University at Buffalo - The State University of New York
Skip to Content
Outer Membrane Protein Expression Profile in Helicobacter pylori Clinical Isolates

Outer Membrane Protein Expression Profile in Helicobacter pylori Clinical Isolates

  1. Rainer Haas1,*
  1. 1Max von Pettenkofer Institute for Hygiene and Medical Microbiology, Ludwig-Maximilians-University Munich, Pettenkoferstrasse 9a, D-80336 Munich, Germany
  2. 2Department of Oral Biology, School of Dental Medicine, State University of New York at Buffalo, Buffalo, New York
  3. 3Department of Medical Biochemistry and Biophysics, Umeå University, SE-901 87 Umeå, Sweden
  4. 4Dr. von Haunersches Kinderspital, Division of Pediatric Gastroenterology, Ludwig-Maximilians-University Munich, Munich, Germany


The gram-negative gastric pathogen Helicobacter pylori is equipped with an extraordinarily large set of outer membrane proteins (OMPs), whose role in the infection process is not well understood. The Hop (Helicobacter outer membrane porins) and Hor (Hop-related proteins) groups constitute a large paralogous family consisting of 33 members. The OMPs AlpA, AlpB, BabA, SabA, and HopZ have been identified as adhesins or adherence-associated proteins. To better understand the relevance of these and other OMPs during infection, we analyzed the expression of eight different omp genes (alpA, alpB, babA, babB, babC, sabA, hopM, and oipA) in a set of 200 patient isolates, mostly from symptomatic children or young adults. Virtually all clinical isolates produced the AlpA and AlpB proteins, supporting their essential function. All other OMPs were produced at extremely variable rates, ranging from 35% to 73%, indicating a function in close adaptation to the individual host or gastric niche. In 11% of the isolates, BabA was produced, and SabA was produced in 5% of the isolates, but the strains failed to bind their cognate substrates. Interleukin-8 (IL-8) expression in gastric cells was strictly dependent on the presence of the cag pathogenicity island, whereas the presence of OipA clearly enhanced IL-8 production. The presence of the translocated effector protein CagA correlated well with BabA and OipA production. In conclusion, we found unexpectedly diverse omp expression profiles in individual H. pylori strains and hypothesize that this reflects the selective pressure for adhesion, which may differ across different hosts as well as within an individual over time.

Helicobacter pylori persistently colonizes the stomachs of one-half of the world population and thereby causes gastric disorders and severe diseases, including active chronic gastritis and gastric or duodenal ulcer. Furthermore, H. pylori is considered a risk factor for the development of gastric cancer and mucosa-associated lymphoid tissue lymphoma (45). Ultrastructural studies of infected human gastric tissue revealed that a minor proportion of the bacteria are intimately attached to the epithelium of the gastric pits, whereas the majority of the H. pylori organisms actively move within the mucus overlying the gastric epithelium (22). Theoretical modeling of the dynamics of H. pylori infection identified adherence to the gastric mucosa as an important mechanism contributing to chronic colonization of the human stomach (7, 24).

One characteristic feature of H. pylori is the high plasticity of its genome, which is caused by both an elevated mutation rate and an extensive exchange of genetic material, leading to free recombination of H. pylori genes (46, 50). The micro- and macrodiversity among H. pylori strains acts as an important driving force for adaptation to the hostile gastric environment and the variable living conditions during the inflammation process and/or during transmission between host individuals (32, 44). This genetic heterogeneity becomes especially apparent in the class of outer membrane protein (OMP) genes. The H. pylori genome contains more than 30 omp genes, which have been divided into the hop (helicobacter outer membrane porins) and hor (hop-related) groups (1). All H. pylori adhesins identified so far are in the Hop group of proteins (reviewed in references 3, 17, and 30).

