|Home | About | Journals | Submit | Contact Us | Français|
The Cpx two-component system is thought to mediate envelope stress responses in many gram-negative bacteria and has been implicated in the pathogenicity of several enteric pathogens. While cues that activate the Escherichia coli Cpx system have been identified, the nature of the molecular signals that stimulate this pathway is not well understood. Here, we investigated stimuli that trigger this system in Vibrio cholerae, a facultative pathogen that adapts to various niches during its life cycle. In contrast to E. coli, there was no basal activity of the V. cholerae Cpx pathway under standard laboratory conditions. Furthermore, several known stimuli of the E. coli pathway did not induce expression of this system in V. cholerae. There were no defects in intestinal growth in V. cholerae cpx mutants, arguing against the idea that this pathway promotes V. cholerae adaptation to conditions in the mammalian host. We discovered that chloride ions activate the V. cholerae Cpx pathway, raising the possibility that this signal transduction system provides a means for V. cholerae to sense and respond to alterations in salinity. We used a genetic approach to screen for mutants in which the Cpx pathway is activated. We found that mutations in genes whose products are required for periplasmic disulfide bond isomerization result in activation of the Cpx pathway, suggesting that periplasmic accumulation of proteins with aberrant disulfide bonds triggers the V. cholerae Cpx pathway.
Two-component signal transduction systems enable bacteria to transfer information about the environment across the cytoplasmic membrane to the cytoplasm, thereby facilitating cellular responses to changing environmental conditions. The sensory components of these signal transduction systems are membrane-embedded protein kinases. When the sensor protein is stimulated by a specific signal, it autophosphorylates a histidine residue and then transfers this phosphate to a conserved aspartate residue in a cytoplasmic response regulator, a transcription factor. Phosphorylation of the response regulator alters its activity and, hence, results in altered expression of the proteins that it regulates, which constitutes the cellular response to the initial signal. While two-component systems are the most widespread signal transduction systems in bacteria, determining the precise signals that stimulate these pathways has proved to be difficult.
The CpxRA two-component system is found in many gram-negative bacteria, especially in gammaproteobacteria. In this two-component system, which has been extensively characterized in Escherichia coli, the inner-membrane-spanning histidine kinase CpxA functions as the sensor and the cytoplasmic CpxR is the response regulator (Fig. (Fig.1A).1A). In the absence of inducing stimuli, CpxA acts upon CpxR as a phosphatase, rendering it inactive. The activation of CpxA's kinase activity results in the phosphorylation of CpxR, which promotes its binding to regulatory sites upstream of target genes (50). CpxR activity is also thought to be regulated by CpxP, which is encoded by a divergently transcribed gene immediately upstream of the cpxRA operon (12, 51). CpxP is a periplasmic protein that likely interacts with the periplasmic domain of CpxA and inhibits its autophosphorylation activity, thereby reducing phosphorylation of CpxR (24). However, the interplay between CpxA and CpxP and the physiological role of CpxP are not fully understood.
Perturbations in the tripartite gram-negative cell envelope, comprised of the outer membrane, periplasm, and inner membrane, are thought to stimulate CpxR activity. While several cues that induce the Cpx pathway are known, including alkaline pH (12), copper (61), changes in the lipid composition of the inner membrane (39), overproduction of the outer-membrane lipoprotein NlpE (55), accumulation of misfolded variants of maltose binding protein (31), and elevated osmolarity (49), the molecular mechanisms through which these stimuli are transmitted to the Cpx pathway are not well understood. In general, misfolded cell envelope proteins are postulated to be the triggers of this pathway. However, not all misfolded periplasmic proteins serve as stimuli (33, 42), a fact clearly demonstrated by studies of the biogenesis of the P pilus in uropathogenic E. coli. P pilus subunits are transported across the periplasm to an outer-membrane pilus assembly site by the chaperone PapD. In the absence of PapD, all of the subunits accumulate in the periplasm, but only accumulated PapE and PapG induce the Cpx pathway (33). Furthermore, a particular structural feature found at the amino terminus of PapE has been shown to be required but not sufficient for the induction of the Cpx system (36).
Stimulation of the Cpx pathway is thought to alleviate “envelope stress.” Phosphorylated CpxR (CpxR~P) activates the transcription of several genes, including those that encode the extracytoplasmic protease DegP, chaperones such as the peptidyl prolyl cis/trans isomerase PpiA, and the disulfide oxidoreductase DsbA (11, 47). Collectively, these proteins may eliminate or repair damaged envelope proteins. CpxR~P also activates the transcription of cpxRA and cpxP, presumably thereby facilitating both amplification of the stress signal and restoration of basal expression levels once the stimulus is removed. However, the function of the Cpx pathway likely extends beyond the maintenance of cell envelope homeostasis (48). In E. coli, CpxR~P represses the expression of motB and aer (15, 16), leading to a reduction in cell swarming. Furthermore, it was reported that activation of the Cpx pathway reduces E. coli's growth in biofilms (21).
The results of several studies suggest that the Cpx pathway is important for the virulence of pathogenic E. coli and closely related pathogens, due at least in part to its influence on the production of virulence-linked extracellular appendages. Disruption of cpxR in enteropathogenic E. coli reduced the production of type IV bundle-forming pili and the adherence of this enteric pathogen to tissue-cultured eukaryotic cells (44). In uropathogenic E. coli, the Cpx pathway controls the expression of factors important for the assembly of P pili, and fewer and shorter P pili were observed in a cpxR null mutant (30). A Salmonella enterica serovar Typhimurium cpxA mutant was attenuated in mice (29). Finally, in Shigella sonnei, the Cpx pathway regulates the expression of activators of the type III secretion system in response to extracellular pH change (43).
