ABSTRACT
Flagellar biogenesis in Helicobacter pylori involves the coordinated expression of flagellar genes with assembly of the flagellum. The H. pylori flagellar genes are organized into three regulons based on the sigma factor needed for their transcription (RpoD [σ80], RpoN [σ54], or FliA [σ28]). Transcription of RpoN-dependent genes is activated by a two-component system consisting of the sensor kinase FlgS and the response regulator FlgR. While the cellular cues sensed by the FlgS/FlgR two-component system remain to be elucidated, previous studies revealed that disrupting certain components of the flagellar export apparatus inhibited transcription of the RpoN regulon. FliO is the least conserved of the membrane-bound components of the export apparatus and has not been annotated for any of the H. pylori genomes sequenced to date. A PSI-BLAST analysis identified a potential H. pylori FliO protein which membrane topology algorithms predict to possess a large N-terminal periplasmic domain that is absent from FliO of Escherichia coli and Salmonella, the paradigms for flagellar structure/function studies. FliO was necessary for flagellar biogenesis as well as wild-type levels of motility and transcription of RpoN-dependent and FliA-dependent flagellar genes in H. pylori strain B128. FliO also appears to be required for wild-type levels of the export apparatus protein FlhA in the membrane. Interestingly, the periplasmic and cytoplasmic domains were somewhat dispensable for flagellar gene regulation and assembly, suggesting that these domains have relatively minor roles in flagellar synthesis.
INTRODUCTION
Helicobacter pylori is a member of the Epsilonproteobacteria that colonizes the human gastric mucosa, where it can cause a variety of diseases, including chronic gastritis, peptic and duodenal ulcers, B cell MALT lymphoma, and gastric adenocarcinoma (1–3). H. pylori is highly motile via a cluster of sheathed polar flagella. Motility of H. pylori is required for virulence, as nonmotile mutants are unable to colonize the gastric mucosa of model animal systems (4, 5).
The bacterial flagellum is a helical propeller driven by a rotary motor and consists of three basic structures, referred to as the basal body, hook, and filament (6). The basal body is a complex, multifaceted structure that contains the flagellar rod, rings, motor, switch complex, and a specialized type III secretion system that transports axial components of the flagellum (e.g., rod, hook, and filament proteins) across the cell membrane (6–8). In Salmonella, a model organism for bacterial flagellum studies, the flagellar protein export apparatus consists of integral membrane proteins (FlhA, FlhB, FliO, FliP, FliQ, and FliR), which form an export pore, and cytoplasmic components (FliI, FliH, and FliJ), which deliver protein substrates to the export pore (9). With the exception of FliO, orthologs for all of the export apparatus proteins have been tentatively identified or annotated in at least some of the over 200 H. pylori genomes that have been sequenced to date.
Where it has been examined, expression of flagellar genes is controlled by a transcriptional hierarchy that coordinates the synthesis of flagellar proteins with assembly of the nascent flagellum (10, 11). Transcription of flagellar genes in H. pylori is governed by all three sigma factors found in the bacterium (see Fig. S1 in the supplemental material). Sigma (σ) factors bind core RNA polymerase and allow the resulting RNA polymerase holoenzymes to recognize specific promoters (12). Transcription of genes required early in flagellar assembly in H. pylori is dependent on the primary σ factor RpoD (σ80), transcription of genes needed later in flagellar biogenesis is dependent on RpoN (σ54), and transcription of genes needed near the end of the assembly process is dependent on FliA (σ28) (13). The organization of H. pylori flagellar genes into three regulons based on the σ factor needed for transcription suggests a framework for a transcriptional hierarchy operating in conjunction with flagellar assembly. The molecular mechanisms by which such a hierarchy might be regulated, however, have yet to be elucidated.
Products of the RpoN-dependent genes in H. pylori include rod proteins, hook protein, hook-associated proteins, the hook length control protein (FliK), a minor flagellin (FlaB), and enzymes required for flagellin glycosylation (13–15). A two-component regulatory system composed of the sensor kinase FlgS and the response regulator FlgR activates transcription of the H. pylori RpoN regulon (13, 14, 16, 17). The H. pylori RpoN regulon is linked with the export apparatus, as mutations in genes encoding various components of the export apparatus generally inhibit expression of RpoN-dependent flagellar genes (13, 18–21). While the mechanism by which the export apparatus influences expression of the H. pylori RpoN regulon is not known, it is likely mediated through the FlgS/FlgR two-component system. In support of this hypothesis, a constitutively active form of FlgR partially restored expression of an RpoN-dependent reporter gene in export apparatus mutants of Campylobacter jejuni (a member of the Epsilonproteobacteria that is closely related to H. pylori) (22).
