|Home | About | Journals | Submit | Contact Us | Français|
The facultative pathogen Vibrio cholerae is the causative agent of the human intestinal disease cholera. Both motility and chemotaxis of V. cholerae have been shown to contribute to the virulence and spread of cholera. The flagellar gene operons are organized into a hierarchy composed of four classes (I to IV) based on their temporal expression patterns. Some regulatory elements involved in flagellar gene expression have been elucidated, but regulation is complex and flagellar biogenesis in V. cholerae is not completely understood. In this study, we determined that the virulence defect of a V. cholerae cheW1 deletion mutant was due to polar effects on the downstream open reading frame VC2058 (flrD). Expression of flrD in trans restored the virulence defect of the cheW1 deletion mutant, and deletion of flrD resulted in a V. cholerae strain attenuated for virulence, as determined by using the infant mouse intestinal colonization model. The flrD mutant strain exhibited decreased transcription of class III and IV flagellar genes and reduced motility. Transcription of the flrD promoter, which lies within the coding sequence of cheW1, is independent of the flagellar transcriptional activators FlrA and RpoN, which activate class II genes, indicating that flrD does not fit into any of the four flagellar gene classes. Genetic epistasis studies revealed that the two-component system FlrBC, which is required for class III and IV flagellar gene transcription, acts downstream of flrD. We hypothesize that the inner membrane protein FlrD interacts with the cytoplasmic FlrBC complex to activate class III and IV gene transcription.
The causative agent of cholera is the gram-negative facultative human pathogen Vibrio cholerae (34). V. cholerae is highly motile via its single sheathed polar flagellum. The sodium driven rotary motor complex of the Vibrio flagellum can turn at rates as high as 1,700 revolutions/s and produces speeds up to 60 μm/s in liquid media (3, 4).
Not surprisingly, the synthesis of an important cellular structure like the flagellum is tightly regulated. Flagellar gene transcription is organized in a transcriptional hierarchy, which has been categorized into four gene classes (50) (see Fig. Fig.7).7). The only class I gene identified so far encodes the σ54 (RpoN)-dependent activator FlrA, which together with the RpoN-holoenzyme of RNA polymerase (RNAP) activates the expression of class II genes. These genes comprise mainly those for structural components, like the MS (membrane/supramembrane) ring and export apparatus components, but also include chemotaxis genes and genes encoding regulatory factors, including FlhG, FlhF, the two-component system FlrBC, and σ28 (FliA). FlhG is a negative regulator of class I gene transcription, while FlhF stimulates class III gene transcription, most likely by interaction with the FlrBC system (15). The response regulator FlrC, along with the RpoN holoenzyme, directly activates transcription of class III genes, which encode components of the basal body and hook, as well as the main flagellin, FlaA. FlrC must be phosphorylated by the sensor histidine kinase FlrB to activate flagellar gene transcription, although the specific stimulating signals remain unknown (14). Unlike most other sensor histidine kinases, FlrB is a cytosolic soluble protein (14). The FliA holoenzyme activates the expression of class IV genes, which encode additional flagellins FlaBCDE, the motor complex, and the anti-sigma factor FlgM. FlgM prevents the association of FliA with RNAP (13); after complete assembly of the basal-body-hook structure, FlgM is secreted through the sheathed flagellum, which facilitates FliA association with RNAP and class IV transcription (13).
The presence of the flagellum and flagellum-mediated motility are crucial factors in several steps of the V. cholerae life cycle, which for the pathogenic strains is marked by transitions between the aquatic ecosystem and the human gastrointestinal tract. Several studies indicate a close association and interaction between V. cholerae and chitinous surfaces (e.g., zooplankton such as copepods) in the aquatic environment (12, 28, 48, 54). In this environment, V. cholerae utilizes chitin as a carbon and nitrogen source and induces natural competence (42, 43). V. cholerae also forms biofilms on chitinous surfaces, which is likely to be an important survival and persistence mechanism within the aquatic environment between epidemic outbreaks (52, 55). Flagellar motility is crucial for early steps during biofilm development, as it accelerates the attachment to abiotic surfaces and mediates spread along these surfaces (64). In later steps of biofilm formation, the flagellar motor complex appears to act as a mechanosensor, signaling the interaction with an abiotic surface and stimulating exopolysaccharide expression (36).
V. cholerae transits from the aquatic reservoir into the human host by oral ingestion of contaminated food or water. After passage through the stomach, the bacteria reach the small intestine, which is the primary site of colonization. Nonmotile mutants of the O1 El Tor biotype, which is currently the dominant clinical isolate, colonize 10 to 25 times less efficiently than wild-type (WT) strains (37, 38, 47). It has been proposed that flagellar motility supports the initial contact with the intestinal epithelial surface and is necessary for the penetration through mucus (9, 18, 19). Recently, it was demonstrated that breakage of the flagellum during mucin penetration allows secretion of the anti-σ28 factor FlgM through the damaged flagellum (13, 38). This decrease of intracellular FlgM releases the alternative sigma factor FliA, which results in activation of FliA-dependent genes and inhibits the HapR-mediated repression of virulence genes. Thus, the flagellar biosynthesis pathway and virulence gene regulation are linked, and this interaction is important for full expression of virulence genes (38). Two recent studies which focused on the late stage of infection indicated that V. cholerae has evolved genetic programs for the transition from the human host into the aquatic environment (49, 60). Detachment and escape from the mucosal surface after successful colonization are controlled by the stationary-phase sigma factor RpoS, which induces flagellar motility genes and activates the mucinase HapA via HapR (49). Interestingly, V. cholerae organisms shed by cholera patients are transiently more infectious than in vitro-cultured bacteria (44). Despite their being highly motile, the hyperinfectious phenotype of stool bacteria is marked by a repression of chemotaxis genes, which enhances survival of the bacteria in a new host (8-10). Consistent with this observation, smoothly swimming counterclockwise (CCW)-biased nonchemotactic mutants of V. cholerae have a significantly lower infectious dose and outcompete the WT strain in the infant mouse colonization model (8, 10). Mathematical modeling of human epidemiological data supports the hyperinfectivity phenomenon as one explanation for the rapid and explosive outbreaks caused by V. cholerae (24).
