Integration of Cyclic di-GMP and Quorum Sensing in the Control of vpsT and aphA in Vibrio cholerae

Abstract

Vibrio cholerae transitions between aquatic environmental reservoirs and infection in the gastrointestinal tracts of human hosts. The second-messenger molecule cyclic di-GMP (c-di-GMP) and quorum sensing (QS) are important signaling systems that enable V. cholerae to alternate between these distinct environments by controlling biofilm formation and virulence factor expression. Here we identify a conserved regulatory mechanism in V. cholerae that integrates c-di-GMP and QS to control the expression of two transcriptional regulators: aphA, an activator of virulence gene expression and an important regulator of the quorum-sensing pathway, and vpsT, a transcriptional activator that induces biofilm formation. Surprisingly, aphA expression was induced by c-di-GMP. Activation of both aphA and vpsT by c-di-GMP requires the transcriptional activator VpsR, which binds to c-di-GMP. The VpsR binding site at each of these promoters overlaps with the binding site of HapR, the master QS regulator at high cell densities. Our results suggest that V. cholerae combines information conveyed by QS and c-di-GMP to appropriately respond and adapt to divergent environments by modulating the expression of key transcriptional regulators.

INTRODUCTION

Bacteria use multiple signaling pathways to monitor and respond appropriately to changing surroundings. Small-molecule chemical signals convey information about the presence, nature, number, and characteristics of the surrounding bacterial species as well as the composition of the environment. Proper responses to changing environments are vital to the survival of bacteria. Vibrio cholerae, the causative agent of cholera, alternates between a motile, virulent state within the host and a sessile, biofilm state in aquatic environmental reservoirs (15). Quorum sensing (QS) and cyclic di-GMP (c-di-GMP) signaling are two chemical signaling systems that control this transition (19).

QS allows bacteria to sense the population density and species composition of the surrounding bacterial consortium through the secretion and detection of chemical signals called autoinducers so as to collectively control behaviors (46). In V. cholerae, in the high-cell-density QS state, both biofilm formation and virulence factor expression are repressed (21, 31). c-di-GMP is a nearly ubiquitous bacterial second messenger that induces biofilm formation and represses motility (19). In contrast to QS, c-di-GMP activates the expression of genes necessary for biofilm formation in V. cholerae (3). However, like QS, c-di-GMP is thought to repress the expression of virulence factors (38, 42).

The QS regulatory pathways that control biofilm formation and virulence factor expression have been largely elucidated. HapR, the master high-cell-density regulator of the QS signaling cascade, represses biofilm formation by directly binding to the biofilm activator vpsT and inhibiting its transcription (45). Additionally, HapR production reduces intracellular c-di-GMP levels (45). Inhibition of virulence factor expression by QS is mediated by HapR repression of the virulence gene-activating protein AphA (26). Interestingly, AphA is also the master regulator of the QS low-cell-density state in V. cholerae and Vibrio harveyi (33). Although much is known about QS control of biofilms and virulence factor expression, the molecular mechanism by which c-di-GMP controls biofilm formation and virulence factor expression is less well understood.

Biofilm formation is important for the survival of V. cholerae in aquatic reservoirs and for disease transmission (50). V. cholerae forms biofilms on biotic surfaces of chitinous animals, such as copepods and zooplankton (22). Biofilms are critical for the development of conditionally viable but nonculturable V. cholerae cells (conditionally viable environmental cells [CVEC]), as evidenced by the fact that mutants with defects in biofilm formation are unable to enter the CVEC state (15, 23). CVEC are thought to be important for the spread of disease, because this state increases the ability of V. cholerae to survive stress and starvation outside the host (15). Biofilm formation also increases the resistance of V. cholerae to the acidic environment in the stomach, which is essential for the passage of bacteria to the small intestine (50). Although mutants defective in biofilm formation are not compromised for colonization in a murine or rabbit disease model, stool from V. cholerae patients contains a mixture of infectious biofilm-like aggregates along with planktonic cells, suggesting that biofilm formation occurs in vivo (15). Furthermore, in vitro biofilm formation increases the infectivity of V. cholerae (39).

The development of biofilms in V. cholerae relies on the expression of two linked operons, termed vpsI and vpsII, that encode proteins essential for exopolysaccharide (EPS) production. Expression of these operons is increased at high c-di-GMP levels (28, 49). At least two transcriptional activators—VpsT, a member of the LuxR family of transcriptional regulators, and VpsR, an NtrC-like transcriptional regulator—positively control biofilm development in V. cholerae by activating the expression of the vps operons (8, 48). We and others have shown that vpsT expression is induced at high c-di-GMP levels and is repressed at a high cell density by the QS regulator HapR (16, 45). Recently, the VpsT protein was shown to be activated upon binding to c-di-GMP, resulting in increased expression of the vpsI and vpsII operons and repression of flagellar assembly genes (27). The mechanism by which vpsT expression itself is induced by c-di-GMP remains unknown; however, it was hypothesized that VpsT bound to c-di-GMP might induce its own expression via a positive-feedback loop (27).

Other transcriptional activators that have been shown to bind to c-di-GMP and regulate gene expression are FleQ and Clp, from Pseudomonas aeruginosa and Xanthomonas campestris, respectively (9, 20). Regulation of gene expression by c-di-GMP has been shown to occur in many bacterial species; however, the molecular mechanisms responsible for this regulation have been elucidated for only a subset of genes (19). Further characterization of transcriptional effectors that bind c-di-GMP in V. cholerae and other bacteria will shed light on how c-di-GMP controls important phenotypes, such as biofilm formation and virulence factor expression.