Here, we investigated the Cpx pathway in Vibrio cholerae. This gram-negative facultative pathogen causes cholera, a severe diarrheal illness that can be fatal if untreated. We tested whether the Cpx pathway is important for V. cholerae pathogenicity, as reported for the other intestinal pathogens mentioned above. However, we found that V. cholerae cpx mutants were not attenuated in the suckling mouse model of cholera. In contrast to the E. coli Cpx pathway, which has significant basal expression when cells are grown in rich medium, there was no detectable basal activity of the V. cholerae Cpx pathway. Furthermore, several cues known to stimulate the E. coli Cpx pathway did not trigger the V. cholerae Cpx pathway. We found that chloride ions specifically activated Cpx. We also report here that mutations in genes whose products are required for periplasmic disulfide bond isomerization in V. cholerae triggered the Cpx pathway. These findings suggest that periplasmic accumulation of proteins with aberrant disulfide bonds may trigger the V. cholerae Cpx pathway and provide new understanding of the activators of the Cpx two-component system.
The sequenced Vibrio cholerae El Tor clinical isolate N16961 (27) was used as the parental strain to create all the V. cholerae strains used in this study. E. coli strain DH5αλpir was used as the host strain for plasmid construction. E. coli SM10λpir was used as the donor to mobilize plasmids into V. cholerae via conjugation. E. coli BL21λDE3 was used for protein overproduction. Unless otherwise noted, cells were grown in Luria-Bertani (LB) medium and stored at −80°C in LB medium containing 20% glycerol.
The antibiotic concentrations used for selection of V. cholerae and E. coli, respectively, were as follows: streptomycin, 200 μg/ml; kanamycin, 50 μg/ml; chloramphenicol, 5 μg/ml and 20 μg/ml; spectinomycin, 100 μg/ml; carbenicillin, 50 μg/ml; and ampicillin, 80 μg/ml and 100 μg/ml. X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactoside) was used at a final concentration of 120 μg/ml or 240 μg/ml.
Plasmid DNA was prepared using Qiagen kits, and chromosomal DNA was prepared using genome DNA kits from Q-Biogene or Wizard genomic kits from Promega. Restriction enzymes and T4 DNA ligase were purchased from New England Biolabs. PCR was performed using a PTC-200 thermal cycler from MJ Research, with Pfx Platinum polymerase from Invitrogen, Pfu Ultra II polymerase from Stratagene, or HotStart Taq Mastermix from Qiagen. The oligonucleotides for PCR amplification were purchased from Operon (Huntsville, AL). DNA sequencing was performed by GeneWiz (South Plainfield, NJ).
A complete list of the plasmids used in this study is shown in Table Table1.1. pCpxR contains the promoterless cpxR gene (vc2692) under the control of the arabinose-inducible PBAD promoter. cpxR was amplified by PCR using N16961 chromosomal DNA as template and primers cpxR9 and cpxR10 (Table (Table2)2) and then inserted into pBAD33 (26) between the KpnI and XbaI restriction sites.
CpxR-His was overproduced using pET28b-TEV-CpxR. This plasmid was constructed by inserting a BsaI-XhoI fragment corresponding to the cpxR coding sequence between the NcoI and XhoI sites of pET28b-TEV (58). The DNA fragment corresponding to cpxR was generated by PCR using primers cpxRH1 and cpxRH2.
The promoter region of cpxP was amplified by PCR using primers PcpxP1 and PcpxP2 and cloned between the BamHI and HindIII sites of the lacZ fusion vector pCB182 (53). The fragment containing the transcriptional fusion between the promoter region of cpxP and the promoterless lacZ gene was then amplified by PCR using primers PcpxP1 and YPR331 and cloned between the BamHI and KpnI sites of pWKS30 (59). The resulting plasmid was designated pP'Z.
The allele exchange vector pJL1PcpxP'Z was constructed by the following steps. The promoterless lacZ gene carried by pCB182 (53) was amplified by PCR using primers lacZ5 and lacZ6 and then inserted between the StuI and NotI sites of pJL1 (7) to generate pJL1lac. The promoterless lacZ gene is now preceded by the following unique restriction sites: BglII, PacI, StuI, NheI, and XhoI. The promoter region of cpxP was then amplified by PCR using primers PcpxP5 and PcpxP6 and inserted between the PacI and StuI sites of pJL1lac.
The allele exchange vectors used to delete cpxR, cpxA, cpxP, dsbE, dsbC, and dsbA, respectively, were all made by using the following scheme. Sequences 5′ and 3′ of each of these loci were amplified by PCR (primers are shown in Table Table3)3) yielding 5′ and 3′ fragments that were subsequently ligated with an antibiotic resistance gene and then inserted into pCVD442 (20), a pir-dependent sacB-containing allele exchange vector. The 5′ fragment for pΔcpxR was purified as an XbaI/BamHI fragment, and the 3′ fragment was purified as an EcoRI/SphI fragment. A kanamycin resistance cassette was amplified by PCR using pKD4 (13) as a template and primers Kan5 and Kan3 and then digested with BamHI and EcoRI. The terminator of transcription from the rrnB operon was amplified from pCB182N (a derivative of pCB182 [53; H. Kimsey, unpublished data]) using primers term3b and term5b, digested with HindIII and EcoRI, and cloned along with the kanamycin resistance cassette in pUC19 (63) between the BamHI and EcoRI sites. The kanamycin resistance gene followed by the transcription terminator was then purified as a BamHI/EcoRI fragment. This fragment, along with the 5′ and 3′ fragments flanking cpxR, was ligated with XbaI- and SphI-digested pCVD442. The pΔcpxA vector was constructed by ligating the BamHI- and EcoRI-digested kanamycin resistance gene followed by the transcription terminator and the 5′ and 3′ regions of cpxA that were amplified by PCR using primer pairs cpxA9/cpxA10 and cpxA11/cpxA12 with XbaI- and SphI-digested pCVD442. The pΔcpxP vector was constructed by ligating the kanamycin resistance gene digested with BamHI and EcoRI and the 5′ and 3′ regions of cpxP that were amplified by PCR using primer pairs cpxP1/cpxP2 and cpxP3/cpxP4 with XbaI- and SphI-digested pCVD442.
