|Home | About | Journals | Submit | Contact Us | Français|
Vfr is a global regulator of virulence factor expression in the human pathogen Pseudomonas aeruginosa. Although indirect evidence suggests that Vfr activity is controlled by cyclic AMP (cAMP), it has been hypothesized that the putative cAMP binding pocket of Vfr may accommodate additional cyclic nucleotides. In this study, we used two different approaches to generate apo-Vfr and examined its ability to bind a representative set of virulence gene promoters in the absence and presence of different allosteric effectors. Of the cyclic nucleotides tested, only cAMP was able to restore DNA binding activity to apo-Vfr. In contrast, cGMP was capable of inhibiting cAMP-Vfr DNA binding. Further, we demonstrate that vfr expression is autoregulated and cAMP dependent and involves Vfr binding to a previously unidentified site within the vfr promoter region. Using a combination of in vitro and in vivo approaches, we show that cAMP is required for Vfr-dependent regulation of a specific subset of virulence genes. In contrast, we discovered that Vfr controls expression of the lasR promoter in a cAMP-independent manner. In summary, our data support a model in which Vfr controls virulence gene expression by distinct (cAMP-dependent and -independent) mechanisms, which may allow P. aeruginosa to fine-tune its virulence program in response to specific host cues or environments.
Pseudomonas aeruginosa is an opportunistic pathogen responsible for a variety of life-threatening infections in immunocompromised individuals and those receiving critical care (12). P. aeruginosa is the primary cause of morbidity and mortality in individuals with cystic fibrosis, in whom it causes chronic lung infection (6). Furthermore, indwelling medical devices, severe wounds, burns, surgery, and corneal abrasion predispose otherwise-healthy individuals to infection by this organism (12). The ability of P. aeruginosa to cause infection depends on the expression of an array of surface-exposed and secreted virulence factors (40). Many of these factors are directly or indirectly controlled by the transcriptional regulator protein Vfr (virulence factor regulator). Vfr positively regulates production of exotoxin A (ETA or ToxA), type IV pili (Tfp), a type III secretion system (T3SS), and the las quorum-sensing system which, in turn, controls the expression of hundreds of additional genes, including multiple virulence factors (2, 4, 44, 54). In addition, Vfr negatively regulates flagellar gene expression (10). A consensus Vfr binding sequence has been proposed (24), and direct binding of Vfr to target promoters has been demonstrated for several genes, including those encoding ToxA (toxA), LasR (the las quorum-sensing regulator), FleQ (the master regulator of flagellar biogenesis), RegA and PtxR (regulators of toxA expression), and CpdA (a cyclic AMP [cAMP] phosphodiesterase) (2, 10, 14, 16, 24). While the global role of Vfr in regulating virulence gene expression has been established, the molecular mechanisms that control Vfr activity and expression are not well understood.
Vfr is a member of the 3′,5′-cAMP receptor protein (CRP) family of transcriptional regulators (55). The best-studied member of this family is Escherichia coli CRP, which primarily regulates genes involved in carbon metabolism (20). CRP functions as a homodimer, and its activity is directly controlled by the allosteric regulator cAMP; CRP undergoes a conformational change upon cAMP binding that enables the protein to interact with target promoters in a DNA sequence-specific manner (5, 26, 34-36, 43). CRP is also capable of binding 3′,5′-cGMP (3, 13, 49); however, structural studies indicate that cGMP does not induce the necessary conformational change required for CRP DNA binding (36). Although previous studies have demonstrated that Vfr and CRP have similar affinities for cAMP in vitro (48), the roles of cAMP and other cyclic nucleotides in Vfr function have not been directly examined.
P. aeruginosa encodes two intracellular adenylate cyclases (CyaA and CyaB) responsible for cAMP synthesis (58). Mutants lacking both cyaA and cyaB exhibit reduced virulence factor expression and are severely attenuated in an adult mouse model of acute pneumonia (47, 58). In addition, whole-genome expression profiling revealed that the transcriptomes of P. aeruginosa mutants defective in cAMP synthesis or lacking vfr are nearly identical, suggesting that Vfr activity is dependent on cAMP availability (58). In support of this notion, a previous study revealed that Vfr is capable of restoring cAMP-dependent gene expression in an E. coli crp mutant. However, CRP cannot complement a P. aeruginosa vfr mutant, suggesting that Vfr and CRP are not functionally interchangeable (48, 55).
The putative cAMP binding pocket of Vfr contains three additional amino acids relative to that of E. coli CRP (4, 55). While most of the residues involved in cAMP binding are conserved, Vfr has a threonine residue at a position equivalent to serine 128 (S128) of E. coli CRP (4). Mutational studies have demonstrated that threonine substitution of CRP S128 results in CRP activation by both cAMP and cGMP (28). Furthermore, a spontaneous Vfr mutant (VfrΔEQERS) lacking 5 amino acids (EQERS) in the cyclic nucleotide binding domain, including two critical cAMP binding residues conserved among other CRP homologs, retained the ability to regulate expression of a subset of virulence factors (4). Based on these observations, it has been proposed that Vfr may respond to cAMP, cGMP and/or other allosteric regulators (4, 55). While there is currently no evidence that P. aeruginosa has the ability to synthesize cGMP (16), it does produce cyclic diguanosine monophosphate (c-di-GMP) and possibly cyclic diadenosine monophosphate (c-di-AMP) (22, 25, 56). Recent findings indicate that some members of the CRP family bind c-di-GMP, which acts as a negative allosteric regulator (27, 51).
While the mechanism of allosteric regulation of Vfr activity is unresolved, there is evidence suggesting that vfr expression is controlled at the level of transcription. A previous study identified two putative Vfr binding sites upstream of the vfr gene (centered at bp −67.5 and bp −39.5 relative to the transcription start site) (24). Furthermore, it was demonstrated through electrophoretic mobility shift assays (EMSAs) that Vfr could specifically bind to a vfr promoter probe containing both putative sites (24). Although the role of these sites in vfr promoter activity has not been tested, direct binding of Vfr to this region suggests that vfr expression is autoregulated, as is the case for E. coli crp (1, 8, 21).