Here we report that aphA expression is induced by c-di-GMP. We determine that this induction occurs via a common regulatory mechanism encoded in both the aphA and vpsT promoters that integrates QS and c-di-GMP signaling. Induction of vpsT and aphA is independent of VpsT but requires VpsR, a transcriptional activator that directly binds the promoters of these genes. VpsR binds c-di-GMP with a dissociation constant (K_d) of 1.6 μM. Both vpsT and aphA are directly repressed by HapR at a binding site that overlaps with the VpsR binding site (29, 45). Finally, we identify additional promoters activated by c-di-GMP that are not dependent on VpsR, suggesting the existence of multiple signal transduction pathways linking c-di-GMP to the regulation of gene expression in V. cholerae.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

All strains, primers, and plasmids used in this study are listed in Tables S1 and S2 in the supplemental material. The V. cholerae strains used in the study were derived from El Tor biotype strain C6706str2 and contain a mutation in vpsL (45). This mutation renders V. cholerae unable to make biofilms. Therefore, even at high c-di-GMP levels, the cells do not flocculate, enabling accurate readings of reporter gene expression. Strains carrying mutations in lrp, vpsT, and vpsR were constructed using the pKAS32 suicide vector as described previously (36). This procedure generated unmarked deletions of the entire coding sequences of vpsT and vpsR. To generate the Δlrp mutant, a gene encoding resistance to tetracycline, tetA, was amplified from pBR322 and was inserted between the DNAs harboring the upstream (550-bp) and downstream (500-bp) lrp-flanking regions in pKAS32. Selection for crossover events was performed in the presence of a plasmid expressing wild-type (WT) lrp (pDS15). These steps were performed because we were unable to generate an lrp deletion mutant by using the standard protocol described by Skorupski and Taylor (36). pDS15 was cured prior to further analysis, and the deletion mutation was stably maintained in the absence of pDS15. A ΔalsS mutant was constructed by natural transformation and homologous recombination of a PCR product harboring the cat open reading frame (ORF) from pKD3 flanked by FLP recombination target (FRT) sites, which was fused between 500 bp upstream and downstream of the alsS coding sequence (10). This method created a chloramphenicol-resistant alsS mutant. The unmarked deletion mutant was generated by flipping out the cat gene by use of plasmid pTL17, which overexpresses the flippase enzyme (30). pTL17 was cured before any assays were performed, and pCMW75 was introduced through biparental mating. All V. cholerae strains were grown in Luria-Bertani (LB) medium. Antibiotics were obtained from Sigma and were used at the following concentrations (in micrograms per milliliter), unless stated otherwise: ampicillin, 100; kanamycin, 100; chloramphenicol, 10; tetracycline, 10. Escherichia coli BW29427 and S17-λpir were used as the donors were in biparental conjugation to mobilize plasmids into V. cholerae (12, 32).

Molecular methods.

DNA manipulation was performed using standard procedures (35). T4 DNA ligase and restriction enzymes were purchased from New England BioLabs (NEB) and Stratagene. PCRs were performed with iProof DNA polymerase (NEB) and Phire DNA polymerase (Finnzymes). Promoter deletion constructs were cloned into the SpeI and BamHI restriction sites of pBBRlux (17) (for primer sequences, see Table S2 in the supplemental material). lrp, vpsT, and vpsR were cloned into pKAS32 using the primers listed in Table S2 in the supplemental material. Overexpression constructs for protein purification were engineered into pTXB1 for HapR, Lrp, and YcgR (Impact protein purification system; NEB) and into pET28b (Novagen) for VpsR according to the manufacturer's instructions.

Screen to identify V. cholerae c-di-GMP-responsive promoters and measurement of gene expression.

A V. cholerae promoter library was previously constructed by inserting SauIIIA-restricted genomic DNA fragments into the BamHI restriction site of pBBRlux (17). From approximately 150,000 clones, 5,000 inserts that encoded promoters as determined by expression of the luciferase operon were rearrayed to generate a V. cholerae promoter-enriched library (17). These clones were pooled, and the plasmids were isolated, retransformed into an E. coli donor strain, and ultimately conjugated into V. cholerae strain CW2034 containing plasmid pCMW75 (45). pCMW75 overexpresses qrgB, a V. harveyi GGDEF protein, under the control of the Ptac promoter (45). Induction of QrgB with isopropyl-β-d-thiogalactopyranoside (IPTG) produces high levels of c-di-GMP in V. cholerae (45). The bioluminescence of 960 isolates from the promoter-enriched library containing pCMW75 was measured on a plate reader (SpectraMax B5; Molecular Devices) in the presence and absence of 0.1 mM IPTG in LB medium after 7 h and was adjusted for growth by concurrent measurement of the optical density at 600 nm (OD₆₀₀). Clones showing significant changes in bioluminescence in response to IPTG were reassayed in triplicate. The fold change was calculated by dividing the induced values by the uninduced values (see Table 1). The inserted genomic DNA was sequenced using primers homologous to sequences upstream and downstream of the SpeI and BamHI restriction sites of pBBRlux. Gene expression studies were performed similarly by assessing the expression of promoter-lux reporter fusions in pBBRlux for aphA and vpsT (and their promoter deletion constructs) and for virulence genes in the presence of plasmids pCMW75 and pCMW98 in the uninduced versus induced (0.1 mM IPTG) state. pCMW98 carries a mutant allele for QrgB in which the active site has been mutated to the amino acid sequence AAEEF. This mutant is unable to synthesize c-di-GMP (45).

Table 1.

Cyclic di-GMP-regulated promoters identified in the screen

Promoter fragment	Gene	Protein description	Fold changea
1:B8	VCA0213	Hypothetical	2.8
1:F6	VC2647	AphA, a central regulator of virulence gene expression and quorum sensing	3.9
2:G12	VC2610	Hypothetical	1.3
4:H4	VC2108	Erythronate-4-phosphate dehydrogenase	1.8
5:A6	VC1899	Hypothetical	2.9
6:C9	VC1673	Transporter family	3.9
9:C11	VCA0055	Conserved hypothetical	4.8

^aCalculated as (luminescence value at high levels of c-di-GMP)/(luminescence value at low levels of c-di-GMP).