To further examine the role of the export apparatus in expression of the H. pylori RpoN regulon, we characterized a potential FliO (HP0583 in H. pylori 26695) in H. pylori strain B128. FliO shows the least conservation among export apparatus proteins and even appears to be absent from some systems (23). Its role in the export apparatus is poorly defined, but deletion of fliO in Salmonella results in a drastic decrease in motility (24). The H. pylori FliO homolog differs from Salmonella FliO in that it is predicted to possess a large N-terminal periplasmic domain that is absent from the Salmonella protein (Fig. 1B). FliO was required for wild-type motility and flagellation, as well as transcription of both RpoN-dependent and FliA-dependent flagellar genes, in H. pylori. Interestingly, FliO variants that lacked either the periplasmic or cytoplasmic domains partially restored flagellation as well as transcription of RpoN-dependent and FliA-dependent flagellar genes, suggesting the transmembrane region of FliO is the most important determinant of the protein for flagellar biogenesis. Deletion of fliO resulted in reduced FlhA levels in the membrane, suggesting that FliO is required for the assembly or stability of other components of the export apparatus.
Genome organization around the hp0583 locus and predicted FliO structural features. (A) fliO (hp0583) is predicted to be in an operon with fliN, hp0582, hp0581, and hp0580. The promoter for this operon (PfliN) is located upstream of fliN (27), which encodes a flagellar motor switch protein. hp0582 encodes a member of the TonB family C-terminal domain, hp0581 encodes a possible dihydro-orotase (pyrC) involved in pyrimidine biosynthesis, and hp0580 encodes a possible neuraminidase/sialidase). (B) The FliO protein in H. pylori has an N-terminal domain not present in Salmonella or E. coli FliO proteins. The Salmonella FliO protein (top) is much shorter than the H. pylori FliO homolog (bottom). Transmembrane regions (TM) are shown in black, N-terminal periplasmic domains (PD) are shown in white, and C-terminal cytoplasmic domains (CD) are shown in gray. The predicted lengths of the proteins are indicated in parentheses.
MATERIALS AND METHODS
Bacterial strains and growth conditions.Escherichia coli DH5α was used for cloning and plasmid construction. E. coli strains were grown at 37°C in Luria-Bertani broth or agar supplemented with ampicillin (100 μg/ml), kanamycin (30 μg/ml), or chloramphenicol (30 μg/ml) when appropriate. H. pylori B128 (kindly provided by Richard Peek, Jr.) was grown at 37°C microaerobically (2% O2, 5% CO2, and 93% N2) on tryptic soy agar (TSA) supplemented with 10% horse serum or under an atmosphere consisting of 8.3% O2, 4.6% CO2, 9.2% H2, and 77.9% N2 in brain heart infusion (BHI) broth supplemented with 0.4% β-cyclodextrin. Kanamycin (30 μg/ml) or chloramphenicol (30 μg/ml) was added to the medium used to culture H. pylori when appropriate.
DNA sequencing.The region surrounding the putative fliO gene in H. pylori B128 was amplified using primers B128 fliO forward and B128 fliO reverse with iProof high-fidelity DNA polymerase (Bio-Rad) and cloned into pGEM-T Easy (Promega). Primers used are listed in Table S1 in the supplemental material. Both strands of the cloned DNA within the resulting plasmid were sequenced. All DNA sequencing was done using a commercial service (Genewiz, Inc., South Plainfield, NJ).
Construction of fliO mutation in H. pylori B128.The fliO homolog in H. pylori B128 (locus tag hpB128_25g10) was deleted as follows. Overlapping PCR was used to generate an amplicon that contained a chloramphenicol acetyltransferase (cat) gene flanked by ∼500-bp regions located upstream and downstream of the fliO homolog. iProof DNA polymerase was used for all PCR procedures for strain construction. Genomic DNA from H. pylori 26695 was prepared using the Wizard Genomic DNA purification kit (Promega) and was used as the template to amplify the regions flanking the fliO homolog, and the cat cassette was amplified from pUC20cat (25) using the cat forward and cat reverse primers. Overlapping PCR via the complementary regions of the amplified cat cassette and the amplicons of the regions flanking fliO generated a product with the cat cassette between the flanking regions. The final amplicon was introduced into H. pylori B128 by natural transformation, and chloramphenicol-resistant mutants were selected on TSA medium supplemented with the antibiotic. The fliO mutation in H. pylori B128 was confirmed by amplifying the regions around the targeted genes and sequencing the resulting amplicons.
Complementation of the ΔfliO mutant.The fliO deletion mutant was complemented by expressing wild-type fliO or truncated versions of fliO on the shuttle vector pHel3 (26). These fliO alleles were placed under the native promoter located upstream of fliN (27). The fliN promoter region was amplified by the primers PfliN forward and PfliN reverse. The PfliN reverse primer contains a sequence complementary to the primer fliO forward. Wild-type fliO was amplified with primers fliO forward and fliO reverse. The fliN promoter and fliO were fused by overlapping PCR. The resulting amplicon was cloned into pGEM-T Easy, sequenced, and subcloned into pHel3 (26) to create pfliO, which was introduced into H. pylori by natural transformation.