The blood group antigen binding adhesin (BabA) binds to ABO histo-blood group antigens and corresponding Lewis b antigens (Leb), which are expressed in the human gastric mucosa of most individuals (23). A further H. pylori OMP, the sialic acid-binding adhesin (SabA), binds to sialylated carbohydrate structures, which are upregulated as part of complex gangliosides in inflamed gastric tissue. SabA was postulated to contribute to the chronic persistence of the infection (4, 29). Two highly homologous OMPs, the adherence-associated lipoproteins A and B (AlpA and AlpB), are also involved in H. pylori adherence to human gastric histo-tissue sections (35, 36), although a corresponding receptor for these proteins is not known. Interestingly, functional and especially intracellular signaling differences in AlpAB proteins between Western and East Asian H. pylori strains have been reported (28). BabB, BabC, and HP0227 are considered members of the bacterial adhesion family, since their genes belong to the hop group of H. pylori omp genes (1), but a function in adherence for these OMPs has not been demonstrated. In addition, H. pylori adherence to extracellular matrix proteins, including laminin and collagen type IV, has been described (48). H. pylori adherence to laminin was recently attributed to the binding of SabA to sialylated moieties on this molecule, whereas fibronectin binding was independent of the SabA and BabA proteins (49). The glycan structures on laminin have not yet been elucidated. It was thus unknown whether sialyl-dimeric Lex is carried by this protein. However, SabA not only recognizes sialyl-dimeric Lex but has also been shown to have a broader sialic acid recognition, including binding to α-2,3-linked Neu5NAc (49) that may be present on laminin (25). The intimate contact between H. pylori and the gastric epithelium enables the bacterium to manipulate signal transduction pathways, resulting in the induction of proinflammatory cytokines, such as interleukin-8 (IL-8), in epithelial cells (11). The IL-8 induction is attributable mainly to the action of the cag type IV secretion system, a 40-kb chromosomal locus, also called the cag pathogenicity island (cag-PAI), which is involved in translocation of the bacterial effector protein CagA into the host cells (34). In many H. pylori strains, the CagA protein itself is not directly involved in IL-8 induction (14), but for several strains a direct contribution of translocated CagA to NF-κB activation and IL-8 induction has been demonstrated (9). In addition, the outer inflammatory protein (OipA), which belongs to the large H. pylori OMP family, was shown to be involved in IL-8 induction in epithelial cells upon H. pylori infection (51, 52). In another study, we reported that an oipA mutant strain induced similar IL-8 values to those in the parental wild-type (wt) strain (33, 51). Thus, the role of the OipA protein in this process remains to be elucidated in more detail.

Another H. pylori OMP-encoding gene, hopQ, exists as two highly divergent alleles. The type I hopQ allele is mainly found associated with East Asian H. pylori strains, in close association with the cagA gene, whereas the type II hopQ allele is commonly found in Western H. pylori strains lacking cagA (10). Interestingly, certain HopQ OMPs are able to attenuate H. pylori adherence to gastric epithelial cells and thereby affect the efficiency of CagA translocation into epithelial cells (27). Thus, the expression pattern of H. pylori omp genes is expected to have an important influence on H. pylori adaptation and virulence.

In order to obtain the first more global overview of the complexity of H. pylori omp gene expression, we concentrated on a set of eight different OMPs and CagA. No data about their expression or possible adhesive or other functions were available, but all selected omp genes belong to the adhesin-encoding hop gene family. Furthermore, we chose our set of 200 clinical H. pylori isolates from a pediatric clientele and tried to correlate the OMP profile and the cag status with adherence properties of the strains and their capacity to induce IL-8 in epithelial cells.


Patient collection, bacterial strains, and culture conditions.H. pylori bacteria were isolated from antral biopsy specimens from 200 patients of the Medical School of the Ludwig-Maximilians-University (Munich, Germany). Most isolates (52%) originated from children and young adults (≤20 years) who sought medical attention because of dyspeptic symptoms. Only 9% of the patients were older than 50 years. Thus, severe pathologies, including gastric and duodenal ulcer (4/200 patients) and gastric carcinoma (8/200 patients), were rare among our patient group, and therefore association studies of omp gene expression status and gastroduodenal disease were not meaningful for this patient group. From biopsy material, single H. pylori colonies were isolated to obtain pure cultures from each patient. These isolates were frozen at −80°C for further investigations. In all following experiments, bacteria were cultivated for only two or three passages on agar plates to minimize the risk of phase-variable switching of omp genes. As well-studied reference strains, we used H. pylori P1 (20), 26695 (47), and J99 (2). H. pylori cells were grown for 1 to 3 days (depending on the growth characteristics of the respective strain) in a microaerobic atmosphere (85% N2, 10% CO2, and 5% O2) at 37°C on serum plates as previously described (31). Escherichia coli strains DH5α (BRL) and 2136 (37) were grown on Luria-Bertani (LB) agar plates or in LB liquid medium supplemented with ampicillin (100 mg/liter), if appropriate.