In the present study, we characterized the role of VC2058 in virulence, motility, and regulation of flagellar synthesis. Due to the functional properties of VC2058 (see below), we have renamed it the flagellar regulatory protein D (flrD). Deletion of flrD results in a strain that is less motile and attenuated for virulence in the infant mouse colonization model, demonstrating a role for this gene in both the motility and virulence of V. cholerae. We further show that flrD is expressed from a promoter located within the coding sequence of cheW1 that is independent of FlrA and RpoN, regulators of class II transcription. Although flrD is expressed independently of the established flagellar hierarchy, it is a positive regulator of class III and IV flagellar genes.
Bacterial strains and plasmids used in this study are listed in Table Table1,1, oligonucleotides are listed in Table Table2.2. Unless noted otherwise, strains were grown with aeration in Luria-Bertani (LB) broth at 37°C. V. cholerae AC51, a spontaneous streptomycin-resistant mutant of V. cholerae O1 El Tor clinical isolate C6709 was used as a WT strain in all experiments (56). For genetic manipulations, Escherichia coli strains DH5α, DH5αλpir, and SM10λpir were used (23, 35, 45). Antibiotics and other supplements were used in the following final concentrations: streptomycin (Sm), 100 μg/ml; ampicillin (Ap), 50 μg/ml in combination with other antibiotics or 100 μg/ml; kanamycin, 50 μg/ml; arabinose, 0.02%; and isopropyl-β-d-thiogalactopyranoside, 0.1 mM.
Chromosomal DNA was prepared using a DNeasy tissue kit, whereas PCR products and digested plasmid DNA were purified using Qiaquick gel extraction and Qiaquick PCR purification kits (Qiagen). PCRs for sequencing and subcloning were carried out using the Easy-A high-fidelity PCR cloning enzyme (Stratagene), Pfu polymerase (Stratagene), or Phusion high-fidelity polymerase (New England Biolabs). For all other reactions, Taq DNA polymerase (New England Biolabs) was used. Automated DNA sequencing was performed either with an automated ABI 3130XL DNA sequencer or with an ABI Prism 310 automated sequencer.
Deletion mutations were generated by SOE (splicing by overlap extension) PCR (27). All deletions are in frame, and the sequences deleted are listed in Table Table1.1. The resulting PCR products were initially cloned into the pCR-Script Amp SK vector (Stratagene), followed by subcloning into the pCVD442-lac plasmid using SphI and SacI restriction sites that had been incorporated into the primers used for PCR amplification. The pCVD442-lac derivatives were transformed into E. coli Sm10λpir and subsequently moved into V. cholerae via conjugation.
To construct plasmid pGPphoAflrD, first the promoterless phoA gene was PCR amplified from Sm10λpir with the oligonucleotide pair phoA-XbaI-5′ and phoA-XbaI-3′. The resulting PCR fragment containing the full coding sequence of phoA, including the Shine-Dalgarno site, was then digested with XbaI and ligated into pGP704 that had been digested similarly, resulting in pGPphoA. Next, an flrD fragment was PCR amplified using oligonucleotide primers VC2058-SacI and VC2058-KpnI, and the resulting fragment, containing the translational stop codon of flrD, was digested with SacI and KpnI and ligated into pGPphoA that had been digested similarly, resulting in pGPphoAFlrD. Thus, the flrD fragment was now located upstream of phoA and separated from it by at least one stop codon. The plasmid was transformed into SM10λpir to allow conjugation into V. cholerae. Insertion of pGPphoAflrD into the flrD locus on the V. cholerae chromosome by homologous recombination results in a transcriptional fusion of phoA to the flrD transcript.
To construct plasmid pGPflaA, an internal fragment of flaA was PCR amplified with the oligonucleotide pair flaA-SacI and flaA-XbaI, digested with SacI and XbaI, and ligated into pGP704 that had been digested similarly.
Conjugation was achieved by cross-streaking the donor E. coli strain carrying the plasmid with the recipient V. cholerae strain on an LB plate and then incubating at 37°C for ~6 h. Plasmid cointegrants were then selected by isolation of colonies on LB Smr/Apr medium. For generation of chromosomal deletions with pCVD442 derivatives, isolated colonies were then grown in LB broth in the absence of antibiotics, followed by growth on LB plates with sucrose to obtain Aps colonies. Correct insertions or chromosomal deletions were confirmed by PCR (data not shown).
