Helicobacter pylori strains harboring the vacAs1, cagA and babA2 have been associated with ulcer disease (UD). We compared the prevalence of these different genotypes and adhesive properties in H. pylori infected patients with UD in four European countries. Genomic DNA was isolated from 314 H. pylori strains: Germany (GER; n= 92), Sweden (SWE, n= 74), Portugal (POR, n= 91) and Finland (FIN, n= 57). The frequencies of babA2 genotype varied from 35% to 60%. Triple-positive strains (vacAs1+, cagA+ and babA2+) were significantly associated with UD in GER and POR and were closely correlated with UD in FIN, but not in SWE. Classification as triple-positive strains had a higher specificity for detection of UD in GER, POR and FIN than type1 or cagA+ strains. In vitro adhesion assays revealed that Swedish strains showed high adhesion properties and were thus correlated with the diagnosis of UD, although PCR detected the babA2 gene at lower frequencies and failed to show a correlation with UD. This finding appears to reflect allelic variations of the babA2 gene in SWE, although adhesive properties of the strains are retained.
Infection with Helicobacter pylori is the major cause of gastroduodenal diseases, including gastritis, peptic ulcer , mucosa-associated lymphoid tissue (MALT) lymphoma and gastric adenocarcinoma . The different outcomes are induced by complex interactions between environmental, host and bacterial virulence factors . The importance of the classical H. pylori virulence factors, i.e. the vacuolating cytotoxin VacA and the cytotoxin associated gene product CagA, has been described in detail [4–6]. H. pylori strains have been divided into type1 (VacA, CagA positive) and type2 (VacA, CagA negative) strains. VacA is a secreted cytotoxin that has the capacity to form vacuoles in epithelial cells . The P34 subunit of VacA acts on mitochondria, inducing the release of cytochrome C. Atherton et al.  have identified the mosaic combination of signal regions (s1, s2) and middle regions (m1, m2). vacAs1 strains produce a large amount of cytotoxin, whereas s2 type strains secrete small amounts or no cytotoxin at all. CagA, which is known as a marker for presence of the cag PAI , is involved in cytoskeletal changes of the host cells. The increased production of IL-8, which is induced by the specific genes of cag PAI, may partially explain the relationship between cag PAI positive strains and more severe diseases induced by the infection.
Adherence factors of H. pylori are further considered to contribute to the pathogenicity of these bacteria. One of the major adhesins, BabA, encoded by the babA2 gene, has been identified as the adhesin responsible for H. pylori binding to the fucosylated blood group antigens Lewisb. BabA is a member of a paralogous family of outer membrane proteins. Its capability of binding to Lewisb facilitates H. pylori colonization of the stomach but may also play a direct role in pathogenesis. It has been demonstrated that bacterial binding to Lewisb is mainly seen in type1 isolates , indicating that no single factor can be considered solely responsible for determining outcome of H. pylori infection. An adherent strain however, might have more clinical impact than a strain, which lacks these properties .
Previous studies of our group have found that in a German population the presence of babA2, as well as the simultaneous presence of vacAs1 and cagA (“triple-positive strains”), was significantly associated with duodenal ulcer , and adherence of H. pylori via BabA appeared to be important for efficient delivery of VacA and CagA. BabA may also play a role in the pathogenesis of severe histological changes, such as atrophy or intestinal metaplasia . In contrast to our results, studies by other groups [15,16] did not reveal any direct association between babA2 and the status of cagA or vacA, and also did not detect any correlation with the clinical outcome. The present study therefore aimed to compare these virulence factors of H. pylori isolated from patients from four European countries to further investigate the correlation of babA2, cagA and vacAs1 genotypes with disease, as well as the effect of genetic diversity of babA2 on its adhesive activity.
Materials and methods
Populations of the patients
Helicobacter pylori strains were isolated from a total of 314 patients suffering from peptic ulcer disease (UD) and gastritis (GA) from four European countries: 91 from Portugal (POR) (UD = 30, GA = 61, male 52%, median age = 67 years, range 27–87); 92 from Germany (GER) (UD = 25, GA = 67, 50% male, median age = 58, ranging from 19 to 91); 74 from Sweden (SWE) (UD = 30, GA = 44, male = 54%, median age = 62, range 28–85) and 57 from Finland (FIN) (UD = 31, GA = 26, 38% male, median age = 55, range 25–83), respectively. All of these patients presenting with abdominal pain or dyspeptic complaints were diagnosed endoscopically after informed consent was obtained. Only patients diagnosed with H. pylori positive duodenal ulcers were investigated that did not take aspirin or non-steroidal anti-inflammatory drugs (NSAID) medication. Patients using proton pump inhibitors (PPI) > double standard dose or patients with history of gastric cancer were excluded.
