Direct regulation of the natural competence regulator gene tfoX by cyclic AMP (cAMP) and cAMP receptor protein (CRP) in Vibrios

TfoX (Sxy) and CRP are two important competence activators. The link between tfoX and CRP has been shown in H. influenza but lacking evidence of direct interaction. Recently a Sxy-dependent CRP (CRP-S) site autoregulating Sxy was reported in E. coli. Here, we show that the cAMP-CRP complex transcriptionally regulates tfoX expression through multiple canonical CRP (CRP-N) sites in Vibrios. This conclusion is supported by an analysis of the tfoX mRNA levels and tfoX transcriptional reporter fusions. The reduced expression of tfoXVC was restored by trans-complementation of crp in ∆crp and by exogenous cAMP in ∆cya. A promoter deletion analysis and the site-directed mutagenesis of the putative CRP-N sites revealed the presence of two functional CRP-N sites. The direct binding of cAMP-CRP to the tfoXVCpromoter was demonstrated by EMSA assays. Additionally, the transcriptional start site (TSS) of tfoXVF in V. fluvialis was determined, and −10/−35 regions were predicted. Further comparison of the tfoX promoter in Vibrios revealed the existence of similar −10 motifs and putative CRP-N sites, indicating the conserved mechanism of CRP regulation on tfoX. Our study demonstrates the direct binding of the cAMP-CRP complex to tfoX promoter, and broadens the understanding of the molecular mechanism regulating tfoX in Vibrios.


Results
Effect of CRP on tfoX gene expression. Yamamoto's works greatly improved the knowledge of the regulation of tfoX VC in V. cholerae by identifying chitin disaccharide (GlcNAc)2 as the minimum inducer of chitin-dependent competence, by mapping the TSS, and the transcriptional and translational cis-acting elements, and by distinguishing the sRNA tfoR regulation 21,22 . Using this knowledge, combined with the prior clues provided by microarray data 27 , we reexamined the promoter region of tfoX VC and found a potential CRP-N binding site centered at − 84.5 relative to the TSS, which contained a perfect TGTGA half-site and a 4/5 matched TCTCA half-site for the DNA binding domains of the active CRP dimer. These findings strongly suggested the possibility that CRP regulates tfoX VC expression.
To determine whether CRP was involved in the expression of tfoX VC , we examined the tfoX VC mRNA level in wild type and isogenic mutant crp via qRT-PCR (Fig. 1A). Deletion of the crp gene resulted in much lower expression of tfoX VC , which confirmed our microarray data. For further confirmation, we performed a complementation test by introducing plasmid pBADCRP7, which expresses biologically active CRP protein from the araBAD promoter. As shown in Fig. 1A, the tfoX VC expression was restored to a much higher level than the WT level in the pBADCRP7-complemented crp mutant. As expected, inclusion of the control vector did not restore the tfoX VC expression.
To verify and expand the above finding that CRP was required for the expression of tfoX, we further examined the tfoX mRNA level in an emerging foodborne pathogen, V. fluvialis, and its isogenic crp mutant. The crp mutant was constructed by allelic exchange as described in the Materials and Methods section. Compared to the wild type pathogen, the tfoX VF mRNA level in the corresponding crp deletion mutant was significantly reduced (Fig. 1B). Taken together, these results indicate that the requirement of CRP for the expression of tfoX is a common feature in Vibrios. However, it was unclear whether this Scientific RepoRts | 5:14921 | DOi: 10.1038/srep14921 requirement is dependent on a direct interaction, or whether it may be a pleiotropic (or secondary) effect, because CRP is a global regulator and affects numerous genes.
Effects of cya mutation on the tfoX gene expression. To determine if the tfoX VC dependency on CRP can be fully accounted for by the cAMP binding to CRP, we measured the tfoX VC expression in WL7259 and WL7259 supplemented with 1.5 or 2.5 mM cAMP in the culture medium. As shown in Fig. 2, there was significantly lower expression of tfoX VC in the WL7259 lacking adenylate cyclase compared to the wildtype strain. Furthermore, the expression of tfoX VC was fully restored by supplementing the medium with exogenous cAMP. The activity of CRP is determined by the intracellular concentration of its allosteric activator, cAMP 31 . Consistent with this, the tfoX VC level in the strain with extra cAMP supplementation was more than 10-fold higher than that of the wild type strain. These results suggest that the expression of tfoX VC is under positive regulation by the cAMP-CRP complex. However, it was still unclear whether the dependence on both CRP and cAMP indicates the presence of a direct effect, and further evidence was required to clarify this point.