SDS-polyacrylamide gel electrophoresis and immunoblotting.For specific detection of OMPs and CagA in H. pylori clinical isolates, bacteria were collected from serum plates and suspended in 300 μl Laemmli buffer. Boiled aliquots of bacterial lysates were subjected to sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis, using a minigel apparatus (Bio-Rad), and blotted onto nitrocellulose membranes at 1 mA/cm2 by use of a semidry blotting system (Biotec Fischer). The filters were blocked with 3% bovine serum albumin in Tris-buffered saline (TBS; 50 mM Tris-HCl, pH 7.5, 150 mM NaCl) and incubated with AK214 (AlpA) (36), AK262 (AlpB) (31), AK275 (BabC), AK276 (BabB), AK277 (BabA) (all described in reference 33), AK278 (SabA) (40), AK281 (HP0227) (this study), AK282 (OipA) (26), or AK257 (CagA) (34) antiserum for at least 4 hours. Either alkaline phosphatase-coupled protein A was used to visualize the antibody bound by decomposition of nitroblue tetrazolium or the blots were developed using horseradish peroxidase-coupled anti-rabbit immunoglobulin G and a Western Lightning detection system (Perkin-Elmer Life Sciences, Boston, MA).

For the generation of an HP0227-specific antiserum, a fusion protein with an N-terminal MS2 polymerase and His tag was produced using the E. coli expression vector pEV40 (37). A fragment of the hp0227 gene (bp 195 to 1107) was amplified by PCR with the primers SO100 (5′-GAGAATTCTTCCTTAGTCAATTTAGCCA-3′) and SO101 (5′-GACACTCGAGTCAAATGCTGTGGAAATTGTTC-3′), with H. pylori 26695 chromosomal DNA as the template. The PCR fragment (EcoRI/XhoI) was cloned into the expression vector pEV40 (EcoRI/SalI) (37), resulting in plasmid pSO212. The fusion protein was expressed by temperature induction of the λPL promoter and purified by Ni2+-nitrilotriacetic acid affinity chromatography according to the manufacturer's protocol (Qiagen, Hilden, Germany). The purified fusion protein was used to immunize a rabbit to obtain the polyclonal antiserum AK281. The antiserum was tested for potential cross-reactivity against other OMPs by using appropriate knockout H. pylori strains (data not shown).

Dot blot overlay assay.Leb-human serum albumin (Leb-HSA) conjugates were purchased from IsoSep AB (Tullinge, Sweden) (10 to 20 mol oligosaccharides/mol HSA). Laminin (from human placenta), fibronectin (from human plasma), collagen type IV, and HSA (fraction V, as a nonglycosylated negative control) originated from Sigma (Deisenhofen, Germany).

Bacterial overlays were performed as previously described (49). Dry nitrocellulose membranes (pore size, 0.45 μm; Schleicher & Schuell, Dassel, Germany) were spotted with 1 μg of Leb-HSA, laminin, fibronectin, or collagen type IV. Membranes (6 cm2) were blocked in 2.5 ml TBS (20 mM Tris-HCl, pH 7.6, 150 mM NaCl, 1 mM CaCl2, 1 mM MgCl2) containing 5% bovine serum albumin (fraction V, essentially free of γ-globulin; Sigma) for 2 hours at 4°C. Bacteria at 2 × 109/ml (optical density at 550 nm [OD550] = 6) in phosphate-buffered saline were labeled with fluorescein by incubation with 100 μg/ml fluorescein-5-isothiocyanate (Molecular Probes, Eugene, OR) for 30 min at room temperature. Labeled bacteria were recovered by centrifugation at 2,300 × g for 7 min, washed three times in phosphate-buffered saline, suspended in 500 μl of blocking buffer, and added to the membrane in blocking buffer. The overlays were incubated for 1 hour at 4°C in the dark without mixing and were washed three times in 10 ml TBS containing 0.05% Tween 20 at room temperature for 5 min on a rotary shaker. The fluorescence of the adherent bacteria was detected by a Fuji FLA3000 imaging system (Raytest, Straubenhardt, Germany).

