Motility is an essential colonization factor for the human gastric pathogen Helicobacter pylori. The H. pylori genome encodes most known flagellar proteins, although a number of key transcription regulators, chaperones, and structural proteins have not yet been identified. Using recently published yeast two-hybrid data we identified HP0958 as a potential motility-associated protein due to its strong interactions with RpoN (σ54) and FliH, a flagellar ATPase regulator. HP0958 exhibits no sequence similarity to any published flagellar genes but contains a carboxy-terminal zinc finger domain that could function in nucleic acid or protein binding. We created a HP0958 mutant by inserting a chloramphenicol resistance marker into the gene using a PCR-based allelic exchange method and the resultant mutant was non-motile as measured by a BacTracker instrument. Electron microscopic analysis revealed that the HP0958 mutant cells were aflagellate and Western blot analysis revealed a dramatic reduction in flagellin and hook protein production. The HP0958 mutant also showed decreased transcription of flgE, flaB and flaA as well as the checkpoint genes flhA and flhF. Expression of flgM was increased relative to the wild-type and both rpoN and fliA (σ28) expression were unchanged. We conclude that HP0958 is essential for normal motility and flagella production, and represents a novel flagellar component in the epsilon proteobacteria.
Helicobacter pylori, a Gram-negative spiral bacterium, is a causative agent for duodenal ulcers and peptic ulcers [1,2], and infection with H. pylori is a risk factor for gastric adenocarcinoma  and for B cell MALT lymphoma . H. pylori infection in Western countries occurs in up to 50% of the population , although at present infection rates are decreasing . A better understanding of H. pylori metabolism and cell biology will facilitate development of new therapies for treating this chronic infection.
Motility is an essential colonization factor for H. pylori in experimental infection models [6–8]. Flagellar genes have been shown to be transcribed in infection of humans and mice [9,10]. In addition to imparting motility, there is an association between H. pylori flagellar biogenesis and virulence mechanisms , and the sialic acid-specific adhesin HpaA is enriched in the flagellar sheath . Recent studies have also shown that, similar to other bacteria, H. pylori flagella induce a cytokine response in host cells through interactions with Toll-like receptors , but the mechanisms are still controversial .
The genomic basis of H. pylori flagellum production is therefore of considerable interest. The genetics and genomics of flagellar biogenesis have been studied in most detail in Salmonella enterica and Escherichia coli. Compared to these paradigms, the H. pylori genome [16,17] contains homologs of most of the expected complement of flagellar genes (reviewed in ). The flagellar filament is composed of a major flagellin FlaA, and a minor flagellin FlaB [19,20], and the hook is composed of FlgE protein . Expression of flagellar genes is controlled by at least three RNA polymerase sigma factors, σ80, σ54 and σ28[16,17,22,23] and the anti-σ28-factor FlgM [24,25]. Suerbaum and colleagues  have shown that FlhA and FlhF are two important transcription regulators of global flagellar biosynthesis. The current paradigm for flagellar transcription control in H. pylori describes three classes; class 1 gene expression is controlled by the σ80 factor, class 2 genes are controlled by the σ54 factor, while class 3 genes are expressed by the σ28-FlgM controlled system [22,26]. There exists a fourth less well-defined set of intermediate class genes, which contain both σ54 and σ28 binding sequences.
We sought to identify new flagellar genes which had not been found in the original genome annotations [16,17]. The protein–protein interaction map of H. pylori has recently been established by a high-throughput yeast two-hybrid (Y2H) strategy . A number of the major flagellar structural proteins, and export proteins, had both predictable and cryptic interaction partners . Our examination of the published Y2H interaction list for H. pylori revealed that of the 34 annotated flagellar proteins, 23 were shown to interact with other proteins. This identified 41 interaction partners of unknown function, which we have begun to investigate by targeted knock-out mutagenesis. One such protein, encoded by HP0958, showed a strong Y2H interaction (A rank) with σ54 (or RpoN) and also FliH . FliH forms a complex with FliI, and regulates its activity in Salmonella. FliI is the ATPase that drives flagellar protein export in H. pylori and other motile bacteria [10,29,30]. We have recently identified the residues involved in the interaction of H. pylori FliI and FliH . The observation that HP0958 interacted with FliH thus afforded the opportunity to logically extend this protein interaction network node. We now describe microscopic, biochemical and transcriptional analysis of a mutant defective in HP0958 production. Our data support the hypothesis that HP0958 has an important role to play in the regulation of transcription of both class 2 and class 3 flagellar genes, possibly by influencing expression of the regulators flhA and flhF.