5′ RACE for identification of the vpsT start site.

RNA was prepared from WT V. cholerae cultures using the RNeasy RNA extraction kit (Qiagen). The transcription start site of vpsT was determined by rapid amplification of 5′ cDNA ends (5′ RACE) (Invitrogen) according to the manufacturer's instructions. The primers used were vpsT-GSP1 and vpsT-GSP2 (see Table S2 in the supplemental material).

Purification of proteins and EMSA.

HapR, Lrp, and YcgR were purified with the Impact protein purification system using the pTXB1 plasmid as described in the manufacturer's instructions (NEB). Purified proteins were stored in 20 mM Tris (pH 7.5), 1 mM EDTA, and 10 mM NaCl, in 20% glycerol. VpsR was purified, by using the pET28B plasmid according to the manufacturer's instructions (Novagen), as a C-terminally 6× histidine tagged fusion protein and was dialyzed in storage buffer (20 mM Tris [pH 7.5], 1 mM EDTA, 1 M NaCl, 0.1 mM dithiothreitol [DTT]). DNA probes for electrophoretic mobility shift assays (EMSA) were prepared using PCR amplification with primers CMW234 and CMW235 (which are complementary to upstream and downstream sequences in pBBRlux that lie adjacent to the SpeI and BamHI restriction sites), tagged at the 5′ ends with FAM (6-carboxyfluorescein). The vpsT promoter fragment constructs shown in Fig. 3B were used as the templates to generate the probes. A 10 nM concentration of the probe was incubated at 30°C or 4°C for 30 min with HapR (42 nM to 800 nM) or VpsR (25 nM to 650 nM), respectively, and 1 μl dI-dC (1 mg/ml stock) in a final volume of 20 μl in the respective protein buffer (45). EMSA were performed on 5% polyacrylamide–Tris-borate-EDTA (TBE) gels, and the results were visualized using a Typhoon FLA 9000 scanner (GE Healthcare Life Sciences).

Fig. 3.

Architectures of the aphA and vpsT promoters. The VpsR, HapR, and Lrp binding sites, transcription start sites, and translation start sites for the aphA and vpsT promoters are shown. The filled bars underneath each promoter diagram indicate the promoter deletion constructs that were made in this study.

Acetoin production assay.

V. cholerae cultures (with or without 0.1 mM IPTG) were grown in MR-VP medium (1) from overnight cultures for 24 h with shaking at 37°C. The Voges-Proskauer test was performed with 100 μl of culture by addition of 30 μl solution A (5% naphthol) followed by addition of 10 μl solution B (40% KOH) (BD Life Sciences). After 5 min, the color development was quantified in a SpectraMax 96-well plate reader at 550 nm (43). Absorbance values at 600 nm were also recorded, and data were analyzed by use of the OD₅₅₀/OD₆₀₀ ratio. The experiment was repeated in triplicate.

Cyclic di-GMP binding assay.

The VpsR and YcgR proteins were purified using a 6× histidine tag with affinity purification and the Impact protein purification kit from NEB, respectively, according to the manufacturer's instructions. ³²P-labeled c-di-GMP was generated using the purified cytoplasmic portion of GGDEF VC2370, which does not contain the first 142 amino acid residues (VC2370−142) (11). Reaction mixtures with a total volume of 100 μl containing 10 μM VC2370 in buffer (75 mM Tris-Cl [pH 7.8], 250 mM NaCl, 25 mM KCl, 10 mM MgCl₂) with 12.5 μM [α-³²P]GTP (800 Ci/mmol; Perkin-Elmer) or 12.5 μM unlabeled GTP were incubated at room temperature for 30 min. Antarctic phosphatase (NEB) was added, and reaction products were incubated for an additional 30 min to remove any residual [α-³²P]GTP. Reaction products were then heated at 100°C for 5 min, subsequently cooled on ice, and spun at 15,000 × g for 10 min to remove denatured protein and collect the supernatants. The amount of c-di-GMP synthesized was assessed by analysis of the control reaction mixture using ultrahigh-pressure liquid chromatography (UPLC)–tandem mass spectrometry (MS-MS) and compared to known c-di-GMP standards (4). For binding reactions, 400 nM protein in binding buffer (10 mM MgCl₂, 20 mM Tris [pH 7.8], and 50 mM NaCl) was incubated with varying amounts of ³²P-labeled c-di-GMP (0.125 μM to 1.14 μM) in a 20-μl volume for 30 min at room temperature. Binding was assessed using a filter-binding technique (20). Preparations were loaded onto a nitrocellulose membrane (pore size, 0.2 μm; Whatman) through a vacuum slot blot (Hybri·Dot Manifold, catalog no. 1050MM; BRL). Sample wells were washed with 3 ml binding buffer in order to wash away unbound ³²P-labeled c-di-GMP. The membrane was removed and dried, and the bound ³²P-labeled c-di-GMP of individual wells was quantified by scintillation counting (cpm/min) (20). For competition experiments with unlabeled c-di-GMP and GTP, 3 μM unlabeled nucleotides were incubated with the preparations after 15 min of incubation with 1 μM ³²P-labeled c-di-GMP in a 20-μl volume and were processed similarly.

RESULTS

Identification of Vibrio cholerae c-di-GMP-responsive promoters.

To determine the molecular mechanism by which changes in the levels of c-di-GMP are coupled to the regulation of gene expression in V. cholerae, we performed a genetic screen to isolate promoters regulated by c-di-GMP. A library of random V. cholerae genomic fragments driving the expression of a promoterless luxCDABE operon in plasmid pBBRlux was used (17). c-di-GMP levels in the cell were modulated by introducing a plasmid encoding QrgB, a V. harveyi GGDEF protein, under the control of an IPTG-inducible Ptac promoter. Strains containing a ΔvpsL mutation were used in this study, because this mutation eliminates biofilm formation. This strategy is essential for accurate reading of reporter gene expression at high levels of c-di-GMP. Without the ability to form biofilms, the cells do not aggregate but remain as a well-dispersed planktonic culture. Induction of QrgB by IPTG leads to increased c-di-GMP levels in V. cholerae (45). Therefore, promoters induced by c-di-GMP show increased luciferase expression following IPTG addition, while promoters repressed by c-di-GMP show decreased luciferase expression following IPTG addition.