CRP activates the transcription of tfoX.
To determine if the CRP-mediated regulation of tfoX occurs at the level of transcription, we constructed transcriptional reporter plasmids by fusing the promoter regions of tfoX VC and tfoX VF to the reporter genes lacZ and luxCDABE, respectively. The − 10 and − 35 motifs of tfoX VC promoter have been determined in Yamamoto's work 21 . The promoter of tfoX VF was predicted with sequence analogy of the tfoX VC promoter and TTS determined in our study (below part). In order to retain the full promoter activity, the sequences extending 408 bp and 381 bp upstream of tfoX VC and tfoX VF translational start sites were respectively amplified as promoter regions. The reporter constructs were transferred into the corresponding wild-type strains and the isogenic crp mutants. Significant differences in β-galactosidase expression (Fig. 3A) and luminescence activity (Fig. 3B) were detected between the WT and ∆crp containing the reporter constructs on both the V. cholerae and V. were grown in LB medium, and cells were collected at the late-log phase. The tfoX VC and tfoX VF mRNA abundances were measured by qRT-PCR. The "WT" bar was set to 1 and used as a reference to calculate subsequent expression values. Error bars indicate the standard deviations of three independent cultures. ***Significantly different from the wild-type strain (t-test, P < 0.05). Deletion analysis of the tfoX VC promoter region. To further delineate the cis DNA sequences in the tfoX promoter region required for CRP activation, we constructed additional transcriptional fusions containing 5′ deletions of the tfoX VC promoter (Fig. 4A). Plasmid p 2 tfoX VC -lacZ contained a DNA sequence extending from − 74 bp upstream of the TSS, lacking the previously predicted CRP-binding sequence (TGTGA-N6-TCTCA); while p 3 tfoX VC -lacZ contained an even shorter sequence starting from the putative − 35 element. As shown in Fig. 4B, p 3 tfoX VC -lacZ exhibited the lowest β-galactosidase activity, and no difference was observed between the WT and ∆crp. Surprisingly, plasmid p 2 tfoX VC -lacZ retained substantially high promoter activity, and a significant difference in β-galactosidase activity was still retained for the WT and ∆crp, indicating that the encompassed promoter region contains DNA sequences necessary for CRP activation.
Further inspection of the promoter region revealed another suboptimal CRP binding site (TGTGAGATAAGTCCAG), with three mismatches from the consensus motif on the one half-site (TCACA). This binding site is centered at − 41.5, encompasses the predicted − 35 hexamer, and thus extends into the region generally bound by the RNA polymerase (RNAP). Based on these results, we inferred that the tfoX VC promoter resembles or belongs to the Class III CRP-dependent promoter, which requires multiple activator molecules for full transcriptional activation. In this case, it requires two CRP molecules. Specifically, one binding site is centered at position − 84.5 as in the class I CRP-dependent promoters; the other is centered at − 41.5, exactly like the class II CRP-dependent galP1 promoter 32 . The tandem binding of CRP causes synergistic effects on transcriptional activation, with the best effects resulting from the binding to a site situated 40 or 50 bp upstream 33 . The 43-bp spacer of the two binding sites (center-to-center distance) of tfoX VC seems to be an optimal distance that produces the best synergistic effects.
Site-directed mutagenesis of the CRP-binding site at the promoter of tfoX VC . The ptfoX VC -lacZ reporter fusion contains the full length functional tfoX promoter region so that it encompasses two putative CRP-binding sites which were designated as CBS1 and CBS2, respectively. To further verify and distinguish the roles of each CRP-binding site in tfoX VC transcription, we substituted the consensus GTG for ACA in the half-site of CBS1, CBS2, and both by site-directed mutagenesis (Fig. 4C). The promoter activities of the resultant mutant fusions ptfoX VC -lacZ-CBS1M, ptfoX VC -lacZ-CBS2M and ptfoX VC -lacZ-CBS1M+ 2M, were determined. As shown in Fig. 4C, the mutation of CBS1 resulted in a major decrease in the β-galactosidase activity of the ptfoX VC -lacZ-CBS1M fusion compared to the wild-type ptfoX VC -lacZ, which is consistent with the results of the p 2 tfoX VC -lacZ construct. As expected, were diluted 100-fold in LB and grown to OD600 0.5. Each culture was divided into three portions, and exogenous cAMP (Sigma Chemical Co.) was added to a final concentration of zero (control), 1.25, or 2.5 mM. The cultures were incubated at 37 °C for 1 h and the tfoX VC mRNA abundance was measured by qRT-PCR. The "WT" bar was set to 1 and used as a reference to calculate subsequent expression values. Error bars indicate the standard deviations of three independent cultures. **Significantly different from the wild-type strain (t-test, P < 0.05). ***Significantly different from the wild-type strain (t-test, P < 0.001).