In this study, we directly assessed the cyclic nucleotide requirement for Vfr activity both in vitro and in vivo. We generated apo-Vfr by two independent methods and demonstrate that Vfr binding to most target promoters specifically requires cAMP. Other cyclic nucleotides did not support Vfr activity. Furthermore, we show that high concentrations of cGMP inhibit the formation of Vfr-DNA complexes. Using DNase I footprinting, we identified a novel Vfr binding site within the vfr promoter distinct from the previously proposed binding sites. In vitro transcription assays were employed to demonstrate that both cAMP and Vfr are required for positive vfr autoregulation. Finally, we provide evidence that the lasR promoter is an exception to the cAMP requirement paradigm, since cAMP was not required for Vfr binding to the lasR promoter in vitro or for activation of the lasR promoter activity in vivo. Taken together, our findings provide new mechanistic insights into the complex cAMP/Vfr signaling pathway that controls P. aeruginosa virulence.
The strains and plasmids used in this study are listed in Table Table1.1. For routine passage, strains were grown at 37°C in LB (Difco). For complementation experiments, pPa-vfr was maintained in P. aeruginosa with 30 μg/ml carbenicillin (Cb). Bacterial growth in broth culture was assessed based on the optical density at 600 nm (OD600).
P. aeruginosa strains PAK cyaA cyaB vfr and PAK lasR vfr were constructed by introducing a deletion allele for vfr (encoded by the pEXGmΔvfr plasmid) onto the chromosome of PAK cyaA cyaB and PAK lasR, respectively, using a previously described method (58). PAK lasR cyaA cyaB was constructed by sequentially introducing deletion alleles for cyaA and cyaB (encoded by pEXGmΔcyaA and pEXGmΔcyaB, respectively) onto the chromosome of PAK lasR.
Chromosomal transcriptional reporters were constructed by PCR amplifying the vfr promoter region (bp −164 to +206 relative to the vfr transcriptional start site ) and the lasR promoter region (bp −264 to +238 relative to the predominant lasR T1 transcriptional start site ) from P. aeruginosa strain PAK chromosomal DNA using oligonucleotides (see Table S1 in the supplemental material; EcoRI and BamHI restriction sites are underlined) tailed with attB1 or attB2 sequences for Gateway cloning into pDONR201 (Invitrogen). Both promoter fragments encompassed known or predicted promoter elements and the translational start sites of the corresponding genes. Promoter fragments were removed from pDONR201 by digestion with EcoRI and BamHI and ligated into the corresponding restriction sites of mini-CTX-lacZ (23). The resulting plasmids were used to integrate the promoter-lacZ fusions onto the chromosome at a vacant CTX phage attachment site of wild-type and mutant P. aeruginosa strains as described previously (23).
A toxA transcriptional reporter plasmid was constructed by PCR amplifying bp −500 to +100 relative to the toxA translational start codon from PAK chromosomal DNA with toxA rep 5′ and toxA rep 3′ oligonucleotides (see Table S1 in the supplemental material; HindIII and BamHI restriction sites are underlined), digesting with HindIII and BamHI, and ligating into the corresponding restriction sites of the plasmid pR-lacZ to create pRtoxA-lacZ. pR-lacZ is a low-copy-number lacZ transcriptional reporter plasmid (gift of Arne Rietsch [Case Western Reserve University]). The pRtoxA-lacZ plasmid was transferred to appropriate strains by conjugation (18) followed by selection on LB agar plates containing 75 μg/ml gentamicin (Gm) and 25 μg/ml irgasan (Irg).
To create a P. aeruginosa vfr expression plasmid (pPa-vfr), the open reading frame of vfr was PCR amplified from strain PAK genomic DNA using primers 5′ vfr and 3′ vfr (see Table S1 in the supplemental material) and cloned into pMMBV1GW (16) using Gateway cloning (Invitrogen) by a previously described method (58).
Plasmid templates used in the in vitro transcription assays were created by cloning the vfr promoter region (bp −164 to +206 relative to the vfr transcriptional start site ) and the lasR promoter region (bp −264 to +30 or −264 to +238 relative to the lasR T1 transcriptional start site ) upstream of the rpoC transcriptional terminator in plasmid pOM90 (38) to create pOM90-vfr, pOM90-lasR(−264 to +30), and pOM90-lasR(−264 to +238). The promoter fragments were PCR amplified from strain PAK chromosomal DNA using oligonucleotides tailed with BamHI and EcoRI restriction sites (see Table S1 in the supplemental material), digested with BamHI and EcoRI, and ligated into the corresponding restriction sites of pOM90. For the in vitro transcription assay, the predicted size of the vfr transcript is 281 nucleotides; the predicted sizes of the lasR transcripts are 102 and 132 nucleotides for the pOM90-lasR(−264 to +30) template and 310 and 340 nucleotides for the pOM90-lasR(−264 to +238) template.
Purification of cAMP-Vfr, CpdA, a mutant derivative of CpdA (CpdA-N93A), and P. aeruginosa RNA polymerase holoenzyme (RNAP) was carried out as described previously (16, 52) (see Fig. S1 in the supplemental material). Apo-Vfr was generated by two independent methods. In the first approach, cAMP-Vfr was incubated with a 5-fold molar excess of purified CpdA or CpdA-N93A for 18 h at 23°C. In some experiments, CpdA was inactivated by heating for 5 min at 95°C prior to incubation with Vfr. In the second approach, bound cAMP was removed from Vfr by denaturing Vfr protein (0.5 ml at 55 μg/ml) by dialysis for 2 h at 4°C against buffer (50 mM Tris-HCl [pH 7.0], 100 mM KCl, 50 mM NaCl, 1 mM dithiothreitol [DTT], 1 mM EDTA, 10% glycerol, 0.5% Tween 20) containing 6 M urea. Vfr was refolded by sequentially reducing the urea concentration to 3 M, 2 M, 1 M, 0.5 M, and 0 M under the dialysis conditions described above in the absence or presence of 50 μM cAMP.