Changes in bioluminescence readings with and without IPTG were measured for 960 independent clones. From these, we identified 7 unique c-di-GMP-responsive promoters (Table 1). The expression of all of the promoters isolated in this screen increased in the presence of c-di-GMP. Four of these promoters drive genes encoding hypothetical proteins; one promoter is located in the ORF for VC1673, a putative transporter protein; one promoter drives the expression of VC2108, erythronate-4-phosphate dehydrogenase; and one promoter maps upstream of the virulence and QS regulator aphA (VC2647). Our screen was not carried to saturation and did not identify promoters that have been previously shown to be regulated by c-di-GMP (3, 27).

The transcription activator AphA, along with its coactivator, AphB, positively regulates virulence in V. cholerae by increasing the expression of tcpPH (25). TcpP, along with TcpH, induces toxT, which activates the expression of genes encoding the two major virulence factors of V. cholerae, cholera toxin (CT) and the toxin-coregulated pilus (TCP) (18). Earlier studies of c-di-GMP in V. cholerae have reported that c-di-GMP negatively regulates virulence (38), although the influence of c-di-GMP on aphA expression has not been examined. Thus, our observation that c-di-GMP activates aphA expression was surprising. However, consistent with our results, aphA functions as the master QS regulator at a low cell density in V. cholerae (33), a state in which the intracellular c-di-GMP concentration is relatively high (45).

To confirm that c-di-GMP induces the transcription of aphA, we reconstructed a transcriptional fusion of the aphA promoter in pBBRlux and examined its expression upon qrgB overexpression. As a control, we also examined expression following overproduction of a qrgB allele encoding a nonfunctional active site (GGEEF → AAEEF). This mutant protein has previously been shown to be incapable of c-di-GMP synthesis (45). Like that of the original aphA clone identified in the screen, expression of aphA was induced by c-di-GMP (Fig. 1). Furthermore, no induction of aphA occurred upon overexpression of the qrgB active-site mutant, confirming that induction occurs through synthesis of c-di-GMP (Fig. 1). Similarly, a construct with a transcriptional fusion of the vpsT promoter to luciferase in pBBRlux was induced by c-di-GMP upon overexpression of wild-type QrgB but not the QrgB active-site mutant derivative (Fig. 1), confirming our previous observation (45).

Fig. 1.

aphA and vpsT are induced at high levels of c-di-GMP. Luciferase production from aphA-lux and vpsT-lux following overexpression of a GGEEF enzyme (QrgB) or the corresponding QrgB AAEEF active-site mutant was determined in the ΔvpsL strain. Shaded bars, noninduced cultures; filled bars, cultures induced by addition of 0.1 mM IPTG. Error bars indicate standard deviations. Relative light units (RLU) were calculated by dividing the raw bioluminescence by the optical density of the culture at 600 nm.

To examine the effects of c-di-GMP induction of aphA on the expression of genes encoding virulence factors, we constructed lux transcriptional fusions of the virulence genes tcpA, ctxA, tcpP, and toxT. The expression of these genes increased 1.2- to 1.5-fold at high concentrations of c-di-GMP when the cells were grown in the virulence-inducing AKI medium (data not shown). The induction of aphA expression by c-di-GMP in AKI medium was similar to the induction observed in LB medium. Therefore, we conclude that while c-di-GMP increases the transcription of the aphA promoter, the induction of downstream virulence genes is modest under the conditions tested here.

c-di-GMP reduces acetoin production through induction of aphA.

In addition to controlling virulence factor expression, AphA regulates numerous other genes (24, 33). AphA reduces acetoin synthesis through repression of the VC1588-VC1593 operon, harboring biosynthetic genes for acetoin and 2,3-butanediol synthesis (24). To test if the induction of aphA by c-di-GMP impacts additional aphA-controlled phenotypes, we measured the acetoin produced at low versus high levels of c-di-GMP using a Voges-Proskauer test (1), which assesses the amount of acetoin produced by bacteria in the medium. Bacteria were grown in MR-VP medium (Difco), a medium that contains a mixture of glucose and buffered peptone. This environment induces V. cholerae to make acetoin in order to combat the acidic effects of metabolic end products (24). Induction of QrgB, which generates high levels of c-di-GMP, reduced acetoin production in the ΔvpsL mutant (Fig. 2). We hypothesized that this reduction was due to increased expression of aphA. Indeed, induction of c-di-GMP synthesis in an aphA mutant strain did not significantly reduce acetoin production, showing that the impact of c-di-GMP on acetoin requires aphA. Overexpression of aphA significantly reduced acetoin production, even when the Ptac promoter driving its expression was not induced with IPTG (Fig. 2). We interpret this result to mean that low levels of aphA expression from the uninduced Ptac promoter on a multicopy plasmid are sufficient to fully repress acetoin production. As expected, acetoin production was also abolished in the mutant with a deletion of alsS, encoding the enzyme alpha-acetolactate, which is essential for this biosynthetic pathway (24). These results indicate that c-di-GMP induction of aphA alters the expression of a subset of the genes controlled by aphA. Because aphA is the major low-cell-density regulator of the QS pathway in V. cholerae (33), induction of aphA by c-di-GMP would be expected to alter the expression of numerous genes and phenotypes, although this remains to be formally examined.