Scientific RepoRts | 5:14921 | DOi: 10.1038/srep14921 ptfoX VC -lacZ-CBS1M+ 2M showed the lowest β-galactosidase activity due to the mutations of both CBS1 and CBS2 in the promoter region. Unexpectedly, mutation of CBS2 also results in a very low level of β-galactosidase activity in ptfoX VC -lacZ-CBS2M, similar to that observed for the double mutant fusion ptfoX VC -lacZ-CBS1M+ 2M. This result indicates that the CBS1 centered at − 84.5 cannot efficiently initiate the transcription of the tfoX VC promoter alone, and CBS2 is required for the synergistic activation of the tfoX VC promoter by CBS1.
To further determine the mutation of either CBS or both in the tfoX promoter has an effect in competence development or on natural transformation in vivo, We subsequently introduced the site specific mutations to CBS1, CBS2 and both on the V. cholerae chromosomal tfoX promoter, and measured the mRNA levels of tfoX, and tfoX-induced genes, pilB, chiA1 and chiA2, which are all required for competence in V. cholerae 5 . As shown in Fig. 4D, the mRNA levels of tfoX, pilB, chiA1 and chiA2 in C7258∆ptfoX-CBS1M, C7258∆ptfoX-CBS2M and C7258∆ptfoX-CBS1M+ 2M were significantly reduced compare to the wildtype strain. These results coincide with the above results of ptfoX VC -lacZ-CBS mutant fusions and indicate the regulation of TfoX-dependent natural transformation by CRP through the CBS sites in vivo.
CRP activates transcription by directly interacting with RNAP and/or acting upon DNA to change its structure to facilitate RNAP binding 34 . At the class I promoter, CRP binds upstream of the promoter and increases the rate of initial binding of RNAP to form a closed promoter complex (RPc). At the class II promoter, CRP binds within the promoter to increase the rate of transition from the closed to open promoter complex (RPo), in which the DNA duplex becomes unwound around the − 10 region and short RNA products are synthesized 35 . We speculate that the loss of the CBS2 site might cause a substantial decrease in the − 35 and − 10 motif binding by RNAP in the tfoX VC promoter and/or may decrease the efficiency of the transition from RPc to RPo. This will need to be investigated in the future studies. Another possibility is that an inverted repeat sequence (termed IR1), which was predicted to be a potential transcriptional operator 21 , overlaps the CBS2 site, so the site-directed mutagenesis of the consensus GTG to ACA in the CBS2 site could change the structure of IR1, thus affecting the transcription of tfoX VC . However, the present results demonstrate that the two putative CRP binding sites both play vital roles in the activation of tfoX VC expression. and WL7258∆lacZ (∆crp) containing the p2tfoXVC-lacZ and p3tfoXVC-lacZ fusion, respectively, were grown at 37 °C to mid-log phase. (C) The promoter activities of wild-type and CBS mutated fusions. C7258∆lacZ containing ptfoXVC-lacZ, ptfoXVC-lacZ-CBS1M, ptfoXVC-lacZ-CBS2M, or ptfoXVC-lacZ-CBS1M+ 2M were grown at 37 °C to the mid-log phase. The β-galactosidase activity was measured as described in the Methods. The mutated bases in fusions were constructed by site-directed mutagenesis and underlined. (D) V. cholerae strains C7258 (WT), C7258∆ptfoX-CBS1M, C7258∆ptfoX-CBS2M and C7258∆ptfoX-CBS1M+ 2M were grown in LB medium to late-log phase. The tfoX VC and pilB, chiA-1 and chiA-2 mRNA abundances were measured by qRT-PCR. Error bars indicate the standard deviations of three independent cultures. The "WT" bar was set to 1 and used as a reference to calculate subsequent expression values. ***Significantly different from the wild-type strain (t-test, P < 0.0003). *Significantly different from the wild-type strain (t-test, P < 0.05).