IL-8 ELISA.Cells of the gastric carcinoma cell line AGS were grown in 24-well plates (2 cm2) at 37°C and 5% CO2 in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. After infection with H. pylori at an OD550 of 0.1 (multiplicity of infection [MOI] = 100) for 4 hours, supernatants from infected cells were collected and bacteria were removed by centrifugation. The concentration of secreted IL-8 was determined using an optimized enzyme-linked immunosorbent assay (ELISA) as previously described (39). The amounts of IL-8 secreted by AGS cells infected with different H. pylori strains were calculated as ratios to the value for strain 26695 (positive control, set to 100%). Noninfected cells served as a negative control. Each strain was assayed in duplicate.

Statistical analyses.Statistical analysis of the correlations in omp gene expression was performed using SAS software (version 9.1.3). The phi coefficient is a measure of association derived from the Pearson chi-square statistic. It has the range −1 ≤ φ ≤ 1 for two-by-two tables. The phi coefficient is computed as follows: Formula where n is the number of cases (15). The amounts of secreted IL-8 were compared among the four groups of H. pylori strains (depending on CagA and OipA status) by the t test, using the Sigmastat software package. P values are indicated in the figures and were considered significant if they were <0.05.


Production of OMPs in clinical H. pylori isolates.Two hundred clinical H. pylori isolates from human gastric biopsies and a panel of well-characterized H. pylori control strains were included in this study (Table 1). After the initial growth of H. pylori from biopsy material, pure cultures were generated from each patient by isolation of single colonies representing a single H. pylori strain. The strains were frozen, and bacterial lysates were prepared. The production of eight different OMPs, namely, AlpA, AlpB, SabA, BabA, BabB, BabC, HP0227, and OipA, was investigated by immunoblot analysis using a panel of highly specific polyclonal rabbit antisera raised against purified fusion proteins of the corresponding OMPs. Also, the production of CagA was included in this study because of its high impact on pathogenesis.


Bacterial strains used in this study

The antisera specifically recognized their corresponding target proteins, as exemplified for nine representative isolates (Fig. 1). The sizes and amounts of the produced OMPs varied considerably between independent isolates. The AlpA and AlpB proteins turned out to be expressed in similar amounts by all 200 strains, suggesting an important or essential function of these proteins (Table 2). All other OMPs were produced at variable frequencies. BabA, BabB, HP0227, OipA, and CagA were detected in approximately 60 to 70% of the isolates, whereas BabC and SabA were produced in only one-third of the strains. Notably, approximately one-third of all strains were negative for both BabA and SabA (Fig. 2A).

FIG. 1.

OMP expression profiles of representative H. pylori isolates. Whole-cell lysates of H. pylori strains were separated in a 10% SDS-polyacrylamide gel (or a 6% gel for CagA detection), blotted on a nitrocellulose membrane, and detected with antisera against eight different OMPs or CagA.

FIG. 2.

Binding of H. pylori control strains to defined receptor structures. (A) Strains were classified according to the production of BabA and SabA proteins. (B) Production of BabA and SabA in well-characterized control strains. (C) Binding of fluorescently labeled H. pylori bacteria of these strains to defined antigens, including Leb-HSA, laminin, collagen type IV, and fibronectin, was explored by dot blot overlay assays. J99 and CCUG17874 are wt strains, and J99(babA) and J99(sabA) are isogenic mutant strains with chromosomal deletions of the genes in parentheses (33, 49).


Production of H. pylori OMPs in clinical isolates

To screen for possible functional linkages between different OMPs, we calculated the φ value for the coexpression of all protein pairs. The most prominent φ values for coexpression were found among the proteins BabA, OipA, and CagA (Table 3), with particular emphasis on the CagA-OipA pair. For instance, 95% of all CagA-positive strains also produced OipA. Furthermore, the production of SabA was weakly but significantly correlated with the presence of CagA, OipA, BabA, and HP0227. Interestingly, the BabB protein was associated with a CagA-negative status.


