ToxR interferes with CRP-dependent transcriptional activation of ompT in Vibrio cholerae

Summary

In pathogenic Vibrio cholerae, the transmembrane DNA-binding protein ToxR co-ordinates the expression of over 20 genes, including those encoding important virulence factors such as cholera toxin and the toxin-co-regulated pilus. The outer membrane protein OmpT is the only member of the ToxR regulon known to be repressed by ToxR. In this study, we examined the environmental conditions that regulate OmpT expression and demonstrated that ompT transcription is upregulated 14-fold when the bacteria enter late log phase from early log phase. Deletion of the crp gene completely abolishes OmpT expression. Comparison of ompT transcription levels in the isogenic crp⁻, toxR⁻ and crp⁻toxR⁻ mutants revealed that (i) in the absence of ToxR, constitutive high-level ompT transcription is dependent on cAMP receptor protein (CRP); (ii) ToxR not only interferes with CRP-dependent ompT activation, but also abolishes the CRP-independent, basal level ompT transcription; thus, the mechanism by which ToxR represses ompT transcription involves both antiactivation and direct repression; (iii) both CRP and ToxR are required for the regulation of OmpT expression by growth phase. To provide further insights into the molecular mechanism of CRP-dependent activation of ompT transcription, we demonstrated that CRP-dependent activation requires a CRP binding site centred at −310 of the ompT promoter, without which the interaction of CRP with other CRP binding site(s) more proximal to the promoter results in repression. Mutations in two regions on CRP (AR1 and AR2) that directly contact RNA polymerase (RNAP) abolish activation, suggesting direct interaction of CRP with RNAP from −310 of the ompT promoter via DNA looping.

Introduction

The Gram-negative bacterium Vibrio cholerae can colonize the human intestine and cause the potentially fatal diarrhoeal disease cholera. The major virulence factors of this pathogen include cholera toxin (CT), which is the direct cause of the profuse watery diarrhoea, and the toxin-co-regulated pilus (TCP), which is required for intestinal colonization (reviewed by Kaper et al., 1995). V. cholerae is also a common inhabitant of the aquatic environment. Environmental sensing and co-ordinate regulation of gene expression is essential for V. cholerae to persist in drastically different environments and cause disease. Among the increasing number of transcription factors described in the regulation of virulence gene expression in V. cholerae, ToxR is the most extensively characterized (reviewed by DiRita, 1992; Skorupski and Taylor, 1997a).

Over 20 genes in V. cholerae are co-ordinately controlled by ToxR and termed the ToxR regulon (Peterson and Mekalanos, 1988; Merrell et al., 2001). The ToxR regulon is organized into two branches in a cascade fashion (DiRita et al., 1991; Champion et al., 1997). In one branch, ToxR works as a co-activator of another transmembrane DNA-binding protein TcpP to activate transcription of ToxT, an AraC family transcription activator (Higgins et al., 1992; Häse and Mekalanos, 1998; Krukonis et al., 2000). ToxT then directly activates transcription of the genes encoding CT and TCP and autoregulates its own gene (Yu and DiRita, 1999). In the other branch, independent of TcpP and ToxT, ToxR differentially regulates the expression of two outer membrane porins, OmpU and OmpT (Chakrabarti et al., 1996), resulting in almost exclusive expression of OmpU in a toxR⁺ strain and exclusive expression of OmpT in a toxR⁻ strain (Taylor et al., 1987; Crawford et al., 1998; Li et al., 2000). The genes encoding CT are part of the genome of a lysogenic phage (Waldor and Mekalanos, 1996), and the genes encoding TCP are part of the Vibrio pathogenicity island, which is also apparently of phage origin (Karaolis et al., 1998; 1999). Therefore, it is generally believed that ToxR first evolved as the regulator of OmpU and OmpT expression in an ancestral V. cholerae strain, and subsequently gained the control of virulence genes when these genes were acquired through horizontal transfer. The essential role of ToxR during V. cholerae pathogenesis in vivo has been demonstrated in human volunteer studies as well as in animal models (Herrington et al., 1988; Lee et al., 1999).

ToxR is a 32.5 kDa protein with a 100-amino-acid periplasmic domain, a single transmembrane domain and an amino-terminal 180-amino-acid cytoplasmic domain that shares homology with the DNA-binding domain of the OmpR family of two-component transcription factors (Miller and Mekalanos, 1984; Miller et al., 1987; Ottemann et al., 1992). With the discovery of additional proteins sharing the topological and functional features of ToxR, including TcpP and CadC of V. cholerae (Häse and Mekalanos, 1998; Merrell and Camilli, 2000), as well as the ToxR homologue in other Vibrio species and Photobacterium species (Welch and Bartlett, 1998; Osorio and Klose, 2000), ToxR has been recognized as the prototype of a unique family of transmembrane regulators.

ToxR interacts directly with long stretches of AT-rich sequences at the ctxAB, toxT, ompU and ompT promoters and functions as a versatile transcription factor. For ctxAB, ToxR binding to the −34 to −143 region of the promoter is sufficient to activate transcription in Escherichia coli, but not in V. cholerae (Miller et al., 1987; Champion et al., 1997; Li et al., 2000). At the ompU promoter, DNase I footprinting revealed the protection of three distinct sites by ToxR. It was proposed that the binding of ToxR to a region extending from −238 to −149 of the promoter is followed by co-operative binding to the downstream regions extending from −116 to −58 and from −53 to −24, where ToxR interacts directly with RNA polymerase (RNAP) and activates transcription (Crawford et al., 1998). For toxT, ToxR binding to the −108 to −65 region of the promoter recruits TcpP to bind downstream at −58 to −44, where TcpP can activate transcription directly (Krukonis et al., 2000; Li et al., 2000). ToxR was proposed as a repressor of OmpT because the expression of this protein is dramatically increased in a toxR mutant (Taylor et al., 1987). We have shown previously that ToxR binds directly to the −30 to −95 region of the ompT promoter, consistent with its role as a repressor. We also observed sequence-dependent activation of ompT transcription in the absence of ToxR and suggested that cAMP receptor protein (CRP) positively regulates ompT transcription (Li et al., 2000). However, how ToxR represses transcription by binding to this site and how ToxR as well as CRP mediate regulated expression of OmpT in response to environmental signals remained to be determined.