Plasmid pA* was generated by amplifying two regions in cpxA using primer pairs cpxA5/cpxA6 and cpxA7/cpxA8 and then joining the two fragments by spliced overlap extension PCR with primers cpxA5 and cpxA8. The final PCR product was then cloned between the XbaI and SphI sites of pCVD442.
Plasmid pΔdsbE contains sequences 5′ and 3′ of dsbE (vc2051) that were amplified by PCR using primer pairs dsbE1/dsbE2 and dsbE3/dsbE4, digested with XbaI/BamHI and BamHI/SphI, respectively, and then ligated between the XbaI and SphI sites of pCVD442.
Sequences in the 5′ and 3′ regions of dsbC (vc2418) were amplified by PCR using primer pairs dsbC1/dsbC2 and dsbC3/dsbC4, digested with XbaI/BamHI and HindIII/SphI, respectively, and then ligated along with a kanamycin resistance cassette (amplified by PCR from plasmid pKD4 using primers Kan5 and Kan3d and digested with BamHI and HindIII) between the XbaI and SphI sites of pCVD442, yielding pΔdsbC.
Plasmid pΔdsbA was constructed as follows. The chloramphenicol resistance gene was amplified by PCR from pKD3 (13) using primers P1 and P2 and digested with BamHI and NheI. Sequences 5′ and 3′ of dsbA (vc0034) were amplified by PCR using primer pairs dsbA1/dsbA2 and dsbA3/dsbA4, digested with XbaI/BamHI and NheI/SphI, respectively, and then ligated along with the chloramphenicol resistance cassette between the XbaI and SphI sites of pCVD442.
Plasmid pGPdsbB was constructed by ligating an internal fragment of dsbB, amplified by PCR using primers dsbB1 and dsbB2, between the BglII and XbaI sites of pGP704 (41). To prevent any potential residual activity of the truncated protein generated by the insertion of pGPdsbB into the dsbB locus, primer dsb2 was designed to change the cysteine at position 133 to a tryptophan.
Plasmids pBRfA-DsbC, pBRfA-TolC, pBRfA-VC1888, pBRfA-VC1898, and pBRfA-HutA, which were used to complement transposon insertion mutations in these six genes, were generated using the Gateway cloning system. We initially constructed a derivative of pBAD43 (a kind gift from Jon Beckwith) to serve as the “destination vector” for the Gateway cloning of the genes of interest. This vector contains the RfA cassette provided with the Gateway cloning kit (Invitrogen). The open reading frames corresponding to V. cholerae dsbC, tolC, vc1888, vc1898, and hutA subcloned into the pDONR were selected from the V. cholerae Gateway library (provided by the Pathogen Functional Genomics Resource Center). We flipped the open reading frame from pDONR to pBRfA by following the Gateway cloning kit protocol (Invitrogen).
DNA sequencing was performed on all plasmids to confirm that the cloned regions were identical to the published sequences.
A list of the V. cholerae strains used here is shown in Table Table3.3. LSP'Z, an N16961 derivative containing a PcpxP'lacZ transcriptional fusion inserted into the V. cholerae lacZ locus, was made by initially mobilizing pJL1PcpxP'Z from SM10λpir into N16961 by conjugation. The transconjugants were grown in LB medium at 37°C and then grown on plates containing sucrose and streptomycin to select for cells that had lost the allele exchange vector after homologous recombination. The insertion of the PcpxP'lacZ transcriptional fusion into the V. cholerae lacZ locus (vc2338) was confirmed by PCR with primers that anneal outside of the V. cholerae lacZ locus.
To generate the deletion mutants, the allele exchange vectors described above, all of which are derivatives of pCVD442, were initially mobilized from SM10λpir via conjugation into N16961 lacZ or LSP'Z. Then, the exconjugants were grown in LB medium at 37°C in the presence of appropriate antibiotics and subsequently grown on sucrose plates, supplemented with chloramphenicol or kanamycin when needed, to select for cells that had lost the allele exchange vector after homologous recombination. The resulting strains were verified by PCR using primers that anneal outside of the deleted region. The following strains were generated in this fashion: LSΔR, LSΔP, LSΔA, LSA*, LSΔRP'Z, LSΔPP'Z, LSΔAP'Z, LSA*P'Z, LSΔdsbEP'Z, LSΔdsbCP'Z, and LSΔdsbAP'Z.
LSdsbBinsP'Z, an LSP'Z derivative with an insertionally inactivated dsbB, was made by mobilizing pGPdsbB from SM10λpir by conjugation into LSP'Z. Exconjugants were selected as streptomycin- and carbenicillin-resistant colonies. Insertion of the plasmid in dsbB was verified by PCR using primers that anneal outside of the locus.
LSΔP, LSΔR, LSΔA, and LSA* (all Lac−) and wild-type N16961 (Lac+) were grown overnight at 37°C in LB medium in the presence of streptomycin, as well as kanamycin for the deletion mutants. The cultures were then diluted 1:1,000 into fresh LB medium. The experiment was carried out in a competition format where one-to-one mixtures of each mutant and the wild-type strain were intragastrically inoculated into 5-day-old CD-1 suckling mice, essentially as described previously (18, 57). Mice were sacrificed 20 h after inoculation, the small intestine was harvested and ground, and serial dilutions were plated on LB medium supplemented with X-Gal and streptomycin.
LSP'Z harboring pBAD33-CpxR was grown in LB medium at 37°C to an optical density at 600 nm (OD600) of ~1.5. The culture was then split in half, and arabinose or glucose was added to the cultures to a final concentration of 0.1%. Aliquots from the two cultures, grown at 37°C, were removed periodically for assay of their β-galactosidase activities as described previously (40). Briefly, 0.5-ml aliquots from the cultures were pelleted and resuspended in 0.5 ml of Z buffer (16.1 g/liter NaH2PO4, 5.5 g/liter Na2HPO4, 0.75 g/liter KCl, 0.246 g/liter MgSO4) and β-mercaptoethanol (40 mM final concentration). The assays were then conducted as follows. Two hundred microliters from each sample was added to 800 μl of Z buffer supplemented with 20 μl of chloroform and 20 μl of 0.1% sodium dodecyl sulfate. After being vortexed for 5 s, the tubes were placed in a heat block at 30°C for 5 min. Two hundred microliters of ONPG (o-nitrophenyl β-d-galactopyranoside) (0.4% in Z buffer) was added to each tube to initiate the reaction. The reaction was then stopped by the addition of 500 μl of Na2CO3. The measurements and the calculations were executed as previously described (40).