2Materials and methods
2.1Bacterial strains, media and growth conditions
Helicobacter pylori strain CCUG 17874 (Culture Collection University of Gothenburg, Sweden; identical to NCTC 11637, the type strain of H. pylori) was used in this study. Bacteria were cultured on chocolate blood agar (CBA) plates (Columbia agar base) with 10% (v/v) heat-inactivated whole horse blood (Charles River Labs) for 48 h at 37°C in an atmosphere containing 5% CO2. H. pylori liquid cultures were grown in brain-heart infusion (BHI) broth (Oxoid) with 10% heat-inactivated calf serum (Sigma), under agitation in microaerobic conditions generated by CampyGen sachets (Oxoid). H. pylori transformants were selected on CBA plates containing chloramphenicol at 10 μg/ml.
For transmission electron microscopy, samples were subjected to negative staining. Whole cells of H. pylori, recovered from a plate of BHI/1% newborn calf serum agar incubated for 24 h, were gently suspended with a loop in 2% ammonium molybdate with 70 μg ml−1 bacitracin as wetting agent to just-visible turbidity. A drop was applied immediately to a copper grid covered with a carbon-coated Formvar film. The excess sample was withdrawn by touching the edge of the grid to a cut edge of Whatman No. 1 filter paper. The grids were examined in a JEOL JEM-1200EX transmission electron microscope operated at an accelerating voltage of 80 kV.
To assess motility, computer image processing technology coupled to phase contrast microscopy (Hobson BacTracker) was performed three times as previously described , using H. pylori cultures grown for 24 h in BHI broth plus 10% calf serum. Curvilinear velocity (CLV) and run time (RT) were measured and statistical differences were calculated using the Student's t test.
2.3Allele exchange mutagenesis
Helicobacter DNA was isolated as previously described . All custom primers were purchased from MWG. Marker exchange mutagenesis of HP0958 was carried out according to a published PCR-based method , using primers listed in Table 1. Briefly, two fragments were PCR amplified from each target gene adjacent to an internal portion of the gene that was to be knocked-out. Each PCR fragment contained a 21 bp overlap consisting of chloramphenicol acetyl transferase (CAT) gene sequence. Following amplification of the CAT gene from pRY109 , it was joined to both target gene fragments in two sequential overlap extension PCRs. The final triple PCR product was concentrated using Pellet Paint (Novagen) and 2–5 μg was used to transform H. pylori as described . Three primer sets were used to confirm integration: 0958-F1 and 0958-R2; 0958-F1 and C2 (cat R primer); 0958-R1 and C1 (cat F primer). A knock-out genotype for the mutant was confirmed by amplifying a 1331 bp fragment from the knock-out gene (compared with 681 bp from the wild-type), using the HP0958-based primers, and the cat-HP0958 amplicons of the mutant were also consistent with the expected genome structure (data not shown).