All expression plasmids using pMMB67EH, pFLAG-MAC, or pBAD18-Kan were constructed in a similar manner. PCR fragments of the respective gene were generated using the oligonucleotide pairs designated in the format x-y-5′ and x-y-3′, in which x stands for the gene and y for the restriction site/enzyme used (Table (Table2).2). PCR fragments were digested with the respective restriction enzymes and ligated into similarly digested pMMB67EH, pFLAG-MAC, or pBAD18-Kan. The M114I mutation in FlrC was generated by SOE PCR, using pBKflrC as a template and oligonucleotide pairs flrC-M114I-up and flrC-SacI-5′ as well as flrC-M114I-down and flrC-XbaI-3′. The two PCR products were used as the template in the second PCR with flrC-SacI-5′ and flrC-XbaI-3′, and the resulting fragment was digested with SacI and XbaI and ligated into similarly digested pBAD18-Kan.
The plasmid series pRS551′cheW′-I through pRS551′cheW′-V was constructed by introducing different upstream fragments of flrD into pRS551. PCR fragments were generated using one of the oligonucleotides cheW-I through cheW-V as a forward primer paired with cheW-BamHI as the reverse primer. Thus, one end of each PCR product was located 28 bp downstream of the flrD start codon. PCR products were digested with EcoRI and BamHI and ligated into pRS551 that had been similarly digested.
Swarm plates composed of 1% tryptone, 0.5% NaCl, and 0.3% agar were used to assess motility of V. cholerae strains. Strains were first grown overnight at 37°C, then a single colony was inoculated by toothpick into the swarm plate, and the plate was incubated for 24 h at room temperature.
Competition assays for intestinal colonization in infant mice (in vivo) and for growth in LB broth (in vitro) were performed with a mixture of mutant (Lac−) and isogenic WT (LacZ+) strains as previously described (11). The competitive index (CI) represents the ratio of mutant to WT CFU recovered at 24 h, normalized to the input ratio. For competition assays involving mutants complemented with an expression plasmid, the corresponding WT strain carried the empty plasmid vector. The pMMB vector and its derivatives are well maintained in V. cholerae, and no significant loss of the plasmids was observed during in vivo passage (46).
RNA purification and quantitative reverse transcriptase PCR (qRT-PCR) were performed as described previously (60). Briefly, RNA was isolated from at least five independent mid-exponential-phase LB broth cultures (optical density at 600 nm [OD600], 0.5 to 0.55). DNA was removed using a DNA-free kit (Ambion), and cDNA was synthesized from 1 μg RNA using a SuperScript first-strand synthesis system for qPCR (Invitrogen). Controls lacking reverse transcriptase were included. qRT-PCR experiments were performed with SYBR brilliant green qPCR Master Mix, utilizing a Stratagene Mv3005P real-time cycler and MxPro qPCR software (Stratagene). Each reaction mixture contained 300 nM primers, 100 ng template, and ROX reference dye. Each independent sample was tested in triplicate. The sequences of the primers used in these studies are in Table Table2,2, labeled in the format x-F and x-R, in which x stands for the respective gene and F and R indicate forward and reverse primers. All primer pairs amplified the target gene with efficiencies of 97% or more (data not shown).
For each sample, the mean cycle threshold of the test transcript was normalized to that of rpoB and to one randomly selected reference sample of the WT. Values of >1 and <1 indicate that the transcript is present in higher and lower numbers, respectively, in the mutant than in the WT.
The +1 transcriptional start site of the flrD gene was identified using a 5′ RACE (rapid amplification of cDNA ends) system (Invitrogen) according to the instruction manual. Briefly, RNA was isolated and treated with DNase as described above. Total RNA (1 to 5 μg) was used to synthesize cDNA from the 5′ end of the flrD mRNA with the gene-specific primer GSP1-flrD. The reaction was performed with the SuperScriptII RT at 42°C for 50 min. A homopolymeric C tail was added to the cDNA with the terminal deoxynucleotidyl transferase. The RACE products were synthesized using the abridged anchor primer AAP and the gene-specific primer GSP2-flrD followed by a second PCR to reamplify the primary product using the universal amplification primer UAP and a nested gene-specific primer GSP3-flrD. The RACE products were directly sequenced to identify transcription start sites.
The fractionation protocol was adapted from the work of Bose and Taylor (6). A late-exponential-phase LB broth culture (500 ml; OD600 ~ 1) was harvested by centrifugation, washed once with LB broth, and resuspended in 10 mM Tris-HCl (pH 8) with Complete protease inhibitor mix (Roche). Cell extracts were obtained by passing the resuspended cell solution through a French pressure cell at 1,000 lb/in2, followed by a low-speed centrifugation at 4,500 × g at 4°C for 20 min to remove intact cells. The whole-cell lysate was then centrifuged at 10,000 × g at 4°C for 30 min for further purification, before membranes were separated from the supernatant by centrifugation at 75,000 × g at 4°C for 30 min. In order to solubilize the inner membrane proteins, the membrane pellet was resuspended in 10 mM Tris-HCl (pH 8), 100 mM NaCl, and 2.5% Sarkosyl, incubated with gentle rocking for 30 min, and then centrifuged at 75,000 × g at 4°C for 1 h. The supernatant was retained as the inner membrane fraction, and the pellet containing the outer membrane was resuspended in 1 ml of 10 mM Tris-Cl (pH 8.0). Protein estimation was carried out using a Lowry protein assay kit (Pierce) with bovine serum albumin as the standard. Whole-cell lysates for the detection of FlaA were prepared as follows: 500 μl of a log-phase culture (OD600 = 0.5) was harvested by centrifugation, and the cell pellet was resuspended and boiled in 100 μl protein sample buffer. Protein concentration was estimated using sodium dodecyl sulfate-polyacrylamide gel electrophoresis followed by Coomassie staining.