Truncated fliO alleles were created by overlapping PCR to express either the N terminus of FliO or the C terminus of FliO along with the transmembrane regions. For the N-terminal deletion (fliOΔN), amino acids 26 to 174 were replaced with the FLAG tag (DYKDDDDK). The fliN promoter and the first 75 bp of fliO were amplified with primers PfliN forward and fliOΔN reverse, and the last 360 bp of fliO was amplified with primers fliOΔN forward and fliO reverse from pfliO. The primers fliOΔN reverse and fliOΔN forward have reverse and complementary sequences that code for the FLAG tag at their 5′ end, which were used in overlapping PCR to create the fliOΔN allele. For the C-terminal deletion (fliOΔC), amino acids 217 to 283 were replaced with the FLAG tag. The fliN promoter and the first 648 bp of fliO were amplified with primers PfliN forward and fliOΔC reverse, and the last 33 bp of fliO were amplified with fliOΔC forward and fliO reverse 2 from pfliO. The primers fliOΔC reverse and fliOΔC forward also have reverse complementary sequences that have the FLAG tag at their 5′ ends which are used in overlapping PCR to create the fliOΔC allele. For the N- and C-terminal deletion, plasmid pfliOΔN was used as the template. Primers PfliN forward and fliOΔC 2 reverse were used to amplify the fliN promoter and fliOTM to introduce a C-terminal deletion within the fliOΔN allele. Amplicons bearing the three fliO alleles were cloned into pGEM-T Easy (Promega), sequenced, and subcloned into pHel3. The resulting plasmids bearing the truncated fliO alleles (pfliOΔN, pfliOΔC, and pfliOTM) were introduced into the ΔfliO mutant by natural transformation.
Motility assay.Motility was assayed using semisolid medium consisting of Mueller-Hinton broth and 0.4% Noble agar. After autoclaving, the medium was supplemented with sterile 10% heat-inactivated horse serum, 10 μM FeSO4, and 20 mM 2-(N-morpholino)ethanesulfonic acid (pH 6.0). Kanamycin (30 μg/ml) and chloramphenicol (30 μg/ml) were supplemented when appropriate. A sterile toothpick was used to inoculate the cells into the agar. Plates were incubated at 37°C under an atmosphere consisting of 2% O2, 5% CO2, and 93% N2. Diameters of the spreading H. pylori cells were measured after 7 days. For each strain, four stabs were analyzed. Statistical significance was determined using the two-sample t test.
Electron microscopy.Strains were grown to mid- to late-log phase in BHI supplemented with 0.4% β-cyclodextrin to an optical density at 600 nm (OD600) of 0.5 to 1.0. Kanamycin (30 μg/ml) was included in the medium for cultures of cells carrying derivatives of pHel3. Cells were pelleted by centrifugation for 1 min at 16,000 × g and then resuspended in half-strength Karnovsky's fixative (2.5% gluteraldehyde, 2% paraformaldehyde, 0.1 M cacodylate buffer). Cells were fixed for 5 min and then transferred to 300-mesh Formvar-coated copper grids. After 5 min, grids were washed with 0.1 M cacodylate buffer, followed by a wash with deionized water. Excess liquid was wicked off with filter paper between washes. One drop of 1% uranyl acetate was applied to the grids for 30 s and then wicked off with filter paper. Grids were washed in deionized water and dried at room temperature overnight. Cells were visualized using an FEI Tecnai20 transmission electron microscope. For each strain, at least 115 cells were included for determining the proportion of flagellated cells and the number of flagella per cell. Statistical analyses were performed using the Mann-Whitney U test to determine whether strains were significantly different from one another with regard to the number of flagella per cell.
RNA extraction and cDNA synthesis.H. pylori cells were grown on TSA supplemented with 10% horse serum for 18 h before harvesting and resuspending into 1 ml of nuclease-free water. Cells were pelleted by centrifugation for 1 min at 16,000 × g and then resuspended into 100 μl of nuclease-free deionized water. The Aurum total RNA minikit (Bio-Rad) was used to isolate RNA, and the RNA solution was treated with the TURBO DNA-free kit (Ambion) to remove any contaminating DNA. RNA was quantified using a BioPhotometer (Eppendorf), and RNA quality was confirmed on a 1.2% agarose gel. Single-strand cDNA was synthesized from 200 ng of RNA using the iScript cDNA synthesis kit (Bio-Rad). The efficiency of cDNA synthesis was assessed by analyzing a series of serial dilutions of each cDNA preparation using quantitative reverse transcription-PCR (qRT-PCR). If cDNA synthesis occurred at 100% efficiency, a theoretical −3.4 change in cycle threshold is expected for each 10-fold dilution of cDNA.
qRT-PCR.Transcript levels of flaA, flaB, and flgE were monitored by qRT-PCR using the Bio-Rad iCycler iQ system. Primers used are listed in Table S2 in the supplemental material. gyrA transcript levels were measured as a reference, as gyrA levels remain constant during the exponential phase (28). The specificity and efficiency of each primer pair was confirmed by PCR using genomic DNA and by qRT-PCR on a serial dilution of wild-type cDNA. Each qRT-PCR, totaling 20 μl, consisted of 10 μl of iQ SYBR green Supermix (Bio-Rad), 5 μl of 100-fold diluted cDNA from the cDNA synthesis reaction, and 200 nM each primer. In separate reactions, 5 μl of distilled H2O was used in placed of diluted cDNA as a no-template control, and 5 μl of 100-fold-diluted RNA was used in place of cDNA as a control to ensure that cDNA samples did not contain contaminating DNA from the RNA extractions. A melting curve analysis was performed at the end of each experiment. Experiments were performed in technical triplicate for three biological replicates of each strain. Gene expression levels were quantified by the 2−ΔΔCT method (29). Statistical significance was determined using the two-sample t test.