The β-galactosidase activities in strains harboring the plasmid-borne PcpxP-lacZ fusion pP'Z were determined in a similar fashion. Strains were initially grown overnight at 37°C in LB medium in appropriate antibiotics and then diluted 1:100 in fresh medium. Aliquots of the cultures were removed periodically, and β-galactosidase activities were measured as described above.
To compare the levels of activity of the PcpxP'lacZ reporter in different strains on plates, strains were patched on LB plates supplemented with X-Gal at a final concentration of 240 μg/ml. Blue color was then assessed by eye.
We used plasmid pSC189, a suicide vector carrying a Himar mini transposon marked with a kanamycin resistance gene (9), to generate a transposon insertion library in LSP'Z. E. coli SM10λpir(pSC189) and LSP'Z were grown overnight in LB medium supplemented with appropriate antibiotics. Two hundred microliters of each overnight culture was washed twice with fresh LB medium to remove any residual antibiotics, resuspended in 50 μl of LB medium, and mixed together. The mixture was spotted onto a nitrocellulose filter (with a pore diameter of 0.45 μm) on an LB plate and incubated at 37°C for 8 h. The mixtures were then resuspended in 2 ml of LB medium and spread on 40 plates (50 μl per plate) supplemented with streptomycin (200 μg/ml), kanamycin (150 μg/ml), and X-Gal (240 μg/ml). Colonies were screened for their color and sensitivity to ampicillin, which reflects loss of the pSC189 plasmid. To map the sites of Himar insertion in strains selected for further analysis, genomic DNA was extracted, digested with SalI and XhoI, ligated into pBluescriptIIKS(+) (Stratagene), and transformed into E. coli cells. Finally, the inserts in the plasmids were sequenced using primer Himar3out. A Himar transposon library in LSΔdsbCP'Z was generated in a similar fashion, but pSC137 (8), a suicide vector carrying a Himar mini transposon marked with a chloramphenicol resistance gene, was used instead of pSC189.
To purify CpxR, we constructed a C-terminally His-tagged version of CpxR by inserting cpxR into pET28b-TEV. The pET28b-TEV-CpxR expression vector was introduced into E. coli strain BL21λDE3, and the resulting strain was grown in 1 liter of LB medium supplemented with kanamycin until an OD600 of ≈0.6 was reached. Isopropyl-β-d-thiogalactopyranoside (IPTG; 1 mM) was then added to the culture, which was grown for an additional 3 h at room temperature. The cells were centrifuged at 4,000 × g for 20 min and resuspended in 20 ml of buffer A (50 mM NaH2PO4·H2O, 300 mM NaCl, 10 mM imidazole, 10% glycerol, pH 8). The cells were disrupted with a French press, and cell debris was removed by centrifugation at 12,000 × g for 20 min. The resulting crude protein extract was loaded onto a 1-ml HisTrap FF crude column (Amersham) that was previously equilibrated with buffer A and installed on a fast liquid protein chromatography system. CpxR-His was eluted with an imidazole gradient (10 to 500 mM) and analyzed on a 12% NuPage gel (Invitrogen). Protein concentrations were determined by using Coomassie Plus protein assay reagent from Pierce.
Strains were patched onto LB agar plates supplemented with the appropriate antibiotics and grown overnight at 25°C. Cells were then scraped up and resuspended in 250 μl of distilled water and kept on ice. A 2.5-μl amount of 10% sodium dodecyl sulfate was added to the samples, which were then left on ice for 10 min. Five microliters of 2.5 mg/ml DNase I was then added to the tubes, and they were incubated at 37°C for 15 min. NuPage LDS sample buffer (Invitrogen) and dithiothreitol (0.01 M final concentration) were then added, and after heating at 95°C for 5 min, aliquots were run on a NuPage Novex 4-to-12% Bis-Tris midi gel. Proteins were transferred to nitrocellulose membranes by using an iBlot apparatus (Invitrogen) and probed with rabbit antisera raised against CpxRHis (Prosci, CA). Horseradish peroxidase-conjugated anti-rabbit antisera (Pierce) and a SuperSignal West Pico chemiluminescent substrate kit (Pierce) were used to detect the bound primary antibodies. Membranes were then incubated with Restore Western stripping buffer (Pierce) to remove bound antibodies. Membranes were subsequently probed with anti-L1 antibodies, and the subsequent steps of the Western blot were carried out as described above.
Using the TIGR annotation of the V. cholerae N16961 genome, we compared the V. cholerae and E. coli cpx loci. The cpx loci in these related gammaproteobacteria are organized similarly. cpxP is oriented divergently from cpxR (Fig. (Fig.1B),1B), and cxpR and cpxA overlap, suggesting that cpxRA constitute an operon, as in E. coli (19). The intergenic region between cpxP and cpxR in V. cholerae contains sequences similar to the CpxR binding sites found at comparable positions in E. coli (16). These sites mediate the activation of cpxRA and cpxP transcription by CpxR. The V. cholerae cpx locus is flanked by vc2690 (encoding a conserved hypothetical protein) and vc2694 (encoding a superoxide dismutase, SodA), which are homologous to genes that surround the E. coli cpx locus. However, in E. coli there is a gene intervening between cpxA and the sodA homologue.