H. pylori culture and isolation of genomic DNA
Gastric biopsies were obtained from antrum of the patients for isolating H. pylori. The specimens were homogenized in 0.5 ml brucella broth with 10% (vol/vol) fetal calf serum (FCS) with a tissue homogenizer. The suspensions were inoculated onto Wilkins Chalgren agar (Oxoid, Basingstoke, UK) containing 10% horse blood, Dent supplement (Oxoid) and 0.4 g/L KNO3. The plates were incubated up to 8 days in an CO2 water jacketed incubator (Forma Scientific Inc.) with a microaerobic atmosphere of 10% CO2, 85% N2, 5% O2 and 100% relative humidity at 37 °C. H. pylori were identified by colony morphology, Christensen urease (Bacto urea base, Difco), oxidase (Oxidase Dry Slide, Difco), catalase, and phase-contrast microscopy. Isolates were resuspened in brucella broth with 10% FCS and 20% glycerol and frozen at +80 °C for storage. Genomic DNA was isolated from harvested bacteria using the QIAamp Tissue Kit (Qiagen, Munich, Germany) according to the manufacturer's instructions. H. pylori isolates cultured for in vitro adherence assay were harvested within 24–48 h, and resuspended in PBS, then diluted to a density of OD600= 1 for immediate use.
Detection of babA2, cagA and vacA genotypes by PCR
PCR amplifications were performed for detecting babA2, cagA and vacAs1/2 genotypes of all isolated strains. Primer sequences for babA2, cagA and vacA have been published previously [9–12]. Amplification was carried out in a total volume of the 25 µl using Taq PCR Master Mix (Qiagen), containing 20 ng H. pylori genomic DNA and 0.5 µl of each primer (20 µM). Reaction was performed in a Primus PCR Cycler (MWG Biotech, Ebersberg, Germany) under the following conditions: 5 min initial denaturation at 94 °C; 30 cycles of 30 s at 94 °C, 30 s at 55 °C and 1 min at 72 °C; and 10 min final extension at 72 °C. PCR products were analyzed on 1% agarose gels stained with ethidium bromide.
In vitro adherence assay
Lewisb antigen conjugated to human serum albumin (Isosep AB, Stockholm) was diluted in 0.2 M carbonate buffer (pH 9.2) to a final concentration of 1 ng/µl. COSTAR 2503 Universal Covalent 96-well microtiter plates were coated with 50 µl/well of above solution or 50 µl/well carbonate buffer without antigen as control), and incubated at 4 °C in the dark over night. The decanted plates were exposed to UV light for 30 s in a Stratalinker (Strategene, Heidelberg, Germany) for antigen crosslink. 100 µl of blocking buffer (0.5% non-fat dry milk, 0.2% Tween20) was added to each well. After incubation for 1 h at room temperature, plates were decanted. Bacterial suspension prepared as described above, diluted 1:2 in blocking buffer containing 10% FCS, and 50 µl were added to each well. After incubation for 2 h at room temperature with gentle agitation (50 rpm), plates were sucked with vacuum and washed with PBS. Bound bacteria were detected with anti-H. pylori mAb (rabbit, 1:1000) and anti-rabbit-HRP Ab (1:1000), followed by ABTS-solution. Extinction was measured in a microplate reader (Bio-Rad) at 405 nm and normalized to controls. Strains were considered as adhesion positive if the ratio of extinctionLewis/extinctioncontrol was >1.5. Alternatively, strains were labeled with Digoxigenin (DIG) and detected by anti-DIG-HRP as described before . In each single experiment, all strains were tested in two coated wells and two controls. Each strain was assayed at least twice in independent experiments.
The χ2 test and the Fisher's exact test were used to compare the differences among the groups. P values < 0.05 were considered as statistically significant.
Prevalence of cagA, vacAs1 and babA2 genotypes
314 patients infected with a single H. pylori strain were enrolled in four European countries and H. pylori genotypes were analyzed. The overall prevalence of cagA and vacAs1 genotype was 77% (243 of 314 isolates) and 79% (248 of 314 isolates), respectively. The biopsies were divided into two groups according to their clinical outcomes: chronic gastritis (GA) and ulcer disease (UD). Considering the correlation between vacAs1, cagA status and the clinical outcome, vacAs1 strains were more frequent in patients with UD than with GA in GER and POR (GER: 100% versus 79%, p= 0.0096; POR: 70% versus 33%, p= 0.048, respectively). Similar observations were made for cagA+ strains (96% versus 73%, p= 0.016; 80% versus 51%, p= 0.007, respectively). In SWE, the prevalence of vacAs1 was equally distributed (100% versus 95%, p= 0.54), while cagA+ strains were more frequent in the DU group than in the GA group (90% versus 73%, p= 0.07). In FIN, the high prevalence of vacAs1 (100%) and cagA (98%) prevented the analysis of a significant difference. The babA2 genotype was detected in 139 of all 314 isolates (44%), with the frequency ranging from 35% to 60%. The babA2+ strains were significantly more frequent in patients with UD compared to GA in GER (88% versus 28%, p < 0.0001) and POR (63% versus 20%, p < 0.001), and were predominant in UD patients in FIN, although not statistically significant (71% versus 46%, p= 0.057). The babA2 prevalence was surprisingly low (40%) in SWE, and showed no significantly correlation with disease (p= 0.51) (results presented in Fig. 1).