Fig. 2.

c-di-GMP represses acetoin production through aphA induction. Acetoin production was assessed by the MR-VP test. Three of the strains examined contained the pTac-qrgB overexpression vector, while the fourth (rightmost bars) had a pTac-aphA overexpression vector. Each strain was tested under noninducing conditions (shaded bars) and after the addition of 0.1 mM IPTG (filled bars). Error bars indicate standard deviations.

aphA and vpsT have similar promoter architectures.

c-di-GMP induces the transcription of vpsT, leading to increased biofilm formation in V. cholerae (3, 45). VpsT is a transcriptional activator of the LuxR, CsgD, and FixJ family (8). The mechanism of c-di-GMP induction of vpsT expression is not known; however, comparison of the promoter architecture of vpsT to that of aphA indicates that they could have similar regulatory control mechanisms. aphA expression is controlled by three global regulatory proteins: HapR, Lrp, and VpsR (29). Lrp, a transcriptional regulator that responds to changes in the cell's metabolic state (5), induces aphA expression by binding to a site between bases −138 and −123 (Fig. 3 A). VpsR, a transcriptional activator of biofilm formation (48), activates aphA expression by binding in the region from −88 to −70 (Fig. 3A). HapR, the master high-cell-density transcription regulator of the quorum-sensing system in V. cholerae, represses aphA expression by binding to the nucleotides located at positions −85 to −57 from the transcription start site (Fig. 3A) (29). HapR represses aphA expression by excluding VpsR binding (29).

A binding site for VpsR was predicted to exist in the vpsT promoter (29). Genetic evidence also suggests that VpsR is essential for vpsT expression, although no direct interaction between VpsR and the vpsT promoter has been shown (8). We also identified a predicted binding site for Lrp, located upstream of the predicted VpsR binding site, on the vpsT promoter. Lrp has not been reported to be a regulator of vpsT. A HapR binding site is also predicted in the vpsT promoter, and we have previously shown that HapR directly binds to the vpsT promoter and represses its expression, although the exact binding site has not yet been identified (45). Importantly, as in the aphA promoter, the predicted binding sites for VpsR and HapR also overlap in the vpsT promoter (Fig. 3). Therefore, we wondered if aphA and vpsT could share a regulatory mechanism that is activated by c-di-GMP.

To examine this possibility, we first determined, by use of 5′ RACE, that the vpsT transcriptional start site is 20 bp upstream of the translational start site at a T nucleotide. We refer to this base as +1 (Fig. 3B). Using this information, we can now define the predicted binding sites of Lrp, HapR, and VpsR relative to the transcription start site as −163 to −148, −144 to −123, and −136 to −118, respectively (Fig. 3B). While the locations of these predicted binding sites relative to one another in vpsT indeed parallel their sites in the aphA promoter, the distance between these binding sites and the transcription start site is greater in the vpsT promoter (118 bases) than in the aphA promoter (57 bases) (Fig. 3). The implications, if any, of this difference in spacing are currently unknown.

c-di-GMP activation of the aphA and vpsT promoters requires VpsR.

To further examine the roles of HapR, VpsR, and Lrp in c-di-GMP regulation of aphA and vpsT, we constructed transcriptional fusions of aphA (comprising the region from bp −396 to the translation start site) and vpsT (comprising the region from bp −482 to the translation start site) in a lux reporter plasmid (Fig. 3). The expression of these two fusion constructs in response to increased c-di-GMP levels was measured in ΔvpsL, ΔvpsL Δlrp::tetA, ΔvpsL ΔhapR, and ΔvpsL ΔvpsR mutant strains. c-di-GMP induction of transcription from these promoters was maintained in the Δlrp::tetA and ΔhapR mutants for both aphA and vpsT (Fig. 4). The expression of aphA was lower in a Δlrp::tetA mutant than in the wild-type strain, because Lrp activates aphA expression (29). No difference in vpsT expression was observed in the Δlrp::tetA mutant, suggesting that Lrp does not regulate vpsT expression under the conditions examined here.

Fig. 4.

VpsR is required for c-di-GMP-mediated induction of aphA and vpsT. The expression of aphA-lux (A) and vpsT-lux (B) constructs was analyzed in ΔvpsL, ΔvpsL Δlrp::tetA, ΔvpsL ΔhapR, and ΔvpsL ΔvpsR mutants containing the Ptac-qrgB overexpression vector under noninducing conditions (shaded bars) and following the addition of 0.1 mM IPTG (filled bars). Error bars indicate standard deviations. Relative light units (RLU) were calculated as described in the legend to Fig. 1.

The expression of both aphA and vpsT was increased in a ΔhapR strain, confirming previous findings that QS regulates these promoters through HapR repression (26, 45). The fold induction of aphA expression in the ΔhapR mutant was reduced, since basal expression under the low-c-di-GMP condition was greatly increased, suggesting that HapR repression of aphA predominates. The fold induction of vpsT expression by c-di-GMP in the ΔhapR mutant, however, was greater than that observed for aphA. Therefore, c-di-GMP induction may play a larger role relative to HapR repression in the regulation of vpsT than in the regulation of aphA. But c-di-GMP induction was consistently observed for both genes in the ΔhapR mutant. Importantly, neither aphA nor vpsT was induced in the ΔvpsR mutant strain. These results suggest that VpsR, but not HapR or Lrp, is essential for the c-di-GMP-mediated induction of aphA and vpsT. Furthermore, Lrp does not regulate vpsT expression under these conditions.

VpsR binds directly to the vpsT promoter.

HapR, Lrp, and VpsR are known to bind the aphA promoter, while only HapR has been shown to bind the vpsT promoter (45). To determine whether VpsR and HapR also bind to the vpsT promoter at the predicted binding site (Fig. 3), these proteins were purified, and electrophoretic mobility shift assays (EMSA) were performed with each protein and a fluorescently labeled vpsT DNA probe. This probe contained the sequence from positions −482 to +20 of the vpsT promoter (Fig. 3B). As we have previously observed, HapR binds to the vpsT promoter (45). Furthermore, VpsR also binds to the vpsT promoter (Fig. 5, top).

Fig. 5.

