SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Helicobacter pylori is thought to regulate gene ex-pression with a very small set of regulatory genes. We identified a previously unannotated open reading frame (ORF) in the H. pylori 26695 genome (HP1122) as a putative H. pylori flgM gene (σ28 factor antagonist) by a motif-based bioinformatic approach. Deletion of HP1122 resulted in a fourfold increase in transcription of the σ28-dependent major flagellin gene flaA, supporting the function of HP1122 as H. pylori FlgM. Helicobacter pylori FlgM lacks a conserved 20-amino-acid N-terminal domain of enterobacterial FlgM proteins, but was able to interact with the Salmonella typhimuriumσ28 (FliA) and inhibit the expression of FliA-dependent genes in Salmonella. Helicobacter pylori FlgM inhibited FliA to the same extent in a Salmonella strain with an intact flagellar export system and in an export-deficient strain. Helicobacter pylori FliA was able to drive transcription of FliA-dependent genes in Salmonella. The effects of mutations in the H. pylori flgM and fliA genes on the H. pylori transcriptome were analysed using whole genome DNA microarrays. The antagonistic roles of FlgM and FliA in controlling the transcription of the major flagellin gene flaA were confirmed, and two additional FliA/FlgM dependent operons (HP472 and HP1051/HP1052) were identified. None of the three genes contained in these operons has a known function in flagellar biogenesis in other bacteria. Like other motile bacteria, H. pylori has a FliA/FlgM pair of sigma and anti-sigma factors, but the genes controlled by these differ markedly from the Salmonella/Escherichia coli paradigm.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Helicobacter pylori persistently colonizes the stomach of one half of the world population. It is one of the most common human pathogens, causing chronic gastritis in all infected persons and, in some individuals, causes more severe diseases including ulcers and gastric cancer. The only known habitat of H. pylori is the gastric mucosa of humans. Helicobacter pylori requires flagella-dependent motility to survive and multiply in this particular ecological niche (Eaton et al., 1996; Mankoski et al., 1999). Some components of the flagellar apparatus of H. pylori have been studied intensely over the last two decades. Many different flagellar components have been identified, including two flagellins that are governed by σ28 and σ54 promoters respectively (Suerbaum et al., 1993; O´Toole et al., 1994; Foynes et al., 1999; Kim et al., 1999; Spohn and Scarlato, 1999). The publication of H. pylori genome sequences has facilitated the identification of additional proteins with putative functions in motility, based on their similarity to known motility proteins of Salmonella typhimurium, Bacillus subtilis and other well characterized bacteria (Tomb et al., 1997; Alm et al., 1999). Reconstruction of the H. pylori motility system, based on the original annotations of the two published genome sequences, showed several differences between H. pylori and the Enterobacteriaceae. One of the most notable differences between the enterobacterial motility regulation cascade and the H. pylori system was the apparent lack of two key regulatory components, the flhCD master operon, and the antisigma (σ28) factor gene flgM. This is consistent with a general scarcity of regulatory proteins in H. pylori. The flagellar regulatory system of S. typhimurium is an intricate network of temporally regulated genes, which are organized into a transcriptional hierarchy (Macnab, 1996; Aizawa, 2000; Chilcott, 2000). At the top of the hierarchy are the flagellar master regulators FlhCD (class 1), which act as positive transcriptional regulators on the class 2 promoters. Class 2 operons encode the majority of the proteins of the flagellar hook and basal body, including the flagella-specific Type III secretion apparatus. The class 3 promoters transcribe the late flagellar genes, including those for adaptor and filament subunits (export substrates), as well as those responsible for chemotaxis and flagellar rotation. Class 3 promoters are under the control of the alternative sigma factor FliA (σ28) and its antagonist, the anti-sigma factor FlgM.

In H. pylori, a number of flagellar genes are under the transcriptional control of a σ54 promoter (Spohn and Scarlato, 1999; Josenhans and Suerbaum, 2000), or co-regulated with housekeeping genes. A flagella-specific sigma factor (σ28 or FliA) is present, but has only been shown to control the major flagellin gene flaA (Leying et al., 1992; Haas et al., 1993), and the fliDST operon that includes the flagellar genes fliS and fliD (Kim et al., 1999). FliS is a putative intrabacterial chaperone for both FlaA and FlaB flagellin subunits (Bennett and Hughes, 2000; Rain et al., 2001; http://pim.hybrigenics. com/pimriderlobby/current/PimRiderLobby.htm.) FliD, by homology to the well characterized flagellar system in Salmonella, is the capping protein of the flagellar filament (Yonekura et al., 2000; 2001).

In the cytoplasm of Salmonella, FlgM binds to the flagellar sigma factor FliA (σ28), and interferes with the interaction of FliA with the core subunits of RNA polymerase and the sigma binding site of the specific promoter regions (Burgess and Anthony, 2001; Chadsey and Hughes, 2001). Upon completion of the flagellar hook-basal body structure, FlgM is an export substrate and is secreted through the axial channel of the flagellar type III secretion apparatus. This relieves the inhibition of FliA-dependent genes (called class 3 flagellar genes in Salmonella; Chilcott and Hughes, 2000; Karlinsey et al., 2000a). Motility is crucial for the bacterium, and regulation of flagellar gene expression is closely linked to global regulatory networks, including regulation of cell division and pathogenicity (Pruss and Matsumura, 1997; Aizawa and Kubori, 1998; Komoriya et al., 1999; Eichelberg and Galan, 2000).

Analysis of the H. pylori genome sequences with a pattern-based, bioinformatic approach has permitted us to identify a candidate FlgM protein. This prompted us to study the role of the FlgM homologue and its counterpart, FliA, during late flagellar biogenesis in H. pylori. The effects of mutations in flgM and fliA on the H. pylori transcriptome were investigated using whole genome DNA microarrays. These data demonstrate that FlgM and FliA play a central role in the regulation of flagellar biogenesis in H. pylori. Several previously undescribed genes were also found to be under the control of FliA and FlgM.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Identification of the H. pylori flgM anti-sigma factor gene

A previously unannotated ORF was identified as a putative FlgM protein in the whole genome sequences of H. pylori 26695 (Tomb et al., 1997) and J99 (Alm et al., 1999) by a bioinformatical approach (see Experimental procedures). The candidate FlgM proteins (HP1122 and JHP1051 in the 26695 and J99 databases; 94% identity) were 76- and 67-amino-acid proteins that showed significant homology with Salmonella and B. subtilis FlgM, and differed from each other in their amino terminus (Fig. 1). A well conserved σ54 consensus promoter sequence (TTGGTA-N6-TGCAA) was situated 60 nucleotides (nt) upstream of the putative start codon of HP1122 and 87 nt upstream of JHP1051. In addition, a σ28 consensus sequence was located 13 nucleotides upstream of the start of HP1122, and 40 nucleotides upstream of the putative start of JHP1051. The spacer region of these σ28 promoters had a regular length of 15 nucleotides and contained a poly T repeat. A putative AGGA ribosome binding site is located 7 nucleotides upstream of the second ATG in HP1122 and of the first ATG of JHP1051. In comparison to the S. typhimurium FlgM protein, the translated H. pylori ORF lacks the N-terminal 20 (HP1122) and 29 (JHP1051) amino acids (Fig. 1). However, parts of the second region (IIa in Salmonella; Chilcott and Hughes, 1998) involved in type III secretion (amino acids 21–43 in Fig. 1) and the σ28 binding domain (amino acids 44–95 in Fig. 1) are both present. The functional amino acid homology to other FlgM proteins was highest in the region between amino acids 66 and 95, which constitutes a large part of the FliA binding-domain in Salmonella (Daughdrill et al., 1997; Chilcott and Hughes, 1998). To determine if the differences between HP1122 and JHP1051, in particular the length differences in the N-terminal domain, are also present in other H. pylori isolates, the flgM nucleotide sequences of 10 H. pylori strains from different geographical regions were analysed and found to segregate into two major groups. The FlgM homologues of three out of four strains of African origin were found to possess a shortened N-terminus (JHP1051-type), whereas all other strains were similar to the 26695 HP1122 sequence (Fig. 1). The putative σ54 and σ28 promoter regions, as well as the ribosome binding site upstream of the second ATG codon, were highly conserved between all strains.