Table 1. Primers used in this study
Annealing position, gene
nt 109–128, HP0517
Forward primer for real-time PCR of housekeeping gene
nt 189–208, HP0517
Reverse primer for real-time PCR of housekeeping gene
nt 24–45, HP0958
Forward primer for left flank of mutagenic construct
nt 320–340, HP0958
Reverse primer for left flank of mutagenic construct
nt 420–441, HP0958
Forward primer for right flank of mutagenic construct
nt 704–724, HP0958
Reverse primer for right flank of mutagenic construct
nt 192–212, HP0957
Forward primer for RT-PCR of HP0957
nt 372–392, HP0957
Forward primer for RT-PCR of HP0957
nt 213–232, HP0601
Forward primer for real-time PCR of flaA
nt 504–523, HP0601
Reverse primer for real-time PCR of flaA
nt 1739–1758, HP0870
Forward primer for real-time PCR of flgE
nt 1863–1882, HP0870
Reverse primer for real-time PCR of flgE
nt 1051–1070, HP0115
Forward primer for real-time PCR of flaB
nt 1164–1183, HP0115
Reverse primer for real-time PCR of flaB
nt 112–131, HP1122
Forward primer for real-time PCR of flgM
nt 201–221, HP1122
Reverse primer for real-time PCR of flgM
nt 927–946, HP1041
Forward primer for real-time PCR of flhA
nt 1012–1031, HP1041
Reverse primer for real-time PCR of flhA
nt 645–664, HP1035
Forward primer for real-time PCR of flhF
nt 728–747, HP1035
Reverse primer for real-time PCR of flhF
nt 435–454, HP0714
Forward primer for real-time PCR of rpoN
nt 504–524, HP0714
Reverse primer for real-time PCR of rpoN
nt 23–42, HP1032
Forward primer for real-time PCR of fliA
nt 107–126, HP1032
Reverse primer for real-time PCR of fliA
cat gene, pRY109
Forward primer for cat gene
cat gene, pRY109
Reverse primer for cat gene
2.4Protein electrophoresis and blotting
Helicobacter pylori cells were harvested from 48 h plate cultures into 1 ml PBS and centrifuged at 20,800g for 1 min. Residual medium was removed from the cell pellet after an additional spin of 1 min at 20,800g. Pellets were resuspended in sterile water, boiled for 10 min and stored at −70°C. Standard conditions were employed for immunoblotting and sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS–PAGE; ), in gels containing 10% acrylamide. Rabbit antisera against H. pylori flagellin and hook protein was prepared as described previously .
2.5Silver staining for lipopolysaccharide (LPS)
Whole cell fractions of wild-type and HP0958 mutant were collected after 48 h of growth and washed with 1 ml PBS. Cells were lysed in Tris buffer containing 2% SDS, 4%β-mercaptoethanol and 10% glycerol (all Sigma) and treated with 25 μg proteinase K (Sigma) for 1 h at 60°C. Samples were separated using 12% SDS–PAGE and LPS was stained using silver nitrate .
Quantitative real-time reverse transcriptase PCR (qRT-PCR) was employed to determine relative transcript amounts of selected flagellar genes. RNA was extracted from H. pylori strain 17874 or HP0958 mutant cells grown in liquid culture for 8 h using the Absolutely RNA RT-PCR Miniprep kit (Stratagene) according to the manufacturer's instructions. RNA thus prepared was analyzed using a BioAnalyzer 2100 instrument (Agilent Technologies, Palo Alto, USA) and showed no sign of degradation. A minimum of 0.5 μg of RNA was reverse transcribed using 20 ng of random primer and the Improm-II reverse transcriptase enzyme (both Promega) as per the manufacturer's recommended protocol. Real-time PCR primers for eight flagellar-associated genes and a housekeeping gene (era HP0517) were designed using the Primer3 online software package (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi). Primer sequences are listed in Table 1. qRT-PCR was performed using an ABI7000 thermocycler (Applied Biosystems). Each 12.5 μl reaction contained 50 nM of each primer, 6.25 μl of 2× Master Mix and 1/60,000 Sybr Green I (both from Biogene). Individual amplification reactions were optimized for single-band specificity and verified by gel analysis of the pilot reaction products. Dissociation curves of all subsequent test reactions were monitored. Reactions were performed in triplicate and critical threshold (ct) values were averaged. Fold change in expression was calculated according to the standard formula 2(En−Rn)−(Et−Rt), where En is the ct of the experimental gene in the normal (wild-type strain) sample, Rn is the ct of the reference gene in the normal sample, Et is the ct of the experimental gene in the treated (knock-out strain) sample and Rt is the ct of the reference gene in the knock-out sample. qRT-PCRs were repeated on three different sets of cultures collected on separate days and fold expression changes were averaged.