DNA promoter probes were generated by PCR using the indicated oligonucleotides (see Table S1 in the supplemental material) and end labeled using 10 μCi of [γ-32P]ATP (GE Healthcare) and 10 U of T4 polynucleotide kinase (New England Biolabs). EMSAs were performed as previously described (7). Briefly, probes (0.25 nM each) were incubated in binding buffer (10 mM Tris [pH 7.5], 50 mM KCl, 1 mM EDTA, 1 mM DTT, 5% glycerol, and 100 μg/ml bovine serum albumin) containing 5 μg/ml poly(2′-deoxyinosinic-2′-deoxycytidylic acid) [poly(dI-dC); Sigma] for 5 min at 25°C. As noted below and in the figure legends, cAMP, cGMP, c-di-GMP, or c-di-AMP was also present. cAMP-Vfr or apo-Vfr protein was then added at concentrations (indicated in figure legends) for a final reaction volume of 20 μl and incubated for an additional 15 min at 25°C. Samples were subjected to electrophoresis on a 5% polyacrylamide glycine gel (10 mM Tris [pH 7.5], 380 mM glycine, 1 mM EDTA) at 4°C. Imaging and data analyses were performed using an FLA-7000 PhosphorImager (Fujifilm) and MultiGauge v3.0 software (Fujifilm). EMSAs were repeated a minimum of two times, and representative gels are shown.
A single end-labeled γ-32P-labeled double-stranded DNA probe was generated by PCR in which one of the oligonucleotide primers was modified (5 Amino-MC6; Integrated DNA Technologies) at the 5′ end to prevent phosphorylation (see Table S1 in the supplemental material). Probes were subsequently labeled as described above. Footprinting reaction mixtures consisted of a single end-labeled probe (10 fmol) with 5 μg/ml poly(dI-dC) (Sigma) in DNase I reaction buffer (10 mM Tris [pH 8.0], 50 mM KCl, 2 mM MgCl2, 0.5 mM DTT, 100 μg/ml bovine serum albumin, 10% glycerol). cAMP-Vfr was added for a final reaction volume of 25 μl, and the mixture was incubated for 15 min at 25°C. DNase I footprinting and DNA sequencing reactions were performed as previously described (31, 41, 45).
Plasmid templates (2 nM) were incubated in the absence or presence of refolded apo-Vfr (100 nM) and/or cAMP (100 nM) in 1× transcription buffer (40 mM Tris-HCl [pH 7.5], 150 mM KCl, 10 mM MgCl2, 0.01% Tween 20, and 1 mM DTT) containing 0.75 mM rATP, rGTP, and rCTP for 10 min at 25°C. Purified P. aeruginosa RNA polymerase holoenzyme, known to be largely σ70 saturated (52), was then added (10 nM) and the mixture was incubated for 5 min at 25°C. Finally, 0.75 mM rUTP and 5 μCi [α-32P]CTP in 1× transcription buffer containing heparin (50 μg/ml) were added for a final reaction volume of 20 μl, and transcription was allowed to proceed for 10 min at 30°C. Reactions were terminated by adding 20 μl of stop buffer (98% formamide, 20 mM EDTA, 0.05% bromophenol blue, and 0.05% xylene cyanol), heated to 95°C for 5 min, and electrophoresed on 5% denaturing urea polyacrylamide gels. Phosphorimaging and densitometry were performed using a FLA-7000 PhosphorImager (Fujifilm) and MultiGage v3.0 software (Fujifilm), respectively.
Overnight cultures of strains containing vfr or lasR promoter-lacZ fusions were diluted 1:100 into LB and grown to an OD600 of 1 or 5, respectively. Cb (30 μg/ml) and isopropyl-β-d-thiogalactopyranoside (IPTG; 40 μM) were added to the medium as indicated in the figure legends. Strains harboring both pRtoxA-lacZ and pPa-vfr were grown overnight in LB containing 30 μg/ml Cb and 15 μg/ml Gm, washed twice with deferrated (33) Bacto tryptic soy broth (DTSB; Becton Dickinson), and inoculated into fresh DTSB containing 30 μg/ml Cb, 15 μg/ml Gm, and 40 μM IPTG. Starting with a culture OD600 of 0.025, bacteria were grown for 8 h, and 1 ml of culture supernatant was collected. For all samples, β-galactosidase activity measurements were carried out as described previously (32), except that optical density determinations were made using a 96-well plate reader (Bio-Tek). Each assay was repeated at least three times.
Bacteria were grown as described for β-galactosidase assays. Whole-cell lysates and culture supernatants were prepared for detection of Vfr and secreted ToxA, respectively, as previously described (16). All Western blot assays were repeated a minimum of three times with independently derived protein samples, and representative blots are shown.
The two-tailed unpaired t test was used for data comparison where appropriate using Prism v5.0b (GraphPad Software).