image

Figure 1. Multiple alignment of FlgM amino acid sequences. Six representative FlgM sequences from diverse bacterial species that were used to identify the H. pylori FlgM homologue are depicted on top of the alignment (names shaded in yellow). The identified FlgM sequences from the H. pylori genomes of 26695 (HP1122) and J99 (JHP1051) are shaded in green. They are 76(26695)- and 67(J99)-amino-acid proteins (calculated molecular mass of 8.6 kDa, pI = 8.96, and 7.5 kDa, pI = 6.06) and show 94% identity. FlgM of H. pylori 26695 shares 19.1%/19.4% amino acid identity and 32.4%/31.94% amino acid similarity with S. typhimurium and B. subtilis FlgM proteins respectively. The HP1122 FlgM carries an additional nine amino acids at the N-terminus compared with JHP1051. Functionally conservative amino acid exchanges between HP1022 FlgM and JHP1051 FlgM were found at four positions throughout the protein (38, 43, 73, 93), the latter two of which are within the putative FliA-binding domain. The lower part of the alignment (shaded in orange) contains the translated amino acid sequences of 10 different isolates of H. pylori from different geographical regions. Two types of strains were identified. Three out of four African strains (CC29c, CC28c, CC48a) and J99 lack the first methionine residue and differ in their N-terminus from the other group of strains, which have been isolated in the United States, Asia, and Europe. Their putative N-terminal methionine is marked (yellow letters on red background). Functional homologies of amino acids in the different proteins are shown by matching colours. The shading of the alignment was created using the program GENEDOC (http://www.psc.edu/biomed/genedoc). The respective nucleotides sequences of the newly sequenced flgM genes of different H. pylori strains are accessible in the databases as AJ421711 to AJ421720.

Download figure to PowerPoint

Construction and characterization of H. pylori flgM and fliA mutants

Isogenic flgM and fliA mutants of H. pylori N6 were constructed by disrupting each gene with a kanamycin resistance cassette (aphA3′-III). These mutants were characterized phenotypically. To determine the effects of FlgM and FliA on the transcription of flaA, Northern blots were performed for mutant and the wild-type strains in the early exponential growth phase. The flaA transcript was not detected by Northern analysis in the fliA mutant strain. Quantitation of the Northern blots revealed that transcription of the σ28-dependent flagellin gene flaA was approxi-mately fourfold higher in the flgM mutant as compared with the wild-type strain. This is consistent with the ex-pected function of FlgM as an antagonist of FliA (Fig. 2).

image

Figure 2. Northern blots of whole RNA preparations of H. pylori N6 wild-type strains and fliA and flgM mutants respectively. The Northern blots were hybridized with DIG-labelled probes specific for H. pylori flaA and 16S rRNA. In each lane, 2 μg of whole RNA was loaded. In the fliA mutant, no flaA transcript is detected. flaA message in the flgM mutant is increased approximately fourfold in comparison with the wild type.

Download figure to PowerPoint

The effect of a flgM mutation on FlaA production was also measured in a H. pylori strain that carries a translational fusion of FlaA to the green fluorescent protein (GFP) in the chromosome (strains H. pylori N6138 (flgM+) and H. pylori N6138 flgM). The fluorescence of 50 individual bacteria of both wild-type and flgM mutant strains were scanned with a highly sensitive CCD camera in three independent experiments and quantitated (see Experimental procedures). Consistent with the Northern blot results, the flgM mutant showed a fourfold overexpression of the FlaA–GFP fusion protein in comparison with the flgM+ strain (283 ± 36 relative units of mean fluorescence intensity for the flgM mutant versus 74 ± 18 units for the flgM+ strain; P < 0.00001).

The morphology of the flgM mutants, as determined by electron microscopy, was slightly different from the wild-type strain, which has a unipolar bundle of four or five flagella (Fig. 3A). Bipolar flagella (a regular number of four or five at one pole, and one or two shorter ones at the other pole) were more frequently observed in the flgM mutants and, in some instances, one flagellum originated from the axial plane of the bacteria. The flgM-mutant bacteria possessed slightly more flagella than the wild-type strain (4.7 ± 1.6, flgM-mutant versus 3.9 ± 0.94, wild-type; 50 bacteria counted) but this difference is not statistically significant. The lengths of individual flagella were not significantly different between the flgM mutant strains and the wild-type (2.3 ± 0.88 μm, flgM-mutant versus 2.66 ± 1.16 μm, wild type; flagella of 50 bacteria measured). Western blot analysis of sheared flagella and cell extracts (not shown) from sheared (flagella-less) whole cells, developed with an anti-H. pylori FlaA antiserum, exhibited similar amounts of FlaA flagellin in both wild-type and flgM mutant strains (Fig. 3C).

image

Figure 3. Phenotypical characterization of H. pylori fliA and flgM mutants.

A. Electron micrograph of flgM mutant. Note the unusual arrangement of 2 and 5 flagella, respectively, at both poles of the bacterium, as well as one flagellum in the middle (white arrows).

B. Electron micrograph of fliA mutant with truncated flagella. For electron microscopy, bacteria were collected on formvar/carbon-coated copper grids directly from plates grown for 24 h. They were negatively stained using 1% phosphotungstate (pH = 7.0). Bars equal 0.5 μm.

C. Western immunoblot of sheared flagellar preparations of H. pylori N6 wild-type strain (HPN6) as well as fliA and flgM mutants, developed with antiserum AK179, raised against native purified H. pylori flagella. In each lane, 0.5 μg of protein was loaded. The wild-type and flgM mutant strains express both flagellin subunits, FlaA and FlaB, whereas the fliA mutant is defective in FlaA production.

Download figure to PowerPoint

In contrast to the flgM mutants, H. pylori fliA mutants possessed quite distinct truncated flagella, morphologically similar to those synthesized by mutants defective in the major flagellin gene flaA (Fig. 3B). FlaA protein was not detected in Western blot analysis of fliA mutants (Fig. 3C). FlaB and the hook subunit protein FlgE were detected in the fliA- and flgM-mutant strains in amounts similar to the wild type. In the fliA-mutant, a flaA-specific transcript was not detected by Northern blot analysis. No major differences in morphology or in protein profiles were observed between fliA and flaA mutants (flaA data not shown ; see Suerbaum et al., 1993).