VpsR and HapR bind to the vpsT promoter at the predicted binding sites. The vpsT promoter deletion constructs shown in Fig. 3 were used to generate fluorescent probes for EMSA with purified VpsR and HapR. A 10 nM concentration of the probe was used in all lanes. Lanes 1 and 6 contain no protein; lanes 2 to 5 contain 25, 120, 360, and 650 nM VpsR, respectively; lanes 7 to 10 contain 42, 230, 620, and 800 nM HapR, respectively.

To determine if the predicted binding sites for HapR and VpsR are correct, we constructed 5′-truncated transcriptional fusions of the vpsT promoter starting at positions −482, −195, −149, and −119 and ending at the +20 translation start site in pBBRlux. These constructs are referred to below as −482T, −195T, −149T, and −119T, respectively (Fig. 3B). Fluorescent probes were generated for each of these derivatives, and EMSA were performed with purified HapR and VpsR (Fig. 5). The results indicate that VpsR and HapR bind to vpsT promoter fragments −482T, −195T, and −149T but do not bind to the −119T fragment. This finding suggests that the HapR and VpsR binding sites at the vpsT promoter do indeed lie between positions −149 and −119, as predicted. In addition, the binding of these proteins is specific, since neither protein bound to the smallest fragment. Importantly, this result suggests that the binding sites for HapR and VpsR overlap, as they do in the aphA promoter (Fig. 3).

The VpsR and HapR binding sites are required for c-di-GMP induction of aphA and vpsT.

The expression of both aphA and vpsT is induced by c-di-GMP, and they share binding sites for both HapR and VpsR. Our mutation analysis showed that VpsR is critical for the induction of the vpsT and aphA promoters by c-di-GMP (Fig. 4). To test if the VpsR binding sites harbored in the aphA and vpsT promoters are important for this regulation, we analyzed c-di-GMP induction of the −428T, −195T, −149T, and −119T vpsT transcriptional fusions in pBBRlux. Similarly, we engineered constructs with corresponding transcriptional fusions of the aphA promoter in pBBRlux, referred to as −396A, −156A, −106A, and −51A (Fig. 3). These aphA constructs contain binding sites for Lrp, HapR, and VpsR (−396A and −156A), for HapR and VpsR only (−106A), or for neither of these regulators (−51A) (29). The expression levels of these aphA-lux and vpsT-lux promoter fusions were determined at varying intracellular c-di-GMP concentrations by inducing QrgB expression via increasing amounts of IPTG. This strategy increases the intracellular c-di-GMP concentrations in a dose-dependent manner from 1 to 10 μM (unpublished observation). In the aphA and vpsT promoter derivatives, promoter fragments encoding the overlapping binding sites for HapR and VpsR were induced by c-di-GMP in an IPTG-dependent manner, while the shortest fragments, −51A and −119T, which lacked the HapR and VpsR binding sites, were not significantly induced (Fig. 6). This result suggests that the VpsR binding site is critical for c-di-GMP induction of aphA or vpsT.

Fig. 6.

Transcriptional responses of aphA and vpsT to increasing levels of c-di-GMP. The aphA (A) and vpsT (B) promoter deletion constructs shown in Fig. 3 were constructed as transcriptional fusions to the luciferase operon and were introduced into a V. cholerae ΔvpsL mutant containing the Ptac-qrgB overexpression plasmid. c-di-GMP levels were increased by adding IPTG at concentrations from 0.45 μM to 1 mM. Error bars indicate standard deviations. Relative light units (RLU) were calculated as described in the legend to Fig. 1.

VpsT is not required for c-di-GMP induction of vpsT or aphA.

Binding of c-di-GMP by VpsT induces a change in its oligomeric state, leading to a form of the protein capable of promoting transcription (27). Genetic evidence has suggested that VpsT activates its own expression (2), and it was hypothesized that induction of vpsT by c-di-GMP could occur through increased autoactivation by VpsT bound to c-di-GMP (27). To investigate whether VpsT mediates c-di-GMP induction of aphA and its own expression, we constructed a ΔvpsT mutant and analyzed c-di-GMP induction of the −396A (aphA-lux) and −482T (vpsT-lux) promoter constructs in the ΔvpsL and ΔvpsL ΔvpsT mutants (Fig. 7). The expression of aphA and its induction by c-di-GMP were unaffected by the deletion of vpsT. This result shows that vpsT does not regulate aphA under the conditions examined here. Similarly, although the overall expression of vpsT was reduced in the ΔvpsT mutant, confirming that vpsT activates its own expression, the vpsT promoter remained inducible by c-di-GMP, as in the WT strain. Therefore, we conclude that VpsT is not required for c-di-GMP-mediated induction of either aphA or vpsT.

Fig. 7.

c-di-GMP activation of aphA and vpsT is independent of VpsT. The expression of aphA-lux (A) and vpsT-lux (B) in the ΔvpsL strain and the ΔvpsL ΔvpsT mutant containing a vector control or the Ptac-qrgB overexpression vector was measured under noninducing conditions (shaded bars) or following induction with 0.1 mM IPTG (filled bar). Error bars indicate standard deviations. Relative light units (RLU) were calculated as described in the legend to Fig. 1.

VpsR binds to c-di-GMP.