The nucleotide sequence and position of Vfr binding sites within target promoters vary substantially (2, 10, 14, 16, 24). To determine the affinity of Vfr for a representative set of target promoters, we used quantitative EMSAs. Vfr protein was expressed in P. aeruginosa and isolated from cellular lysates using a cAMP-agarose affinity purification column (see Fig. S1 in the supplemental material). Specific promoter probes (Pvfr, PtoxA, PregA, PptxR, PlasR, and PcpdA) encompassing known or predicted Vfr binding sites (2, 14, 16, 24) were generated by PCR. For each binding assay, cAMP-Vfr was incubated with the specific promoter probe (~200 bp), a nonspecific control probe (~160 bp), and poly(dI-dC) and analyzed by native polyacrylamide gel electrophoresis and phosphorimaging. The addition of cAMP-Vfr had no effect on mobility of the nonspecific DNA probe (Fig. (Fig.1).1). In contrast, mobilities of the Pvfr, PtoxA, PregA, PptxR, PlasR, and PcpdA promoter probes were retarded in a Vfr concentration-dependent fashion, indicating the formation of specific protein-DNA complexes (Fig. (Fig.1).1). The fraction of probe that shifted as a function of Vfr concentration was used to calculate the apparent equilibrium constant (Keq) for cAMP-Vfr binding. The apparent Keq for the promoter probes varied over an 18-fold range (Table (Table2),2), with cAMP-Vfr having the highest affinity for PlasR and PcpdA and the lowest affinity for PtoxA. Interestingly, our results showed two distinct PlasR shift products that were not previously reported (2). The higher-mobility product (complex I) was the predominant form at the highest cAMP-Vfr concentration tested (113 nM), and the lower-mobility shift product (complex II) was observed as the cAMP-Vfr concentration was reduced. At the lowest cAMP-Vfr concentration examined (4 nM), complex II was the dominant shift product detected. The reason for the two distinct PlasR shift products is unclear; however, the subtle change in mobility likely reflects a difference in the conformation of the protein-DNA complex rather than variation in the number of Vfr molecules bound per probe.
To determine the role of cAMP in Vfr function, we used two different approaches to generate cAMP-free Vfr (apo-Vfr). As a first approach, we incubated the cAMP-Vfr complex overnight with a 5-fold molar excess of purified P. aeruginosa CpdA (see Fig. S1 in the supplemental material), a cAMP phosphodiesterase that degrades cAMP (16). To determine whether CpdA treatment affected the ability of Vfr to form complexes with its target promoters, the Vfr and CpdA mixture was used in EMSA experiments in the absence or presence of exogenous cAMP. CpdA treatment eliminated the formation of Vfr shift complexes with the Pvfr, PregA, PptxR, and PcpdA promoter probes (Fig. (Fig.2,2, lane 3). Interestingly, CpdA-treated Vfr was unable to form complex I with PlasR but retained the ability to form complex II (discussed below). When cAMP-Vfr was incubated overnight in the absence of CpdA (lane 2), with heat-inactivated CpdA (lane 4), or with a catalytically inactive CpdA mutant (CpdA-N93A; lane 8) (16), Vfr retained DNA binding activity. The latter results indicate that the loss of Vfr-dependent DNA binding was specifically associated with CpdA activity. The ability of CpdA-treated Vfr to shift the Pvfr, PregA, PptxR, and PcpdA probes was rescued by the addition of cAMP (50 μM) to the DNA binding reaction mixtures (lane 5). These results indicate that cAMP is necessary for Vfr binding to Pvfr, PregA, PptxR, and PcpdA.
To further assess the contribution of cAMP to Vfr function, we generated apo-Vfr by a second, nonenzymatic approach. Specifically, cAMP-Vfr was denatured in the presence of 6 M urea and cAMP was removed by dialysis. Vfr was then refolded by sequentially reducing the urea concentration in the absence or presence of 50 μM cAMP, and the DNA binding activity of recovered Vfr was examined by EMSA. Whereas Vfr refolded in the presence of cAMP retained the ability to bind the Pvfr probe (Fig. (Fig.3,3, lane 2), Vfr refolded in the absence of cAMP was unable to shift Pvfr (lane 3). These results are consistent with those obtained with CpdA-treated Vfr (Fig. (Fig.2)2) and suggest that refolding of Vfr in the absence of cAMP generates apo-Vfr. Cyclic AMP restored the DNA binding activity of refolded apo-Vfr for Pvfr in a concentration-dependent manner (Fig. (Fig.3,3, lanes 4 to 11). A 50% shift of the Pvfr probe occurred at ~50 nM cAMP (lane 8), and the maximal shift was achieved at 200 nM cAMP (lane 6). Taken together, these results demonstrate that for a representative set of Vfr-dependent promoters, cAMP is required for the DNA binding activity of Vfr.
In contrast to the Pvfr, PregA, PptxR, and PcpdA probes, our EMSA studies suggest that both cAMP-Vfr and apo-Vfr bind to the PlasR probe. As mentioned above (Fig. (Fig.2,2, lane 3), incubation of cAMP-Vfr with CpdA eliminated formation of the higher-mobility Vfr-PlasR complex (complex I); however, CpdA treatment had no effect on the formation of the lower-mobility complex (complex II). The addition of excess cAMP (50 μM) specifically restored formation of complex I (Fig. (Fig.2,2, lane 3 versus 5). Similar results were observed with refolded apo-Vfr; in the absence of exogenous cAMP only complex II was detected (Fig. (Fig.3,3, lane 3). The addition of exogenous cAMP to refolded apo-Vfr resulted in the formation of complex I in a concentration-dependent manner. For refolded apo-Vfr, ~25 nM cAMP was sufficient to shift 50% of the PlasR probe (lane 9), and the maximal shift was achieved between 50 and 100 nM cAMP (Fig. (Fig.3).3). These results indicate that Vfr is capable of forming distinct complexes with PlasR in a cAMP concentration-dependent manner and suggest that cAMP-Vfr and apo-Vfr may be responsible for the different PlasR shift products (complex I and II, respectively). As such, it appears that PlasR is a unique promoter in that Vfr can bind the PlasR probe in vitro without being fully saturated with cAMP. Although the two Vfr-PlasR shift products were also detected in EMSAs using Vfr presumed to be cAMP saturated (Fig. (Fig.1),1), it is possible that Vfr was not fully occupied with cAMP at the lower protein concentrations as a consequence of dilution. To test the possibility that cAMP dissociates from the cAMP-Vfr complex upon dilution, we examined the affinity of cAMP-Vfr for PlasR and PcpdA at various dilutions by EMSA in the presence of excess cAMP (50 μM). The Keq for cAMP-Vfr binding to PlasR and PcpdA was unaffected by the addition of excess cAMP (see Fig. S2 in the supplemental material), and both conditions gave results identical to those in Fig. Fig.1,1, indicating that the purified cAMP-Vfr complex is fully saturated with cAMP even when diluted to low nanomolar concentrations. Taken together, these results indicate that Vfr forms distinct complexes with PlasR in both a Vfr and cAMP concentration-dependent manner.