Scientific RepoRts | 5:14921 | DOi: 10.1038/srep14921 Determination of the TSS of tfoX VF and a comparative analysis of the tfoX promoter from different Vibrio species. The molecular structural features, such as the − 10 and − 35 motifs, transcriptional and translational cis-acting elements, secondary structure of the tfoX VC mRNA, and TSS have already been determined 21,22 . The original annotated open reading frame (ORF) of tfoX VC was 609 bp long, with GUG as the start codon and GGGA as the predicted Shine-Dalgarno (SD) sequence. Through a series of mutational analyses of potential start codons based on fusion-plasmid constructions, Yamamoto revised the start codon and SD sequence of tfoX VC to be ATG and GAAG, respectively, which are located 21 and 14 nucleotides downstream of the original sequences 21,22 . To assess the conservation of the molecular structural and regulatory characteristics of tfoX in different Vibrio species, we determined the TSS of tfoX VF and compared the tfoX promoters in different Vibrio species whose genome sequences are available from the NCBI database and for which the homologs of tfoX have been annotated.
The open reading frame (ORF) of tfoX VF shared 74.32% nucleotide identity with tfoX VC , but the intergenic regions showed much lower homology (55.87%). Using 5′ RACE, we identified a transcriptional start site 126 nucleotides upstream of the putative ATG start codon predicted according to the similarity to tfoX VC . Putative σ 70 -specific − 10 (TATGAT) and − 35 (TCCGGA) motifs separated by 19 bp were found in the upstream region of the TSS (Fig. 5), which coincides with the fact that competence genes are regulated by σ 70 type promoters 36 . The − 10 sequence has only one mismatch from the − 10 motifs of E. coli (TATAAT) and tfoX VC (TATCAT). The − 35 sequence has four mismatches from the E. coli consensus sequence TTGACA, but was a match at five of six nucleotides in the − 35 region of tfoX VC (TCCAGA).
As shown in Fig. 5, conserved − 10 motifs and two putative CRP-binding sites were found around similar locations in the tfoX promoter in V. fluvialis, V. cholerae, V. mimicus, V. furnissii, V. vulnificus, V. anguillarum, V. harveyi, V. campbellii, V. alginolyticus, and V. splendidus. The − 35 regions were less conserved. This was not unexpected, because no CRP-dependent promoter has a good − 35 sequence, and some even lack good − 10 sequences 37 . The center-to-center distances between the two putative CRP binding sites were 43, 42 or 41 bp in different species. Three putative CRP-binding sites were discernible in the promoter region of tfoX in V. fischeri. Notably, the spacer sequence between the two half-sites of CRP binding site 3 in V. fischeri was seven bases, rather than the conventional six bases present in standard CRP half-sites. In general, the distal CRP-binding sites are more conserved and can roughly be classified into four classes based on the sequence homology, although the corresponding regions in V. nigripulchritudo and V. paraheamlyticus display higher variation. It is difficult to predict the putative − 35 and/or − 10 elements in similar locations in these species compared with the above species, and only one potential CRP binding site was found. The different numbers of binding sites and the variations in the binding sequences may contribute to the fine-tuning of tfoX expression in response to the changes in the environment, and may reflect dynamic evolutionary histories of the acquisition and or loss of CRP-binding sites. CRP binds directly to the tfoX promoter in vitro. All of the above-described results suggest that tfoX expression is regulated at the transcriptional level via cAMP-CRP complex-mediated activation, and there are two cAMP-CRP binding sites at the tfoX promoter region. To verify whether there is cAMP-CRP binding to the putative sites, we measured the binding of purified CRP to a 152-bp DNA fragment of tfoX VC encompassing the two CRP binding sites. The fragment was labeled with biotin at the 5′ end and used as a probe. As shown in Fig. 6B, the addition of CRP (0.11 μ M) resulted in a shift of the 152-bp DNA fragment to slower mobility. When the amount of CRP was increased (> 0.88 μ M), the shifted band slowed to an even higher position. The binding was abolished when CRP or cAMP was excluded from the reaction mixture. Inclusion of the same, but unlabelled, 152-bp DNA fragment greatly competed with the labeled probe for the binding sites in a dose-dependent manner. The addition of 100-fold of the competitor cold probe changed the higher-shifted band to the lower one, while the addition of 300-fold of the cold probe completely abolished the retarded-band, and the labeled probe in the reaction mixture was released as the free probe. On one hand, these results confirmed the specific binding of CRP to the tfoX VC promoter, and on the other hand, suggested that one binding site exhibits lower-affinity binding. This is in agreement with the fact that the proximal CBS2 differs from the consensus sequence in three conserved positions, while CBS1 differs from the consensus sequence in only one conserved position, thus resulting in different binding affinity.
