.

Supplementary Material

01

SI Table 1. Bacterial strains used in this study.