Immunoblotting.For detection of FlhA, H. pylori membrane fractions were collected as described previously (30). Cells were grown for 3 days on TSA plates supplemented with 10% horse serum and antibiotics where appropriate and resuspended into a buffer containing 10% sucrose, 20 mM HEPES, and 1 mM EDTA, pH 7.4 (buffer A). Cells were lysed by three passages through a French pressure cell at 10,000 kPa. Unlysed cells and cellular debris were removed by centrifugation for 15 min at 6,000 × g. Membranes were separated from cytoplasmic proteins by centrifugation for 60 min at 100,000 × g and collected into a 0.25-ml pad of 30% sucrose, 20 mM HEPES, and 1 mM EDTA, pH 7.4. The membrane fraction was diluted in buffer A, and the centrifugation was repeated. The cytoplasmic fraction was precipitated using trichloroacetic acid as described previously (31). Protein concentrations of the fractions were determined using the bicinchoninic acid protein assay. Membrane fractions were analyzed by Western blotting using affinity-purified FlhA antibodies (30) or KatA antiserum (kindly provided by Stéphane Benoit). Goat anti-rabbit horseradish peroxidase-conjugated antibody (Bio-Rad) was used as a secondary antibody. Antigen-antibody complexes were detected by chemiluminescence using the SuperSignal West Pico luminol/enhancer solution and SuperSignal West stable peroxide solution (Thermo Scientific). Blots were visualized using the FluoroChem E imager (ProteinSimple). Cross-reacting proteins detected in Western blot analysis were quantified by densitometry using ImageJ (http://rsbweb.nih.gov/ij/) for three biological replicates, and statistical significance was determined using the two-sample t test.
RESULTS
Identification of an FliO homolog in H. pylori.Of the six integral membrane proteins that comprise the flagellar protein export apparatus, FliO is the least conserved and may even be absent from some bacteria. FliO homologs have not been identified in any of the more than 200 H. pylori strains whose genomes have been sequenced to date. However, a gene downstream of fliN is annotated as a member of the FliO protein family (pfam04347) in Helicobacter bilis, Helicobacter cinaedi, and Helicobacter hepaticus. In addition, Pallen and coworkers tentatively identified an fliO homolog located immediately downstream of fliN (encoding a flagellar motor switch protein) in C. jejuni by analyzing the open reading frame using PSI-BLAST (23). The annotation of the C. jejuni fliO homolog is consistent with the synteny of fliN and fliO in other bacteria, including E. coli and Salmonella. Immediately downstream of fliN in H. pylori 26695 is an open reading frame (hp0583) which we identified as encoding a potential FliO homolog from PSI-BLAST analysis (Fig. 1A). Of the FliO homologs found in the 247 sequenced H. pylori genomes in the Joint Genome Institute's Integrated Microbial Genomes (IMG) database, FliO homologs among H. pylori strains shared 84 to 100% identity and 88 to 100% similarity with the corrected FliO sequence from H. pylori B128 (see below).
Potential fliO homologs are similarly located immediately downstream of fliN in other Helicobacter species. An examination of sequence information for the 18 Helicobacter species in the National Center for Biotechnology Information (NCBI) and the IMG databases, however, revealed that the putative FliO proteins in these bacteria vary markedly in their sequences. Only about half of these FliO proteins (8 out of 17) had significant homology with H. pylori FliO over their entire length (expect [E] values of <1e−10 from a BLAST analysis for the entire 293-amino-acid sequence of H. pylori FliO). These FliO homologs share 25 to 79% identity and 45 to 86% similarity with H. pylori FliO. The expect values for comparison of H. pylori FliO to the FliO homologs from the remaining Helicobacter species were >1e−3 in the BLAST analysis. A BLAST analysis of the N-terminal region and the C-terminal region of FliO (see below) indicated that neither domain was more conserved than the other. Out of the 8 species with significant homology to the H. pylori FliO protein, 5 species contained homology with the N-terminal region of FliO with an E value of <2e−4, and 6 species contained homology with the C-terminal region of FliO with an E value of <2e−5. In contrast to the FliO homologs, sequences of the other integral membrane components of the flagellar protein export apparatus (FlhA, FlhB, FliP, FliQ, and FliR) and FliN were highly conserved among the various Helicobacter species. Comparing sequences of these proteins from H. pylori to their counterparts from the 17 other Helicobacter species revealed robust sequence conservation over the entire lengths of the proteins, with FliR displaying the least conservation (46 to 96% identity and 67 to 98% similarity) and FliQ displaying the highest conservation (71 to 98% identity and 85 to 98% similarity).