VpsR belongs to the NtrC family of transcriptional regulators, harboring an aspartate residue in the amino terminus that serves as a phosphorylation site (48). No cognate kinase for VpsR phosphorylation has been identified yet. Like other NtrC regulators, VpsR encodes consensus sequences for interaction with sigma 54 (48). However, expression of the vpsT and aphA promoters is not affected by deletion of rpoN, the gene encoding sigma 54 (37). VpsR harbors predicted ATP binding and helix-turn-helix DNA binding domains, like other members of the NtrC family (6, 48). FleQ, an NtrC-like regulator in Pseudomonas aeruginosa, directly binds c-di-GMP to regulate pel gene expression (20). To determine whether VpsR also binds to c-di-GMP, a filter-binding assay was performed. In this experiment, purified protein incubated with ³²P-labeled c-di-GMP was bound to a nitrocellulose membrane by using a slot blot apparatus and was washed extensively, and bound radioactivity was quantified using scintillation counting (20). We observed a dose-dependent increase in the level of binding when 400 nM purified VpsR was incubated with varying amounts of ³²P-labeled c-di-GMP (Fig. 8 A). In the absence of protein, minimal binding was observed. Saturation binding of VpsR to c-di-GMP was examined four times independently to determine the affinity of binding of VpsR to c-di-GMP. Averaging the data from these four experiments yielded a dissociation constant (K_d) of 1.6 μM for this interaction, with a standard deviation of 0.66. The data for saturation binding were analyzed with GraphPad Prism (version 5.0, 2007; GraphPad Software, Inc., San Diego, CA) using the specific binding equation.

Fig. 8.

VpsR binds c-di-GMP. (A) Purified VpsR (400 nM) (circles) and buffer control (squares) reaction mixtures were incubated with varying concentrations of ³²P-labeled c-di-GMP to generate a saturation binding curve. The data were analyzed with GraphPad Prism, version 5.0, using nonlinear regression analysis. This experiment was repeated four times, and a representative curve is shown. (B) The purified proteins (400 nM) indicated on the x axis were incubated with 1 μM ³²P-labeled c-di-GMP with or without 3 μM unlabeled c-di-GMP and GTP. This experiment was repeated three times with similar results.

To determine if the binding of c-di-GMP to VpsR is specific, we performed a similar filter-binding assay examining the binding of one concentration of ³²P-labeled c-di-GMP to VpsR, bovine serum albumin (BSA), and the protein YcgR in the presence of excess unlabeled c-di-GMP or GTP. YcgR is a PilZ domain protein that binds directly to c-di-GMP to control motility (14, 34). YcgR bound to the greatest amount of ³²P-labeled c-di-GMP, while BSA did not retain any radioactivity. VpsR retained 2-fold less ³²P-labeled c-di-GMP than YcgR (Fig. 8B). It is difficult to compare total binding directly, however, since we do not know if VpsR binds c-di-GMP as a higher-order oligomer or if the relative activity of each protein preparation is similar. The addition of a 3-fold excess of unlabeled c-di-GMP reduced the binding of both VpsR and YcgR, while a 3-fold excess of unlabeled GTP had no effect (Fig. 8B). This experiment was repeated three times with similar results. Thus, we conclude that the observed binding of c-di-GMP to VpsR is specific.

Addition of c-di-GMP does not change the in vitro DNA binding profile of VpsR for the vpsT or aphA promoter under the conditions we tested (data not shown). This result differs from the findings for the other c-di-GMP binding transcriptional regulators reported—FleQ, Clp, and VpsT—all of which differentially bind DNA in vitro in the presence and absence of c-di-GMP (9, 20, 27). Therefore, we conclude that VpsR binds c-di-GMP; however, how this binding affects VpsR function at these promoters remains to be determined.

VpsR is not required for c-di-GMP-mediated induction of additional V. cholerae promoters.

In addition to the promoter controlling aphA, our screen identified six other promoters that are activated by c-di-GMP (Table 1). To determine if VpsR is required for the c-di-GMP-mediated regulation of these promoters, they were introduced (as promoter-lux fusions) into the ΔvpsL and ΔvpsL ΔvpsR mutants containing inducible qrgB. Here we present results for two isolates (with promoter fragments 1:B8 and 5:A6, regulating the VCA0213 and VC1899 genes, respectively) as representatives (Fig. 9). The other four promoters behaved identically to these two (data not shown). In contrast to what we found for aphA and vpsT, none of the other six promoters required VpsR for c-di-GMP induction, indicating that their c-di-GMP-regulation occurs via a VpsR-independent mechanism. Because VpsT is not expressed in a ΔvpsR mutant (8), we also infer that VpsT is not required for the induction of these six promoters. Taken together, our results indicate that multiple signal transduction pathways in V. cholerae link c-di-GMP to transcription regulation.

Fig. 9.

VpsR is not required for c-di-GMP-mediated activation of other V. cholerae promoters. The expression of two promoter constructs, VCA0213 (A) and VC1899 (B), described in Table 1, was analyzed in the ΔvpsL and ΔvpsL ΔvpsR mutants containing the Ptac-qrgB overexpression vector under noninducing conditions (shaded bars) and following the addition of 0.1 mM IPTG (solid bars). Error bars indicate standard deviations. Relative light units (RLU) were calculated as described in the legend to Fig. 1.

DISCUSSION

Here we show that the integration of QS and c-di-GMP in the control of aphA and vpsT occurs through a common mechanism. These two transcriptional regulators function as checkpoints at the apices of cascades regulating entry into the low-cell-density QS state, the virulence cascade, and biofilm developmental pathways. Our results lead to four important conclusions. First, the expression of aphA, a central regulator of the virulence cascade and the master low-cell-density regulator of the QS pathway, is induced by c-di-GMP. This finding was unexpected, because c-di-GMP is thought to play a negative role in virulence gene expression (38); however, it is consistent with aphA functioning at low a cell density, since the level of c-di-GMP is high in this state (45). Second, a shared regulatory mechanism controls c-di-GMP-mediated induction of aphA and vpsT. Although both promoters interact with HapR and VpsR, only VpsR is required for c-di-GMP induction of their transcription. Importantly, c-di-GMP induction was not dependent on VpsT. Third, VpsR directly binds c-di-GMP. Fourth, we have identified six additional promoters whose transcription is induced by high levels of c-di-GMP independently of VpsR. We are examining the regulation of these genes to identify additional c-di-GMP-dependent transcriptional regulators of V. cholerae.