To address whether cGMP can regulate Vfr function, CpdA-generated apo-Vfr was incubated with an excess of cGMP (1 mM). Whereas cAMP restores DNA binding activity of apo-Vfr, cGMP was unable to restore Vfr binding to the Pvfr, PregA, PptxR, or PcpdA probes (Fig. (Fig.2,2, lane 6). Furthermore, cGMP did not support the formation of Vfr-PlasR complex I and appeared to reduce formation of complex II, suggesting that cGMP inhibits Vfr DNA binding activity. To further examine this possibility, we conducted competition experiments by simultaneously adding both cAMP and cGMP to CpdA-generated apo-Vfr protein. The presence of 50 μM cAMP and excess cGMP (1 mM) prevented Vfr from binding to the Pvfr, PregA, PptxR, and PcpdA probes and inhibited formation of PlasR complex I (Fig. (Fig.2,2, lane 5 versus 7). While addition of cGMP alone inhibited formation of Vfr-PlasR complex II, it appears that complex II was produced when both cGMP and cAMP were present. Currently, we are unable to explain this discrepancy. We next examined the effect of cGMP on Vfr binding to Pvfr, PtoxA, and PlasR promoter probes using refolded apo-Vfr. Similar to the results with CpdA-generated apo-Vfr, the addition of cGMP alone was not sufficient to restore binding of apo-Vfr to Pvfr and PtoxA, and it inhibited formation of Vfr-PlasR complex II (Fig. (Fig.4,4, lane 4). Again, consistent with the above results, the presence of both cAMP (50 μM) and excess cGMP (1 mM) prevented Vfr from binding to the Pvfr and PtoxA probes and formation of Vfr-PlasR complex I (Fig. (Fig.4,4, lane 7).
To determine the amount of cGMP required for inhibition of Vfr DNA binding activity, we conducted a titration experiment using the Pvfr probe. Refolded apo-Vfr shifted 90% of Pvfr in the presence of 100 nM cAMP (Fig. (Fig.5,5, lane 3), and 50% inhibition of the shift complex required an ~2,500-fold molar excess of cGMP (250 μM; lane 6). Vfr binding activity was unaffected in the presence of a 300-fold molar excess of cGMP (30 μM; lane 9). While these results indicate that cGMP can act as an inhibitor of cAMP-dependent Vfr DNA binding activity in vitro, the high concentration of cGMP required is unlikely to be achieved in vivo (see Discussion). Furthermore, it remains to be determined whether cGMP acts as a competitive or allosteric inhibitor.
The inhibitory effect of cGMP raised the question as to whether other cyclic nucleotides affect the DNA binding activity of Vfr. When added alone (1 mM) to refolded apo-Vfr, neither c-di-GMP nor c-di-AMP supported the formation of Vfr-DNA complexes with the Pvfr, PtoxA, or PlasR promoter probes (Fig. (Fig.4,4, lanes 5 and 6). In addition, neither signaling molecule inhibited the formation of Vfr-DNA complexes in the presence of cAMP (lanes 8 and 9). Thus, our data indicate that these nucleotides are unlikely to play a direct role in regulating Vfr function.
The vfr promoter region is predicted to encompass two putative Vfr binding sites, centered at bp −67.5 and −39.5 relative to the transcriptional start site (24) (Fig. (Fig.6A).6A). However, binding to either site has not been experimentally proven. Although both sites were present in the Pvfr probe examined by EMSA (Fig. (Fig.11 and and6A),6A), our data and those of a previous study (24) demonstrate that only a single Vfr-dependent shift product is formed. While not definitive, these results suggest that Vfr may bind to a single site within Pvfr. To address this issue, DNase I footprinting of Pvfr was performed. In the presence of cAMP-Vfr, a 32-bp sequence (bp −73 to −42 relative to the vfr transcriptional start site) was protected from DNase I cleavage (Fig. (Fig.6B).6B). The protected region partially overlapped the two previously predicted binding sites and the intervening sequence (Fig. (Fig.6A).6A). Within the protected region, enhanced cleavage sites were detected at bp −61 and bp −50/−51 (Fig. (Fig.6B).6B). The size of the footprint and spacing of the DNase I hypersensitivity sites are consistent with Vfr footprints of the toxA, regA, ptxR, fleQ, lasR, and cpdA promoters (2, 10, 14, 16, 24) and suggest that Vfr also binds to a single site within Pvfr. We identified a putative Vfr binding site (5′-GGATCACAGTC:CTGATAGCTGC) within the protected region by aligning the position of the DNase I hypersensitivity sites with those found in other published Vfr DNase I footprints (Fig. (Fig.6B;6B; see also Fig. S3 in the supplemental material). Enhanced DNase I cleavage is associated with distortion of the helical DNA structure and occurs at positions 5 and 6 within the conserved half-sites of CRP and Vfr binding sequences (24) (see Fig. S3). CRP has been shown to induce ~40° bends at equivalent positions in its target promoters (42), a distortion that likely accounts for increased DNase I sensitivity.
The putative Vfr binding site (5′-GGATCACAGTC:CTGATAGCTGC; underlined sequence portions are the conserved half-sites to which Vfr is predicted to bind) identified in the vfr promoter shares limited conservation with the proposed Vfr consensus binding sequence (5′-ANWWTGNGAWNY:AGWTCACAT, where dimeric Vfr is predicted to bind two half-sites) (24); however, the upstream half-site (TCACA) within the vfr promoter is identical to the consensus downstream half-site, suggesting that Vfr may tolerate a high degree of variability in the organization of its target binding sites. While further analysis is needed to confirm the exact Vfr binding site, our data support the notion that Vfr binds to its own promoter at a single site.