Alignment of genomic regions was performed using The Institute for Genome Research (TIGR) Comprehensive Microbial Resource (CMR; ), CampyDB , and draft H. mustelae genome sequence from the Pathogen Sequencing Unit of the Wellcome Trust Sanger Institute (http://www.sanger.ac.uk/Projects/H_mustelae/). BLAST searches  of the non-redundant protein database at NCBI (http://www.ncbi.nlm.nih.gov/BLAST/) using standard parameters and the HP0958 protein sequence yielded a number of highly significant hits. Iterative PSI-BLAST searches  using the identified homologs of HP0958 uncovered several other members of this protein family. We then used T-Coffee  to create a multiple alignment. Secondary structure was predicted by a combination of manual inspection of the multiple sequence alignment followed by automated prediction using the PHD algorithm at the Predict Protein Server .
3Results and discussion
3.1Bioinformatic analysis of HP0958
We have begun to use the protein interaction network of H. pylori to identify and functionally investigate proteins interacting with flagellum components, that might represent known flagellar proteins that had escaped annotation in H. pylori, or novel flagellar proteins specific to H. pylori. One such protein, HP0958, has a predicted size of 29,691 Da and has orthologs in several related species, none of which have yet been annotated (Fig. 1). These include; HH0639 of Helicobacter hepaticus (70% similar), WS2117 of Wolinella succinogenes (70% similar), Cj0706 of Campylobacter jejuni (64% similar), and 60% similarity with the product of an unnamed ORF in the draft genome sequence of H. mustelae. More distant homologs are also present outside the epsilon proteobacteria (see below). Although there is no extended synteny in the wider genomic regions of the five epsilon proteobacteria, the respective HP0958 ortholog is followed by kdtA (3-deoxy-d-manno-octulosonic-acid transferase) and a putative rluD gene (pseudouridine synthase) gene in all five genomes (Fig. 1). Four of the five genomes contain a gene for a conserved hypothetical protein immediately upstream of the HP0958 ortholog. In W. succinogenes, the HP0959 and HP0958 orthologs appeared to be fused to form WS2117. In three of the aligned genomes (Fig. 1), the glyQ gene (glycyl-tRNA synthetase, alpha subunit) is closely linked. These observations suggest selection for conservation of gene order in a group of bacteria in which genomic diversity is a distinguishing feature .
A multiple alignment was generated of HP0958 and presumptive orthologs identified by PSI-BLAST. This alignment (Fig. 2) revealed the presence of four absolutely conserved cysteine residues near the C-terminus of the polypeptide. The spacing of these cysteine residues is highly reminiscent of the spacings of four conserved cysteine residues in the zinc ribbon family of proteins . Many of these proteins are involved in transcriptional regulation. In addition to the presence of the four conserved cysteine residues, amino acid sequences bordering the cysteines are highly suggestive of β-sheet forming amino acid residues (Fig. 2), consistent with our tentative assignment of HP0958 to the zinc ribbon protein superfamily. Taken together, these bioinformatic analyses led us to hypothesize that HP0958 may have a regulatory role to play in flagellar biosynthesis in H. pylori, possibly through protein–nucleic acid or protein–protein interactions.
3.2An HP0958 mutant is non-motile
An insertion mutant was created by allele replacement with a fragment in which residues 321–419 of HP0958 had been deleted, and replaced with the chloramphenicol resistance gene of pRY109. Chloramphenicol-resistant transformants were verified for the expected insertional mutation in HP0958 by PCR, using HP0958-specific and HP0958-cat specific primer pairs (not shown). Disruption of HP0958 could have polar effects; as shown in Fig. 1, it is linked to HP0957 (kdtA) and HP0956 (rluD). To rule out this possibility, we performed RT-PCR for HP0958 and HP0957. The data showed that HP0958 was expressed only in the wild-type strain and not in the mutant (not shown). Furthermore, amplicons were generated from both wild-type and mutant using the HP0957 primers. Aware that the HP0957 amplicon in the mutant strain may be due to transcriptional run-off from expression of the chloramphenicol resistance gene, we examined the LPS phenotype of the HP0958 mutant to ensure KdtA function was intact. KdtA catalyses the transfer of 3-deoxy-d-manno-octulosonic acid to lipid A, a step which is necessary for LPS biosynthesis . Silver staining of proteinase K-treated whole cell extracts revealed no difference in the LPS profile of the wild-type and HP0958 mutant, indicating that KdtA activity was intact in the mutant (Fig. 3). This result was not unexpected as it has been reported that kdtA is an essential gene in gram-negative bacteria . Nevertheless, it did rule out polarity as a confounding factor for interpreting the phenotype of the HP0958 mutant.