Because H. pylori 26695 is nonmotile due to a frameshift mutation in fliP (32), H. pylori B128 was chosen for our studies. The predicted open reading frame located downstream of fliN in H. pylori B128 (locus tag hpB128_25g10) is substantially shorter than its counterpart in H. pylori 26695, corresponding to amino acid residues Met-13 through Pro-142 of the H. pylori 26695 FliO homolog. To determine if the H. pylori B128 protein is indeed truncated, the region surrounding the predicted reading frame was amplified using a high-fidelity DNA polymerase and sequenced. The sequence of the amplified DNA differed from that of the corresponding region in the draft sequence of the H. pylori B128 genome by two nucleotides out of 918 bp of DNA sequence. The draft genome sequence contains a TA base pair at position 391 which was not present in the sequence of the amplicon. This additional base pair results in a TAA stop codon immediately following codon 130 and corresponds to Pro-142 of the H. pylori 26695 FliO homolog. The genome sequence also contains an AT base pair located 192 nucleotides downstream of the other insertion and likewise was not present in the amplicon sequence. Some flagellar genes, such as flgR from C. jejuni, are known to be subject to phase variation resulting from the loss or gain of a nucleotide in the homopolymeric tract (33). Neither of the differences between the fliO sequences from the draft genome sequence and that which we determined, however, was within a homopolymeric tract. Thus, the differences in the sequences do not appear to be due to phase variation but instead may be due to sequencing errors in the draft sequence of the H. pylori B128 genome.
The revised sequence for the region surrounding the H. pylori B128 fliO homolog revealed an open reading frame encoding a protein of 293 amino acid residues which shared 92% amino acid identity with the H. pylori 26695 FliO homolog over its entire length. H. pylori FliO is significantly larger than that of E. coli or Salmonella, which is only 125 amino acid residues in length. There are two potential AUG start codons 36 nucleotides apart in fliO from H. pylori strains B128 and 26695 as well as most of the other 240 H. pylori strains whose genome sequences are in the IMG database. It is unclear which of the AUG sequences is the actual start codon. For many of the H. pylori FliO proteins in the IMG database, the first AUG is indicated as the start codon (51 out of 246), while the second AUG is indicated as the start codon for the remaining H. pylori FliO proteins. For H. pylori 26695 and B128, the upstream AUG overlaps fliN by four nucleotides. There is not a well-defined ribosome binding site upstream of either of the potential start codons for H. pylori B128 fliO (sequences upstream of the first and second AUG codons are 5′-TTATCTCGCTAAAAATTC-3′ and 5′-TTACTGAGCGCTACTTTG-3′, respectively). Given the ambiguity in assigning the start codon of fliO, we included the first AUG codon in all plasmid constructs used for expression of full-length or truncated versions of fliO.
The topology of the H. pylori B128 FliO protein was analyzed using two topology prediction programs, TopPred (http://mobyle.pasteur.fr/cgi-bin/portal.py?#forms::toppred) and TMpred (http://www.ch.embnet.org/software/TMPRED_form.html). The programs predicted two transmembrane regions corresponding to amino acid residues 1 to 21 and 187 to 207 or 1 to 18 and 188 to 206, respectively. TMpred predicted the C-terminal region of the H. pylori B128 FliO homolog to be located on the cytoplasmic side of the membrane and the N-terminal region of the protein to be located on the periplasmic side of the membrane (Fig. 1B). This is consistent with the topology of the Salmonella FliO protein, which was determined by Barker and coworkers using membrane topology prediction programs and chimeric proteins in which FliO was fused to alkaline phosphatase or green fluorescent protein (24). The predicted periplasmic domain of the H. pylori FliO homolog (∼160 amino acid residues) is much larger than that in FliO from E. coli and Salmonella (22 amino acid residues).
The H. pylori B128 FliO homolog is required for normal motility and flagellar biogenesis.The H. pylori B128 fliO homolog was disrupted with the cat cassette, and the phenotype of the resulting mutant was analyzed. The motility of the ΔfliO mutant was significantly reduced (Fig. 2B), but at least some of the cells appeared to be motile, as evidenced by the small halo around the point of inoculation, which was not observed with the nonmotile ΔflhA mutant (Fig. 2G) (30). Transmission electron microscopy revealed that most of the ΔfliO mutant cells were aflagellated (93.5%), although some cells possessed either a single flagellum (6%) or two flagella (0.5%) (Fig. 3B). In contrast, the majority of wild-type cells were flagellated (93%), with most of the cells possessing two to four flagella (Fig. 3A). These results suggested that FliO is required for optimal flagellar biogenesis but is not absolutely essential, since some ΔfliO mutant cells were flagellated.
Effects of fliO alleles on motility. Motility of the ΔfliO mutant and fliO variants were assessed by stab inoculation on 0.4% agar plates. Measurements indicate the diameter of the halo around the site of inoculation after 7 days of incubation in microaerobic conditions. For each strain, four different inoculations were analyzed. The motility of the wild-type strain (WT) was significantly different from that of the other strains (P < 0.002). Similarly, the motility of the ΔfliO mutant differed significantly from those of each of the complemented strains shown (P < 0.002) but not the ΔfliO/pfliOTM strain (data not shown). A ΔflhA mutant was used as a nonmotile control. Although the ΔfliO mutant clearly shows a halo around the point of inoculation that is not seen for the ΔflhA mutant, the size of the growth rings for the two strains did not differ significantly (P = 0.14).