We report here the novel observation that growth phase-dependent regulation of the expression of OmpT requires the interplay of CRP, which functions as a direct activator, ToxR, which functions as an antiactivator as well as a repressor, and possibly an additional factor(s). We also demonstrate the direct interaction of CRP with a CRP binding site centred at −310 of the ompT promoter, which is required for the activation of transcription. These results suggest DNA looping as part of the mechanism by which CRP activates ompT transcription. Therefore, we propose that the binding of ToxR to the ompT promoter antagonizes CRP-dependent transcription by disrupting the formation of a transcription complex involving CRP.

Results

Regulation of OmpT expression by environmental signals

Despite the original observation that OmpT is upregulated by high osmolarity and downregulated by the supplementation of amino acids in minimal medium (Miller and Mekalanos, 1988), the regulation of OmpT expression has not been studied extensively. This is partly because of the intrinsically low level of OmpT in a wild-type strain resulting from ToxR-mediated repression (Fig. 1A, in LB). Our previous observation that CRP is a potential activator of ompT gene expression implies that CRP may contribute to the upregulation of OmpT in response to certain environmental signals (Li et al., 2000). To define the optimal conditions for OmpT expression so that the contribution of CRP as an activator can be assessed, we grew V. cholerae classical strain 395 to different growth phases in rich or minimal media, varying the osmolarity and temperature. Whole-cell lysates of the bacteria were separated by SDS–PAGE, and OmpT was visualized by Coomassie blue staining (Fig. 1A) or Western blotting with anti-OmpT serum (Figs 1B and C). The relative amounts of OmpU and OmpT in rich medium (LB broth) and in M9 glycerol medium are shown in Fig. 1A. As reported previously, in LB broth, OmpU appeared as a predominant protein, and OmpT was nearly undetectable, as were any differences in OmpT expression at different growth phases. In M9 glycerol medium, OmpT expression increased to a level even higher than that of OmpU. These results demonstrate that, even in the presence of the repressor ToxR, OmpT levels vary substantially between minimal medium and rich medium. Furthermore, using OmpT antiserum enabled us to detect a substantial change in OmpT expression levels across the growth curve (Fig. 1B), which was not readily detectable by Coomassie blue staining (Fig. 1A). Temperature had no significant effect on OmpT expression (Fig. 1B, 30°C versus 37°C). To examine the effect of osmolarity on OmpT expression, cells were grown in 1% tryptone broth containing various amounts of NaCl, KCl and melibiose. As shown in Fig. 1C, during both mid-log and early stationary phases, OmpT expression was increased further when the NaCl concentration was increased from 0.0625 M to 0.125 M. Further increase in NaCl concentration had no effect on OmpT expression. The osmoregulation of OmpT expression is consistent with the observations of Miller and Mekalanos (1988). These results confirm the previous observation that OmpT is upregulated by osmolarity, demonstrate the difference in OmpT expression in rich medium versus minimal medium and reveal for the first time that OmpT expression is substantially regulated by growth phase in rich medium.

Fig. 1.

Regulated OmpT expression under different conditions.

A. SDS–PAGE of whole-cell lysates of V. cholerae 395 grown in LB broth for 2 h (M, for mid-log phase), 4 h (L, for late log phase) and overnight (S, for stationary phase). Cells were grown overnight in M9 glycerol medium. The positions of OmpU and OmpT are indicated.

B. V. cholerae 395 were grown in LB broth at 30°C or 37°C, cells were collected at 2 h, 4 h and overnight, and whole-cell lysates were separated on an SDS–polyacrylamide gel, blotted to PVDF membrane and detected with OmpT antiserum. The cell density of each sample is also indicated.

C. V. cholerae 395 were grown in 1% tryptone broth with the addition of different concentrations of NaCl as indicated. OmpT expression was detected with OmpT antibody as described above; only two samples with the indicated cell densities were collected for each NaCl concentration, representing the mid- and late log phases of growth.

CRP is required for OmpT expression

To evaluate the contribution of CRP to the upregulation of OmpT expression, an isogenic crp mutant of V. cholerae 395, KSK377 (Skorupski and Taylor, 1997b), was also examined under the above conditions. In KSK377, OmpT expression was diminished to an undetectable level and did not respond to any condition tested (Fig. 2A, crp⁻). The V. cholerae crp gene was amplified by polymerase chain reaction (PCR) and cloned into a low-copy-number plasmid pACYC184 under the control of the tet promoter, yielding pCYL51. When pCYL51 was introduced into KSK377, the expression of OmpT was largely restored (Fig. 2A, crp^−/+). As ToxR is present on the chromosome of KSK377, these results demonstrate that ToxR alone is not sufficient to couple environmental signals to OmpT expression levels, and CRP is required for OmpT regulation.

Fig. 2.

Contribution of CRP and ToxR to the regulation of OmpT expression.