For further confirmation, we performed EMSA with the fragments containing single CRP binding sites. Consistent with the more conserved features of CBS1 and the EMSA results described above, the 75 bp fragment with CBS1 was shifted to a single band with slower mobility following the addition of 0.11 μ M of CRP, while mutagenesis of the conserved GTG completely abolished the binding (Fig. 6C). While the 97 bp fragment with CBS2 was not shifted even when up to 0.88 μ M of CRP was added, smeared bands were formed when more than 1.76 μ M CRP were added in the assay mixture, which may indicate the instability of the DNA-CRP complex formed in vitro due to lower affinity of binding  (Fig. 6D). Although the EMSA assays indicated that CBS2 is a low affinity binding site, the site-directed mutagenesis analysis demonstrated that this binding site also plays a vital role in the initiation of the transcription of tfoX VC . It should also be kept in mind that it is impossible to rule out the possibility that the low affinity binding of CBS2 was due to the in vitro binding conditions in the EMSA assay, which do not exactly mimic the in vivo condition.
With V. cholerae CRP, which has a high amino acid identity (99.05%) to the CRP in V. fluvialis, we also found the direct and specific binding of CRP to the tfoX VF promoter in an EMSA assay (data not shown), further demonstrating the conserved direct regulatory effects of the cAMP-CRP complex on the competence regulator tfoX in Vibrio species.

Discussion
Competence development for DNA uptake by microorganisms is tightly regulated, and cAMP-CRP plays a central role in competence regulation in response to nutritional stress. The direct binding of CRP to the CRE site on the promoter region of competence-associated genes was predicted 8 and experimentally confirmed 10 . The second essential regulator of competence, TfoX, was proposed to act cooperatively with CRP by directing it to CRE sites to allow maximal expression of the competence genes 10 , although it lacks the helix-turn-helix DNA binding motif 12 and remains recalcitrant to overexpression and purification 8 . Recently Jaskolska and coworker demonstrated that TfoX is unstable and degraded by Lon protease in E.coli 38 . Previously, Zulty & Barcak 39 and Cameron et al. 20 showed a strong induction of sxy expression after the addition of 1 mM cAMP to H. influenzae, although evidence for direct binding is not yet available. Additionally, the opposite result was also reported, i.e., cya mutation has no effect on sxy expression 10 . While we were in the preparation of the manuscript, Jaskolska et al. published their data showing that in E. coli, Sxy is positively autoregulated at the transcriptional level by a mechanism requiring cAMP-CRP and the CRP-S site in the sxy promoter 38 . It is interesting that Vibrio species possess the CRP-N sites instead of CRP-S sites in the tfoX promoter regions. Experimental evidences have demonstrated that the CRP-N sites are higher affinity sites than the CRP-S sites 10,12,36 . The presence of CRP-N sites in tfoX could start higher expression of TfoX and in turn more efficiently induce the development of competence, thus to favor the survival of Vibrio species in the aquatic system which is normally nutrition-limited. This finding further indicates that though the members of gamma-proteobacteria families, such as Enterobacteriaceae, Pseudomonadaceae and Vibrionaceae share a common regulatory mechanism in competence development 10,12 , the divergence or variation in details exists which may facilitate to refine the regulation in different genetic backgrounds and or survival environments. We reasoned that this may due to the long-term evolution pressure selection between the different native habitant environments.
In V. cholerae, chitin induces natural competence 5 , and cAMP and CRP are required for efficient chitin colonization, degradation and increased competence gene expression 19 . The regulation of tfoX by CRP, particularly in Vibrio species, had not been established or investigated. In the present study, we experimentally demonstrated that cAMP and CRP positively regulate the expression of tfoX in V. cholerae and V. fluvialis. A set of promoter deletion fusions and site-directed mutagenesis analyses confirmed the functional existence of two CRP binding sites in the tfoX VC promoter region, and direct binding was further demonstrated by an EMSA in vitro.