Effect of fliO mutations on flagellation in H. pylori. H. pylori strains bearing various fliO alleles were examined by transmission electron microscopy after negative staining to determine the number of flagella per cell. At least 115 cells were counted for each strain. Bars indicate percentages of cells that possessed the specified number of flagella. All strains were significantly different from one another (P < 0.0001), with the exception of WT/pfliO and ΔfliO/pfliO (P = 0.98) and ΔfliO/pfliOΔN and ΔfliO/pfliOΔC (P = 0.71).
We attempted to complement the ΔfliO mutation by expressing fliO from the hp0405 locus in the H. pylori chromosome, but we were unable to achieve significant restoration of motility (data not shown). We postulate that the failure to complement the ΔfliO mutation with a copy of fliO introduced into the hp0405 locus is due to contextual issues which resulted in poor expression of the gene. Introducing a wild-type copy of fliO on the shuttle vector pHel3 (pfliO) into the ΔfliO mutant (ΔfliO/pfliO) resulted in a partial restoration of motility in the soft agar (Fig. 2D). Introduction of pfliO into the wild-type strain (WT/pfliO) reduced the motility of the strain in soft agar to that of the ΔfliO/pfliO strain (Fig. 2C), whereas introduction of just pHel3 into the wild-type strain had no effect on motility (data not shown). The copy number of pHel3 is ∼10 (26), and the additional copies of fliO may result in elevated levels of FliO, which could interfere with assembly of the export apparatus or regulation of the transcriptional machinery controlling flagellar gene expression. Alternatively, the plasmid-borne fliO may interfere with motility by titrating transcription factors needed for the normal regulation of other flagellar genes. The ΔfliO/pfliO and WT/pfliO strains were flagellated to similar extents, with ∼70% of the cells possessing from one to five flagella (Fig. 3C and D). Taken together, these results verify that deletion of the H. pylori fliO homolog was responsible for decreased flagellar biogenesis.
Barker and coworkers showed that incubation of a Salmonella ΔfliO mutant in soft agar gives rise to motile pseudorevertants that contain extragenic bypass mutations in fliP (24). To determine if the flagellated H. pylori ΔfliO cells similarly possessed extragenic bypass mutations, we tried to enrich for cells in which a higher proportion of the population was flagellated. To do so, the ΔfliO mutant was subjected to repeated rounds of inoculation in soft agar, allowing the cultures to grow for several days, and then picking bacteria from the periphery of the halo and reinoculating these cells into fresh soft agar. The ΔfliO mutant displayed a slight increase in motility during the enrichment (see Fig. S2 in the supplemental material), but there was not a concomitant increase in the proportion of flagellated cells (data not shown). These observations suggest that the flagellated ΔfliO cells do not have stable extragenic bypass mutations that allow them to produce flagella, rather that the export apparatus stochastically assembles with low efficacy into a fully functional form in the absence of FliO.
FliO is required for wild-type levels of flagellar gene expression.To determine if disruption of fliO affected flagellar gene expression, transcript levels of two RpoN-dependent flagellar genes, flaB and flgE, and one FliA-dependent flagellar gene, flaA, were examined in wild-type and ΔfliO mutant strains. Compared to the wild type, the ΔfliO mutant displayed an ∼24-fold reduction in flaB and flgE transcript levels (Fig. 4A and B) and an ∼7-fold reduction in flaA transcript levels (Fig. 4C). The fliO mutant displayed wild-type levels of flgR, flgS, and fliA transcripts (Fig. 4D) and RpoN protein (Fig. 4E), indicating that the lower levels of flaB, flgE, and flaA transcripts in the mutant were not due to reduced expression of regulatory genes known to be required for the transcription of the RpoN and FliA regulons. Introduction of pfliO into the ΔfliO mutant resulted in the partial restoration of flaB, flgE, and flaA transcript levels (Fig. 4A to C). Introducing pfliO into the wild-type strain depressed flaB and flgE transcript levels such that they were similar to those for the complemented ΔfliO mutants (Fig. 4A and B).
Effects of fliO alleles on transcription of specific flagellar genes. Transcript levels of flgE (A), flaB (B), and flaA (C) were assessed by qRT-PCR and normalized to transcript levels of gyrA. The expression of each strain relative to the wild type is plotted on the y axis, with wild-type levels indicated by the dashed line at 1.0. All transcript levels measured in the ΔfliO mutant were significantly different from those for the wild type (P < 0.05). All transcript levels measured in the WT/pfliO, ΔfliO/pfliO, ΔfliO/pfliOΔN, and ΔfliO/pfliOΔC strains are significantly different from that of the ΔfliO:cat mutant (P < 0.05). (D) Transcript levels of regulatory genes flgS, flgR, and fliA in the ΔfliO mutant were determined by qRT-PCR and normalized to transcript levels of gyrA. The relative levels of expression of each gene compared to the wild type (dashed line) are plotted on the y axis and were not significantly different from wild-type levels (P > 0.05). (E) RpoN levels in the WT and ΔrpoN and ΔfliO mutants were accessed by Western blotting using the soluble fraction. RpoN levels in the ΔfliO mutant were not significantly different from that in the wild type (P > 0.05).