A. V. cholerae 395 (wt), isogenic crp mutant KSK377 (crp⁻) and KSK377 complemented with pCYL51 containing cloned crp (crp^−/+) were grown in LB broth at 30°C or 37°C. Whole-cell lysates were prepared, and OmpT expression was detected by Western blot as described in the legend to Fig. 1B.

B. V. cholerae 395 (wt), toxR mutant EK307 (toxR⁻), double mutant SCL100 (toxR⁻crp⁻) and double mutant complemented with plasmid pCYL51 (toxR⁻crp⁻/crp⁺) were grown in LB broth at 37°C. OmpT expression was detected as described above. M, L and S indicate mid-log, late log and stationary phases of growth respectively.

C. Quantification of ompT transcription by RNAse protection assay. V. cholerae 395 (toxR⁺crp⁺), EK307 (toxR⁻crp⁺), KSK377 (toxR⁺crp⁻) and SCL100 (toxR⁻crp⁻) were grown in LB broth at 37°C. Bacteria were harvested at the time points indicated, and total RNA was prepared. Total RNA (1 μg) from each sample was used in the RPA to hybridize with a 344 bp ompT-specific riboprobe. After digestion of the unhybridized probe with RNase, the amount of hybridized probe, which represents the amount of ompT-specific transcripts in the total RNA, was visualized by 4% denaturing PAGE followed by autoradiography. Fold activation of each sample is relative to SCL100 at 2 h of growth.

Owing to the presence of the repressor ToxR in the crp strain, it remained to be tested whether the diminution in OmpT expression in this strain resulted from loss of direct activation by CRP or lack of antirepression by CRP. Using phage CP-T1ts-mediated generalized transduction (Hava and Camilli, 2001), crp was inactivated in a toxR mutant of 395, EK307 (Krukonis et al., 2000), generating the double mutant SCL100 (toxR⁻crp⁻). The growth phase-dependent regulation of OmpT expression in wild-type 395 was then compared with that in EK307 (toxR⁻), KSK377 (crp⁻) and SCL100 (toxR⁻crp⁻). As shown in Fig. 2B, OmpT expression levels in the double mutant were slightly higher than in the crp mutant but much lower than in the toxR mutant. Furthermore, complementation of SCL100 with CRP results in a substantial increase in OmpT levels (toxR⁻crp⁻/crp). These results demonstrate that CRP does not regulate ompT transcription simply by antagonizing the effect of the repressor ToxR, but plays an active role in directly stimulating a low basal level OmpT expression.

Combinatorial control of ompT transcription by ToxR and CRP

To confirm the CRP-dependent activation of OmpT expression at the transcriptional level and to quantify the contribution of ToxR and CRP to ompT transcription, RNase protection assays (RPAs) were performed to measure the amount of ompT-specific transcripts at each growth phase. As shown in Fig. 2C, ompT transcription profiles were largely consistent with OmpT expression profiles except that, from late log phase to stationary phase (8 h), in both 395 (wt) and EK307 (toxR⁻), ompT mRNA levels decreased, yet OmpT protein levels increased further. Decreased ompT transcription was also seen at 6 h (data not shown), demonstrating the transient nature of transcriptional activation of ompT as cells entered late log phase. Increases in OmpT levels at stationary phase, despite the decrease in ompT mRNA, reflect the stability of OmpT protein and the possibility of post-transcriptional regulation.

The ompT mRNA levels in the double mutant, which remained essentially constant across the growth phase (Fig. 2C, toxR⁻crp⁻), were set as the basal level to quantify ompT transcripts in other strains. In KSK377 (toxR⁺crp⁻), the ompT mRNA level remained constant across the growth curve and was approximately eightfold lower than in the toxR⁻crp⁻ mutant, reflecting the ToxR-dependent repression of basal level ompT transcription. In EK307 (crp⁺toxR⁻), at both mid-log and late log phase, ompT mRNA was eightfold higher than in the double mutant, as a result of the CRP-mediated activation of transcription. Apparently, ompT transcription in neither crp⁻ nor toxR⁻ strains showed significant growth phase-dependent regulation, which was only observed in wild-type strain 395. In the presence of both ToxR and CRP, at early log phase (2 h, OD₆₀₀ = 0.2), the ompT mRNA level was fourfold less than in the double mutant (basal level), indicating that, under this condition, repression by ToxR dominated activation by CRP. ToxR was not only abolishing the CRP-dependent activation of the ompT promoter, but also partially interfering with basal level transcription. When wild-type cells entered late log phase (4 h, OD₆₀₀ = 0.6–0.8), the ompT mRNA level was 3.5-fold higher than in the double mutant, implying that, at this growth stage, ToxR could only partially interfere with the CRP-dependent transcription. In total, for the wild-type strain, transcription of ompT at late log phase was 14-fold higher than at mid-log phase, as a result of the combinatorial control of CRP and ToxR, even though neither ToxR or CRP alone seemed to respond to the growth phase signal(s).