Activation of aphA and vpsT by c-di-GMP is mediated through a mechanism involving VpsR, demonstrating that they share a regulatory pathway. Furthermore, the expression of both genes is repressed by the QS regulator HapR (26, 45). Therefore, joint regulation of HapR and VpsR represents a central control module integrating QS- and c-di-GMP-mediated regulation. Binding of HapR at the aphA promoter excludes VpsR binding (29). Further experimentation is required to determine whether HapR similarly excludes VpsR binding at the vpsT promoter. This control module appears to combine important information about the surrounding bacterial community and the local environment. We propose that the integration of the information from these two major sensory pathways at the apices of the biofilm, low-cell-density QS state, and virulence cascades functions similarly to a regulatory checkpoint. We are currently identifying other genes/promoters with similarly arranged HapR and VpsR binding sites to determine the extent of c-di-GMP and QS cross-wiring in other V. cholerae developmental pathways.

We investigated whether VpsR might function by binding to c-di-GMP to further activate the expression of aphA and vpsT. We found that VpsR binds c-di-GMP with a K_d of 1.6 μM, and we hypothesize that this binding is important for the c-di-GMP-mediated induction of the vpsT and aphA promoters. Interestingly, VpsT and VpsR have similar binding affinities (27). The intracellular c-di-GMP concentration in V. cholerae ranges from 10 μM at a low cell density to 1 μM at a high cell density (unpublished observations). Thus, the binding affinities of VpsR and VpsT fall within the normal physiological concentrations of c-di-GMP in the cell, although we expect that the in vitro binding affinity of VpsR determined here is less than the true in vivo value, due to an incomplete replication of the in vivo environment in our filter-binding assay. The binding of c-di-GMP to both of the major transcriptional activators of biofilm formation in V. cholerae, VpsT and VpsR, increases the positive induction of biofilm development genes and possibly amplifies the response of this second messenger.

c-di-GMP has been reported previously to exert a negative effect on the virulence of V. cholerae (38, 42). Two pathogenic V. cholerae biotypes have been described: the classical and El Tor biotypes (7). The first six pandemics of V. cholerae were caused by the classical biotype, while the seventh and current pandemic is caused by El Tor strains (13). In the classical biotype, the phosphodiesterase (PDE) VieA functions to reduce the c-di-GMP levels in vivo (42). The PDE activity of VieA is required for full expression of toxT and for the production of CT. Mutation of vieA causes a 10-fold decrease in V. cholerae colonization in an infant mouse model (40). The regulation of virulence and colonization by c-di-GMP in the El Tor biotype (the biotype studied here) is not as well understood, because VieA does not play a role in controlling c-di-GMP levels. Rather, CdpA, a PDE in V. cholerae El Tor C6706, is suggested to modulate c-di-GMP levels during infection by repressing biofilm formation and positively increasing CT production (39). However, cdpA mutation has no effect on colonization in the murine model. Consistent with the lack of a major negative role for c-di-GMP in El Tor infection, overexpression of VdcA, a V. cholerae GGDEF protein, reduced colonization levels only 3-fold from that for the wild-type strain (39). The effect of VdcA overexpression in vivo on ToxT expression was determined using the recombination-based in vivo expression technology (RIVET) system. These experiments showed that a subpopulation of V. cholerae expressed virulence at both high and low levels of c-di-GMP. Interestingly, both the wild-type El Tor strain and the VdcA overexpression strain exhibited heterogeneous toxT expression, suggesting that V. cholerae may exist in multiple development stages in vivo (39).

Since these previous results have suggested that c-di-GMP inhibits virulence in early stages of infection, our discovery that aphA is activated by c-di-GMP was unexpected. AphA was recently shown to be an important low-cell-density master regulator in V. cholerae and V. harveyi, controlling the expression of genes involved in motility, type III secretion, acetoin production, and multiple hypothetical genes (24, 33). Our results showing that the regulation of aphA by c-di-GMP leads to altered production of acetoin suggest that other genes may be jointly controlled by c-di-GMP and QS through induction of aphA expression.

AphA controls the expression of the tcpPH operon, which leads to the activation of toxT expression. However, under laboratory conditions, we did not observe significant induction of the transcription of virulence genes known to lie downstream of AphA (such as toxT) at high c-di-GMP levels. One possibility is that c-di-GMP induction of aphA occurs under many conditions, while AphA activation of the remainder of the virulence cascade may be context dependent, occurring only in specific environments, such as the host environment. Alternatively, c-di-GMP could negatively control virulence gene expression independently of aphA at additional points in the virulence cascade.

Interestingly, AphA itself induces vpsT expression and biofilm formation, further linking these two central regulators (33, 47). The AphA binding site at the vpsT promoter is located 240 bases upstream of the translation start site of vpsT. Our observation that truncated promoter fragments that do not contain this sequence, specifically −195T and −149T (Fig. 3B), maintain c-di-GMP induction suggests that AphA is not required for this process. However, we would predict that c-di-GMP induction of AphA would further amplify vpsT expression. Further work on the connections between biofilms and virulence in V. cholerae is required to characterize the interplay between the biofilm and virulence signal transduction pathways and the potential in vivo role for c-di-GMP in these processes.

In V. cholerae, biofilm formation and virulence have been reported to be inversely regulated, and c-di-GMP has been suggested to inhibit in vivo disease development. Our results showing that expression of the virulence regulator aphA is activated by c-di-GMP hint that c-di-GMP may, in some cases, have a positive function during in vivo infection. Furthermore, because aphA is the master low-cell-density regulator of the QS pathway, induction of aphA by c-di-GMP has a significant impact on the low-cell-density state in V. cholerae. Clearly, the QS and c-di-GMP signaling pathways in V. cholerae, controlling biofilm formation, virulence factor expression, and numerous other phenotypes, are intricately intertwined at many levels. We hypothesize that these connections allow V. cholerae to sense and combine information about the extracellular community and the surrounding environment so as to adapt optimally to ever-changing conditions.