Although the V. cholerae and E. coli cpx loci are organized similarly, there are differences between the predicted amino acid sequences of the Cpx proteins in these two bacterial species (Fig. (Fig.1B).1B). The CpxR proteins are 60.3% identical, whereas the CpxP proteins are only 21.6% identical. Overall, the CpxA proteins are 43.6% identical, but the degree of conservation varies significantly between their constituent domains. The cytoplasmic domains of the CpxA proteins, thought to interact with CpxR, are 54.3% identical, whereas the periplasmic loops, which presumably interact with CpxP and perhaps sense activating stimuli, are only 20.7% identical (Fig. (Fig.1B).1B). The relative lack of conservation of CpxP and the periplasmic domain of CpxA in the two species may reflect differences between the signals that activate the Cpx signal transduction system in these two gram-negative organisms.
We constructed deletion mutants of the three genes that comprise the cpx locus in the sequenced V. cholerae El Tor clinical isolate N16961. Each of the three cpx genes was replaced with a kanamycin resistance cassette, yielding the following strains: LSΔR (N16961 ΔcpxR::Kn), LSΔA (N16961 ΔcpxA::Kn), and LSΔP (N16961 ΔcpxP::Kn). The ΔcpxR::Kn mutation in LSΔR probably abolishes transcription of cpxA as well. We also generated LSA*, an N16961 derivative that lacks part of the periplasmic region of CpxA (amino acids 97 to 133); based on studies of CpxA activity in E. coli (50), the mutant CpxA in LSA* (CpxA*) should constitutively phosphorylate CpxR.
We constructed a reporter of CpxR activity to explore the conditions that activate the V. cholerae Cpx pathway. Genome-wide microarray-based studies had revealed that cpxP is one of the most-transcribed genes when cpxR is overexpressed in V. cholerae N16961 (not shown). We therefore fused the promoter region of cpxP to a promoterless lacZ to generate a cpxP transcription reporter on plasmid pWKS30 (designated pP'Z). This PcpxP-lacZ fusion was also introduced into the V. cholerae lacZ locus, yielding LSP'Z, a strain which has an intact cpx locus as well as a chromosomal reporter of cpxP transcription. Unexpectedly, there was virtually no detectable β-galactosidase activity in all phases of growth of wild-type V. cholerae harboring pP'Z (Fig. (Fig.2A)2A) or LSP'Z (Fig. (Fig.2D).2D). A cpxR-lacZ transcriptional fusion was similarly inactive (data not shown), confirming that the Cpx pathway has little basal activity in V. cholerae. In contrast, a PcpxP-lacZ reporter in E. coli has significant basal activity (12). Expression of the V. cholerae cpxP reporter was detected when CpxR was overexpressed from the arabinose-inducible promoter PBAD (Fig. (Fig.2B),2B), confirming that cpxP is under the control of CpxR and that its transcription reflects cpxR expression or CpxR activity in the cell. There was no detectable expression of a cpxP reporter in the absence of cpxR (in LSΔR) (Fig. 2A and D). There was also no detectable activity of the reporter in the cpxP deletion mutant, LSΔP (Fig. 2A and D). In contrast, deletion of cpxP in E. coli leads to a modest elevation of transcription of cpxP, which is thought to result from eliminating CpxP's negative influence on CpxA activity (17). In the cpxA background, the β-galactosidase activity of the cpxP reporter reached 15 Miller units during stationary phase (Fig. 2A and D). This result likely reflects phosphorylation of CpxR by small phosphate carriers, such as acetyl-phosphate, or cross talk from other kinases (10); such phosphorylation is presumably removed by CpxA's phosphatase activity in wild-type cells. Finally, the PcpxP-lacZ fusion was most active in the cpxA* background (Fig. 2C and D), suggesting that the cpxA* mutant exhibits constitutive expression of cpxR; thus, the periplasmic domain of V. cholerae CpxA appears to be critical for regulating the phosphorylation of CpxR, as in E. coli. Overall, these observations suggest that similarities underlie the regulation of the Cpx pathway in E. coli and V. cholerae; however, the inactive state of the V. cholerae Cpx pathway under standard laboratory growth conditions suggests that distinct cues trigger this pathway in these two organisms.
Since the Cpx pathway was not active in standard laboratory medium, we wondered whether this envelope stress response pathway was activated during infection. To begin to address this possibility, we used the suckling mouse model of V. cholerae pathogenicity to evaluate if any cpx genes are important for V. cholerae intestinal colonization. These assays were carried out as competition experiments in which one-to-one mixtures of the wild-type strain N16961 and either LSΔP, LSΔR, LSΔA, or LSA* were intragastrically inoculated into 5-day old suckling mice. After 20 h of in vivo or in vitro growth, the ratios of the wild type to the mutant strains were determined. None of the cpx mutants was attenuated for growth in the suckling mouse (Fig. (Fig.3),3), indicating that the Cpx pathway is not required for V. cholerae intestinal colonization and that constitutive activation of the pathway (in LSA*) does not alter in vivo fitness. Furthermore, the three cpx deletion mutants had growth similar to that of the wild-type strain in LB medium in monoculture (Fig. (Fig.2A)2A) or in competition with the wild-type strain in LB medium (data not shown). Even though the in vivo competition experiments do not directly assess whether the V. cholerae Cpx pathway is activated in the intestine, the absence of intestinal growth defects in the cpx mutants suggests that the Cpx signal transduction pathway does not play a major role in V. cholerae's adaptation to conditions in the mammalian intestine.
We took advantage of the chromosomal PcpxP-lacZ transcription reporter in LSP'Z to screen for conditions that stimulate the Cpx pathway. Pilot experiments revealed that plate-based assays were more sensitive than broth culture-based assays for assessing the activation of the Cpx pathway. Thus, surface contact may promote activation of the V. cholerae Cpx pathway, as previously described for E. coli (46). As noted above, there is no detectable activity of the V. cholerae Cpx pathway on standard laboratory medium. No activity of the PcpxP-lacZ reporter was seen when LSP'Z was plated on agar plates made with M9 medium, LB medium, or yeast extract plus tryptone (Fig. (Fig.4;4; also data not shown). Furthermore, at least some of the stimuli that are known to trigger the E. coli Cpx pathway, including overexpression of E. coli NlpE, an outer-membrane lipoprotein, and alterations in osmolarity (using sucrose or sorbitol) did not activate the reporter (data not shown). However, CuSO4, which triggers the E. coli Cpx system via an unknown mechanism (61, 62), activated the PcxpP-lacZ fusion (Fig. (Fig.4A).4A). The addition of l-cysteine, which is oxidized by copper (23), to plates along with CuSO4 blocked activation of the Cpx pathway (Fig. (Fig.4A).4A). Thus, it is possible that the inducing effect of CuSO4 is mediated by the compound's oxidizing activity.