Cis elements required for CRP-dependent activation of ompT transcription

To begin to understand the interplay of CRP and ToxR at the ompT promoter, we first investigated the mechanism of CRP-dependent activation. Our previous promoter deletion studies suggested that CRP-dependent activation requires the sequence between −165 and −490 of the ompT promoter (Li et al., 2000). Among the three putative CRP binding sites predicted for the ompT promoter, only CRP site 1 is within this region, centred at −310 (Fig. 3A). To test the role of this CRP binding site in directly mediating the CRP-dependent activation of ompT transcription, β-galactosidase activities of a full-length ompT–lacZ fusion plasmid (pT490) were compared with a fusion with an internal deletion of CRP binding site 1 (pT490D3, see Fig. 3A). As shown in Fig. 3B, the β-galactosidase activity profile of the full-length fusion (490) in all four isogenic strains resembled the ompT mRNA profile determined by RPAs. Specifically, the transcriptional activity of this fusion in the double mutant (toxR⁻crp⁻/490) represented a basal level that was not affected by either the activator CRP or the repressor ToxR. This level of transcription was activated by CRP as shown by toxR⁻/490, and repressed by ToxR as shown by crp⁻/490. Growth phase-dependent regulation was most obvious in wt/490 in the presence of both CRP and ToxR. The β-galactosidase activities of pT490D3, however, showed a different profile. Basal level transcription (toxR⁻crp⁻/490D3) was similar to that of 490 and so was the repression by ToxR (crp⁻/490D3), as the ToxR binding site was still intact in this fusion. Yet CRP failed to activate transcription from this fusion, presumably because of the loss of direct CRP interaction with CRP site 1, as fusion 490 only differed from 490D3 by containing the 22 bp CRP binding site and an additional 9 bp to the left and 5 bp to the right of this site. Furthermore, a decrease in β-galactosidase activity was observed (toxR⁻/490D3), suggesting CRP-dependent repression of this fusion and the possibility that CRP interacts with the rest of the promoter in the absence of CRP site 1. This lack of activation by CRP also led to loss of transcriptional activity in the presence of both ToxR and CRP (compare wt/490D3 with wt/490).

Fig. 3.

Cis elements for CRP-dependent activation and repression of ompT transcription.

A. Regulatory elements of the ompT promoter. Three putative CRP binding sites are underlined. The −35 and −10 region of the promoter are represented by small case letters. The ToxR binding site is in bold. The bases that are deleted in fusion 490D3 are italicized.

B. β-Galactosidase activities (Miller units) of the full-length ompT–lacZ fusion (490) and the same fusion with an internal deletion of CRP site 1 (490D3) in SCL100 (toxR⁻crp⁻), EK307 (toxR⁻), KSK377 (crp⁻) and 395 (wt). Open bars indicate transcription activities at mid-log phase (2 h), and solid bars indicate transcription activities at late log phase (4 h).

C. β-Galactosidase activities of ompT promoter deletion fusions in KSK377 (open bars) and KSK377 complemented with pCYL51 containing crp (solid bars). All the fusions have a 3′ end-point at +104 of the promoter. The 5′ ends of the fusions are at −490, −165, −57 and −44 as indicated.

These results clearly demonstrate that the CRP binding site at −310 of the ompT promoter plays a critical role in the activation of ompT transcription. In the absence of this site, CRP seems to interact with the rest of the promoter in a different fashion so that it functions as a repressor. This is consistent with the existence of two more putative CRP binding sites centred at −85 and −7, respectively, of the promoter region (Fig. 3A). To localize the minimal sequences that mediate CRP-dependent repression of ompT transcription in the absence of CRP site 1, transcription from several smaller ompT promoter–lacZ fusions in the absence and presence of CRP were compared in strain KSK377 (crp⁻) (Fig. 3C). In this strain, a minimal promoter fusion with 5′ end at −44 exhibited little β-galactosidase activity, whereas extension of the 5′ end to −57 gave an activity similar to that of the full-length fusion in SCL100 (toxR⁻crp⁻/490, Fig. 3B). This indicates that the sequences between −44 and −57 of the ompT promoter are sufficient for basal level transcriptional activation by RNAP. This basal level activity was not affected by the presence of ToxR in KSK377, as the ToxR binding site (−30 to −95) was disrupted. Extension of the 5′ end of the fusion to −165 and −490 led to almost complete repression of transcription, presumably mediated by ToxR (compare among the white bars). Complementation of KSK377 by plasmids encoding crp had different effects on different fusions (solid bars). For the full-length fusion (−490), overexpression of CRP from the plasmid overcame the repression effect of chromosomally encoded ToxR and led to high levels of β-galactosidase activity, consistent with the function of CRP as an activator. However, approximately fourfold repression by CRP was observed for fusion −57, which only contains CRP site 3 at −7, suggesting that this site is sufficient to mediate CRP-dependent repression. It is noteworthy that fusion −165 was already repressed sixfold by ToxR in the absence of CRP compared with fusion −57. This fusion contains putative CRP site 2 at −85 in addition to site 3 at −7. Slight but repeatable repression by CRP was observed for this fusion in addition to the ToxR-mediated repression. These results confirm our observation that, without CRP site 1, CRP functions as a repressor rather than an activator for the same promoter, through interaction with the downstream CRP binding site(s). However, in addition to CRP site 1, whether the other two CRP binding sites are also required for CRP-dependent activation of ompT transcription remains to be established.

Direct interaction of CRP with the ompT promoter

To demonstrate the direct interaction of CRP with the multiple CRP binding sites of the ompT promoter in vitro, gel mobility shift assays were performed using purified CRP protein from V. cholerae. As the CRP binding sites on the ompT promoter were predicted based on the consensus CRP binding site from E. coli, the promoter region of the flhDC gene (encoding the master regulator of the flagella operon), which is known to interact directly with CRP, was used as a positive control. V. cholerae CRP was shown to bind specifically to a 200 bp fragment of this promoter that contains a consensus CRP binding site (Soutourina et al., 1999) (Fig. 4A). A fragment containing −490 to +104 of the ompT promoter (490) and the same promoter fragment with an internal deletion of CRP site 1 (490D3) were then incubated with different amounts of V. cholerae CRP protein (Fig. 4B). At 2 μM CRP, fragment 490 was completely shifted. This is most probably mediated by CRP site 1 as, at the same concentration of CRP, fragment 490D3 was not shifted. When the CRP concentration was increased fourfold, complete shift of 490D3 was also observed, suggesting weaker binding of CRP to the downstream sites. For this fragment, intermediate species of CRP–DNA complexes were also observed, indicating the sequential binding of CRP site 2 and 3. These results corroborate the direct interaction of CRP at the three sites on the ompT promoter.