H. pylori FlgM and FliA can functionally complement flgM and fliA defective mutants of S. typhimurium

As the H. pylori FlgM proteins lacked the N-terminal amino acid sequences of enterobacterial FlgM, and showed only weak amino acid identity with well characterized FlgM proteins, we asked the question whether the H. pylori FlgM and FliA proteins were functional in Salmonella. The H. pylori flgM and fliA genes were cloned into the enterobacterial expression vector pTrc99A, under the control of an inducible promoter. A C-terminal FLAG-tag was added to H. pylori FlgM to permit detection of the overexpressed protein with antibodies recognizing the FLAG epitope. The tagged H. pylori FlgM was expressed in a S. typhimurium strain lacking the native flgM gene (TH5158). Using anti-FLAG monoclonal antibody M2, H. pylori FlgM was clearly visible as a ≈7 kDa band in Western immunoblots of whole cell lysates after induction with 0.1 mM or 0.4 mM IPTG (Fig. 4A). Having demonstrated that Salmonella can express H. pylori FlgM, we tested for the ability of the tagged H. pylori FlgM protein to complement a S. typhimurium strain lacking FlgM. The S. typhimurium strains used for the complementation experiment contain a motA::MudJ–lacZ transposon fusion (Casadaban and Cohen, 1979; Hughes and Roth, 1984) that allows for the expression of β-galactosidase (LacZ) as a reporter protein under the control of the FliA/FlgM dependent class 3 motAB promoter. We used two different S. typhimurium motA::MudJ strains, one of which is export-competent (TH5157), and one of which is not (as a result of a deletion of basal body genes, strain TH5158). If H. pylori FlgM can interact with S. typhimurium FliA, it would inhibit FliA-dependent, motA-driven LacZ ex-pression at all time-points in the export-deficient S. typhimurium strain. In the export-competent strain, we expected the inhibitory effect of H. pylori FlgM on motA–lacZ transcription to be relieved in the case of export, if the intracellular concentration of FlgM was below the saturation limit of FliA binding.

image

Figure 4. Expression of H. pylori FlgM in S. typhimurium, and complementation of S. typhimurium flgM mutants by H. pylori FlgM.

A. Western blot of S. typhimurium strain TH5158 (export-deficient) whole cell lysates containing the H. pylori FlgM expression plasmid pCJ106. As a control strain, TH5158 transformed with plasmid vector pTrc99A alone was used. Cells were grown in LB medium without IPTG (n.i.), with 0.1 mM IPTG, and with 0.4 mM IPTG for induction of FlgM production. 1 μg protein of whole cell lysates was loaded per lane. The immunoblot was developed using anti-FLAG-tag M2 monoclonal antibody at a dilution of 1:5000.

B. Results of β-galactosidase assays of S. typhimurium strain TH5158 (strain-deficient in the flagellar export apparatus), transformed with control plasmid pTrc99A (shaded columns) or H. pylori FlgM expression plasmid pCJ106 (white columns). Culture conditions and IPTG concentrations were as in the Western blot experiment (0, 0.1 and 0.4 mM IPTG was added, as indicated below the x-axis). All measurements were made in triplicate. Mean values and standard deviations in Miller units of two independent experiments are depicted on the y-axis.

C. Results of β-galactosidase assays of S. typhimurium strain TH5157 (flagellar apparatus export competent). Cells were grown without IPTG, and then induced with 0.1 mM IPTG. Mean values and standard deviations of two independent experiments are depicted in Miller units.

Download figure to PowerPoint

In the export-deficient strain TH5158, H. pylori FlgM decreased motA-dependent LacZ (β-galactosidase) expression by more than one third, suggesting that FlgM can interact with the heterologous Salmonella FliA protein and inhibit FliA-dependent transcription (Fig. 4B). In the absence of induction, only minimal H. pylori FlgM expression could be detected in Western blots, and lacZ reporter gene expression did not change significantly compared with the level of the control. In the export-competent S. typhimurium strain TH5157, the heterologous FlgM led to a significant reduction of β-galactosidase activity, even in the absence of IPTG. This result might indicate that H. pylori FlgM cannot be exported efficiently in S. typhimurium.

The FliA complementation was carried out in S. typhimurium strain TH4100, a fliA transposon mutant that has an additional fliC::MudJ fusion (lacZ reporter gene under the control of the class 3 fliC flagellin promoter; Gillen and Hughes, 1991). Helicobacter pylori FliA expressed from an inducible plasmid in TH4100 led to strong LacZ activity, which could be easily visualized as growth of red colonies on MacConkey lactose agar, whereas the control strain only formed white colonies (Fig. 5A). Quantitative assays of LacZ activity showed that, even in the absence of IPTG, the heterologous FliA protein fully induced LacZ expression from the fliC promoter (Fig. 5B).

image

Figure 5. Complementation of a fliA-deficient mutant of S. typhimurium (strain TH4100) with H. pylori FliA expressed from plasmid pCJ109.

A. Two independent clones (pCJ109-1 and -2) of TH4100(pCJ109) were streaked on McConkey lactose indicator plates and appear with a red (Lac+) colony phenotype, indicating that fliC-promoter-driven β-galactosidase expression is induced by the heterologous FliA, whereas two control clones of TH4100(pTrc99A) grow as white (Lac) colonies (pTrc99A-1 and -2).

B. Quantitative β-galactosidase measurements in S. typhimurium TH4100 complemented with H. pylori FliA (pCJ109). Cultures were made in either the absence (0) or in the presence of IPTG (0.1 and 0.4 mM, as indicated below the x-axis. Shaded columns and white columns represent the control strain TH4100(pTrc99A) and the complementation strain TH4100(pCJ109) respectively.

Download figure to PowerPoint

Transcriptome analysis of H. pylori fliA and flgM mutants with complete genome DNA microarrays

Helicobacter pylori genes previously known to have a σ28 promoter were the major flagellin gene flaA, and an operon (HP0751 to HP0754) containing the flagellar genes fliS and fliD. FliS and FliD are late flagellar genes in Salmonella and are under the control of class 3 promoters. We chose a microarray-based comprehensive transcript approach to further analyse the function of the dual regulatory system of FliA and FlgM, and to identify additional genes controlled by these regulators. Oligonucleotide microarrays (see Experimental procedures) were competitively hybridized with two probes labelled with different fluorescent dyes (Cy5 and Cy3). Probes were cDNAs generated from whole mRNA preparations of the H. pylori N6 wild-type strain and isogenic fliA and flgM mutant strains. All strains were grown in liquid culture and harvested at the early exponential growth phase (OD600 of ~1.0). Eight microarray slides were competitively hybridized with one wild-type probe and one mutant probe. Hybridization for every mutant was carried out four times : two times with the same dye combination and different biological samples (harvested from independent cultures), and twice with inverse dye labelling. These experiments were performed and evaluated as indicated in Experimental procedures.