Equal amounts of proteins from whole-cell lysates, as well as corresponding cell equivalents of cellular fractions, were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred onto a nitrocellulose membrane (Invitrogen). Immunoblot analysis was performed as described previously (59). The membrane was blocked by incubation with 10% skim milk in TBS (10 mM Tris-Cl [pH 7.5], 150 mM NaCl) for 2 h at room temperature. The membrane was washed twice in TBS-TT (20 mM Tris-Cl [pH 7.5], 500 mM NaCl, 0.05% Tween 20, 0.2% Triton X-100) and once in TBS for 10 min each. Primary antibodies used were mouse polyclonal FlaA antibodies (this study), rabbit polyclonal OmpT antibodies (51), or mouse monoclonal anti-FLAG M2 antibody (Sigma). The appropriate primary antibody diluted 1:1,000 in 10% skim milk in TBS was added to the transfer membrane and incubated with gentle rocking at 4°C overnight, followed by washing as described above. The membrane was then incubated for 1 h in TBS 10% skim milk containing the appropriate secondary antibody (horseradish peroxidase-conjugated sheep anti-mouse or donkey anti-rabbit immunoglobulin [GE Healthcare UK Limited]). The membrane was then washed four times in TBS-TT for 10 min each. Chemiluminescence was detected by using the ECL Plus Western blotting detection system (GE Healthcare UK Limited) and exposure to X-ray film (Kodak).
To determine the enzymatic activities for the transcriptional phoA and lacZ fusions, alkaline phosphatase and β-galactosidase assays were performed as described previously (39, 57, 58). The activities are expressed in Miller units, calculated as (A405 × 1,000)/(A600 × ml × min).
Data were analyzed using the Mann-Whitney U test, by a Kruskal-Wallis test followed by post-hoc Dunn's multiple comparisons, or by a Wilcoxon signed rank test against a hypothetical value of 1. Differences were considered significant for P values of <0.05.
As reported previously, nonchemotactic CCW-biased mutants of V. cholerae (e.g., the cheY3 or cheR mutant) have a competitive advantage over the WT during infection of infant mice (8, 10). Consistent with these findings, a nonchemotactic CCW-biased mutant with a point mutation in cheW1 [cheW1(G69D)] outcompetes WT in vivo (10). In E. coli, this point mutation disrupts the interaction of CheW with CheA and the Tar complex, resulting in a smooth swimming nonchemotactic phenotype (7). Additionally, it has been demonstrated that the cheW1(G69D) strain colonizes the intestine, similar to another nonchemotactic CCW-biased mutant, the ΔcheY3 mutant (10). These results with the cheW1(G69D) mutant are consistent with the phenotype expected of a nonchemotactic CCW-biased mutant. Interestingly, V. cholerae strains with in-frame deletions of cheW1 did not exhibit the same phenotype as the strain containing the cheW1 point mutation. For example, a ΔcheW1 mutant exhibits a fivefold colonization defect in vivo in competition against a ΔcheY3 strain (Fig. (Fig.1).1). This defect in colonization is not due to a growth defect of the ΔcheW1 strain, since the ΔcheW1 and ΔcheY3 strains compete equally for in vitro growth. Additionally, chemotaxis and colonization defects of a ΔcheW1 mutant could not be restored by expression of cheW1 in trans (10) (data not shown). This argues for a polar effect of the ΔcheW1 deletion on downstream genes, since these phenotypes are absent in the cheW1(G69D) point mutant strain. We hypothesized that the deletion of cheW1 reduces expression of the downstream located flrD, which in turn leads to reduced fitness in vivo.
In support of this hypothesis, the colonization defect of ΔcheW1 strain was significantly rescued by the expression of flrD in trans on a low-copy-number plasmid but not by the plasmid vector alone (Fig. (Fig.1).1). To investigate the function of FlrD in more detail, a V. cholerae strain with an in-frame deletion in flrD (ΔflrD) was constructed. The ΔflrD strain exhibited a fourfold colonization defect compared to the WT strain in the infant mouse intestine (Fig. (Fig.1).1). This defect could be complemented back to WT levels of colonization by expression of flrD in trans in the ΔflrD strain, while colonization was not enhanced in the ΔflrD strain carrying the plasmid vector alone. It should be noted that the ΔflrD strain had a small but significant (twofold) fitness advantage over the WT in an in vitro competition assay (P < 0.05, using a Wilcoxon signed rank test).