Fig. 4.

Binding of V. cholerae CRP to the promoter region of ompT.

A. Specific binding of V. cholerae CRP to an E. coli CRP binding site. The promoter region of flhDC (PflhDC), which is known to interact directly with CRP, was used as a positive control, and a NotI–Sau3A fragment from pCR-TOPO-Blunt (vector) was used as a negative control to test the ability of purified V. cholerae CRP to bind specifically to the E. coli CRP binding site.

B. A DNA fragment containing −490 to +104 of the ompT promoter (490) and the same fragment with a deletion of the CRP site 1 at −310 (490D3) were used as probe. The filled bars indicate the CRP sites present in the two fragments. CRP concentrations (μM) are indicated in both blots.

Direct interactions of CRP with RNAP at the ompT promoter

Two mechanisms account for CRP-dependent transcription from simple promoters that only have one CRP binding site (reviewed by Busby and Ebright, 1999). At class I promoters, the location of the CRP binding site varies: it can be centred near −93, −83, −72 or −62. A surface-exposed loop (activation region 1, or AR1) of CRP bound to a site at any of these locations can interact directly with the C-terminal domain of the α-subunit (α-CTD) of RNAP, thereby stabilizing the binding of RNAP to promoter DNA (Niu et al., 1996). At class II promoters, the CRP binding site overlaps the site for RNAP, apparently replacing the −35 element. In addition to AR1, another region of CRP called activation region 2 (AR2) is required for the activation of transcription (Rhodius et al., 1997). At such promoters, AR1 functions the same as at a class I promoter by recruiting RNAP, whereas the interaction between AR2 and the N-terminal domain of the α-subunit (α-NTD) of RNAP facilitates open complex formation. Various combinations of these two mechanisms can be found in promoters with multiple CRP binding sites. They show synergistic activation by two CRPs that both function through a class I mechanism or a combination of class I and class II mechanisms (reviewed by Rhodius and Busby, 1998). At some promoters, however, CRP functions not through direct contact with RNAP, but via direct protein–protein interaction with a second activator, which then interacts directly with RNAP. At such promoters, the AR1 and AR2 interactions play little or no role in transcriptional activation (Rhodius and Busby, 1998).

No previous study has been reported how CRP can activate transcription from as far away as −310, as occurs in the ompT promoter. We investigated whether CRP-dependent ompT activation involves direct interaction of CRP with RNAP through AR1 and AR2. As shown in Fig. 5A, the growth phase-dependent expression of OmpT in the wild-type strain 395, which was abolished in the crp mutant KSK377, was rescued when either V. cholerae CRP (pCYL51) or E. coli CRP (pDCRP) was introduced on a plasmid. However, plasmids expressing E. coli CRP with a single amino acid mutation in AR1 (pDCRP158A) or AR2 (pDCRP101E) failed to complement OmpT expression of this strain. Similar results were obtained when β-galactosidase activities of the full-length ompT–lacZ fusion pTA-490 were measured in these strains. As shown in Fig. 5B, wild-type E. coli CRP could complement the V. cholerae crp mutant in activating transcription from the ompT–lacZ fusion, whereas a single amino acid mutation in AR1 and, to a lesser extent, AR2 abolished this activation. These results demonstrate that both AR1 and AR2 are required for CRP-dependent activation of ompT transcription, presumably through direct CRP–RNAP interaction.

Fig. 5.

Both AR1 and AR2 are required for CRP-dependent activation of ompT transcription.

A. Western blot detection of OmpT expression level in 395 (wt), KSK377 (crp⁻) and KSK377 complemented with pCYL51 (V. cholerae CRP), pDCRP (E. coli CRP), pDCRP158A (AR1) or pDCRP101E (AR2). Cells were grown in LB broth at 37°C and harvested at 2 h, 4 h and 8 h. OmpT levels were detected as described above.

B. β-Galactosidase activities (Miller units) of the full-length ompT–lacZ fusion (−490 to +104) were measured in 395 (wt), KSK377 (crp⁻), KSK377 complemented with pDCRP (CRP), KSK377 complemented with pDCRP158A (AR1) or KSK377 complemented with pDCRP101E (AR2).

C. β-Galactosidase activities of the ompT–lacZ fusions −57 to +104 and −165 to +104 were measured in KSK377 (crp⁻) or KSK377 complemented by different CRP constructs from E. coli as described above. The presence of different CRP binding sites in each fusion is indicated underneath (B and C).

Figure 5C shows the effect of AR1 and AR2 mutations on the CRP-dependent repression of transcription from fusions pTA-57 and pTA-165 (in the absence of CRP site 1). In contrast to the situation for activation, CRP-dependent repression only involves AR1, as CRP with a mutation in AR2 still repressed transcription from these fusions. These results are consistent with the hypothesis that, in the absence of CRP site 1, CRP interacts with RNAP differently, resulting in repression rather than activation of transcription. The mechanism by which the binding of CRP to −310 contributes to a differential CRP–RNAP interaction at the promoter region, which probably involves DNA looping, remains to be investigated.