Correlations among OMP proteins and CagA

Adherence properties of H. pylori isolates.The clinical isolates were further analyzed with regard to their properties of adherence to defined proteins or glycoconjugates. We divided the strains into groups according to their BabA and SabA expression, which resulted in a total of 92 representative strains (Fig. 2A) to be analyzed in the adherence assay. In this assay, the capacity of strains to bind Leb, sialylated glycoconjugates (laminin), fibronectin, and collagen type IV was assessed. Well-defined H. pylori reference strain J99 and defined J99 babA and sabA mutant strains, as well as strain CCUG17874, for which the BabA and SabA binding status is known, were used as controls to evaluate the dot blot assay for its specificity in BabA and SabA binding. The control strains showed specific binding to their substrates that was dependent on their cognate adhesin (Fig. 2B and C). Next, all 92 isolates were examined in the dot blot assay, as exemplified for a set of six selected H. pylori isolates (Fig. 3A). In most isolates, expression of BabA or SabA was associated with binding to the expected receptor, i.e., Leb or laminin, respectively. However, strains P1 and 26695, which are well characterized in terms of their binding characteristics, and three clinical isolates did not recognize the Leb glycoconjugate, although a BabA protein was produced (Fig. 3B). In a recent study, it was shown that a terminal modification of the Leb antigen with galactose (BLeb) or N-acetylgalactose (ALeb) is able to modulate recognition by the BabA protein. We tested the set of BabA-positive strains for binding to ALeb and BLeb glycoconjugates. As a positive control, the wt strain J99 bound to both unmodified Leb and ALeb antigen, whereas the BLeb conjugate was recognized only very weakly. Three clinical isolates (no. 161, 222, and 336) were unable to bind to any Leb structure, irrespective of terminal modifications (Fig. 3B). In addition to the strains producing nonbinding BabA proteins, two isolates did not bind to laminin, in spite of clear detection of a SabA protein in the immunoblot (Fig. 3C). Collectively, these data suggest that some H. pylori isolates produce BabA- and/or SabA-immunoreactive proteins that apparently do not allow binding to their originally identified target structures.

FIG. 3.

Binding of H. pylori clinical isolates to defined receptor structures. (A) Binding of fluorescently labeled H. pylori (Hp) bacteria of six representative isolates to defined antigens, including Leb-HSA, laminin, fibronectin, and collagen type IV, was explored by dot blot overlay assays. Production of BabA, SabA, and CagA is indicated below. (B) Characterization of some strains expressing inactive BabA proteins. No binding to Leb-HSA or the derivatives ALeb-HSA and BLeb-HSA could be observed. (C) Production of “inactive” SabA proteins in two H. pylori isolates.

To screen our strain collection for further putative adhesins involved in binding to extracellular matrix proteins, we also explored the binding capacity of the 92 selected H. pylori isolates for fibronectin and type IV collagen. All 92 strains investigated bound efficiently to fibronectin, irrespective of their OMP pattern or Cag status (Fig. 3A). In contrast, type IV collagen was recognized by only two strains, and very weakly (Fig. 3A). The frequencies of adherence of the 92 strains to Leb, laminin, collagen type IV, and fibronectin are summarized in Table 4.


Adherence of clinical H. pylori isolates to defined antigens in dot blot arraysa

Influence of Cag status and OipA expression on IL-8 induction.In previous studies, it has been shown that a positive Cag status and a functional cag-PAI is necessary for IL-8 induction in epithelial cells upon H. pylori infection (11, 14). In addition, the OipA protein has been proposed to be involved in this process (51, 52). To further characterize the role of the OipA protein in stimulation of IL-8 secretion from epithelial cells, we infected AGS cells with 102 representative clinical isolates, which were categorized according to the production of CagA (representing a cag-positive status) and OipA. Since the absolute IL-8 concentration in the supernatant varied among independent experiments, the values were normalized in relation to the IL-8-inducing reference strain H. pylori 26695. Due to the high correlation of CagA and OipA statuses, the groups containing singly positive strains (CagA+ OipA or CagA OipA+) were smaller than the double-positive or double-negative groups (Fig. 4). As expected, the IL-8 values induced by Cag-positive strains were significantly higher than those induced by Cag-negative strains. OipA-positive strains did not induce higher IL-8 amounts than OipA-negative strains in a Cag-negative background. However, within the group of Cag-positive strains, the production of OipA was correlated with twofold higher IL-8 values than those for OipA-negative strains (Fig. 4). These results suggest that the OipA protein alone is not able to stimulate AGS cells with respect to IL-8 secretion, but OipA expression seems to support Cag-induced cytokine secretion. The mechanism behind this phenomenon needs to be clarified in future studies.

FIG. 4.

IL-8 production from AGS cells. AGS cells were infected with representative H. pylori strains classified according to CagA and OipA production (MOI = 100; infection was done for 4 hours). The IL-8 concentration in the supernatant was measured by ELISA. Each value was calculated as a ratio to the amount of IL-8 produced by the IL-8-inducing wt strain 26695, which was set to 100%. The line in each bar represents the 50th percentile (median) of the group; the ends of the bars indicate the 25th and 75th percentiles. Error bars show the 10th and 90th percentiles, and single dots include 5th and 95th percentile outliers.