Correlation of type1 strains and triple-positive strains with the clinical outcome
The strains harboring both vacAs1 and cagA genotypes were detected in 71% of isolates (224 of 314 isolates). Significant associations were found between type1 strains and UD in GER (p= 0.006) and POR (p= 0.0007), but not in SWE (p= 0.07) and FIN (p= 0.204). Type1 strains also harboring babA2 (triple-positive strains) were significantly associated with UD in GER (p < 0.0001) and POR (p < 0.0001) and were closely correlated with UD in FIN (p= 0.057), but not in SWE (p= 0.86). Comparing type1 and triple-positive strains, both were predominant in UD in GER (p= 0.006 and p < 0.0001, respectively) and POR (p < 0.007 and p < 0.0001, respectively). In FIN, however, there was no association between type1 strains and UD (p= 0.204), while triple-positive strains were closely associated with UD (p= 0.057). In SWE, neither type1 nor triple-positive strains were significantly associated with UD (p= 0.07 and 0.86, respectively) (see Fig. 2).
Comparison of the babA2 genotype and Adhesion to Lewisb
The abilities of binding to Lewisb of the strains were evaluated by in vitro adhesion assay. Two strains were not determined because of contamination in culture. Out of 139 babA2 positive strains, 105 strains showed adhesive abilities, 34 strains did not bind to Lewisb (35+ versus 6+ in GER, 17+ versus 14+ in POR, 25+ versus 9+ in FIN and 28+ versus 5+ in SWE). Out of 173 babA2-strains, 145 strains were negative in the adhesion assay, whereas 28 strains (two in GER, two in POR, seven in FIN and 17 in SWE) showed binding ability (Fig. 3). When comparing the presence of babA2 genotype and the strains' capability for Lewisb binding, the babA2 genotype was more frequent than expression of functional BabA protein in vitro in all countries except SWE. In contrast, adhesion to Lewisb was more frequent (63%) than prevalence of babA2 (45%) in SWE.
The H. pylori blood group antigen-binding adhesin BabA mediates adherence of H. pylori to Lewisb epitopes expressed on human gastric epithelial cells. This outer membrane protein appears to play a role in efficient delivery of bacterial virulence factors that damage the gastric epithelium by triggering inflammatory reactions. An initial study reported a highly significant correlation between babA2 gene and prevalence of ulcer and adenocarcinoma in a German population . Further, biopsy based studies demonstrated that, in H. pylori infected patients, babA2 was significantly correlated with higher degrees of mucosal inflammation, glandular atrophy and intestinal metaplasia [14,17]. Thoreson et al.  reported a higher prevalence of BabA in Swedish patients with doudenal ulcer or duodenitis compared to asymptomatic subjects. Yu et al.  reported that in Yantai County, where the incidence of gastric cancer is among the highest in China, the babA2+ strains were present in 79.8% of the studied patients, and were associated with higher degrees of lymphocytic infiltration, glandular atrophy and intestinal metaplasia. Interestingly, only BabA was a candidate for a universal virulence factor, underlining the importance of this adherence factor.
To objectify the variation of babA presence and disease correlation observed in these studies, we conducted a multicenter trail in four different European countries and investigated the prevalence of babA2, vacAs1 and cagA genotypes and their correlation with clinical outcomes. In GER and POR, vacAs1+, cagA+ and type1 strains were more frequent in patients with UD. When characterizing the bacteria by the additional presence of babA2 in type1 strains, classification as triple-positive strains had a higher discriminating value for the detection of UD compared type1 strains alone in all countries except SWE. For instance in FIN, type1 strains could not be related to UD because of the high prevalence of vac As1 (100%) and cag A (98%) genotypes in both two groups, whereas babA2+ strains were closely correlated with UD (p= 0.057), albeit not significant due to the limited number of cases. Therefore, determination of babA2 genotype in type1 strains can improve the differentiation between UD and GA. These data further confirmed the previously reported importance of BabA.