A sequence comparison of the tfoX promoter region revealed the existence of optimal and suboptimal CRP-N binding sites in other Vibrio species homologs, suggesting that the transcriptional regulation of tfoX by CRP is a common feature in Vibrionaceae. The ecologies and natural hosts of Vibrio species vary, but free-living in sea water and attachment to zooplanktons are shared life stages. In general, sea water is a nutritionally-limited environment, where chitin is the most abundant nutrition alternative. The chitin utilization pathway has been known to be conserved in the Vibrionaceae 40 . Nutritional stress induces elevation of cAMP-CRP, and the availability of chitin efficiently activates tfoX expression. TfoX, as the early activated competence regulator, synergistically directs cAMP-CRP to the CRP-S sites on the competence regulon to further promote the competence development, thus allowing the bacteria to take up free DNA from the environment. The uptaken DNA can be used as either an energy source or to repair damaged DNA, or may be used to acquire new alleles/genes, which accounts for the intensive genetic diversity and the mosaic genome structure in Vibrio species revealed by recent genomic sequencing efforts 41 . It was thought that the reuse of nucleotides from DNA degradation in the cytoplasm may be more significant than other genetic benefits, at least in the short term 6 .
We also noticed that the degree of CRP binding sequence conservation varies and can roughly be classified into four different groups (Fig. 5). Generally, the distal CBS1 site is more conserved, with a match of at least eight of 10 nucleotides with the consensus binding site. In constrast, the half-site of the proximal CBS2 site obviously displays more sequence diversity, which may result in transient CRP-DNA interactions. It is tempting and logical to speculate that the regulatory features (such as the intensity and duration of binding) are distinct due to the differential binding affinity of CRP caused by sequence variations in different Vibrio species and between different binding sites. For example, V. fischeri possesses three regions that approximate the CRP consensus sequence. Even though V. fischeri has three CRP binding sites, the transformation efficiency has been reported to be 100-fold lower compared to that of V. cholerae 13 , indicating that the physiological significance of more binding sites needs to be established. The length of the spacer between the two CRP binding half-sites is usually 6 bp or 8 bp 42,43 . As mentioned before, the spacer region between the half-sites of the third CRP binding site in V. fischeri is 7 bp, which could potentially lead to a dramatic reduction in the binding affinity. The study by Pyles and Lee demonstrated that changing the spacer length from 7 bp to 8 bp increased the binding affinity of CRP for DNA 44 . Of course, considering the complexity of the competence-associated machinery components and their regulation, it cannot be stated this is the only reason for the 100-fold lower transformation.
In conclusion, our study provided a definitive analysis of the role of the cAMP-CRP in tfoX expression by experimentally demonstrating that the cAMP-CRP complex directly activates its transcriptional expression. These results, together with previous data 5,19 , demonstrate that the natural competence of Vibrios is subject to catabolite repression by the global transcriptional regulator, CRP, and this regulation is effected through direct control of both the vital competence regulator gene and the competence component genes. In addition, CRP indirectly regulates competence through quorum-sensing by activating primary autoinducer synthesis gene, cqsA 27,28 . The multiple forms of regulation at different layers or pathways mediated by CRP may maximize the input of simple environmental signals to induce or promote the development of competence for the efficient uptake of extracellular DNA as a nutritional supply or for other purposes, thus greatly favoring the survival and environmental fitness of the organism.