Discussion

Co-ordinated regulation of the two branches of the ToxR regulon by CRP

The ToxR regulon has been recognized as a paradigm for co-ordinate regulation of gene expression in bacterial pathogens, yet how ToxR contributes to environmental sensing is poorly understood (Dziejman et al., 1999). For the ToxT-dependent branch, it has become clear that, in addition to ToxR, a variety of other regulators are involved in coupling environmental cues to different levels of the cascade (Fig. 6A). AphA and AphB, the latter being a LysR family regulator, were identified as activators of TcpP (Kovacikova and Skorupski, 1999). They couple pH and temperature signals to the expression of CT and TCP via TcpP and ToxT (Carroll et al., 1997); cAMP–CRP couples carbon source to CT and TCP expression by binding to the promoter of tcpP and interfering with AphAB-dependent activation (Kovacikova and Skorupski, 2001). Much less is known about the combinatorial control of the ToxT-independent branch of the ToxR regulon, except that ToxR may act as a sensor for osmolarity (Miller and Mekalanos, 1988). This study shows for the first time that OmpT expression is substantially regulated by growth phase in rich medium and that ToxR is not the only factor responsible for coupling environmental signals to OmpT expression. Expression of OmpT was also found to be downregulated by acid shock in a ToxR-and CRP-independent manner, suggesting the existence of another yet to be identified regulator (Merrell et al., 2001). Therefore, a complex interaction among ToxR, CRP and perhaps an additional factor(s) seems to be required to fine tune the expression of OmpT.

Fig. 6.

A. Co-ordinate expression of CT, TCP, OmpU and OmpT by ToxR and CRP. ToxR directly activates OmpU expression and represses OmpT expression. It is also an indirect activator of CT and TCP gene expression. CRP directly activates OmpT expression and indirectly represses CT and TCP expression. Some other well-characterized intermediate regulators, AphAB, TcpP and ToxT, are shown. +, activation of trancription; −, repression of transcription.? indicates that the mechanism of signal sensing is not known.

B. A simplified model of the regulation of the ompT promoter by CRP and ToxR. Activation of ompT transcription requires the formation of a nucleoprotein complex containing multiple CRP molecules and some unknown factor(s) (X). Repression of ompT transcription involves disruption of the activation loop by ToxR bound between CRP site 1 (−310) and the rest of the promoter, overlapping CRP binding site 2 (−85). Environmental signals affect ompT transcriptional level by modulating the relative contribution of CRP and ToxR to transcription directly or indirectly.

The involvement of CRP in the regulation of OmpT is consistent with the function of CRP as a prominent global regulator in enteric bacteria that controls the expression of a wide variety of genes in response to carbon and energy sources in the environment (reviewed by Botsford and Harman, 1992). It is noteworthy that, although ToxR activates CT and TCP but represses OmpT, CRP does exactly the opposite, i.e. represses CT and TCP but activates OmpT (Fig. 6A). Apparently, V. cholerae adopts an efficient system to co-ordinate the expression of CT and TCP with that of OmpT. The molecular mechanism of this highly reciprocal regulation is of great interest. The advantage of this co-ordinate regulation can only be speculated on at this time, as the roles of OmpU and OmpT in the fitness and pathogenesis of V. cholerae have not been fully elucidated (Provenzano et al., 2001).

Despite the low level of OmpT expression observed under laboratory conditions in classical strain 395, post-challenge sera from volunteers infected with this strain strongly recognize OmpT (Sperandio et al., 1995). The observation that OmpT is significantly upregulated in minimal medium and late in the growth phase suggests that OmpT plays an important role in the survival of V. cholerae under stressed conditions, both in vivo and in vitro. Our attempts to investigate the in vivo expression of ompT using recombination-based in vivo expression technology (RIVET) (Lee et al., 1999) have been unsuccessful (data not shown). However, in a screen of V. cholerae strains for OmpT expression using OmpT antiserum, all the 11 clinical strains express the same size OmpT as classical strain 395, compared with only four out of 20 non-pathogenic environmental strains. The remaining environmental strains either express a truncated form of OmpT (six out of 20) or produce no detectable OmpT (C. C. Li and J. B. Kaper, unpublished). These results suggest an evolutionary pressure for the pathogenic strains to maintain a conserved OmpT sequence; therefore, strictly regulated expression of OmpT might contribute to V. cholerae pathogenicity.

Mechanism of CRP-dependent activation at the ompT promoter

The genetic data we obtained, using isogenic mutants of classical strain 395, establish the critical role of CRP in the activation of ompT transcription and reveal that the ompT promoter is not subjected to simple negative regulation by ToxR. Yet the mechanism of CRP-dependent activation of ompT transcription demonstrates several levels of complexity.

Our data demonstrate that CRP activates transcription from −310 of the ompT promoter. This is a substantially greater distance than any other known CRP-dependent promoter. Interestingly, the closest homologue of OmpT, OmpH in the deep-sea bacterium Photobacterium profundum, was also found to be under catabolite repression and a putative CRP binding site was predicted at −330 of the ompH promoter, although functional data have not yet been reported (Bartlett and Welch, 1995). The greatest distance reported previously was for the papA promoter of uropathogenic E coli, in which binding of CRP centred at −215.5 of the promoter is essential for the activation of transcription (Weyand et al., 2001). The formation of a DNA loop involving two additional activators, Lrp and PapI, helps to eliminate the distance between CRP bound at −215.5 and the transcription apparatus, so that AR1 of CRP can contact the α-CTD of RNAP directly. The ability of transcriptional regulatory proteins to act at a distance via DNA looping is characteristic of both prokaryotic and eukaryotic promoters (Schleif, 1992). CRP bound to −310 of the ompT promoter has the potential to be brought closer to the transcription apparatus by intrinsic DNA bending (Bellomy and Record, 1990) or by additional scaffolding proteins, similar to the situation at the papA promoter. Our initial in vitro transcription assay suggests that CRP-dependent activation in vivo might involve an additional factor(s) (data not shown). Identification of the putative factor(s) that mediates DNA looping will be critical to elucidate the mechanism by which CRP communicates to the transcription apparatus from −310 of the ompT promoter.