Coexpression patterns of H. pylori omp genes in 200 clinical isolates.In order to gain an impression of how frequently the omp genes and cagA are coexpressed in 200 independent H. pylori patient isolates, we determined the distances between H. pylori strains based on their omp expression patterns and performed a clustering analysis. The resulting dendrograms are presented as a heat map (Fig. 5). H. pylori isolates that show related or even identical patterns of omp gene expression are clustered closely together (compare rows in Fig. 5), whereas the coexpression patterns for certain omp genes are also clearly seen (compare columns). Thus, AlpA and AlpB are most strongly coproduced, but CagA and OipA also show a strong tendency to be coproduced. BabA is related to the CagA-OipA pattern, whereas HP0227 is more related to AlpA-AlpB in its profile. BabB, SabA, and HP0317 form another cluster with low relatedness in the expression profile (see Table 3 for correlations in omp gene expression).

FIG. 5.

Graphic representation of omp gene expression patterns of 200 H. pylori isolates by use of a heat map. Two hundred H. pylori isolates are shown in a two-dimensional representation in which the presence of a certain OMP (AlpA, AlpB, HP0227, OipA, BabA, BabB, HP0317, SabA, and CagA) is indicated by a gray line and its absence is shown by a white line. The relatedness of certain isolates concerning a common expression pattern is shown in one dimension (rows), and the coexpression patterns of all omp genes are shown in the other dimension (columns).


H. pylori isolates are very well adapted to their hosts, which allows them to persist for years and decades in their individual niches. In addition, they have to be flexible enough to rapidly accommodate to a new host during infection. The high genetic variability between different H. pylori strains is based on microdiversity (on the gene level) as well as macrodiversity (on the genomic level) (32). The large group of OMPs is probably of considerable importance for optimal adaptation of H. pylori to its host. We are not aware of a comparable comprehensive investigation of OMP expression covering such a large set of defined H. pylori isolates, especially from young people, as that described here. Many studies rely on H. pylori isolates from patients suffering from more severe gastroduodenal disease. Our collection is different in this respect, since it covers mostly children and young adults. Thus, our collection of strains more closely represents the early natural conditions of the infection, rather than a special adaptation of H. pylori due to antibiotic treatments or an adaptation to conditions of severe gastroduodenal disease.

Many studies on H. pylori omp gene expression have relied on reverse transcription-PCR data. The presence of omp mRNA does not necessarily correlate with the presence of the OMP, since (i) many omp genes are contingency genes and (ii) it is becoming obvious that posttranscriptional regulation in bacteria is a frequent event (e.g., small RNAs in bacteria regulate omp genes, as described for Escherichia coli and Salmonella species) (8). Thus, we believe that our omp expression data are more reliable and very important for characterization of the OMP status of a single H. pylori isolate. Furthermore, a complex expression pattern of different OMP or adhesin proteins as shown here is truly novel, not just for H. pylori but also for adhesin expression in general, based on a bacterial species isolated from a single type of target tissue.

H. pylori uses several strategies to generate diversity in the large group of OMPs. One mechanism relies on slipped-strand mispairing (SSM), which involves the deletion or insertion of nucleotides in homopolymeric tracts located in the gene promoter region or the 5′ gene sequence (coding repeats) (43). The sabA gene, as well as babB and oipA, is subject to SSM regulation (6, 29). These nucleotide changes allow a rapid and flexible on-off switch of the corresponding genes on the transcriptional (promoter switch) or translational (coding repeat) level. In addition to SSM, SabA protein production is also controlled by the ArsRS two-component signal transduction regulatory system on the transcriptional level (19). This dual-control mechanism might explain the low rate (38%) of SabA-producing H. pylori strains in our collection and might be a mechanism for H. pylori to escape the immune system in inflamed tissue. But natural transformation competence mediated by the comB system or intragenomic recombination might also play a major role in dynamic on-off switches of H. pylori omp genes. Especially the group of bab genes (babA, babB, and babC) show high homologies in the 5′ and 3′ regions, whereas the central part is rather diverse (38, 47). Thus, frequent recombination between bab genes seems to occur. In a rhesus macaque model, animals were infected with a BabA-producing H. pylori strain. Reisolated strains had either switched off babA expression by SSM or recombined the babB gene into the babA locus, with both resulting in a loss of Leb binding (42). Furthermore, diverse genotype profiles of babA and babB reflect selective pressures for adhesion, which may differ across different hosts and within an individual over time (12). Thus, our results support the data from the literature and the view of highly dynamic changes in omp gene expression occurring during gastric adaptation of H. pylori in vivo.