Methods
Strains and media. The V. cholerae and V. fluvialis mutants used in this study were constructed using the El Tor biotype strain, C7258 (Peru′ isolate, 1991), and the clinical strain, 85003 45 , as wildtype precursors, respectively. The construction of strains WL7258 (C7258∆crp), WL7259 (C7258∆cya), C7258∆lacZ and WL7258∆lacZ has been described previously 27,28,46 . pBADCRP7 was introduced into WL7258∆lacZ by electroporation 47 . Escherichia coli DH5α λpir and S17-1λpir were used for cloning purposes. The V. cholerae and V. fluvialis strains were grown in LB broth containing 1% NaCl at 37 °C with agitation (250 rpm). Culture media were supplemented with ampicillin (Amp, 100 μ g/ml), kanamycin (Km, 75 μ g/ml), chloramphenicol (Cm, 5 μ g/ml), streptomycin (Sm, 100 μ g/ml), or polymyxin B (PolB, 100 units/ml) as required. All strains, plasmids and primers used in this study are listed in Table 1. The amplicons were mixed in equimolar concentration and used as the template to amplify the chromosomal fragment containing the crp deletion using primer pair VF-CRP-F1-up-SalI/VF-CRP-F2-dn-SacI. The resulting fragment was cloned into suicide plasmid pWM91 to generate pWM-VF∆crp. pWM-VF∆crp was constructed in E. coli S17-1λpir and transferred to strain 85003 by conjugation. Exconjugants were selected in LB medium containing Amp and Sm, and were streaked on LB agar containing 15% (w/v) sucrose. Sucrose-resistant colonies were tested for Amp sensitivity, and the deletion of crp was confirmed by DNA sequencing.

Construction of transcriptional reporter plasmids.
To construct the ptfoX VC -lacZ transcriptional fusion, we cloned a 430-bp fragment containing the tfoX VC promoter region amplified with primers pVC1153-up-PstI/pVC1153-dn-HindIII into pTT3 48 downstream of the strong rrnBT1T2 transcription terminator to generate pTT-tfoX VC . Next, a 900 bp KpnI-HindIII fragment containing the rrnBT 1 T 2 terminator and the tfoX VC promoter from pTT-tfoX VC was ligated into the big HindIII-KpnI fragment of phaplac7, which contains a promoterless lacZ gene, the pBR322 origin of replication, and the bla gene 48 . The resultant plasmid, ptfoX VC -lacZ, was introduced into C7258∆lacZ and WL7258∆lacZ by electroporation 47 . Additional DNA fragments harboring 5′ deletions of the tfoX VC promoter were amplified with primer pairs of pVC1153-up2-PstI/pVC1153-dn-HindIII and pVC1153-up3-PstI /pVC1153-dn-HindIII, and transcriptional fusions p2TTtfoX VC and p3TTtfoX VC were constructed in a similar manner. To construct the bioluminescence based ptfox VF -lux fusion, we cloned a 384 bp fragment containing the tfoX VF promoter region amplified with primers VF-sxy-prom-up-SacI/VF-sxy-prom-dn-BamHI into reporter plasmid pBBRlux, which contains a promoterless luxCDABE operon 49 . ptfox VF -lux was constructed in DH5α λpir and transferred into S17-1λpir, and then mobilized into V. fluvialis strains 85003 and 85003∆crp by conjugation.
Quantitative reverse transcription PCR (qRT-PCR). The V. cholerae and V. fluvialis strains were grown in LB medium to the late-log phase. Cells were collected by centrifugation at 4 °C and immediately subjected to RNA extraction. Total RNA was extracted using the Trizol reagent (Ambion), followed by treatment with a TURBO DNA-free TM kit (Ambion) to remove chromosomal DNA contamination. The purity and integrity of RNA samples was verified by UV spectrophotometry and agarose gel electrophoresis. Using 1 μ g of total RNA per sample, cDNA was synthesized using random hexamer primers with SuperScript TM III reverse transcriptase (Invitrogen) according to the manufacturer's instructions. qRT-PCR was performed using SYBR Green (TaKaRa) on the Bio-Rad CFX96 Real-Time PCR detection system. The relative expression values (R) using recA mRNA as a reference were calculated using the equation R = 2 −(ΔCq target −ΔCq reference) , where Cq is the fractional threshold cycle. The following primer  Determination of the 5′-end of tfoX VF mRNA. We used 5′ RACE (rapid amplification of cDNA ends) to determine the TSS of tfoX VF . V. fluvialis strain 85003 was grown in LB containing 1% NaCl at 37 °C with agitation to OD600 1.5. Total RNA was extracted as described in the preceding section. The cDNA was generated using the SMARTer ™ RACE cDNA Amplification Kit (Clontech) according to the manufacturer's instructions. The cDNA was then amplified by PCR using a kit provided with UPM (universal primer) and the gene specific primer, VF-sxy-race. The PCR product obtained was gel-purified and cloned into the pMD ® 18-T vector. Ten clones were sequenced using the primers M13-R and M13-F to determine the TSS.