Consistent with the Northern blot results, flaA transcription was significantly reduced in the fliA mutant and increased in the flgM mutant, confirming the antagonistic roles of FliA and FlgM in the control of flaA transcription (Fig. 6). Two additional operons showed a similar FliA/FlgM-dependent regulation pattern (see below). The expression of the HP0751-HP0754 operon was not significantly changed, probably as a result of the activity of a consensus σ54 promoter (TTAGAA-N4-TCGCA) located upstream of the σ28 promoter of this operon. A table with all gene transcript ratios is available as supplementary material (see Supplementary material).

image

Figure 6. Results of microarray transcriptome analyses of H. pylori fliA and flgM mutants. Microarrays were competitively hybridized with the wild-type strain and the mutants. Ratios were calculated from normalized mean pixel intensities of scans at both wavelengths (Cy3 and Cy5), corresponding to the mutant strain and control hybridization signals (see Experimental procedures). The results of four slides are included in each graph. Columns represent mean values and standard deviations of the ratios obtained in all four hybridizations. Genes found downregulated below a threshold of 0.5 in the fliA mutant (A) and upregulated > twofold in the flgM mutants (B) (grey columns) were compared with three control groups of genes without transcript changes. The first control group (white columns) were genes involved in tryptophan biosynthesis (housekeeping genes). The second control group (hatched columns) were σ54-dependent genes (class 2) of the flagellar biosynthesis pathway in H. pylori. The third control group (wave pattern columns) were genes involved in chemotaxis and motility that are controlled by σ28 in Salmonella (class 3) but not in H. pylori. In the fliA mutant experiments (A), flgM (HP1122) is included in the graph (black column). The flgM gene is most likely dependent on two promoters, σ54 and σ28, for its transcription.

Download figure to PowerPoint

Two novel operons, directly dependent on FliA and FlgM, identified by microarray-based transcriptional analyses

In addition to flaA, two novel operons were found to be under the control of the FliA/FlgM system (Fig. 6). Both operons are present in the whole genomes of the H. pylori strains 26695 and J99. These operons were downregulated in the fliA-mutant (mean signal ratio of the competi-tive hybridization fliA-mutant versus wild type < 0.5) and upregulated (>twofold higher transcription) in the flgM-mutant compared with the wild-type strain. One of the two operons is monocistronic and contains the gene HP0472 (JHP424). HP0472 (OMP11) belongs to the large paralogous family of H. pylori outer membrane proteins (OMPs) and is specific for H. pylori (it has no other homologues in the databases). The second novel FliA/FlgM-dependent operon contains two genes, HP1051 and HP1052 (JHP373/374). HP1052 (JHP373), the first gene in the operon is predicted to encode an enzyme involved in lipopolysaccharide (lipid A) synthesis (UDP-acyl-N-acetylglycosamine-deacetylase; homologous to envA or lpxC genes of other bacteria). This gene is essential for growth and involved in cell division in E. coli, but its function has not been studied in H. pylori (Normark, 1970). HP1051 (JHP374) encodes a hypothetical protein with weak homologies to a glycopeptide-endopeptidase of B. subtilis. When we analysed the genomic regions upstream of HP0472 (JHP424) and HP1051/1052 (JHP373/374), well conserved σ28 consensus promoters with regular spacer length of 15 nucleotides between their –10 and –35 boxes could be identified (Fig. 7).

image

Figure 7. Alignment of promoter sequences of the operons in H. pylori is identified to be under the control of the FliA/FlgM late flagellar regulators. The E. coliσ28 consensus is given for comparison. In addition to the already known operons, flaA (HP0601) and operon HP0751 to HP0754 (including fliS and fliD genes), two novel operons preceded by well conserved σ28 promoters, were identified by transcriptome microarray analyses to be under the direct control of the FliA/FlgM regulon (HP0472, HP1051 and 1052). flgM itself has a well-conserved putative σ28 promoter sequence in addition to its σ54 promoter. The –10 boxes of the σ28 promoters are localized at a distance between 21 and 87 nucleotides upstream of the respective start codon. Nucleotides conserved throughout the different sequences are shaded in black ; nucleotides conserved in at least three sequences are shaded in grey. The alignment was created using GENEDOC software.

Download figure to PowerPoint

The H. pylori flgM gene was also found to possess a conserved σ28 promoter in addition to its σ54 promoter (Fig. 7). Even if there was no clear downregulation of flgM transcription in the fliA mutant array hybridizations, flgM belonged to the class of genes in the lower range of ratio values (0.6 transcript ratio between fliA mutant and wild-type strain; Fig. 6A). Transcription from the σ54 promoter of flgM might account for residual transcription of this gene in the fliA mutant.

In addition to genes directly under the control of FliA and FlgM, one gene, HP0602, was identified that was strongly upregulated in the flgM mutant but not significantly downregulated in the fliA mutant. This gene, predicted to encode a base-excision repair glycosylase (3-methyladenine DNA glycosylase) (O´Rourke et al., 2000), is located directly downstream from the flaA gene (HP0601) in the same transcriptional orientation, but is apparently not in the same operon with flaA. It does not possess a σ28 consensus promoter. flaA has a downstream, rho-independent transcription termination signal (Leying et al., 1992). It is not clear why transcription of the downstream gene HP0602 should be so strongly influenced in the flgM mutant and what the biological sense of such an influence of flaA over HP0602 could be.

The transcription of many genes involved in late flagellar biogenesis and motility is unaffected in H. pylori fliA and flgM mutants

We analysed two control sets of genes from the micro-array hybridization experiments to perform statistical comparisons with putative FliA-dependent genes. The first control set was a randomly chosen group of ‘housekeeping’ genes of the tryptophan metabolism (trpS, trpA to trpE; Fig. 6); the second set were H. pylori‘class 2’ flagellar genes that are under the control of the σ54 promoter/NtrC-like activator system. Transcription of the tryptophan metabolism genes did not change in the fliA or flgM mutant when compared with the wild-type. None of the known σ54-dependent genes of H. pylori (e.g. flaB, flgE, flgBC, flgG1, flgL, flgK; see Fig. 6) was deregulated in either the fliA or the flgM mutants, indicating that no transcriptional feedback regulation takes place on this level. In addition, a third group of genes was analysed, which consisted of late motility and flagellar structural genes that are under the control of class 3 (σ28) promoters in Salmonella, such as the motAB operon, che genes and chemotaxis receptor genes of the tlp family (only four homologues present in H. pylori). Genes of this group did not display any change in transcriptional activity in H. pylori fliA or flgM mutants (Fig. 6). Apparently their transcription is not related to the FliA/FlgM system in H. pylori.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