We wondered if the high salt concentrations that V. cholerae might encounter in estuarine environments could stimulate the cpxP reporter and therefore assayed its activity on plates made with yeast extract, tryptone, and Instant Ocean salts. The PcpxP-lacZ fusion was stimulated under these conditions (Fig. (Fig.4B),4B), and this upregulation required cpxR (Fig. (Fig.4B),4B), suggesting that PcpxP-lacZ transcription reflects CpxR activity under these conditions. Since the main ingredient in Instant Ocean is NaCl (~450 mM), we tested whether the addition of various concentrations of NaCl to the basal medium of yeast extract and tryptone was sufficient to stimulate the reporter. The PcpxP-lacZ fusion was activated when the NaCl concentration was 250 mM or above (Fig. (Fig.4B;4B; also data not shown). The reporter was not active when LSP'Z was streaked onto plates containing either 1 M sorbitol or 600 mM sucrose (not shown), suggesting that the osmolarity of the NaCl in the plate does not account for the stimulation of the Cpx pathway. Further analysis suggested that the Cl− ions in NaCl account for the activation of the PcpxP-lacZ fusion. We found that plates prepared with Na2SO4 or K2SO4 did not stimulate the reporter, whereas plates containing KCl did (Fig. (Fig.4B).4B). The most-pronounced stimulation of the fusion was observed on plates containing MgCl2 (Fig. (Fig.4B).4B). There could be an additive effect of Mg2+ and Cl− in the activation of the fusion, since there was slight stimulation of the reporter on plates containing MgSO4. In all cases, cpxR was required for the stimulation of the PcpxP-lacZ fusion (Fig. (Fig.4B).4B). In aggregate, these observations suggest that Cl− and, to a lesser extent, Mg2+ stimulate the V. cholerae Cpx pathway.
To learn more about the stimuli that trigger the V. cholerae Cpx pathway, we also used a genetic screen to identify V. cholerae mutants in which the Cpx pathway was active. This screen took advantage of our observation that LSP'Z, the wild-type V. cholerae strain that harbors the chromosomal PcpxP-lacZ fusion, forms white colonies on LB plates containing X-Gal. This strain was mutagenized with the Himar transposon (9), and the resulting mutants were screened for clones that were now blue on LB X-Gal plates. From two independently generated Himar transposon insertion libraries, each containing more than 40,000 insertion mutants, 32 clones, shown in Fig. Fig.5A,5A, were picked for further analyses. Transposon mapping revealed that 26 different genes were disrupted by the transposon (Table (Table4).4). Three genes, cpxA, tolC, and dsbD, had at least two insertions in different nucleotides, and in two cases, these insertion mutants came from different libraries, suggesting that the mutagenesis was fairly comprehensive.
Elevated CpxR levels (and presumably activities) appear to account for at least part of the elevation in cpxP transcription in all of the mutant strains shown in Fig. Fig.5A.5A. As shown in the immunoblot in Fig. Fig.5C,5C, all of the mutants contained more CpxR than the wild-type strain. As expected, CpxR levels in LSA*, the strain that harbors constitutive CpxA, were elevated (Fig. (Fig.5C).5C). There was a rough correlation between the amount of stimulation of the cpxP reporter and the amount of CpxR detected in the Western blot (Fig. 5A to C). Together, the results in Fig. 5A to C suggest that in all cases, at least some of the elevation in cpxP transcription in the mutants is due to activation of the Cpx pathway and suggest that CpxR activates its own expression.
In the mutants selected based on our screen, the transposon insertions disrupted genes that belong to 12 gene categories, according to the TIGR classification of gene function (Table (Table4).4). Four different genes encoding putative methyl-accepting chemotaxis proteins were disrupted, as was tcpI, which also has a methyl-accepting domain and which negatively regulates the synthesis of the major pilin subunit of toxin-coregulated pilus, TcpA. At present, the reason for the relatively large number of hits in this family of genes is unknown. cpxA was interrupted in five of the mutants. Two of these five mutants, E04 and D06, probably represent sibs. These mutations likely act in a fashion similar to that of a cpxA deletion, which activated the Cpx pathway (LSΔA) (Fig. 2A and D). Two insertions were in genes coding for hypothetical proteins (vc1888 and vc1492). Disruptions of genes that code for transport proteins, including hutA, encoding a heme transport protein (27a), and tolC, encoding a multidrug efflux pump (for a review, see reference 35), also trigger the Cpx pathway. Finally, three genes within the energy metabolism category were interrupted in our screen. One of these encodes DsbD, a periplasmic thiol-disulfide interchange protein that catalyzes the reduction of the active sites of DsbC, DsbG, and DsbE (also known as CcmG) in E. coli (for a review, see reference 32).
Interestingly, the majority of the transposon insertions appear to be in genes that code for proteins that localize to the cell envelope. Lopez-Campistrous et al. reported that 76.3% of E. coli proteins are cytoplasmic, 17.7% localize to the inner membrane, 3.7% are periplasmic, and 2.3% localize to the outer membrane (38). There have been no similar comprehensive analyses of the subcellular localization of V. cholerae proteins. However, the subcellular distribution of proteins in these related gammaproteobacteria is likely similar, so we used the estimates for the E. coli proteins to gauge whether there was any bias in the putative localization of proteins encoded by the genes that answered our screen. Of the proteins with known or predicted localization that came out of our screen, 31.8% are predicted to be cytoplasmic, whereas 50% localize in the inner membrane, 4.5% are periplasmic, and 13.6% localize in the outer membrane (Fig. (Fig.5D).5D). Based on the E. coli estimates, only ~24% of V. cholerae proteins should localize to the cell envelope, yet more than 68% of the target proteins found in our screen likely localize to this compartment. This analysis suggests that processes that occur in the cell envelope are likely the principal activators of the Cpx pathway.