The understanding of CRP-dependent regulation of ompT transcription is further complicated by the existence of two more downstream CRP binding sites. Although both AR1 and AR2 are required for CRP-dependent activation of ompT transcription from CRP site 1, CRP-dependent repression of the same promoter in the absence of this site only requires AR1. These results are consistent with the following model of differential regulation of the ompT promoter by CRP. In the absence of CRP site 1 or when CRP cannot bind to this site as a result of certain environmental regulation, CRP bound to the promoter proximal site(s) interacts with RNAP through AR1. This interaction may facilitate the closed complex formation, but prevents transcription initiation. In the presence of CRP site 1 and possibly an additional scaffolding protein(s), the CRP dimer bound to this site either contacts RNAP directly or affects the existing CRP–RNAP interaction, so that an AR2–RNAP interaction can occur, which promotes open complex formation and transcription initiation. However, further studies are necessary to determine the relevance of CRP-dependent repression to the regulation of OmpT expression, and to differentiate the relative contribution of CRP bound to any one of these sites to activation or repression. Specifically, CRP mutants with altered binding preferences will be useful in demonstrating whether CRP contacts RNAP directly from −310 of the ompT promoter during activation.

ToxR as a repressor of ompT transcription

Our genetic data suggest that ToxR-dependent repression of ompT transcription involves antagonizing the activator CRP. As ToxR was shown to bind to the −30 to −95 region of the ompT promoter (Li et al., 2000), the current model of CRP-dependent activation of ompT transcription also implies that ToxR may function as an antiactivator by disrupting the formation of a higher order activation complex involving CRP. This represents a new feature of ToxR as a transcriptional regulator. The study of the mechanisms of ToxR-dependent regulation of different promoters has been focused on how ToxR binds to the four target promoters and the potential interactions between ToxR and RNAP that will facilitate or inhibit transcription (Crawford et al., 1998; Li et al., 2000). Krukonis et al. (2000) recently showed that, at the toxT promoter, ToxR acts as a co-activator of TcpP, demonstrating for the first time the possibility that ToxR-dependent transcription regulation involves other transcription factors in addition to RNAP. Our finding that ToxR represses ompT transcription through an antiactivation mechanism further illuminates the important aspect of ToxR as a global transcription regulator. That is, similar to the strategy of many eukaryotic transcription factors, ToxR enhances its capacity to co-ordinate gene expression at multiple promoters by interacting with different partners at different promoters. Therefore, ToxR not only acts as an activator and a co-activator, but also as a repressor and an antiactivator.

In conclusion, the ompT promoter represents another example of a prokaryotic promoter under complex control by multiple factors. The experiments reported here confirm and extend the environmental regulation of OmpT and demonstrate the fundamental features of the regulation of ompT transcription by CRP and ToxR. As summarized in Fig. 6B, a basal level of transcription from the ompT promoter is mediated by the sequences between −44 and −57 of the ompT promoter. In the absence of ToxR or under inducing conditions, CRP further activates ompT transcription through a looping mechanism in which CRP bound at a site at −310 can be brought close enough to influence RNAP bound at the ompT promoter. Under repressing conditions, ToxR bound to −30 to −95 of the promoter disrupts the activation loop formed by CRP and possibly some unidentified factor(s) and abolishes the basal level ompT transcription. The existence of an unknown regulator(s) at the ompT promoter is consistent with the observation that neither CRP nor ToxR alone seems to respond to the growth phase-dependent signal(s) that modulate(s) OmpT expression. It is possible that other factors act as direct sensors for various environmental signals and regulate the relative contribution of CRP and ToxR to ompT transcription through protein–protein interaction, resulting in a larger range of fluctuation in ompT transcription level than either regulator can achieve alone.

Experimental procedures

Bacterial strains and plasmids (Table 1)

Table 1.

Strains and plasmids used in this study.

V. cholerae strains	Description	Source or reference
395	O1 classical	Laboratory collection
KSK377	395, crp::kan	Skorupski and Taylor (1997b)
EK307	395, Δ toxR	Krukonis et al. (2000)
SCL100	EK307, crp::kan	This study
Plasmids
pACYC184	Low-copy-number cloning vector, TetR CmR	New England Biolabs
pCYL51	V. cholerae crp cloned into pACYC184	This study
pDCRP	Wild-type E. coli crp cloned into pBR322	Bell et al. (1990)
pDCRP159A	pDCRP with defective activation region 1 (AR1)	West et al. (1993)
pDCRP101E	pDCRP with defective activation region 2 (AR2)	Rhodius et al. (1997)
pGEM-T	PCR cloning vector, ApR	Promega
pGMT-490	−490 to +104 of PompT cloned into pGEM-T	Li et al. (2000)
pGMT-490D3	pGMT-490 with an internal deletion of CRP site 1	This study
pTL61T	lacZ transcriptional fusion vector; ApR	Linn and Pierre (1990)
pT-44	−44 to +104 of the PompT cloned into pTL61T	Li et al. (2000)
pT-57	−57 to +104 of the PompT cloned into pTL61T	This study
pT-165	−165 to +104 of the PompT cloned into pTL61T	Li et al. (2000)
pT-490	−490 to +104 of the PompT cloned into pTL61T	Li et al. (2000)
pTA-57	ompT:: lacZ of pT-57 cloned into pACYC184	This study
pTA-165	ompT:: lacZ of pT-165 cloned into pACYC184	This study
pTA-490	ompT:: lacZ of pT-490 cloned into pACYC184	This study