Growth rates and plate morphology of the HP0958 mutant were comparable to that of the wild-type strain (data not shown) but the mutant was non-motile as judged by phase-contrast light microscopy and motility agar plates (data not shown). To corroborate this, the mutant was also examined by BacTracker, a digitized bacterial motility tracking system  (Fig. 4). The HP0958 mutant was non-motile, as shown by measurements of CLV (the length of the track traveled by a bacterium divided by the time taken to travel that distance) and RT (the time taken for a bacterium to complete a run between two stops), while the wild-type strain 17874 grown and examined in parallel showed normal spiraling, tumbling, and straight runs (Fig. 4). CLV and RT values measured for the HP0958 mutant are similar to those previously calculated for Brownian motion .
3.3Electron microscopic analysis of the HP0958 mutant
To investigate whether the lack of motility in the HP0958 mutant was due to aberrant flagella structure or composition, negatively stained cells of wild-type and mutant were examined by electron microscopy (Fig. 5). In contrast to the wild-type, which produced normal flagella, cells of the HP0958 mutant were completely aflagellate. Moreover, there were no free full-length or partial flagellar filaments visible in the HP0958 preparations. Other than lack of flagella, the HP0958 mutant cells appeared to have a normal gross morphology (Fig. 5).
3.4Decreased flagellar protein in the HP0958 mutant
It was unclear whether the lack of flagella on the surface of the HP0958 mutant was due to a deficiency of protein production, or a failure to export these components to the cell surface. To answer this question, we examined whole cell lysates of wild-type and HP0958 mutant by immunoblotting using anti-flagellin anti-serum . The anti-serum reacts with both the major flagellin FlaA, the minor flagellin FlaB, as well as the flagellar hook protein, FlgE. The wild-type cells had normal levels of Fla and FlgE, whereas the HP0958 mutant expressed only a very small amount of flagellin and no detectable FlgE (Fig. 6). We could not clearly distinguish between FlaA and FlaB due to the similarity in molecular weights (FlaA = 53 kDa and FlaB = 54 kDa) but it appears most likely that the HP0958 mutant has small amounts of only FlaA. The absence of flagella on the surface of the HP0958 mutant is therefore due (at least in part) to an almost complete lack of expression of these proteins and not simply a failure of the mutant to export them.
3.5Transcription of flagellum-related genes in an HP0958 mutant
The current model for the regulation of flagellar biosynthesis in H. pylori involves four classes of genes. The expression of class 1 early genes is under the control of an unknown signal and involves important regulators such as flgR, which is an activator of the alternative sigma factor σ54 and flhA. Activation of σ54 in turn initiates transcription of class 2 middle genes, a group that includes structural proteins such as FlaB and FlgE. Following completion of middle gene expression, the FliK homolog HP0906 triggers a switch in expression to class 3 late genes under the control of FliA (σ28) . Furthermore, there exists an intermediate class of regulatory and structural genes that contain both σ54 and σ28 regulatory elements . Many elements of this complex pathway are yet to be resolved.
To further investigate the basis for the lack of flagellar protein production by the HP0958 mutant, we examined the transcription levels of a number of important genes in each of the four classes described above using real-time reverse transcriptase PCR. As noted above, the HP0958 mutant showed similar growth characteristics to the wild-type strain, and era expression was unchanged between the two strains. Transcription of the class 2 genes flaB and flgE was reduced in the HP0958 mutant (Table 2). flaB expression was decreased approximately fivefold and flgE threefold compared to that of the wild-type. Similarly, expression of the class 3 gene flaA was reduced approximately fivefold compared to wild-type. Reduced expression of these three important structural flagellar proteins alone would certainly account for the lack of flagella observed on the HP0958 mutant, but this finding could be an indirect effect, and it did not provide any further insights into the possible function of HP0958.