We were therefore interested in analyzing the status of omp gene expression of a larger set (n = 200) of fresh clinical H. pylori isolates to possibly gain an overview about the frequencies of production of certain OMPs or adhesins and the phenotype of adherence to defined substrates of the corresponding strains. Since the adhesion analysis of BabA and SabA binding and IL-8 assays are labor-intensive, we restricted the analysis to 92 strains for BabA and SabA adherence assays and to 102 strains producing CagA and OipA for induction of IL-8 expression in epithelial cells. The strains selected nicely represented the different BabA and SabA expression states and were therefore considered sufficient for these assays. In total, we analyzed the expression of eight omp genes and also cagA, an important marker of disease induction. All isolates tested produced AlpA and AlpB proteins, but only about 60% produced BabA, BabB, or CagA (Table 2). BabC and HP0227, which are also members of the adhesion family, were produced by 35% and 73% of isolates, respectively. OipA was produced by 68% of isolates, and SabA was produced by 38% of isolates.

The percentages of European isolates producing BabA, ranging from 40% to 70%, were similar to those in other studies (18, 23), whereas the rates for non-European countries, such as Japan, Korea, Columbia, and the United States, were higher (70% to 100%) (41, 53). The expression states for omp genes in the cited studies were based on PCR analysis, which does not give a clear picture about the production of the adhesin protein and its function (16). We therefore used specific antibodies recognizing the presence of the cognate protein. Our data show an unexpectedly high level of complexity of omp gene expression profiles of individual patient isolates (see Fig. 5 for an overview of the total omp gene expression profiles of isolates). It will be important in future to study whether a defined OMP profile is stable in a given host or whether it still changes over time. A clear correlation in the expression profile was, however, found for oipA and babA expression. Expression of cagA clearly correlated with oipA and less stringently correlated with babA expression (Fig. 5; Table 3). All other combinations of omp genes were expressed in a less stringent correlation pattern (Fig. 5; Table 3).

We next were interested in the binding capabilities of the strains for defined substrates, such as Leb glycoconjugate (for BabA binding), laminin (for SabA binding), and more specialized substrates, such as ALeb and BLeb. As expected, most of the BabA-producing isolates bound the Leb glycoconjugate, and most SabA-producing strains were able to bind to laminin (Table 4). Interestingly, however, 11% and 5% of BabA-positive and SabA-positive strains, respectively, were unable to bind their expected substrate molecules in dot blot assays. Several previous studies also reported marked variation in the ability of BabA proteins from different strains to bind Leb or a lack of detectable binding of certain BabA proteins to Leb (5, 6, 21). We reinvestigated these nonbinding strains by using the more quantitative Leb radioimmunoassay. The nonbinding phenotype could be reproduced for strain 222, whereas strains 161 and 336 now showed binding to Leb. Both SabA-producing strains (167 and 216) did not show binding in the radioimmunoassay for sialyl-Lex (data not shown). Possibly, strains 161 and 336 are better at binding to soluble Leb glycoproteins but have difficulties in binding to surfaces, such as ELISA plates or epithelial cells. Solid-phase binding is usually more complicated than binding to proteins or receptors in solution. Furthermore, since the reevaluated strains could be tested only after further freeze-thawing and several subculturing steps, we also cannot rule out that minor changes in the protein expression level or the BabA/SabA gene sequence occurred compared to the original isolates. In the past, BabA-producing strains have been subgrouped into “specialists” and “generalists,” depending on their mode of BabA binding to Leb antigens (5). Generalists bind independent of a terminal modification (Leb, ALeb, and BLeb), whereas specialists show binding to nonsubstituted Leb only. Notably, in addition to the lab strains 26695 and P1, we identified at least one fresh clinical isolate that was unable to bind to any Leb structure, irrespective of terminal modifications, suggesting that this strain cannot be classified according to the generalist/specialist classification (Fig. 3B).