Since transposon insertion mutations do not necessarily result in a simple loss of function for the disrupted gene, we assessed whether PcpxP-lacZ activity was reduced to wild-type levels in the insertion mutants when wild-type copies of the disrupted genes were expressed in trans. Seven mutant strains (A08, B10, C05, C08, D03, D09, and E01) were transformed with plasmids carrying wild-type copies of the genes of interest under the control of the arabinose-inducible promoter PBAD. In the first four strains (two independent dsbD::Himar mutants and two independent tolC::Himar mutants), expression of the wild-type gene markedly reduced the activity of the PcpxP-lacZ fusion (Fig. (Fig.6).6). Thus, activation of the V. cholerae Cpx pathway in the dsbD and tolC backgrounds appears to be due to the absence of DsbD and TolC, respectively, in these strains. In contrast, the PcpxP-lacZ reporter remained active when the appropriate wild-type genes were induced in B10 (vc1898::Himar), D09 (vc1888::Himar), and E01 (hutA::Himar). It is possible that the levels of expression of the complementing proteins were not physiologic; alternatively, activation of the Cpx pathway in these strains may not be due to the absence of the gene products in question. For example, the transposon insertions may have polar effects on the expression of downstream genes or may lead to the synthesis of truncated proteins that stimulate the Cpx pathway.
We investigated the mechanism by which the absence of DsbD stimulates the Cpx pathway in greater detail. DsbD is an inner-membrane protein that promotes the proper folding of secreted proteins by facilitating isomerization of disulfide bonds (for a review, see reference 32). Studies of E. coli and other bacteria have shown that DsbD keeps DsbC and DsbG, two disulfide bond isomerases (2, 4, 32, 52) in a reduced state, thereby maintaining the capacity of these proteins to convert incorrect disulfide bonds to their proper forms. DsbD also promotes the maturation of cytochrome c through its interactions with DsbE (also known as CcmG) (3, 14, 56). DsbC (VC2418) and DsbE (VC2051) are 37.5% and 54.8% identical to their respective E. coli counterparts; no V. cholerae homologue of DsbG was identifiable using BLAST. To better understand the role of DsbD in minimizing activation of the Cpx pathway, we assessed the activity of the PcpxP-lacZ reporter in a variety of V. cholerae strains with impaired capacities to form proper disulfide bonds, in the presence and absence of cpxR. The reporter was activated by mutation of dsbC but not dsbE (Fig. (Fig.7A),7A), suggesting that disruption of the DsbCD redox pathway but not the DsbED pathway activates V. cholerae's Cpx response. As expected, activation was dependent upon CpxR. Notably, we found that strains lacking DsbA or DsbB, which introduce the majority of disulfide bonds into periplasmic proteins and hence generate substrates for DsbC, did not have a constitutively active Cpx pathway (Fig. (Fig.7A).7A). Together, these findings suggest that at least some proteins become triggers of the Cpx pathway only when they contain incorrect disulfide bonds and not simply when they are misfolded (e.g., due to the absence of disulfide bonds). Furthermore, appropriate oxidation of these activating proteins is dependent upon DsbC and DsbD.
To further explore the role of DsbC in the activation of the Cpx pathway, we mutagenized LSP'Z ΔdsbC, which is blue on X-Gal plates, and screened for mutants that had reduced expression of PcpxP-lacZ. One of the clones with reduced PcpxP-lacZ expression contained an insertion in dsbA (LSP'Z ΔdsbC dsbA::Himar) (Fig. (Fig.7B).7B). This is consistent with the model presented above; the aberrant disulfide bonds that accumulate in dsbC mutants and appear to stimulate the Cpx pathway should not be present in strains lacking both dsbC and dsbA. Another insertion mutant in the LSP'Z ΔdsbC background with reduced PcpxP-lacZ expression contained a transposon in cpxP (Fig. (Fig.7B).7B). This result suggests that the signal that passes from a protein containing aberrant disulfide bonds to CpxR requires CpxP for its transmission, although the mechanistic details of this process have not yet been determined.
Cpx two-component systems appear to be present in many bacterial phyla, but almost all studies of this pathway have been carried out with E. coli and closely related species. Our studies suggest that this pathway may fulfill a different role in the physiology of V. cholerae than of E. coli. Unlike Cpx systems in E. coli, there is no basal activation of the V. cholerae Cpx pathway, and this two-component system is not required for V. cholerae pathogenicity. In E. coli, this system is thought to sense and respond to envelope stress; certain misfolded cell envelope proteins are postulated to stimulate this pathway (for a review, see reference 22), but the precise molecular triggers are not well defined. Furthermore, we found that the V. cholerae Cpx pathway does not respond to several of the known activators of the E. coli pathway. The V. cholerae Cpx pathway may monitor salinity, since we discovered that chloride ions activate this system. In addition, our results suggest that the V. cholerae Cpx system may detect misfolded envelope proteins containing aberrant disulfide bonds, particularly proteins whose proper folding is mediated by DsbC.
Differences in the predicted polypeptide sequences of the likely sensor domain of CpxA in V. cholerae and E. coli suggested that the activators of the Cpx pathway in these gram-negative organisms might differ. However, it is difficult to explain why the basal expression of this system is so low in V. cholerae. N16961, the sequenced V. cholerae strain we used in our studies, has 162 predicted sensor proteins and 60 predicted response regulators (SENTRA database), and it is possible that some other signal transduction system is redundant with Cpx in this pathogen. Redundant or alternative signal transduction pathways could also explain the lack of mutation-associated phenotypes that we detected in V. cholerae cpx deletion mutants. E. coli cpx deletion mutants have been reported to have aberrant motility and defects in virulence and production of surface appendages (3, 14, 30, 44), but we did not observe similar phenotypes in V. cholerae mutants.