Table 2. Flagellar gene expression in the HP0958 mutant compared with wild-type, as determined by quantitative real-time PCR
Flagellar gene classa
Fold expression relative to wild-type
Fold expression compared to wild-type was calculated using era as reference gene, as described in Section 2. Expression values are means of three independent biological replicates ± standard error of the mean.
aFlagellar gene classification as per Suerbaum and colleagues .
We next measured expression of the transcriptional regulators of the class 2 and 3 genes, namely fliA (σ28) and rpoN (σ54). Expression of both sigma factors was unaltered in the HP0958 mutant strain (Table 2), indicating that changes in other transcriptional regulators might be responsible for the decreased expression of the class 2 and 3 genes. Suerbaum and colleagues [26,50] have confirmed that flhA and flhF are important for flagellar biosynthesis in H. pylori. They have shown that knock-out mutants of either of these genes have reduced expression of both gene classes. In our current study, both flhA and flhF were down-regulated approximately fourfold in the HP0958 mutant described here (Table 2). In agreement with this data, fliA and rpoN expression were unchanged in the flhA and flhF mutants of Niehus et al. . We next measured the expression of FlgM, the fliA antagonist, in the HP0958 mutant. Expression of this intermediate class gene was upregulated ninefold compared to the wild-type. An increase in FlgM in the HP0958 mutant would account for a concomitant decrease in class 3 (σ28) expression as FlgM binds to σ28 and prevents it interacting with RNA polymerase . It is unclear whether flgM is upregulated because of the absence of HP0958 protein, or because of an indirect feedback mechanism “turning off” FliA regulated expression due to disruption of the flagellar regulatory network.
Expression of genes within a flagellar class was not uniformly affected in the H0958 mutant; but appeared to depend on the function of the protein. The eight genes examined in the present study can be broadly classified as regulatory (rpoN, flhA, flgM, flhF, fliA) or structural (flaB, flgE, flaA) irrespective of gene class. We use the term “regulatory” with the reservation that any upstream gene that interferes with the checkpoint determined by structural progression will have such an apparent “regulatory” function. It is clear that all the structural genes were downregulated; the expression levels of the sigma factors rpoN and fliA were unchanged; the anti-σ factor flgM was upregulated; and the regulators flhA and flhF were upregulated. Definition of the function of HP0958 will help to explain this phenomenon.
We have shown for the first time that HP0958, a protein of previously unknown function, is essential for normal flagellum production and motility in H. pylori. Based upon significant levels of residue identity with HP0958, it is likely that the orthologs presented in Fig. 1 will have similar functions in the respective organisms. Although the exact function of HP0958 is still unclear, a number of intriguing possibilities exist. The identification of a putative nucleic acid-binding motif suggests HP0958 may possibly have a role to play in gene transcription. Our data show that, in the absence of HP0958, there is a downregulation of a number of flagellar structural genes that is most likely due to lowered expression of the important flagellar proteins flhA and flhF. It may be that HP0958 functions upstream of these activators and facilitates their expression either in a direct or indirect manner. Alternatively HP0958 may function via protein–protein interactions, and the yeast-two hybrid data suggest that it has strong interactions with both σ54 and FliH. It is conceivable that if HP0958 interacts with σ54, it could inhibit or enhance binding of the sigma factor to either RNA polymerases or nucleic acid. Moreover, as FliH regulates the activity of a flagellar export ATPase, the absence of HP0958 may have a deleterious effect on flagellar-related export ATPase activity, resulting in a feedback mechanism turning off expression of flagellar genes or their regulators. Certainly there also exists the possibility that HP0958 may exhibit more than one of these activities. Work is underway in our lab using global transcript analysis, to elucidate the exact role of HP0958 in flagellar biosynthesis.
Work in PWOT's laboratory is supported by the Irish Research Council for Science, Engineering and Technology, and by Science Foundation Ireland to the Alimentary Pharmabiotic Center. CWP was supported by BBSRC and the Darwin Trust of Edinburgh.