We have shown recently for SabA that binding to sialylated glycoconjugates may be rather polymorphic (4). Receptor mapping revealed that the NeuAcα2-3Gal disaccharide constitutes the minimal sialylated binding epitope required for SabA binding (4), but clinical isolates demonstrated polymorphism in sialyl binding, as also shown in this study. This variability may be important for H. pylori to adapt its binding properties to the probably changing epithelial glycosylation patterns during chronic inflammation, as well as to change its individual host.

In addition to Leb and sialyl binding, collagen type IV and laminin have also been described as targets for H. pylori adhesion (48). Interestingly, it is the protein rather than the carbohydrate structures of fibronectin which is relevant for binding to H. pylori (49). All of the isolates tested in this study bound to fibronectin, indicating that none of the specific omp expression profiles tested here would be responsible for fibronectin binding. In contrast, only 2 of 92 isolates tested bound to collagen type IV, both of which produced neither BabA nor SabA. Since BabA/SabA-negative strains bound to collagen, we cannot exclude that BabA or SabA adhesins somehow shield the type IV collagen binding adhesin from reaching its target. In conclusion, the binding activity of H. pylori for fibronectin might be rather important, since it was found in all isolates tested. Nonetheless, none of the OMPs studied here seem to be responsible for binding to fibronectin or collagen under the conditions used. It remains to be clarified in further studies whether any other OMPs, or eventually other surface structures, such as carbohydrate structures (e.g., lipopolysaccharide), might be involved.

OMPs not only are important as adhesins but also are able to induce signal transduction events in the host cell. H. pylori stimulates the secretion of IL-8 in epithelial cells, which is mostly dependent on the cag-PAI. Interestingly, the AlpAB adhesins were also recently associated with the induction of interleukins, such as IL-6 and IL-8, in gastric epithelial cells. The IL-8 induction was found only with East Asian strains, not with Western H. pylori strains (28). In this study, we investigated the influence of oipA expression on IL-8 induction in a cag-PAI-positive and a cag-PAI-negative background. Generally, CagA- and OipA-producing strains (double positive) were strongest in IL-8 induction, whereas double-negative strains induced only a very little IL-8 (Fig. 4). CagA-positive but OipA-negative strains induced 30% to 50% the amount of IL-8 induced by a double-positive reference strain (26695), whereas CagA-negative, OipA-positive strains reached only 20% IL-8 induction. Thus, earlier observations concerning a 50% reduction of IL-8 secretion upon oipA knockout in a cag-PAI-positive background could be verified by using independent H. pylori patient isolates producing or not producing OipA (52). In addition, we showed here with a set of independent clinical isolates (Fig. 4) (n = 10) that OipA without a functional cag-PAI is not able to induce IL-8. Our data are based on the comparative analysis of H. pylori wt strains rather than the comparison of isogenic mutant strains, but we clearly demonstrated the absence or presence of the protein between strains. In a recent study using genetic and functional genomic analyses of hopH gene polymorphisms (13), the authors did not find a significant effect of the H. pylori B128 wt strain versus an isogenic hopH mutant strain in IL-8 induction. Whether the hopH gene was functionally expressed remains unclear. Although the authors showed a significant effect of a hopH mutant strain on adherence to gastric Kato-3 cells, which could be restored by complementation analysis, the production of a functional HopH protein in an immunoblot was not verified in their study.


This work was supported by the Deutsche Forschungsgemeinschaft (projects OD 21/1-1 to S.O., HA 2697/10-1 to R.H., and SFB585/B5 to S.R.) and by grants from Vetenskapsrådet/VRM (T.B.) and Cancerfonden (T.B.).

We thank Stefan Oscarson, University College Dublin, for providing a BLeb conjugate. We thank Lukas Windhager, LFE Bioinformatics, LMU Munich, for presentation of the expression data as a heat map.


    • Received 31 March 2009.
    • Returned for modification 5 May 2009.
    • Accepted 15 June 2009.
  • *Corresponding author. Mailing address: Max von Pettenkofer Institute for Hygiene and Medical Microbiology, Pettenkoferstrasse 9a, D-80336 Munich, Germany. Phone: (49)-89-5160 5255. Fax: (49)-89-5160 5223. E-mail: haas{at}
  • S.O. and K.S. contributed equally to this work.

  • Published ahead of print on 22 June 2009.

  • Editor: S. R. Blanke