To determine whether Vfr binding is sufficient for activation of Pvfr, in vitro transcription assays were performed in the absence and presence of cAMP and using refolded apo-Vfr, RNA polymerase isolated from P. aeruginosa that is largely σ70 saturated (see Fig. S1 in the supplemental material), and a plasmid template carrying the vfr promoter region (bp −164 to +206 relative to the transcriptional start) (Fig. (Fig.6A).6A). A σ70-dependent transcript (~320 nucleotides [nt]), produced from the plasmid backbone, was detected under all conditions and served as a positive transcription control (Fig. (Fig.6C).6C). Whereas a faint band, corresponding in size to the predicted plasmid-encoded vfr transcript (~281 nt), was detected in the absence of cAMP or Vfr, the presence of both refolded apo-Vfr and cAMP (100 nM) increased the amount of the vfr transcript ~11-fold as determined by densitometry. In addition, we also observed a cAMP-Vfr-dependent transcript (~115 nt) generated from the plasmid backbone that fortuitously served as an internal control for cAMP-Vfr-dependent transcription. These data, along with the EMSA and DNase I footprinting results, suggest that binding of cAMP-Vfr to a single site in the vfr promoter is sufficient to activate Pvfr in vitro.
The finding that Vfr activates vfr transcription in vitro suggests that Vfr is subject to autoregulation in vivo. To test this hypothesis, β-galactosidase activity was measured in wild-type and vfr mutant strains from a lacZ transcriptional reporter carrying the same vfr promoter fragment used for the in vitro transcription assays (Fig. (Fig.6A).6A). The reporter was introduced at the P. aeruginosa chromosomal CTX phage attachment site (23). The wild-type strain displayed ~6-fold more reporter activity than the vfr mutant (P < 0.0001), indicating that Vfr positively regulates its own expression (Fig. (Fig.7A7A).
To address the in vivo role of cAMP in Vfr regulation, vfr promoter activity was measured in strains lacking one or both of the endogenous adenylate cyclases (CyaA and CyaB) (Fig. (Fig.7A).7A). While β-galactosidase activity in the cyaA mutant was unaffected, activity in the cyaB mutant was reduced by more than 50% compared to the wild type (P < 0.0001). Furthermore, activity in the cyaA cyaB double mutant was significantly reduced (P < 0.0001) to a level similar to that observed for the vfr mutant. Reduction in vfr promoter activity corresponded to reduced Vfr protein levels, as determined by Western blot analysis (Fig. (Fig.7A,7A, bottom panel). Thus, vfr expression is autoregulated and dependent on cAMP levels in vivo. Based on our EMSA studies, the simplest interpretation is that cAMP is required for Vfr DNA binding activity in vivo. Alternatively, cAMP binding may stabilize Vfr and protect it from degradation.
To distinguish whether cAMP affects Vfr transcriptional activity or Vfr protein stability, we compared vfr promoter activity when Vfr was ectopically expressed from an IPTG-inducible tac promoter to equivalent levels in the vfr mutant and in a triple mutant strain lacking both adenylate cyclases and chromosomal vfr (cyaAB vfr) (Fig. (Fig.7B).7B). When expressed to levels similar to those observed in a wild-type strain (Fig. (Fig.7B,7B, bottom panel), plasmid-encoded Vfr was sufficient to fully restore vfr promoter activity in the vfr mutant. In contrast, plasmid-encoded Vfr did not substantially increase vfr promoter activity in the cyaAB vfr triple mutant, demonstrating that intracellular cAMP is required for Vfr activity in vivo. Taken together, our results indicate that cAMP modulates the cellular levels of Vfr protein via Vfr autoregulation.
To examine the role of cAMP in regulating Vfr-dependent virulence factor expression, we evaluated the transcriptional activity of the toxA promoter. The regulation of toxA expression is complex and involves numerous regulators, including the products of the Vfr-dependent regA and ptxR genes (11, 14). In addition, our EMSA experiments and previous studies suggest that toxA expression is also directly regulated by Vfr (11, 15, 24, 55); however, the role of cAMP in toxA expression in vivo has not been tested. β-Galactosidase assays were used to measure toxA promoter activity in strains harboring a plasmid-borne toxA transcriptional reporter (pRtoxA-lacZ). Consistent with published data (11), toxA promoter activity was significantly reduced (P < 0.0001) in the vfr mutant, and ectopic expression of Vfr in this strain background was sufficient to restore transcriptional activity to that of the wild-type strain (Fig. (Fig.8).8). Expression from the toxA promoter was also significantly reduced (P < 0.0001) in the cyaAB vfr triple mutant; however, complementation with plasmid-encoded Vfr was not sufficient to restore wild-type promoter activity. To determine whether toxA expression correlated with secretion of the toxA gene product, ToxA, we assessed protein levels in culture supernatants by Western blot analysis (Fig. (Fig.8,8, bottom panel). Consistent with previously published data (11, 16, 55), the vfr mutant did not secrete detectable levels of ToxA, and plasmid-encoded Vfr complemented the vfr mutant phenotype (Fig. (Fig.8,8, bottom panel). In contrast, ectopic expression of Vfr in the cyaAB vfr triple mutant was not sufficient to restore ToxA secretion. Thus, like the vfr promoter, cAMP directly affects Vfr-dependent transcription of toxA.