H. pylori possesses an unusual FlgM flagellar anti-sigma factor

In silico analysis has permitted us to identify a H. pylori flgM gene that had not been found in the annotations of the whole genome sequences. The predicted H. pylori FlgM proteins of the two sequenced genomes differ in length, and sequence analyses of 10 independent clinical isolates, from six different geographical regions, have confirmed the existence of a shorter and a longer type of FlgM. However, the N-terminus of mature FlgM has not been experimentally determined in any strain, and both types of strains may produce proteins starting with the same amino acid, or in JHP1051-type FlgM, an alter-native start codon (TTG; for leucine) may be used. Irrespective of the exact start of the proteins, both types lack the conserved N-terminal domain of other bacterial FlgM proteins. The function of this N-terminal domain is not known. Amino acids at the N-terminus (amino acids 3–11, notably positions 3 and 9) have been suggested to be involved in FlgM export in Salmonella, but they are not absolutely required for export (Chilcott and Hughes, 1998). The current hypothesis is that the N-terminal domain of enterobacterial FlgM is involved in recognition by the flagellar secretion apparatus, and that the 5′-region of the flgM transcript is involved in FlgN-dependent translation coupled to export (Hueck, 1998; Karlinsey et al., 2000a). A gene homologous to the secretion chaperone gene flgN seems to be absent in H. pylori, which is also noteworthy as FlgN is a chaperone for the late flagellar proteins FlgL and FlgK (hook-filament adaptor proteins) that are both present in H. pylori (Fraser et al., 1999). As H. pylori apparently lacks a FlgM-specific chaperone, the N-terminal domain would not be required for secretion chaperone-dependent export. There is ample evidence that the motility regulatory system and the ordering of flagellar export substrates of H. pylori is organized differently from Salmonella. For example, the FliK protein, which, together with FlhB and other proteins, mediates substrate specificity switching of the export apparatus (Minamino and Macnab, 2000a; Makishima et al., 2001), is absent, and there is no feedback inhibition of transcription of the flaA flagellin gene in H. pylori in the absence of a completed hook (O´Toole et al., 1994). The coupling of flagellar assembly to gene expression in H. pylori and the substrate specificity switching are probably achieved by a mechanism distinct from the enteric system.

Our identification of HP1122 as H. pylori FlgM is in full agreement with results published recently by Rain and co-workers, who have constructed a protein interaction map of H. pylori, based on the results of yeast two-hybrid experiments. This interaction map shows only one single but very strong interaction of HP1122 (FlgM) with the C-terminal region of FliA (P < 1e-500) (Colland et al., 2001; Rain et al., 2001; http://pim.hybrigenics.com/pimriderlobby/current/PimRiderLobby.htm). The biological assays in Salmonella support the hypothesis that the two H. pylori proteins are a high-affinity pair of sigma and anti-sigma factors as in Salmonella.

Helicobacter pylori flgM is not organized in an operon with other flagellar-related genes. Well conserved σ54- and σ28-like promoters upstream of flgM, and the results of our microarray analyses, indicate that H. pylori flgM might be controlled on two levels ; by class 2 and class 3 promoters. This is similar to Salmonella, in which flgM is also transcribed from two promoters, belonging to class 2 (σ70/FlhDC) and class 3 (σ28) of the flagellar hierarchy (Gillen and Hughes, 1993). The putative σ28 promoter sequence of flgM is most conserved in the –10 region. The spacer region contains a poly T repeat, which might allow changes in spacer length by slipped-strand mispairing mutagenesis and, thereby, serve to modulate late flgM transcription. In Salmonella, FliA-dependent transcription is repressed, when FlgM expression increases before hook-basal body completion. Salmonella class 3 σ28 promoters only become activated in a late growth phase-dependent fashion when FlgM can be exported from the cell via the completed axial flagellar channel consisting of the inner export apparatus, rod and hook (Gillen and Hughes, 1993; Chilcott and Hughes, 2000; Karlinsey et al., 2000b; Kalir et al., 2001). At this point, the σ28-dependent flgM transcript is made, translated with the help of FlgN, and directly secreted. The possible dependence of H. pylori flgM on σ28-driven transcription may indicate that a similar mechanism of translation, coupled to export and feedback regulation of FlgM and other late flagellar genes, exists in H. pylori.

Function of the antagonistic regulators FliA and FlgM of H. pylori

FlgM of H. pylori has been found to be different from previously known FlgM proteins, whereas H. pylori FliA is very similar to other σ28 transcriptional activators. Analyses of H. pylori flgM and fliA mutants have confirmed that the role of these regulators in the transcription of flaA flagellin corresponds to the paradigm Salmonella system. However, several questions remain regarding the physiological role of H. pylori FlgM. Salmonella FlgM mutants have been described by one group to produce more flagella than the wild type (Kutsukake and Iino, 1994). We did not find a significant difference in flagellar length or number between H. pylori flgM mutant and wild type in continuously growing cultures. As flagellar biogenesis seems to be the net result of a complex interplay of temporally regulated functions during the cell cycle, and FlgM is only a negative regulator of FliA in early flagellar biogenesis, many more parameters, e.g. efficiency of transcription, translation and secretion, and the complete organization of the cell cycle in a bacterium lacking the flagellar master regulators FlhCD, have to be taken into account to understand these apparent differences.

In Salmonella, the ability of FlgM to be secreted through the flagellar central channel is paramount to its function in coupling transcriptional regulation of late genes to flagellar assembly (see above). In bacteria with a flagellar sheath such as H. pylori, the export of secreted proteins via the flagellar central channel represents a challenge that would have to be solved differently from the flagellar system of enteric bacteria. Cross-complementation assays suggest that H. pylori FlgM is not secreted in Salmonella. There are two possible explanations for this observation. First, H. pylori FlgM cannot be secreted in the heterologous system because the components responsible for proper targetting and secretion are incompatible between the two bacteria (e.g. because of the truncated N-terminus of H. pylori FlgM). Second, H. pylori FlgM might not be a secreted protein. If FlgM is not secreted, it might accumulate in the bacterial cytoplasm bound to other molecules, undergo recycling or be proteolytically degraded. These questions need to be addressed in the future.

H. pylori FlgM and FliA are active in S. typhimurium

The cross-complementation experiments demonstrate that H. pylori FlgM, despite its shorter N-terminus and low degree of sequence similarity with enterobacterial FlgM proteins, is capable of binding Salmonella FliA and inhibiting FliA-dependent transcription in the Salmonella flagellar type III secretion system. Helicobacter pylori FliA was also active in Salmonella. However, the complementation in the case of H. pylori FlgM was not complete. We hypothesize that the partial complementation phenotype is linked to the mode and affinity of FlgM–FliA interaction and to FlgM secretion, which will be clarified in further studies.

This work is the first demonstration of functional activity of H. pylori regulatory proteins in Salmonella. The success of these complementation experiments raises the attractive possibility of performing functional studies of H. pylori proteins using the tractable and well characterized Salmonella system.

Detection of H. pylori genes under FliA/FlgM control by global transcriptome analysis with DNA microarrays

The paradigm σ28 promoter of H. pylori is the flaA promoter, which is transcribed in a growth phase-dependent manner (Leying et al., 1992 ; Haas et al., 1993; own unpublished experiments). This promoter has an unusually short spacer region between the –10 and –35 boxes (13 instead of 15 nucleotides; Fig. 7). It has been suggested that this might make the promoter particularly sensitive to changes of DNA topology. A second operon, also under the control of σ28, consists of the four genes, HP0751 to HP0754 (Kim et al., 1999). The promoter of this operon, which includes the flagellar genes fliS and fliD, shows well conserved σ28 consensus –10 and –35 boxes and has a regular spacer length of 15 nucleotides (Fig. 7). Both known σ28 promoters had been confirmed to be functional by primer extension analysis (Leying et al., 1992; Kim et al., 1999). Previously, it was unclear whether other genes are under the control of the FliA–FlgM regulatory system. However, it seemed unlikely that in a bacterium with only three sigma factors one of them should control only two operons.