The polar effect of a deletion of cheW1 on flrD expression suggests there is an active promoter within the cheW1 coding region which is required for transcription of flrD. To demonstrate the existence of an active promoter element within cheW1, we constructed chromosomal transcriptional fusions of a promoterless phoA reporter gene to flrD in the WT, ΔcheW1, and cheW1(G69D) strains. The resulting PhoA activity reflects the expression levels of flrD in the respective strains (Fig. (Fig.2A).2A). The WT and the cheW1(G69D) strains exhibited comparable levels of PhoA activity, which indicates similar flrD expression levels in these strains. In contrast, the ΔcheW1 strain showed significantly decreased levels of PhoA activity in comparison to the WT. These results argue for the existence of an active promoter within cheW1 that drives expression of flrD and that is absent in the deletion mutant.
To characterize the flrD promoter in more detail, different-sized fragments of the flrD upstream region were analyzed for their promoter activity. We constructed plasmids containing the upstream region fragments (′cheW′-I to -V) fused to lacZ and then measured transcription levels of lacZ in the WT strain carrying these plasmids (Fig. (Fig.2B).2B). All fragments end 28 bp downstream of the flrD start codon, but they start at different positions in cheW1. The largest fragment (′cheW′-I; 525 bp) comprises the entire coding region of cheW1. The two largest fragments, 486 and 525 bp, fused to lacZ expressed the highest β-galactosidase activity, while the smaller fragments, 353, 232, and 153 bp, fused to lacZ showed significantly lower β-galactosidase activity. These results demonstrate the presence of an active promoter within cheW1 that drives transcription of flrD. Additionally, 5′ RACE was carried out to determine the transcriptional start site of the flrD mRNA. The results place the +1 transcriptional start site 282 bp upstream of the flrD start codon. Additionally, a highly conserved putative −10 region with the sequence TATTAT was identified upstream of the +1 start site at the proper distance. Thus, the +1 start site was present in the fragments ′cheW′-I to -III. Since this transcriptional start site is located very close to the 5′ end of the ′cheW′-III fragment, it is not surprising that this fragment exhibited only low promoter activities. Additional upstream elements and the correct DNA topology might be important for full promoter activity, and these are obviously present in the two largest fragments, ′cheW′-I and -II. In summary, flrD should no longer be assigned to the upstream operon starting with flhA.
The flagellar gene operon upstream of flrD, encompassing genes from flhA to cheW1, has previously been shown to be RpoN and FlrA dependent and has therefore been assigned to flagellar gene class II (50). We measured transcription levels of flhA by qRT-PCR, which confirmed that flhA transcripts were significantly reduced in rpoN and flrA mutants compared to the WT (Fig. (Fig.2C).2C). In contrast, transcript levels of flrD were not significantly altered in rpoN or flrA mutants compared to the WT (Fig. (Fig.2D).2D). This indicates that the promoter of flrD is RpoN and FlrA independent and that flrD does not belong to the class II flagellar genes. This is also further evidence that flrD is not part of the class II operon starting with flhA but rather has its own differentially regulated promoter.
The proximity of flrD to other flagellar and chemotaxis genes prompted us to determine whether flrD is involved in motility and/or chemotaxis. We inoculated the strains into soft-agar plates, which are used to detect defects in motility and chemotaxis (Fig. (Fig.3A).3A). The chemotactic motility of the WT strain produces a characteristic swarm as the bacteria swim away from the point of inoculum. In comparison to the WT, the ΔflrD strain produced a swarm with a much smaller diameter. This defect was complemented by expression of flrD in trans but not by the vector alone. To determine whether the reduced swarming ability of the ΔflrD strain in soft agar was due to motility or chemotaxis, we measured the expression of the major flagellin FlaA in the WT and ΔflrD strains by immunoblot analysis with an anti-FlaA antibody, using whole-cell lysates (Fig. (Fig.3B).3B). FlaA (40.2 kDa) was detected in the whole-cell lysate of the WT strain but not in the whole-cell lysate of ΔflrD. Expression of FlaA was restored in the ΔflrD strain by expression of flrD in trans but not by the plasmid vector alone. These results demonstrate reduced expression of the class III flagellin FlaA in the ΔflrD strain, suggesting an effect of FlrD on the flagellar transcription hierarchy.
To determine the effect of FlrD on the flagellar transcription hierarchy, we measured transcription of all four flagellar gene classes, using representative plasmids with flagellar gene promoter-lacZ transcriptional fusions in AC66 and ΔflrD (Fig. (Fig.3C).3C). One representative promoter-lacZ fusion for each flagellar gene class was used. While flrA was transcribed at comparable levels in both strains, the ΔflrD strain showed a moderate increase in flrB transcription. This indicates that FlrD is not required for class I or class II gene transcription. In contrast, transcription of the class III gene flaA and the class IV gene flaB was significantly decreased in the ΔflrD strain compared to transcription of these genes in AC66, suggesting that FlrD enhances class III and IV flagellar gene transcription.