In contrast to the Pvfr, PtoxA, PptxR, PregA, and PcpdA probes, our EMSA studies suggest that both apo-Vfr and cAMP-Vfr can bind the PlasR probe in vitro. To evaluate the role of cAMP in Vfr-dependent activation of lasR in vivo, we constructed a chromosomal transcriptional reporter in which the lasR promoter (bp −264 to +238 relative to the lasR T1 transcriptional start site) (Fig. (Fig.9A)9A) was fused to lacZ. β-Galactosidase activity from the lasR promoter reporter was then compared for the wild-type strain, the vfr mutant, and the double adenylate cyclase (cyaAB) mutant. As previously reported (2), there was a significant reduction (P < 0.0001) in reporter activity in the vfr mutant compared to the wild-type strain (Fig. (Fig.9B).9B). In contrast, lasR promoter activity was unaffected in the cyaAB double mutant. These data indicate that in vivo expression of lasR does not require cAMP synthesis. We previously showed that deletion of the P. aeruginosa cAMP phosphodiesterase gene cpdA results in a 30-fold increase in intracellular cAMP and a 12-fold increase in vfr expression (16). To evaluate the effects of increased cAMP and Vfr protein levels on lasR expression, we measured lasR promoter reporter activity in an isogenic cpdA mutant. Like the cyaAB mutant, the cpdA mutant retained wild-type levels of reporter activity (Fig. (Fig.9B).9B). Thus, the results of our in vitro studies correlate with our in vivo finding that cAMP synthesis is not required for Vfr to bind the lasR promoter. Furthermore, our results indicate that while Vfr is required for lasR expression, promoter activity is unaffected over a wide range of Vfr protein levels (Fig. (Fig.9B,9B, bottom panel) and cAMP concentrations. To further confirm these results, we demonstrated that plasmid-encoded Vfr is sufficient to restore lasR promoter activity in both a vfr mutant and a cyaAB vfr triple mutant (Fig. (Fig.9C9C).
To determine whether binding of apo-Vfr or cAMP-Vfr is sufficient for activation of PlasR, we conducted in vitro transcription assays in the absence and presence of refolded apo-Vfr and/or cAMP and using P. aeruginosa RNA polymerase, which was largely σ70 saturated, and a plasmid template carrying the lasR promoter region (bp −264 to +30 relative to the lasR T1 transcriptional start site) (Fig. (Fig.9A).9A). No lasR-specific transcripts (expected sizes of 102 and 132 nt) were detected under any of the reaction conditions used (Fig. (Fig.6C).6C). However, both cAMP-Vfr-dependent and -independent control transcripts were produced from the vector backbone, indicating that the transcription reactions worked as expected. To rule out the possibility that lasR transcription is initiated at a site downstream of the T1 and T2 sites identified by S1 nuclease protection assays (2), we created a second in vitro transcription template carrying the same full-length promoter fragment (bp −264 to +238 relative to the T1 transcriptional start site) as used in the in vivo promoter reporter assays (Fig. (Fig.9A).9A). Again, both control transcripts were produced from the vector backbone, but lasR-specific transcripts were not detected (data not shown). Taken together, these results suggest that an additional factor(s) is required for Vfr-dependent activation of lasR or that an alternative σ factor (other than σ70) is necessary for the recruitment of RNA polymerase to the lasR promoter.
Expression from the lasR promoter was previously reported to be cell density dependent, and a putative lux box associated with quorum-sensing-dependent gene expression was identified within the lasR promoter (2). Given the possibility that lasR may be autoregulated, we examined Vfr-dependent expression of the lasR promoter reporter in a lasR mutant (Fig. (Fig.9D).9D). β-Galactosidase activity in the lasR mutant was unaffected compared to the wild-type strain. Furthermore, reporter activity in a lasR vfr double mutant was indistinguishable from the vfr mutant (Fig. 9A and D). To determine if LasR facilitates Vfr-dependent expression from the lasR promoter in vivo in the absence of cAMP, we assessed promoter reporter activity in a lasR cyaA cyaB triple mutant. Again, lasR deletion did not affect promoter reporter activity, indicating that a factor other than LasR is likely to contribute to Vfr-dependent transcription from the lasR promoter.
In this study, we demonstrate that cAMP acts as a positive regulator of Vfr by promoting Vfr DNA binding to multiple virulence gene promoters. Although we demonstrated that the cyclic nucleotide specificity of Vfr is similar to that of E. coli CRP, it is not clear whether the changes in protein structure that occur in CRP in response to cAMP binding also occur in Vfr. In CRP, binding of cAMP causes a series of transitions in the protein structure that ultimately contribute to a repositioning of the DNA binding domain to an orientation compatible for specific interactions with DNA (34-36, 43). Although Vfr and CRP are highly homologous (67% identical and 91% similar) (55), it is possible that particular residue differences, such as those within the Vfr nucleotide binding pocket (4, 55), induce conformational effects that are different than those occurring in CRP, resulting in unique interactions between Vfr and RNAP and/or target promoter DNA. Ultimately, structural studies comparing cAMP-Vfr and apo-Vfr are needed to determine the nature of the allosteric change induced by cAMP and may provide insight as to why CRP cannot functionally substitute for Vfr (48, 55). Further, we cannot formally rule out the possibility that Vfr is controlled by an effector that does not regulate CRP and is different from the cyclic nucleotides tested in this study.
We observed that Vfr-DNA binding activity was inhibited by cGMP, which is a property that is also exhibited by CRP (3, 13, 49). However, the relevance of Vfr regulation by cGMP in vivo is questionable, since inhibition of Vfr-DNA binding activity required a 2,500-fold excess of cGMP when cAMP levels were within a biologically relevant range (16, 17). Overall, there have been few reports of the presence of cGMP in bacteria, and thus a physiological role of this nucleotide in prokaryotes has yet to be defined (29). The P. aeruginosa genome encodes a single enzyme (CyaB) with homology to guanylate cyclase, but we have shown that CyaB has adenylate cyclase activity and possesses critical substrate-determining residues consistent with ATP being the preferred substrate (17, 30). Although the P. aeruginosa cyclic nucleotide phosphodiesterase CpdA was shown to exhibit cGMP phosphodiesterase activity in vitro, intracellular cGMP levels remained below the limit of detection (≤0.01 μM) in mutants lacking cpdA (16). Taken together, these findings suggest that cGMP is unlikely to play a biological role in Vfr control.