Site-directed mutagenesis of the putative CRP binding sites of the tfoX VC promoter. The two putative CRP binding sites in the promoter region of tfoX VC were mutated using the Hieff Mut TM site-directed mutagenesis kit (Shanghai YEASEN Biotechnology Co., Ltd.) according to the manufacturer's instructions. The above constructed ptfoX VC -lacZ plasmid containing the full-length functional tfoX promoter region was used as the template, and mutagenesis was induced using primers carrying the substituted nucleotides. Primer pairs VC1153CBS1M-for/VC1153CBS1M-rev and VC1153CBS2M-for/ VC1153CBS2M-rev were used to substitute three bases in the half-site of putative CRP binding sites 1 and 2, respectively. To simultaneously mutate the two CRP binding sites, primer combinations VC1153CBS1M-for/VC1153CBS2M-rev and VC1153CBS1M-rev/ VC1153CBS2M-for were used. The resultant plasmids, ptfoX VC -lacZ-CBS1M, ptfoX VC -lacZ-CBS2M and ptfoX VC -lacZ-CBS1+ 2M, were sequenced to confirm the mutations. Subsequently, the obtained plasmid was introduced into V. cholerae strain C7258∆lacZ by electroporation. Mutations in CBS1M, CBS2M and CBS1M+ 2M were further introduced into the chromosomal promoter region of tfoX of C7258 through suicide plasmid-mediated allelic exchange. For this purpose, the F3-dn-SphI: Table 1. Strains, plasmids and primers used in this study.
Expression and purification of His-CRP. E. coli TOP10 harboring the recombinant plasmid pBAD-CRP7 50 were cultured at 37 °C in LB medium. The 6xHis-tagged CRP was induced by 0.2% (w/v) arabinose and purified by Ni-IDA affinity chromatography (Novagen) under native conditions according to the manufacturer's instructions. The purity was analyzed by SDS-PAGE (Fig. 6A), and the concentration was determined with a Pierce BCA Protein Assay kit.
Electrophoretic mobility shift assays (EMSA). EMSA was performed as described previously 51 . The 152 bp, 75 bp and 97 bp fragments of the tfoX VC promoter regions were amplified with the biotin-labeled primer pairs Shift-up1976-96/Shift-dn2107-27, Shift-up1976-96/VC1153-p1shift-dn, and VC1153-p2shift-up/Shift-dn2107-27 respectively, and were used as probes. The 152 bp probe covers fragments ranging from − 127 to 24 relative to the reported TSS 21 and encompasses two putative CRP binding sites. The same fragments without a biotin label were used as competing cold probes and were added in 100-300-fold excess of the labeled probes. The 75 bp and 97 bp probes extend from residues − 127 to − 53 and − 74 to 24, respectively, and each contains a single CRP binding site. Binding reactions were performed by mixing each biotin-labeled probe with increasing quantities of purified CRP in 10 μ l of reaction volume with binding buffer (50 mM Tris-HCl [pH7.8], 250 mM KCl, 5 mM MgCl 2 , 2.5 mM EDTA, 0.5 mM MnCl 2 , 2.5 mM DTT, 1 μ g Poly (dI.dC) and 100 μ M cAMP). The reaction mixture was incubated at room temperature for 30 min, and then separated on a 6% (for 152 bp probe) or 10% (for 75 bp and 97 bp probes) native polyacrylamide gel after adding loading buffer. The separated DNA and DNA-protein samples were transferred onto nylon membranes using the Mini Trans-Blot Electrophoretic Transfer cell (Bio-Rad), and were detected with the Chemiluminescent Nucleic Acid Detection Module (Thermo Scientific) following the manufacturer's instructions. β-Galactosidase assays and bioluminescence assays. Overnight cultures of V. cholerae strains C7258∆lacZ and WL7258∆lacZ containing tfoX VC -lacZ fusion plasmids were diluted 100-fold into fresh LB, and were grown to mid-exponential phase. The specific activities of β-galactosidase are expressed in Miller units [1000 (OD420/t v OD600)], where t is the reaction time and v is the volume of enzyme extract per reaction 52 . Similarly, V. fluvialis strains 85003 and VF∆crp with ptfox VF -lux were cultured to the mid-exponential phase. The absorbance and bioluminescence were quantified thereafter. The bioluminescence was measured on opaque-wall 96-well microtiter plates (ostar 3917) with a Tecan Infinite M200 Pro luminometer. The promoter activity is expressed as Lux/OD600.