DNA microarray analysis is an extremely powerful tool for the study of complex regulatory events and networks (Richmond et al., 1999; Zimmer et al., 2000). In this study, we have used DNA microarrays to identify genes under the control of the flagellar sigma factor FliA and its antagonist FlgM. This is the first study that exploits DNA microarray hybridization to characterize complex transcriptional effects in H. pylori. The results have confirmed that flaA is a late flagellar gene that is exclusively controlled by FliA/FlgM. Additional operons regulated by FliA/FlgM were identified (see below). It is noteworthy that a number of genes known to be class 3 in Salmonella were transcribed independently of FliA in H. pylori. Exclusive dependence of the previously described operon HP0751 to HP0574 on σ28 could not be confirmed by this work (see Results).

Two novel operons under the control of FliA/FlgM

Using the microarray system, we identified two novel operons under the control of FliA/FlgM. The FliA/FlgM-dependent regulation of both these operons has also been confirmed by Northern blot analysis (data not shown). The first operon contains a single gene, HP0472, which is annotated in the complete genomes as an outer membrane protein (OMP11). This OMP belongs to the large family of 32 OMP paralogues in the H. pylori genome. Genes homologous to HP0472/OMP11 have not been found in other bacteria. Some of the H. pylori OMPs have been found to function as porins or in adhesion (porins HopABCE, BabAB) (Ilver et al., 1998; Odenbreit et al., 1999; Alm et al., 2000), but the role of HP0472 has not been investigated yet. Transcription of other genes coding for the large H. pylori protein family of OMP paralogues (all 32 genes) was not affected in H. pylori fliA or flgM mutants. The H. pylori flagellar filament is enveloped by a membraneous flagellar sheath that is an extension of the bacterial outer membrane possessing distinct biochemical properties (Geis et al., 1993). The hypothesis that OMP11 is a protein specific for the membrane flagellar sheath is presently under investigation in our laboratory.

The second fliA-dependent operon consists of two genes, HP1052 and HP1051. The second ORF of this operon (1051) encodes a hypothetical protein with weak homology to a glycopeptide-endopeptidase of B. subtilis. The first ORF of the operon encodes the H. pylori homologue of EnvA/LpxC, an UDP-acyl-N-acetylglucosamine-deacetylase. In E. coli, EnvA is an essential protein that catalyses the second step of LipidA biosynthesis (Wolf-Watz and Normark, 1976; Young et al., 1995). In E. coli and other bacteria (Neisseria, Pseudomonas), envA lpxC is located in a gene cluster with ftsA and ftsZ, as well as other genes involved in cell division (Lutkenhaus and Wu, 1980; Beall and Lutkenhaus, 1987; Francis et al., 2000). Previously, EnvA/LpxC have not been shown to be under flagellar regulatory control. It appears that in H. pylori these two genes, HP1051/HP1052, are under the control of a σ28-dependent promoter and an additional promoter, which ensures high expression concomitant with cell division. Upstream of the operon is a stretch of five genes coding for unknown proteins. The distance between the last of these upstream ORFs and HP1052 is very small (the last codon of HP1053 and the start codon of HP1052 overlap by 1 nucleotide). The putative σ28 promoter lies within the 3′-end of the upstream ORF, HP1053. In the closely related bacterial genus Campylobacter (see Campylobacter jejuni genome sequence, http://www.sanger.ac.uk/Projects/C_jejuni/), the envA gene is also located together with a second gene that is highly homologous to HP1051, and is not associated with cell division genes. However, the cluster of putative ORFs, localized upstream of HP1052/1051 in H. pylori is not present in C. jejuni. Further characterization of the promoters and targeted promoter mutagenesis will elucidate the regulation of the HP1052 operon. Mutational analyses will clarify whether the HP1052 and 1051 genes are essential for H. pylori growth, and if they have a function in the biogenesis or elongation of the flagellar sheath. Figure 8 shows a refined model of the complex regula-tory interactions within the H. pylori flagellar biogenesis system, based on the currently available data. Helicobacter pylori flagellar regulation can be ordered clearly into three hierarchical classes, the latter two being dependent on σ54 and σ28, similar to the system in Vibrio cholerae (Prouty et al., 2001). The family of flagellar class 3 genes in H. pylori has been expanded (see Fig. 8) and still keeps some functional surprises in store.

image

Figure 8. Model of regulation pathways in flagellar biosynthesis of H. pylori. Apparently, three classes of genes exist in the flagellar pathway of H. pylori. There is no flhCD master operon. The transcription activation pathway of class 1 genes is not known. Class 1 genes include genes of the basal body and export apparatus, as well as genes invoved in chemotaxis and flagellar rotation. Among the class 1 genes is also the response regulator gene flgR, encoding a NtrC homologue that forms a two-component system together with the sensor kinase encoded by gene HP0244 (Spohn and Scarlato, 1999; Beier and Frank, 2000). FlgR is an activator of class 2 genes, which in turn are controlled by σ54 in H. pylori. Class 2 genes comprise the gene coding for the anti-sigma factor FlgM and probably include the gene for the late flagellar sigma factor σ28. The class 3 family of late flagellar genes, which now has at least five members, are under the dual control of σ28 and FlgM. Some genes have both σ28 and σ54 promoters (class 2 and class 3). Known genes of the flagellar pathway that have not been assigned to any of the three classes so far are not included in the model.

Download figure to PowerPoint

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Bacterial strains and growth media

Strains H. pylori N6 and 26695 were used in the experiments. A total of 10 H. pylori strains from different geographical regions (names in Fig. 1) were used for the comparative flgM sequence analysis (for strain descriptions see: Suerbaum et al., 1998; Josenhans et al., 2000). Bacteria were grown on blood agar plates (blood agar base 2, Oxoid) containing 10% horse blood, vancomycin (10 mg l–1), polymyxin B (2500 U l–1), trimethoprim (5 mg l–1) and amphotericin B (4 mg l–1). Plates were incubated at 37°C under microaerobic conditions. Selective antibiotics (kanamycin, 20 μg ml–1 ; chloramphenicol 10 μg ml–1) were added as re-quired. Alternatively, H. pylori was grown in liquid culture in brain–heart infusion broth (Oxoid) supplemented with 10% heat-inactivated horse serum and antibiotics as required. Escherichia coli strains DH5α and MC1061 were used for the cloning experiments. They were grown in Luria–Bertani (LB) medium containing kanamycin (20 μg l–1) or ampicillin (50 μg l–1). Salmonella typhimurium strains (see Table 1) were grown in LB medium or in M 9 + CA minimal medium (Minamino and Macnab, 2000b), supplemented with selective antibiotics as required.

Table 1. S. typhimurium strains and plasmids used in this work.
Strain/plasmidVectorSize (kb)DescriptionReference
Strain
 TH4100   fliA5059::Tn10dTc; fliC5050::MudJ; fljBe,n,xvh2 Gillen and Hughes (1991)
 TH5157   motA5461::MudJ; flgM5777::Tn10dCmThis work
 TH5158   motA5461::MudJ; flgM5777::Tn10dCm; ΔflgHI958This work
Plasmid
 pILL600 5.8Source of km cassette Labigne-Roussel et al. (1988)
 pHel2 5.0cloning vector H. pylori Heuermann and Haas (1998)
 pILL570 5.3cloning vector Labigne et al. (1991)
 pSUS1405pILL5706.5 H. pylori 26695 fliAThis work
 pSUS1409pILL5707.9 H. pylori 26695 fliA and km cassette disruptionThis work
 pCJ100pUC183.9 H. pylori 26695 flgMThis work
 pCJ103pUC185.1 H. pylori 26695 flgM and km cassette disruptionThis work
 PCJ105pHel25.4 H. pylori 26695 flgM and short FLAG-tagThis work
 pCJ106pTrc99A4.4 H. pylori 26695 flgM and long FLAG-tagThis work
 pCJ109pTrc99A5.0 H. pylori 26695 fliAThis work

Cloning and DNA sequence determination

Cloning and PCR amplification were performed according to standard procedures. Direct DNA sequence determination of PCR products was performed as described previously (Suerbaum et al., 1998). The plasmids and oligonucleotides that were used are listed in Table 1 or indicated in the methods.