The C-terminal and N-terminal domains of FliO are not individually essential for flagellar biogenesis.Barker and coworkers reported that overexpression of either the C-terminal domain of FliO or the first 95 amino acid residues of FliO restored motility in a Salmonella fliO deletion mutant (24). We wished to determine if truncated versions of FliO could similarly restore motility in the H. pylori ΔfliO mutant. Therefore, we constructed plasmids containing alleles encoding truncated versions of FliO in which the N-terminal domain (residues 26 to 174; pfliOΔN) or the C-terminal domain (residues 217 to 283; pfliOΔC) was deleted and examined the ability of the FliO variants to restore flagellar biogenesis in the ΔfliO mutant. A FLAG tag was introduced into the truncated FliO proteins encoded by the fliOΔN and fliOΔC alleles to verify that the proteins were expressed and incorporated into the cell membrane. However, we were unable to identify the truncated FliO proteins in immunoblots of membrane fractions prepared from strains bearing the pfliOΔN or pfliOΔC allele (data not shown). Our failure to detect the truncated FliO proteins likely was due to the low level at which these proteins were expressed and the presence of other cross-reacting proteins in the region of the gel where we expected the FliO variants to migrate. Despite our failure to detect the truncated FliO proteins, introduction of the plasmid-borne copies of either the fliOΔN or the fliOΔC allele into the ΔfliO mutant partially restored motility (Fig. 2E and F) and increased the proportion of flagellated cells from 6.5% to ∼40% (Fig. 3E and F). Despite being indistinguishable in the degree to which they were flagellated, the ΔfliO strain complemented with pfliOΔC appeared to be more motile than the ΔfliO strain complemented with pfliOΔN (Fig. 2E and F). The reason for this difference in motility is not known, but it indicates a role for the predicted periplasmic domain of FliO in chemotaxis or flagellar function. Introduction of the plasmid-borne fliOTM allele, in which both the N- and C-terminal regions were deleted, did not restore motility (data not shown). It is not known if the FliO variant was stably expressed and inserted into the membrane, so studies with the ΔfliO mutant bearing the fliOTM allele were not pursued further. Transcription of flaB, flgE, and flaA was restored in the ΔfliO mutant complemented with either the fliOΔN or fliOΔC allele to levels that were comparable to that observed with pfliO (Fig. 4A to C). Taken together, these data indicate that the C-terminal and N-terminal regions of FliO are individually dispensable for regulating transcription of the RpoN and FliA regulons.
FliO is needed for the assembly or stability of FlhA in the flagellar protein export apparatus.We wished to determine if loss of FliO affected the expression and/or localization of other integral membrane components of the flagellar protein export apparatus, which could account for the defects in flagellar biogenesis and gene expression seen in the ΔfliO mutant. To test this hypothesis, incorporation of FlhA into the membrane of the wild-type and ΔfliO strains, as well as the strains expressing the fliOΔC or fliOΔN allele, was analyzed by Western blotting. FlhA was detected in membrane fractions from wild-type cells but not membrane fractions from the ΔflhA mutant (Fig. 5, lanes 1 and 2). FlhA levels in the ΔfliO mutant were reduced ∼2-fold compared to the level for the wild type (Fig. 5, lane 3). FlhA levels in the WT/pfliO strain were similar to that in the wild type (Fig. 5, lane 4), and FlhA levels were restored to close to wild-type levels in the ΔfliO strain that expressed either the full-length fliO or the fliOΔN or fliOΔC allele (Fig. 4, lanes 5 to 7). FlhA was not detected in the soluble fraction from any of the strains (data not shown). To confirm that comparable amounts of membrane proteins were analyzed for each of the samples, we examined catalase (KatA) levels in each sample by Western blotting. KatA associates with the cell membrane through interaction with KapA, a twin-arginine target protein (34, 35). KatA was not detected in the soluble fractions of the cell extracts (data not shown); thus, it served as a good control for protein load. KatA levels within membrane fractions of the ΔfliO mutant and wild type did not differ significantly (P > 0.1), indicating that the amounts of total membrane protein analyzed for these samples were comparable (Fig. 5). flhA transcript levels in the ΔfliO mutant were similar to wild-type levels, suggesting that the changes in FlhA levels in the ΔfliO mutant membrane were not due to decreased transcription of flhA (data not shown). Taken together, these data indicate that FliO is needed for wild-type accumulation of FlhA in the export apparatus, perhaps by influencing the stability of FlhA or its assembly into the export apparatus. The transmembrane portion of FliO appears to be responsible primarily for the stability of FlhA, since FliO variants which lack either the N-terminal or C-terminal domain support near-normal accumulation of FlhA in the membrane. It is possible that levels of other components of the export apparatus are similarly reduced in the ΔfliO mutant.
Detection of FlhA in wild-type and mutant strains. Membrane fractions prepared from various H. pylori strains were analyzed by Western blotting using affinity-purified antibodies directed against a peptide corresponding to the N-terminal 25 amino acid residues of FlhA (top). Twenty μg of protein was loaded into each lane. FlhA from each lane was quantified using ImageJ. FlhA levels in the ΔfliO mutants were 2- to 3-fold lower than wild-type levels (P < 0.03), whereas FlhA levels in the complemented strains were not significantly different from those in the wild type (P > 0.4). The same membrane fractions were analyzed by Western blotting against KatA as a loading control (bottom). Four μg of protein was loaded into each lane, and KatA levels were quantified using ImageJ. KatA levels in the membrane fractions of the ΔfliO mutant were not significantly different from those in the wild type (P > 0.1).