To confirm the effects of FlrD on the flagellar transcription hierarchy, we performed qRT-PCR of selected flagellar genes in the ΔflrD and WT strains (Fig. (Fig.4).4). Again, the two strains showed comparable transcription levels of the class I gene flrA (Fig. (Fig.4A),4A), while class II flrB transcript levels were slightly, but not significantly, elevated in the ΔflrD strain (Fig. (Fig.4B).4B). These results confirm that neither class I nor class II flagellar genes require FlrD for expression. In contrast, transcription levels of class III and IV genes were confirmed to be significantly decreased in the ΔflrD strain compared to the WT via qRT-PCR (Fig. 4C to F). Consistent with the results obtained with promoter-lacZ fusions, flaA (class III) and flaB (class IV) gene transcripts were significantly lower in the ΔflrD strain than the WT (Fig. 4C and E). We additionally measured transcripts for another class III gene, flaG, and another class IV gene, motY (Fig. 4D and F). These genes were also transcribed at significantly lower levels in the ΔflrD strain than the WT. These genes are located at different positions in the V. cholerae genome and consequently have their own promoters. Thus, FlrD is required for class III and IV gene expression in general, rather than being necessary for the activation of a specific flagellar gene operon.
In addition, qRT-PCR allowed us to study complementation of the decreased transcription of class III and IV genes. Transcription levels of flaA, flaG, flaB, and motY were restored in ΔflrD by expression of flrD in trans, while the empty vector alone had no effect. We also considered the possibility of a regulatory mechanism, whereby FlrD could act as an activator of the upstream located class II operon starting with flhA. However, transcription levels of two representative genes of the respective operon, flhA and fliA, were not significantly different in the ΔflrD strain (mean ± standard deviation, 1.5 ± 1.1 and 1.7 ± 1.3) and the WT (0.8 ± 0.3 and 1.2 ± 0.5).
Computational analysis of the protein sequence of FlrD revealed a putative transmembrane region predicted by several topology prediction programs (16, 25, 62, 63). Consistently, the topology models predict a transmembrane region from amino acids (aa) 18 to 33, with the N terminus located in the periplasm and the C terminus (aa 34 to 167) located in the cytoplasm. In order to investigate the membrane localization of FlrD, we constructed an expression vector that expressed FlrD with an N-terminal FLAG tag, which allowed detection of the protein by immunoblot analysis using an anti-FLAG antibody. The FLAG tag did not adversely affect the activity of FlrD, since expression of FLAG-FlrD in trans still rescued the motility defect of the ΔflrD strain, whereas the vector alone (pFLAG) was unable to rescue the motility defect of this strain (Fig. (Fig.5A).5A). To determine whether membrane localization is important for FlrD activity, we also constructed an expression vector that expressed the truncated protein FLAG-FlrD-31, from which the N-terminal 31 aa, containing the putative transmembrane domain, had been removed. This truncated form of FlrD failed to rescue the motility defect of the ΔflrD strain (Fig. (Fig.5A),5A), indicating that membrane localization is critical for FlrD function. The ΔflrD strain expressing either FLAG-FlrD or FLAG-FlrD-31 was used for fractionation experiments to determine the localization of the respective proteins (Fig. (Fig.5B).5B). Immunoblot analysis of whole-cell lysates indicated that the two proteins were expressed at comparable levels. No degradation products were observed. Fractionation of inner and outer membrane compartments demonstrated that FLAG-FlrD is localized to the inner membrane, as predicted by the topology prediction programs. In contrast, the truncated FLAG-FlrD-31 failed to localize to the inner membrane and remained in the cytoplasm (Fig. (Fig.5B).5B). The outer membrane porin OmpT served as a loading control. OmpT should be detectable in all three fractions, since it is synthesized in the cytoplasm and from there transported to the outer membrane via the inner membrane and periplasm. Consequently, OmpT was present at high levels in the whole-cell lysate samples and outer membrane fractions but only at low levels in the inner membrane fractions.
The two-component regulatory system FlrBC directly activates transcription of class III flagellar genes, and class III gene expression is required for subsequent high-level transcription of class IV flagellar genes (33, 50). Our data indicate that FlrD, like FlrBC, is required for the transcription of class III and class IV flagellar genes. Due to the absence of a recognizable DNA-binding motif in FlrD, we speculated that FlrD might act upstream of FlrBC at the class II-class III checkpoint. FlrC requires phosphorylation by FlrB to activate transcription of class III genes (14), and thus, FlrD may influence the phosphorylation state of FlrC. We reasoned that high-level expression of FlrC or the FlrB-independent version FlrC-M114I in the ΔflrD strain should overcome the motility defect. The methionine-114-to-isoleucine (M114I) mutation allows constitutive FlrB-independent transcription of flagellar genes by FlrC but still requires phosphorylation of the aspartate (D54) by a noncognate kinase or acetyl-phosphate to activate flagellar gene transcription (14). FlrC and FlrC-M114I were expressed from a plasmid in the WT, ΔflrC, and ΔflrD strains. These strains were analyzed for their swarming ability (Fig. (Fig.6A).6A). Furthermore, class III gene expression was evaluated by determining the level of FlaA expression in these strains by immunoblot analysis using an anti-FlaA antibody (Fig. (Fig.6B).6B). The vector control had no effect on motility or on FlaA expression in any strain. As described above, the ΔflrD strain exhibited decreased motility compared to the WT, whereas the ΔflrC strain showed no swarming ability at all.