Escherichia coli DH5α was used for cloning. All plasmids were introduced into V. cholerae strains by electroporation. V. cholerae strains used in this study are isogenic with the classical strain 395 or the El Tor strain C6709-1: KSK377 (395 crp::kan81) (Skorupski and Taylor, 1997b), EK307 (395ΔtoxR) (Krukonis et al., 2000). The crp::kan81 locus in KSK377 was also introduced into EK307 in this study by CP-T1-mediated transduction (Hava and Camilli, 2001) to obtain SCL100 (395 ΔtoxR crp::kan81). Strains were grown in LB broth at 37°C, unless otherwise specified. Antibiotics were used at the following concentrations: ampicillin, 100 μg ml⁻¹; kanamycin, 20 μg ml⁻¹, except for the selection of transductants, in which 50 μg ml⁻¹ was used; chloramphenicol, 10 μg ml⁻¹.

SDS–PAGE and immunoblotting

Overnight cultures of V. cholerae strains were diluted 1 : 100 and grown at 37°C or 30°C in LB broth, M9 glycerol medium or in 1% tryptone broth containing various amounts of NaCl, KCl or melibiose for assessing the effect of osmolarity. Cells were collected at mid-logarithmic phase (2 h, OD₆₀₀ ≈ 0.2), late logarithmic phase (4 h, OD600 = 0.5–0.8) and stationary phase (8 h, OD >.5). Samples were normalized by OD₆₀₀ and protein concentration. Equal amounts of protein were separated by 14% SDS–PAGE and either stained with Coomassie brilliant blue or transferred to polyvinylidene difluoride (PVDF) membrane (0.45 μm; Millipore) and probed with OmpT polyclonal antisera using the enhanced chemiluminescence (ECL) detection system (Amersham Pharmacia).

RNase protection assays

Total RNA was harvested from strains grown in LB broth at 2 h, 4 h and 8 h as described previously (Merrell and Camilli, 1999). A 344 bp ompT riboprobe template was generated by PCR using primers OmpTF (5′-ATTGGTTCTGGTTCT TCG-3′) and OmpTR (5′-TTTGCATTATCTTCTGGA-3′). The amplification product was ligated to pGEM-T (Promega), proper orientation was confirmed and the riboprobe was synthesized using the Maxiscript kit (Ambion) and 50 μCi of [³²P]-UTP (NEN) as described previously (Merrell and Camilli, 1999). RPAs were done using the RPAII kit with 1 μg of RNA as described by the manufacturer (Ambion). The products of RNase protection were separated on 4% denaturing polyacrylamide gels and exposed to phosphor screens (Kodak). Quantification and peak analysis of bands were conducted using a phosphorimager and the IMAGEQUANT program (Molecular Dynamics).

β-Galactosidase assays on ompT–lacZ fusions

ompT–lacZ fusions (pT-490, −165, −57 and −44) with different 5′ end-points and the same 3′ end-points at +104, relative to the transcriptional start site of ompT, were constructed as described previously (Li et al., 2000). To construct pT-490D3, which is pT-490 with an internal deletion of CRP site 1, the plasmid pGEM-T (Promega) containing the ompT promoter (−490 to +104) was used as the template in an inverse PCR reaction. The two PCR primers, K1591 (5′ACTGAACCTTCAGAGACTTA-3′) and K1594 (5′-AGCATAC GATTTCGTGAAGT-3′), flank the putative CRP binding site and read outwards. The ≈ 4 kb PCR product was self-ligated and transformed into DH5α. After confirming the deletion of the CRP site in the resulting plasmid by DNA sequencing, the promoter fragment was recovered as a XbaI–HindIII fragment and inserted into pTL61T (ColE1 origin of replication) as described previously (Li et al., 2000). To study the effect of AR1 and AR2 mutations in CRP on the transcription of ompT, pBR322-based plasmids (ColE1 origin of replication) containing crp were introduced. Compatible fusion plasmids pTA-490, −165 and −57 were constructed by subcloning of a ScaI–NruI fragment containing terminator T14::ompT::lacZ from plasmids pT-490, pT165 and pT-57, respectively, into the EcoRV site of pACYC184 (p15A origin of replication). β-Galactosidase activities of these fusions in different genetic backgrounds were measured as described previously (Li et al., 2000), except that cells were collected at different points on the growth curve as necessary.

Gel mobility shift

Vibrio cholerae CRP protein was purified from KSK377 harbouring plasmid pCYL51 according to the protocol provided by R. Ebright (personal communication) using a 5 ml cAMP affinity column (Sigma) and a Biogel P6DG desalting column (Qiagen). The purity of the protein was verified on an SDS–PAGE gel, and the protein concentration was measured by Bio-Rad protein assay. DNA probes were radiolabelled by Klenow filling reactions as described previously (Li et al., 2000). Binding reactions (10 μl) contain 1× transcription buffer (Ambion), 100 mM KCl, 0.01 mg ml⁻¹ poly-(dI–dC), 0.25 mM cAMP, 3000 c.p.m. of labelled probe and various amounts of CRP. The reactions were carried out at 37°C for 30 min, separated on a 5% non-denaturing polyacrylamide gel and visualized by autoradiography.