Construction of the H. pylori flgM mutant

A genomic fragment of H. pylori 26695, encompassing ORFs 1121–1123 (1226 bp), was amplified by polymerase chain reaction (PCR) using primers OLHPFlgM1 (AATagatctGAA GACAATCAAACCATTCAAGC) and OLHPFlgM2 (ATTagatct CAAGCAAGAAACATTCAGGC) and cloned into plasmid pUC18 (BglII–BamH1). The resulting recombinant plasmid pCJ100 was further amplified in an inverse PCR reaction using oligonucleotides OLHPFlgM3 (AATggatccCATGAGAC TTCTCACAAATTGG) and OLHPFlgM4 (ATTggatccCATTCC ATTACAGCAAGTATCG), creating a near complete deletion of the ORF HP1122. Only the last 45 nucleotides of the gene were retained (size of deleted fragment, 179 bp). This fragment was ligated (BamH1) with the aphA3′-III gene (kanamycin resistance cassette) of C. jejuni to replace the deleted fragment of flgM and serve as a selective marker. Natural transformation of the resulting suicide plasmid pCJ103 (transcriptional orientation of the km-cassette parallel to the flgM gene) into H. pylori was carried out as described previously (Haas et al., 1993). Recombinant clones in H. pylori were selected on kanamycin-containing plates and isolated. Four clones out of each experiment were genetically characterized by PCR with primers OLHPFlgM1 and FlgM2, as well as oligonucleotides binding to the aphA3′ gene. All clones proved to carry the genomic disruption of the flgM gene as a result of a double-crossover event.

Construction of the H. pylori fliA mutant

The fliA gene of H. pylori 26695 was amplified in a PCR, using primers OLHPFliA1 and OLHPFliA2 (TAAagatctC TAATAGTCTGTCAATCATAGG; TAAagatctCAGAGAAAAAA GAAGTGGC; amplified fragment of 1150 nucleotides, containing BglII restriction sites, underlined) and cloned into the plasmid vector pILL570 cut with BglII. An EcoRI site within the fliA gene was deleted by cutting and filling the 5′-overhanging ends using Klenow polymerase. The resulting plasmid after blunt-end religation, pSUS1405, was inversely amplified using primers OLHPFliA3 and OLHPFliA4 (AATga attcCATCTTGCTCAATCGCATTG; ATAgaattcGAAGCAGAA GAGTTTTAGAGC; EcoRI sites). Thereby, a 17 bp deletion in the fliA gene was created, which was closed by inserting the kanamycin resistance cassette (aphA3′-III) prepared from pILL600 using EcoRI sites. The resulting suicide plasmid pSUS1409, containing the cassette in the same transcriptional orientation as the fliA gene, was introduced into H. pylori by natural transformation. Four kanamycin-resistant mutant clones were isolated and characterized by PCR, protein profiles and electron microscopy.

Construction of plasmid for expression and detection of FlgM in H. pylori

The H. pylori flgM gene was PCR-amplified, including its own promoter region, using two primers (OLHPFlgM5, ATAggatcc GCAATCGTTTGGCTCTGTGG; and OLHPFlgM6, AAAgga tccGAAATCCTTGTAATCCATGCTTATCCCCAATAAATCC; BamH1 sites), the one at the 5′-end containing six codons (underlined) for a shortened FLAG-tag, fused to the C-t erminus of the protein. The fragment was cloned into pHel2 (BglII-cut). The plasmid was characterized by restriction mapping and sequence determination (pCJ105).

Plasmid clones for expression and detection of H. pylori FlgM and FliA in S. typhimurium

For expression of H. pylori FliA and FlgM, the H. pylori fliA gene was amplified from H. pylori 26695 complete DNA (primers OLHPFliA2, see above; and OLHPFliA3, AATagatct ACGACTTCTAAAAGCGCATC, BglII sites). The H. pylori flgM gene was amplified from plasmid pCJ105 with primers OLHPFlgM10 (AATggatccTCACTTATCGTCGTCATCCTTG TAATCCATGCT, C-terminally elongated FLAG-tag and an additional stop codon directly downstream of the FLAG-tag, BamH1 site) and OLHPFlgM11 (ATAggatccCTTGCTGTA ATGGAATGAATATC, BamH1 site). The own promoter re-gions of the H. pylori fliA and flgM genes were excluded. The purified PCR products were cloned into the enteric expression plasmid pTrc99A (Amersham Pharmacia Biotech) downstream of the trp-lac promoter. The resulting plasmids pCJ106 (HPflgM-FLAG) and pCJ109 (HPfliA) were characterized by restriction mapping and sequence determination.

SDS–PAGE and Western blotting

SDS–PAGE and Western blotting were performed as described (Leying et al., 1992).

β-Galactosidase (LacZ) assays in S. typhimurium

Strains for the assay were grown overnight, diluted 1:25 in fresh LB medium with antibiotics as required (initial OD600 of about 0.1). Growth was continued at 37°C until the culture reached an OD600 of about 0.4. Then, cultures were induced by adding IPTG to a final concentration of 0.1 and 0.4 mM. As a control, non-induced cultures were used. After induction, cultures were incubated for another 2 h at 37°C with aeration. Final OD600 were measured, the bacteria were harvested on ice and frozen as suspensions. They were kept at –80°C until further use. β-Galactosidase activity was measured according to the original method by Miller and colleagues (Miller et al., 1968), with slight modifications. Briefly, bacterial suspensions were mixed with Z-buffer, SDS, chloroform and β-mercaptoethanol (1:9). They were vigorously vortexed and lysed for 5 min, then ONPG substrate was added. After thorough mixing, each sample was incubated for 2 and 5 min at 30°C, stopped by adding 1 M Na2CO3, then enzyme kinetics were measured spectrophotometrically at 420 nm. All measurements were made in triplicate. Enzyme activities were calculated in Miller units.

Electron microscopy

Plate-grown bacteria were resuspended in PBS buffer and negatively stained on copper EM grids coated with formvar and carbon, using 1% phosphotungstate (pH = 7.0). EM grids were viewed in a Zeiss EM900 microscope at an acceleration voltage of 80 kV.

Fluorescence microscopy and quantification

Fluorescence microscopy was performed using a Olympus BX40 microscope with a mercury light source and a narrow band pass filter combination optimized for GFP (green fluorescent protein) detection (excitation 490 nm/emission 515 nm). For quantification of the fluorescence signal, the microscope was fitted with a CCD camera (Princeton Instruments, 1300 × 1024 pixels, pixel size 6.7 μm) and the MetaMorph® (Universal Imaging Corporation) optical data analysis package. Mean fluorescence intensities of 50 bacteria of three different experiments were individually measured for each strain (fluorescent strain and mutant). The mean values and standard deviation were calculated. The result displayed is the mean fluorescence intensity per pixel for the different bacterial strains.