Overexpression of either FlrC or FlrC-M114I in the WT results in higher FlaA levels, demonstrating that levels of FlaA can increase even in the WT. However, the swarming ability of the WT expressing either type of FlrC in trans was slightly reduced, suggesting that overexpression of FlaA is disadvantageous for motility. Consistent with previous studies, the ΔflrC strain lacks FlaA expression, which can be restored by expression of either FlrC or FlrC-M114I in trans (14, 33). Although FlaA can be detected in whole-cell lysates of the ΔflrC strain expressing FlrC or FlrC-M114I, the motility defect is only partially complemented by FlrC but almost completely rescued by FlrC-M114I. Consistent with previous studies, expression of FlrC or FlrC-M114I in the ΔflrC strain rescued motility partially or almost completely (14, 33) (Fig. (Fig.6A).6A). Expression of FlrC or FlrC-M114I in the ΔflrD strain also resulted in increased motility that was similar to WT levels, demonstrating that FlrBC is epistatic to FlrD. Measurement of FlaA levels in these strains by immunoblotting revealed that expression of FlrC or FlrC-M114I in trans in the ΔflrC and ΔflrD strains restores FlaA expression. The FlrB-independent form FlrC-M114I led to even higher levels of FlaA expression than the native FlrC (Fig. (Fig.6B).6B). These results demonstrate that the motility defect of the ΔflrD strain can be overcome by overexpression of the activator of class III genes, FlrC, indicating that FlrD acts upstream of FlrBC with respect to the activation of class III and IV flagellar gene transcription.
To date, more than 40 gene products are known to participate in V. cholerae flagellar synthesis. The corresponding genes are organized into four classes and transcribed in a hierarchical fashion. In this study, we investigated the role of the uncharacterized gene flrD, which is located immediately downstream of cheW1, in a flagellar synthesis gene cluster. flrD was previously assigned to the flagellar class II gene operon, which starts with flhA (50). However, our results demonstrate that flrD is not part of this operon but rather is expressed from its own RpoN- and FlrA-independent promoter and thus appears to be transcribed independently of the four flagellar gene classes. Nevertheless, we also showed that FlrD positively regulates class III and IV flagellar gene transcription and contributes to intestinal colonization in the infant mouse model.
In addition to flagellar genes, the class II operon starting with flhA also encodes chemotaxis genes, including cheY3, cheZ, cheA, cheB, and cheW1. Interestingly, V. cholerae shed by cholera patients in planktonic form transiently represses chemotaxis genes while maintaining motility to increase infectivity (10, 44). These nonchemotactic bacteria, obtained from fresh stool samples from cholera patients, have a competitive advantage against in vitro-grown bacteria in infant mice and have a lower infectious dose. One hypothesis to explain the competitive advantage of V. cholerae found in stool samples suggests that nonchemotactic V. cholerae organisms avoid antimicrobial factors by their inefficient penetration of the intestinal crypts, whereas chemotactic bacteria follow the gradient of chemoattractants into the intervillous spaces (9, 19, 20). In support of this hypothesis, nonchemotactic, CCW-biased mutants such as the cheY3, cheR2, and cheW1(G69D) strains are hyperinfectious and outcompete the WT strain about 30-fold during infection of the infant mouse intestine (8, 10).
Given the competitive advantage of nonchemotactic mutants for intestinal colonization, it was a surprise to find that a strain with a deletion of cheW1, which also results in a nonchemotactic CCW-biased phenotype, exhibits a colonization defect compared to other nonchemotactic mutants, e.g., the ΔcheY3 mutant. Expression of cheW1 in trans did not restore WT levels of motility and colonization to the ΔcheW1 mutant, but it could restore normal chemotaxis and colonization to the cheW1(G69D) point mutant (10). Intrigued by this inconsistent result, we investigated the phenotype of the cheW1 deletion in more detail. Using a transcriptional phoA fusion to the downstream-located flrD, we demonstrated that a deletion of cheW1, but not the G69D point mutation of cheW1, causes a decrease in transcription of flrD. Since a deletion of flrD by itself already results in a colonization defect in the infant mouse, the polar effect of ΔcheW1 on flrD transcription is most likely responsible for the otherwise inexplicable reduced colonization capability of a cheW1 deletion mutant. This is reinforced by the restored colonization fitness of ΔcheW1 expressing flrD in trans. Thus, the current model concerning the hyperinfectious phenotype of nonchemotactic CCW-biased mutants is still valid.
The results also indicated that flrD is not part of the flagellar class II operon starting with flhA, and further analysis revealed that an active promoter for flrD lies within cheW1. According to the β-galactosidase activities and results from the 5′ RACE obtained in this study, the transcriptional start site of an active promoter element is located 282 bp upstream of the flrD start codon. Interestingly, the largest fragment analyzed for promoter activity contains a RpoN promoter consensus sequence with the conserved GG and GC elements in a 12-nucleotide distance (5). Such RpoN promoter consensus sequences are absent in all smaller fragments. Since the largest and second largest fragment showed quite similar promoter activities, the putative RpoN promoter does not significantly contribute to the transcription activation of flrD. Furthermore, we demonstrated by qRT-PCR analysis that flrD transcription was not altered in rpoN and flrA mutants. Hence, flrD is not a class II flagellar gene.
The data obtained using chromosomal flrD-phoA transcriptional fusions indicate that even in the ΔcheW1, mutant transcriptional activity can be observed. Thus, the promoter element in cheW1 might be only one of several. However, according to our results the promoter element in cheW1 is the most important one for flrD expression under the in vitro and in vivo conditions we investigated. Data obtained from previous studies also suggests a very complex regulation of the flhA-cheW1 gene cluster. Besides RpoN and FlrA, RpoS was shown to be required for full expression of the gene cluster during stationary phase (49). This indicates that different promoters might activate the gene cluster in different growth phases. Hyperinfectious V. cholerae strains show high levels of expression of motility genes but reduced expression of genes required for bacterial chemotaxis (44). Thus, the flagellar and chemotaxis genes in the flhA-cheW1 gene cluster have to be differentially regulated during the life cycle of V. cholerae.