Our findings indicate that Cl− concentrations greater that 250 mM stimulate the V. cholerae Cpx pathway. Additionally, we found that a cpxR mutant had significantly reduced survival when shifted from low to high salt conditions (data not shown), raising the possibility that the Cpx system enables V. cholerae adaptation to upshifts in salinity. A similar role for Cl− in E. coli has not been assessed in reported studies. On the other hand, upshifts in osmolarity, which activate the E. coli Cpx pathway, did not do so for V. cholerae. We have not assessed if Cl− directly alters the activity of one of the components of the Cpx pathway or instead acts indirectly by influencing factors that affect Cpx signal transduction. Biochemical studies have demonstrated that chloride can function as an allosteric regulator of the Lac repressor (6) and can directly bind and activate E. coli glutamate decarboxylase (25). Fleischer et al. (24) reconstituted the E. coli Cpx system in vitro and showed that autophosphorylation of CpxA was enhanced by KCl and, to a lesser extent, other solutes containing chloride, like NaCl, RbCl, or NH4Cl. Interestingly, these authors also found that sucrose and trehalose did not affect CpxA activities in vitro, raising the possibility that osmolarity stimulates the E. coli Cpx pathway indirectly. If osmolarity stimulates Cpx activity indirectly in E. coli, it is possible that the factors that enable the Cpx system to sense osmolarity are missing in V. cholerae.
The results of our transposon mutagenesis studies are consistent with the idea that the V. cholerae Cpx pathway primarily mediates a cellular response to perturbations in the cell envelope, as in E. coli. Most of the genes that answered our screen for mutants in which the Cpx system is activated encode proteins that likely localize to the cell envelope. This strong bias toward proteins in the cell envelope suggests that this compartment is the site where most of the processes that lead to Cpx activation arise. Cytoplasmic proteins also influence Cpx activation since approximately one third of the genes that answered our screen encode proteins that were predicted to localize to that compartment. These proteins could influence processes that occur in the cell envelope and/or the Cpx pathway itself.
The absence of TolC activated the V. cholerae Cpx pathway. TolC is the outer-membrane component of several efflux systems. These systems remove a variety of toxic compounds from the cell, including antibiotics, antimicrobial peptides, and detergents. TolC is also needed for the secretion of several proteins, including hemolysin and proteases (for a review, see references 5 and 35). Thus, activation of the Cpx pathway in the tolC mutant likely results from the accumulation in the cytoplasm or periplasm of products (1, 45) that are ordinarily expelled from the cell.
Genetic analyses of CpxR activation suggest that the formation of aberrant disulfide bonds within the periplasmic compartment is also a potent inducer of the Cpx pathway. Mutations within dsbC and dsbD, which prevent isomerization of periplasmic disulfide bonds, stimulated CpxR activity, but mutations within dsbA and dsbB, which prevent the introduction of disulfides, did not. Both sets of mutations are likely to lead to the accumulation of misfolded proteins in the periplasm; however, in the dsbA and dsbB mutant strains, these proteins should have fewer disulfide bonds, whereas those in the dsbC and dsbD mutant strains should contain incorrect disulfides. These data suggest that at least some proteins that stimulate the Cpx pathway do so via aberrant disulfides and that misfolding alone may not suffice. Our finding that a mutation of dsbA abolished Cpx activation in a dsbC mutant provides additional support for this model. Incorrect disulfide bonds could result in the exposure of specific structures or sequences that stimulate the Cpx pathway. Work with E. coli suggests that there may be specific recognition of protein structures or sequences by the Cpx system (33, 42). Alternatively, it is possible that periplasmic proteins with aberrant disulfide bonds harm some as-yet-undefined component of the cell envelope and thereby indirectly trigger the Cpx pathway.
The formation of aberrant disulfide bonds may also underlie copper's stimulatory effect on the Cpx pathway. We found that this metal, a nonspecific oxidant thought to promote the formation of periplasmic disulfide bonds between cysteines at a higher rate than DsbA (28, 34), is a potent inducer of the Cpx response in V. cholerae, and similar results have been reported for E. coli (61). Strikingly, the normally inactive Cpx pathway in the dsbA dsbC mutant was activated when the mutant was grown on plates containing CuSO4 (data not shown), indicating that CuSO4 can bypass the requirement for dsbA in Cpx activation. Thus, our identification of inappropriate periplasmic protein oxidation as a trigger for the Cpx pathway may provide an explanation for earlier observations regarding Cpx activation.
The mechanism by which improperly oxidized proteins trigger the Cpx pathway has not yet been fully characterized; however, it appears to differ from several previously described stimuli in at least one key respect. We found that CpxP is required for at least some of the activation of the V. cholerae Cpx system in the dsbC background. In contrast, induction of the E. coli Cpx pathway by several stimuli was found to be independent of CpxP (17). In V. cholerae, perhaps CpxP interacts with misoxidized proteins, either directly or indirectly, to promote activation of the Cpx system. Alternatively, CpxP could be a substrate for DsbC; however, this protein only contains two cysteine residues, making this possibility less likely. In aggregate, our findings suggest that studies of the molecular triggers of the Cpx pathway and the substrates of DsbC may be mutually informative.
We thank Jon Beckwith and his lab for the gift of pBAD43 and helpful discussions and the Kolter lab for help with plate photographs. We are grateful to Brigid Davis for excellent suggestions on the manuscript. The V. cholerae Gateway library was obtained through NIAID's Pathogen Functional Genomics Resource Center, managed and funded by the Division of Microbiology and Infectious Disease, NIAID, NIH, DHHS, and operated by The Institute for Genomic Research (TIGR).
This work was supported by funds from NIH (AI-42347) and the Howard Hughes Medical Institute.
Published ahead of print on 19 June 2009.