DISCUSSION
FliO exhibits the lowest conservation among the integral membrane components of the flagellar protein export apparatus. FliO apparently is absent from some systems, such as Aquifex aeolicus or the lateral flagellar systems of Chromobacterium violaceum and Vibrio parahaemolyticus (23). Given the low degree of conservation of FliO sequences even within the Helicobacter genus, it is possible that these systems that are thought to lack FliO have highly divergent FliO homologs which have been overlooked. FliO homologs in some bacteria, including Bacillus species and Borrelia species, are annotated as FliZ and are substantially larger than FliO from E. coli or Salmonella. FliO homologs in H. pylori and C. jejuni similarly are larger than their counterparts in Salmonella and E. coli and are predicted to possess large periplasmic domains that are absent from E. coli or Salmonella FliO.
Loss of FliO in H. pylori appears to affect the organization or stability of other components within the export apparatus, as levels of FlhA were reduced ∼2-fold in the ΔfliO mutant (Fig. 5). Salmonella FlhA and FliO have been shown to interact with each other in affinity blots (36), and FliO may be necessary to stabilize FlhA in the membrane. The ability of mutations in fliP to suppress the ΔfliO mutation in Salmonella (24) suggests that FliO and FliP interact within the export apparatus. Further support for the interaction between these proteins comes from the existence of an fliOP gene fusion in Buchnera aphidicola (37). Although B. aphidicola does not produce flagella because it lacks the genes needed for filament formation, it does form hook-basal body complexes which may be used for export of other proteins (38, 39).
Although the mechanisms H. pylori and related bacteria use to couple transcription of the RpoN-dependent flagellar genes with assembly of the flagellum remain to be defined, a considerable amount of evidence has linked expression of the RpoN regulon in H. pylori and C. jejuni with the flagellar protein export apparatus. Inactivation of fliQ or flhB in H. pylori results in reduced levels of the minor flagellin FlaB and the hook protein FlgE, both of which are dependent on RpoN for their expression (19, 40). Using DNA microarray assays, Niehus and coworkers (13) demonstrated that FlhA is required for transcription of the RpoN regulon, and deletion of flhB similarly inhibits expression of RpoN-dependent reporter genes (31). Hendrixson and DiRita reported that deleting any one of several export apparatus components (FlhA, FlhB, FliP, and FliR) inhibited expression of an RpoN-dependent flagellar reporter gene in C. jejuni (41).
We postulate that the export apparatus influences expression of the H. pylori RpoN-dependent flagellar genes through interactions with one or more of the regulatory proteins that control transcription of the RpoN regulon. A likely candidate for such interactions is the sensor kinase FlgS, which, upon stimulation, would support the phosphorylation of FlgR, allowing it to activate transcription of the RpoN regulon. Boll and Hendrixson proposed similar models for regulation of the RpoN regulon in C. jejuni and further showed that FlgS immunoprecipitates with FliF (flagellar MS ring protein) and FliG (a component of the C ring) (42). They proposed two models for how FliF, FliG, and the export apparatus activate FlgS autokinase activity. In the first model, the formation of the MS and C rings creates a cytoplasmic platform that is sensed by FlgS. In this model, other proteins, such as FlhA, may contribute to the interaction between FlgS, FliG, and FliF (42). Alternatively, formation of the MS and C rings may be required for assembly of a conformation of the export apparatus capable of interacting productively with FlgS (42). Our observation that loss of FliO negatively impacts the level of at least one other export apparatus component (i.e., FlhA) of the phenotype of the H. pylori ΔfliO mutant is consistent with either of the models proposed by Boll and Hendrixson.
The results presented here show that FliO is unique among the export apparatus components in that while it is required for wild-type transcription of RpoN-dependent flagellar genes, it is not as essential as other export apparatus components, such as FlhA or FlhB (22, 30, 31), where their loss abolishes flagellar biogenesis and motility. The molecular basis by which the flagellar protein export apparatus influences transcription of the RpoN-dependent and FliA-dependent genes in H. pylori is complex. Characterizing the roles of FliO and other components of the export apparatus is critical for understanding how H. pylori coordinates flagellar gene expression with assembly of the flagellum.
ACKNOWLEDGMENTS
We thank Rob Maier for use of laboratory equipment, Stéphane Benoit for the KatA antibody and advice on qRT-PCR, Jonathan McMurry for comments on the manuscript, Jan Mrázek for helpful discussions, and John Shields for technical assistance with electron microscopy.
This work was supported by award MCB-1244242 from the National Science Foundation.
FOOTNOTES
- Received 8 November 2013.
- Accepted 9 May 2014.
- Accepted manuscript posted online 16 May 2014.
- Address correspondence to Timothy R. Hoover, trhoover{at}uga.edu.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JB.01332-13.
REFERENCES
- Copyright © 2014, American Society for Microbiology. All Rights Reserved.