A potential control point in regulating the activity of the cAMP/Vfr signaling pathway is at the level of vfr expression. We have provided both in vivo and in vitro evidence that Vfr regulates transcription from its own promoter and that vfr expression is cAMP dependent. Our data demonstrate that Vfr is a positive regulatory factor with respect to its own transcription. In contrast, the E. coli crp promoter is negatively autoregulated by CRP, except at high concentrations of cAMP, where positive autoregulation has been reported (1, 8, 21). The different modes of autoregulation displayed by the crp and vfr promoters are presumably due to their different promoter architectures and reflect their specific regulatory roles in E. coli and P. aeruginosa (catabolite repression versus virulence factor expression, respectively). Our DNase I footprinting result indicates that cAMP-Vfr binds to a single unique site centered approximately 58 bp upstream of the vfr transcriptional start site. This spacing is similar to that of the secondary CRP binding site (CRPII) responsible for positive crp autoregulation (21), raising the possibility that transcriptional activation by Vfr and CRP at these sites occurs by a similar mechanism. In addition, the fact that Vfr binds a single unique region within the vfr promoter in vitro was surprising, given how closely the two predicted Vfr binding sites matched the consensus sequence (24); this result suggests that some Vfr binding sites may be difficult to predict based on bioinformatic data alone.
Our observation that cAMP was not required for Vfr binding to the lasR promoter in vitro or for lasR promoter activity in vivo provides an exception to our overall finding that expression of multiple virulence genes requires cAMP for Vfr activation. While we cannot formally rule out the possibility that P. aeruginosa produces low levels of cAMP by some other mechanism, intracellular levels of cAMP are greatly reduced in a cyaAB mutant compared to a wild-type strain (16, 17, 58). Neither apo-Vfr nor cAMP-Vfr was sufficient to activate transcription from the lasR promoter in vitro, suggesting that an additional factor(s) is required for Vfr-dependent expression of lasR, such as an additional transcriptional regulator or an alternative sigma factor (different from σ70). We ruled out the possibility that LasR itself regulates lasR and/or promotes Vfr-dependent activation, as we observed no effect on lasR reporter activity in strains lacking lasR. An alternative candidate is GacA, the transcriptional regulator of the GacS/GacA signaling cascade involved in extracellular polysaccharide synthesis and biofilm formation (19, 53). GacA was shown to regulate lasR expression in vivo by an unknown mechanism (37), and further experiments are needed to establish whether this regulator and/or other downstream factors are critical for regulation of lasR by apo-Vfr.
Although the mechanism for apo-Vfr activation of lasR remains to be determined, our findings may account for the previously reported phenotypes of a Vfr mutant in which 5 amino acid residues from the putative cAMP binding domain were deleted (VfrΔEQERS) (4). Beatson et al. showed that plasmid-based expression of VfrΔEQERS in a vfr mutant restored quorum-sensing-dependent production of elastase but not Tfp-dependent twitching motility. To explain this phenomenon, those authors proposed that VfrΔEQERS might be responding to cGMP or another effector to differentially regulate gene expression. Taking into account the findings presented in our study, it is more likely that VfrΔEQERS, like apo-Vfr, activates lasR expression, which in turn would support the subsequent expression of downstream quorum-sensing factors like elastase. Nevertheless, the results from both studies still raise the question as to why the las quorum-sensing system is dependent on Vfr but not cAMP. It is possible that the simultaneous activation of the cAMP-dependent and quorum-sensing-dependent regulons is counterproductive under conditions that alter intracellular cAMP levels.
Since Vfr activity is ultimately dependent upon the cellular concentration of cAMP (except in the case of lasR regulation), understanding the mechanisms by which cAMP levels are controlled is critical to expanding our knowledge of virulence factor regulation in P. aeruginosa. Currently, the environmental signals that trigger P. aeruginosa to upregulate or downregulate cAMP/Vfr-dependent virulence factor production are not known. Given that multiple cAMP-dependent virulence factors are required for acute P. aeruginosa infection and that acute virulence factor inhibition is associated with chronic P. aeruginosa infection, we hypothesize that bacterial cAMP synthesis is stimulated upon encountering the host environment but then is subject to downregulation upon onset of the chronic state. We predict that intracellular cAMP fluctuations occur in response to spatio-temporal host signals that are unique to the environments encountered during acute and chronic infection. Further, cAMP downregulation may occur prior to (or in addition to) the well-documented genetic inactivation of virulence factors and their regulators (such as Vfr) that occurs among chronic infection isolates (46). In terms of cAMP activation, we have shown that the Chp chemotaxis-like chemosensory signal transduction system controls cAMP levels via modulation of CyaB adenylate cyclase activity (17). Once cAMP is synthesized, the results of the present study suggest that Vfr regulates virulence promoters in either a cAMP-dependent (ptxR, regA, and toxA) or cAMP-independent (lasR) manner. The cAMP-dependent branch of the pathway, in combination with vfr autoregulation, may serve as a signal amplification loop, whereby subtle changes in intracellular cAMP can have large effects on gene expression. The cAMP-independent branch may allow differential regulation of Vfr-dependent (but not cAMP-dependent) systems, such as quorum sensing. Ultimately, the cAMP signaling cascade is predicted to be reset following CyaB activation by the cAMP phosphodiesterase CpdA. Further investigation is needed to define environmental cues responsible for activating cAMP/Vfr signaling and the mechanisms by which these signals are integrated into the pathway in an effort to understand the specific impacts on virulence regulation in different phases of P. aeruginosa infection.
This work was supported by grants from the Cystic Fibrosis Foundation (to M.C.W.) and the National Institutes of Health (AI069116 to M.C.W. and AI055042 to T.L.Y.). E.L.F. was supported by a Pathogenesis Training Grant from the University of North Carolina Center for Infectious Diseases.
We thank members of the Wolfgang and Yahr laboratories for their constructive suggestions and critical review of the manuscript. We thank Katrina Forest for anti-Vfr serum.
Published ahead of print on 21 May 2010.
†Supplemental material for this article may be found at http://jb.asm.org/.