The ΔflrD mutant exhibits decreased transcription of class III and IV flagellar genes, which causes reduced expression of the major flagellin FlaA and a motility defect on soft-agar plates. Analysis by electron microscopy revealed no obvious differences between the flagellum in the ΔflrD mutant and that in the WT (data not shown). However, as demonstrated in this study, the expression of class III and IV flagellar genes and motility is only reduced, not completely abolished, in ΔflrD. Thus, an effect of flrD on the morphology of the flagellum might not be readily identifiable by electron microscopy. Since flagellar motility contributes to virulence by facilitating attachment and penetration of the mucosal surface in the small intestine (18, 19), the reduced motility of the ΔflrD and ΔcheW1 mutants is a likely explanation for the reduced colonization fitness in the infant mouse intestine. The involvement of flrD, which is not a class I or class II gene, in class III and IV gene expression places it into a novel gene class of the flagellar hierarchy, as shown in the proposed model (Fig. (Fig.7).7). Due to the lack of a conserved DNA-binding motif in FlrD, it is unlikely that FlrD directly activates transcription of class III and IV flagellar genes by binding to the promoter regions. Furthermore, our results indicate that the expression levels of important activators of class III and IV gene transcription, like the two-component system FlrBC, FlhF, and the alternative sigma factor FliA, are not altered in an flrD mutant.
The response regulator FlrC is the direct activator of class III genes and is also required for high-level transcription of class IV genes (50). Hence, it can be hypothesized that FlrD stimulates class III and IV gene transcription via the two-component system FlrBC. Our results demonstrate that transcription of flrB and consequently flrC is not dependent on FlrD. Thus, FlrD and FlrBC most likely interact on the protein level (Fig. (Fig.7).7). We showed that FlrD is a transmembrane protein, and localization to the inner membrane appears to be necessary for its function. The histidine kinase FlrB is a soluble cytosolic protein (14), and the specific conditions that activate FlrB are still unknown. However, completion of the MS ring-switch-export apparatus is the most likely signal for activation of FlrBC (15), but such a regulatory mechanism would probably require an interaction of the cytosolic FlrBC system with a membrane component. A Sequence Similarity DataBase search analysis allocated parts of FlrD to the PilN fimbrial assembly protein family (amino acids 10 to 148) and revealed a conserved HAMP domain (amino acids 20 to 86) (2, 29-31, 40). HAMP domains are usually found in integral membrane proteins that are part of signal transduction pathways. It is suggested that HAMP domains play a role in regulating the phosphorylation or methylation of receptors by transmitting conformational changes from the periplasm to the cytoplasm (2). This makes FlrD an ideal candidate for transferring signals, like the assembly of the MS ring-switch-export apparatus, to the FlrBC two-component system.
If this hypothesis is correct, activation of FlrC should overcome the motility defects in the ΔflrD strain. Accordingly, we analyzed whether FlrC is epistatic to FlrD by overexpression of the response regulator FlrC, as well as a FlrB-independent point mutant, FlrC-M114I (14). It is known that merely increasing the dosage of a response regulator can mimic the effect of activation by phosphorylation (21, 26). Indeed, high-level expression of FlrC and especially FlrC-M114I rescued the motility defect of ΔflrD on soft-agar plates and restored expression of the flagellar class III protein FlaA. Our results suggest a potential interaction between the inner membrane protein FlrD and the two-component system FlrBC. It cannot be ruled out that FlrD does not directly interact with FlrBC, and instead another factor may be involved that interacts with FlrBC. A potential candidate for this would be FlhF, which has been previously characterized as an enhancer of class III and IV transcription (Fig. (Fig.7)7) (15).
Bioinformatic analysis using standard protein BLAST (BLASTP 2.2.21) revealed that FlrD is highly conserved within the order Vibrionales and conserved in other Vibrio spp., such as Vibrio parahaemolyticus, Vibrio harveyi, Vibrio vulnificus, and Vibrio fischeri, with identities of 50% or higher (1). Furthermore, FlrD homologues can also be found in other closely related Gammaproteobacteria, such as Photobacterium spp., Aeromonas spp., Shewanella spp., and Pseudomonas spp. (identities of 30% or higher), whereas FlrD seems not to be conserved in other bacteria, like E. coli. This is not surprising, since there are other genes like flhF and flhG, which are present in Vibrio spp. (and other bacteria), but are also not found in E. coli (41). Further studies are needed to elucidate the complex transcriptional regulation of flrD and the flhA-cheW1 gene region as well as the protein-protein interactions involved in the class II-class III checkpoint in more detail.
This work was supported by a grant from the Heinrich Jörg-Stiftung, Karl-Franzens Universität Graz, Austria, to S.S., by grant FWF-W901-B05 to J.R., and by NIH grant AI055058 to A.C. A.C. is a Howard Hughes Medical Institute Investigator.
Published ahead of print on 18 September 2009.