In one country we did not find a clear association with disease when genotyping strains by PCR: the prevalence of babA2 in UD patients in SWE was surprisingly lower than in UD groups of the other three European countries (40% versus 63–88%). Neither type1 strains nor triple-positive strains could be related to UD in this population. Similarly, in Asia, in spite of the prevalence of babA2 positive strains (36.1% in a Korean group versus 81–90% in a Japanese group), none of the two studies showed any significant difference among the subgroups of different severe gastric diseases and non-ulcer dyspepsia patients [15,16]. One possible explanation might be that the status of Lewisb expression on gastric epithelium is variable in different populations; however, several studies, including different populations from Asia and western countries, showed that Lewisb epitope is present in the vast majority of the people (ranges from 75–95%) , [20,21]. Thus, Lewisb status probably does not influence the outcome of H. pylori infection. Another explanation for this lack of correlation may be allelic variations of the babA2 gene. Genetic diversity of H. pylori is a well-known phenomenon, and the degree of variation for babA paralogues was investigated by Pride et al. . They reported conservation of the 5′ and 3′ regions and variability of the middle regions of the babA gene.
In all studies mentioned above, the same primers for typing babA2 genotype were used. Although the upstream primer is located in the 5′ conserved region, single base mutations or insertions may exist in this region and hinder effective amplification. The downstream primer is located in a region of estimated greater variability, which was intended to achieve discrimination between different BabA homologues (i.e. other outer membrane proteins), but could lead to underestimation of the prevalence of babA2 positive strains. Thereby, our current data are in seeming contradiction to results from Thoreson et al.  in Swedish patients, who found BabA positive strains more frequently in strains from DU. This might only reflect different patients selections; however, it is more likely that the differential methods applied lead to the diverging results, since Thoreson et al. analyzed the expression of BabA protein using biotinylated Lewisb oligosaccharides. We therefore further evaluated the binding abilities to Lewisb of all strains in our study by in vitro adhesion assay to compare the possible difference between the genotype and phenotype. Adhesion to Lewisb was detected in 28 of 145 babA2 negative patients in our study. Since the babA2 gene is necessary for Lewisb binding , the primers we used obviously did not amplify every existent babA2 gene. However, changing the location of the downstream primer or using degenerate primers lead to a notably increase in false positives, i.e. in amplification of babA homologues (data not shown), and was therefore resigned. The difference between genotype and adhesion was particularly notable in Swedish strains, where adhesion to Lewisb was more frequent (63%) than babA2 presence (45%). This may be due to the specific geographic variation of the circulating bacterial lineages, and indicates a tight phylogenetic relationship during geographic isolation of the strain population isolated from Swedish patients. On the other hand, we found that 34 of 139 babA2 positive strains did not bind to Lewisb antigen, and this phenomenon was particularly notable in POR, where 14 (45%) babA2 positive strains showed no binding to Lewisb. It is interesting to note that this is in agreement with data obtained from Spanish strains  and thus most likely reflects true lower Lewisb binding prevalence in southern European countries like POR (55%) compared to northern countries like SWE (84%).
These findings indicate that BabA expression can be regulated in individual strains, as also shown by Solnick et al. . In parallel with the regulation through CT dinucleotide repeats in the 5′ region of babA2 , regulation of protein expression has been also observed during changes of culturing conditions (data not shown). Also, recombination between homologues or duplicate genes would allow synthesis of functional BabA protein to readily be switched on or off, and thus provide a mechanism for adaptation to environmental conditions . Hennig et al.  could show that variation among expressed BabA proteins is mainly in the middle region detected as amino acids polymorphism. Additionally the same variation is detected for expression of BabA protein and Lewisb binding activity . While BabA might initially be important for colonization and induction of inflammation, the tight bacterial-host contact could turn into disadvantage during later stages of disease, when mucosal adaptation and responses are leading to a less hostile microenvironment. In such situation, lowering its adhesion properties may allow the bacterium to temporary detach from the mucosa and avoid toxic products and chemical stress. This, in turn, could favor survival of BabA expressing strains, leading to chronicity and more severe disease. Thus, BabA phenotype variability might reflect the variability in the behavior of individual H. pylori sub-strains within the population of bacteria over time. In summary, the current data confirm the results of previous studies in other European populations, suggesting that determination of babA2 in a subgroup of type1 strains can improve the differentiation between UD and GA. It seems that babA2 genotype might be a useful marker to identify the patients with a higher risk for UDs at least in some European countries. In specific geographic regions, allelic variation of the babA gene may lead to underestimation of the frequency of babA positive strains when using PCR based approaches and should be evaluated in further studies. Furthermore, our in vitro adhesion assay suggests that BabA expression can be regulated, and may be useful in populations where allelic variations are frequently observed.
The authors thank Nina Neumayer for excellent technical assistance. Sponsored by DFG 411/7–1 (Heisenberg Program to C.P.); Bayerischer Habilitationsförderpreis to M.G.; a grant from Astra Zeneca, Wedel, Germany; and Gastroenterology Foundation, Munich, Germany. The Swedish Medical Research Council (11218), the Swedish Cancer Society (4101-B00-03XAB).