Whole RNA preparation from H. pylori

RNA was isolated from liquid cultures of H. pylori. Bacteria were harvested, and spun down for 30 s at maximum speed in an Eppendorf centrifuge. Bacterial pellets were quick-frozen in liquid nitrogen and stored at –80°C until the RNA preparation was carried out. Before RNA isolation, bacteria (approximately 12 × 108) were resuspended in 850 μl of the cell lysis buffer of the Qiagen RNeasy kit (10 μl ml–1 of β-mercaptoethanol), mixed with glass beads in 2 ml FastRNA tubes-Blue (BIO 101) and lysed for 45 s at a speed of 6.5 units in a bead beater FP120 shaker (Savant). Further preparation was done according to the standard Qiagen RNeasy protocol. After column elution, RNA was treated with RNase-free DNase I (Roche) and, in case of use for microarray hybridizations, again purified with the Qiagen RNeasy column kit. RNA preparations were quality-controlled by a PCR reaction using DNA-based primers (OLHPFlaA4 and OLHPFlaAPV4, upstream of the flaA gene; sequences available on request) in a non-transcribed region. RNA was further quality-checked on agarose gels, and RNA amounts were spectrophotometrically quantified using a GeneQuant device (Bio-Rad).

Northern blotting

Whole RNA preparations (2 to 20 μg) were used to perform a denaturing agarose gel electrophoresis (0.24 M formaldehyde) for 2–3 h. Afterwards, the gel was treated with 50 mM NaOH and 0.1 M Tris-HCl (pH 8) for 5 min each and was vacuum-blotted on to positively charged nylon membranes (Roche) in a vacuum-blotting device (Keutz) for 1 h in 6x SSC. Hybridization and washing steps were performed according to the DIG chemiluminescence detection protocol (Roche). Dig-labelled probes were generated by PCR with primers specific for flaA (OLHPFlaA4 and OLHPFlaA9) and 16S rDNA (OL16S1 and OL16S2), with incorporation of DIG-labelled d-UTP nucleotides during the reaction. They were adjusted to 200 ng ml–1 of hybridization buffer. Hybridization signals were detected on chemiluminescence detection film (Hyperfilm® MP) for 1 and 5 min.

Oligonucleotide whole genome array of H. pylori

The PAN®H. pylori array (MWG Biotech) was used for the transcriptome analyses. It contains 1877 H. pylori-specific oligonucleotides spotted on epoxy-coated glass slides. A single oligonucleotide spot (50 mers) is representative for each gene. A total of 1545 oligonucleotides are derived from the complete genome of the strain H. pylori 26695 (Tomb et al., 1997), and 332 additional genes are specific for the strain J99 (Alm et al., 1999) and 100% of the genes of the known H. pylori genomes are represented on the array. The oligonucleotides for the microarray were designed as follows. Oligonucleotides were preferentially selected in the 3′-region of each coding sequence. The oligos were also designed according to physical parameters (GC content, annealing temperature, secondary structure, primer dimer formation, unique binding sites of the sequence). Tests of self-complementarity were performed in silico. According to a scoring matrix, a score of self-complementarity was calculated. All oligos were checked so that they did not contain any stretches of more than 15 nucleotides that are homologous to any other sequence in the genome. BLAST against MWG´s proprietary species-specific CodeSeq database was used for specificity testing. All oligos were checked in extensive alignments based on BLAST and global Smith–Waterman searches to exclude cross-hybridization. The oligomers were synthesized with MWG´s HPSF® technology, ensuring high percentage of full-length oligos and no contaminants, such as salt, metal, etc. Quality control after the oligo synthesis was done by trityl monitoring and MALDI-TOF analysis to ensure >90% purity. Bar-coding and tracking ensured 100% identity of all plated products.

Hybridization of microarrays

First, 25 and 50 μg of RNA were reverse transcribed using random primers and Superscript II reverse transcriptase (Invitrogen), and the cDNA was concomitantly labelled using the dyes Cy3 and Cy5, according to the design of the respective experiment. The RNA was removed after reverse transcription by alkaline hydrolysis. Slides were pretreated by blocking with 4x SSC, 0.5% SDS, 1% BSA for 2 h, washing five times in bi-distilled water and blow drying. The hybridization mixture containing the Cy-labelled cDNAs in hybridization buffer (50% Formamide, 6x SSC, 0.5% SDS, 50 mM NaPO4, pH = 8.0, 5x Denhardts’ solution) was denatured for 5 min and co-incubated with the slides for 16 h at 42°C. Washing was carried out in three subsequent steps with increased washing stringency: 2x SSC, 0.1% SDS, followed by 1x SSC, 0.1%SDS, then 0.5x SSC, for 10 min each step at room temperature. Finally, each slide was spun dry and scanned at both wavelengths using a model 418 or 428 Affymetrix confocal laser scanner at six different fixed intensities.

Bioinformatics

Sequence analyses and multiple sequence alignments and comparisons were performed with the GCG Wisconsin package (Version 10.2, Genetics Computer Group). The H. pylori flgM gene was identified in the genomes of H. pylori 26695 and J99 using the following approach: a profile of six conserved motifs was compiled from an unaligned set of 12 known FlgM sequences (available on request) from the databases with the program MEME (Bailey and Elkan, 1994). This motif profile was used to search the SPTREMBL database with the program MOTIFSEARCH. Two H. pylori proteins, HP1122 and JHP1051, were among the 10 highest scoring hits.

The microarrays were analysed using IMAGENE 4.0® and GENESIGHTLITE® software (BioDiscovery) for spot location, array alignment and background subtraction. Cy5 and Cy3 intensities were adjusted for local background and normalized to the median intensity over all spots in the respective channel in which it was scanned. Cy3/Cy5 ratio normalization was carried out by averaging over the ratio of all spots. Microsoft EXCEL was used for statistical analysis of the IMAGENE output files. For each mutant, the arithmetical means of the mean signal ratios for each spot was calculated for all four experiments. A change of transcription was considered significant when the average mean signal ratio was smaller than 0.5 or greater than 2. GD EXPRESSIONIST® (GeneData) was used for Cluster analyses. All genes with very low signal intensity (less than twofold of the background signal) were excluded from the evaluation. To account for biological variability in the experiment, competitive microarray hybridization was also done using RNAs from two independently grown cultures of the wild-type strain. All genes in the biological variation experiment remained within a cut-off range between 0.5 and 2.0.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

We thank Phil Aldridge, Heather Bonifield, Joyce Karlinsey and Allison Stack for helpful discussions and critically reading this manuscript. We are grateful to Robert Bell for his contribution to the evaluation of the microarray results. Matthias Frosch is gratefully acknowledged for continuous support and critical discussions, and, together with Guido Dietrich, for setting up a state-of-the-art microarray facility. This work was supported by grants Su 133/2–2 and 2–3 from the Deutsche Forschungsgemeinschaft. K.T.H. was supported by grant GM56141 from the National Institutes of Health.

Supplementary material

The following material is available from

Table S1. Gene transcript ratios.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Table S1. Gene transcript ratios

FilenameFormatSizeDescription
MMI_2765_sm_table.pdf4139KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.