Introduction

Helicobacter pylori is a spiral bacterium that chronically colonizes the human stomach, and is currently recognized to be the etiologic factor responsible for gastritis, gastroduodenal ulcer, gastric cancer, and mucosa-associated lymphoid tissue (MALT) lymphoma [1, 2]. Infection with H. pylori almost always results in chronic gastritis, but severe diseases such as peptic ulcer and gastric cancer occur in only a small proportion of infected patients, suggesting that clinical outcomes are determined by the interaction of bacterial virulence, host, and environmental factors [2, 3]. To date, several H. pylori virulence factors associated with severe clinical outcome have been reported, including vacA, babA, iceA, oipA, dupA, and, most notably, cagPAI [2, 4–9].

cagPAI is a cluster of about 30 genes spanning approximately 40 kb that have been acquired through horizontal transmission from unknown extraneous sources and integrated into the H. pylori chromosome (Fig. 1a) [6, 10]. cagPAI contains several genes encoding component proteins of the type IV secretion system (TFSS), a syringe-like structure responsible for the translocation of CagA protein and peptidoglycan from H. pylori into the host cell [6, 10–13]. Moreover, cagPAI is involved in the induction of many proinflammatory cytokines, including interleukin (IL)-8, from gastric epithelial cells [6, 14, 15]. Systematic mutagenesis studies have revealed that 17 out of 27 genes examined in cagPAI are essential for CagA translocation, while 14 genes are indispensable for the full induction of IL-8 [6, 14, 15], indicating the importance of cagPAI intactness in the pathogenesis of H. pylori-associated diseases. Therefore, it can be speculated that partial deletions within cagPAI are likely to affect bacterial virulence and, thus, clinical outcome.

Fig. 1
figure 1

cagPAI structure in Helicobacter pylori isolates from Vietnam. acagPAI consists of about 30 genes (from cagζ to cagA), several of which are homologous with vir genes of Agrobacterium tumefaciens. Relative locations of the probes used in this study are presented as small lines under each gene. bcagPAI status of 103 Vietnamese H. pylori strains and the internal deletion patterns in 12 strains. c The endpoints of deletion were elaborated in two representative strains, Del-61 and Del-146, by alignment with the sequence of H. pylori 26695

Full size image

H. pylori is categorized as cagPAI-negative or -positive, based mostly on the presence or absence of the cagA gene as a marker of the whole cagPAI stretch [6]. However, cagPAI is usually subject to internal deletions [16], suggesting that cagPAI-positive strains can be further divided into intact cagPAI (i.e., those with full-length cagPAI) and partial cagPAI groups (i.e., those lacking one to several genes). It is accepted that strains possessing cagPAI are more toxic and more associated with severe diseases than those lacking it [2, 3, 17]. Nevertheless, the virulence of H. pylori containing partial cagPAI and its association with clinical outcome has not been extensively studied. Moreover, there is currently no information about the cagPAI status of H. pylori from Vietnam. Therefore, the present study was performed to investigate the relationship between clinical outcome and the cagPAI status of H. pylori strains isolated from Vietnamese patients.

Materials and methods

Patients

A total of 103 Vietnamese patients (47 males and 56 females), aged 14 to 83 years (mean age, 45 years) were enrolled. Local ethical approval and written informed consent from all participants were obtained before the study. During gastroduodenoscopy, five biopsy samples were taken from each patient, including two from the antrum, two from the corpus, and one from the upper part of the lesser curvature. Twenty-five patients were endoscopically diagnosed as having peptic ulcer (13 with duodenal ulcers, six with gastric ulcers, and six with gastroduodenal ulcers) and 78 had chronic gastritis as determined by histological examination.

After endoscopy, blood samples from each patient were collected on the same day for the measurement of serum pepsinogen 1, pepsinogen 2, and gastrin.

Histology

Three biopsy specimens from the antrum, corpus, and upper part of the lesser curvature of each patient were examined by an experienced pathologist (T.U.), who was unaware of the characteristics of the H. pylori strains. For each biopsy specimen, the grades of neutrophil infiltration, mononuclear cell infiltration, atrophy, intestinal metaplasia, and H. pylori density were scored on the basis of the updated Sydney System (0, none; 1, mild; 2, moderate; 3, severe) [18].

H. pylori culture and genomic DNA extraction

Two biopsy specimens (one from the antrum and one from the corpus) were homogenized and inoculated onto Mueller Hinton II Agar medium (Becton Dickinson, Franklin Lakes, NJ, USA) supplemented with 7% horse blood without antibiotics. The culture plates were incubated for up to 10 days at 37°C under microaerophilic conditions (10% O2, 5% CO2, and 85% N2). Isolated strains were stored at −80°C in Brucella broth (Difco, Franklin Lakes, NJ, USA) containing 10% dimethylsulfoxide and 10% horse serum.

For DNA extraction, H. pylori was subcultured from the stock. Colonies on agar plates were harvested, and genomic DNA was extracted as described previously [19].

For in vitro experiments, we also used H. pylori strain TN2GF4 (here denoted TN2) and its isogenic mutants, ΔcagA, ΔcagE, ΔcagG, ΔcagPAI, ΔoipA, and ΔbabA, which were used in our previous studies [20–24]. The strain TN2 isogenic cagY mutant (TN2ΔcagY) was generated from the parental strain as described previously [25]. Briefly, genes upstream (hp525–526) and a gene downstream (hp528) of the cagY gene (hp527) were amplified and the corresponding polymerase chain reaction (PCR) products were cloned into pT7Blue vector (Novagen, Madison, WI, USA) to generate pT7hp525–526 and pT7hp528, respectively. A chloramphenicol (cm) cassette was inserted into pT7hp525–526, resulting in pT7hp525–526-cm. Next, the fragment hp525–526-cm was cut out from pT7hp525–526-cm and inserted into pT7hp528, resulting in the plasmid pT7hp525–526-cm-528. Finally, this plasmid was used to inactivate cagY by natural transformation. Inactivation of cagY was confirmed by both PCR and Southern blotting.

Detection of cagPAI genes

The presence or absence of 30 cagPAI genes was investigated by PCR using 32 sets of primers listed in Table 1. Unless stated otherwise, all primers were designed based on the nucleotide sequence of strain NCTC 11638 (GenBank accession numbers AF282852.1 and AF282853.1) using FastPCR software (downloaded from http://www.biocenter.helsinki.fi/bi/Programs/fastpcr.htm).

To avoid false-negative results of PCR due to variations in the primer annealing sites, dot-blotting was performed as described previously [26]. Briefly, 300 ng of sample DNA was mixed with denaturing buffer (0.8 N NaOH; 1.5 M NaCl) and transferred to a Hybond N+ membrane (Amersham Biosciences, Buckinghamshire, UK) using a 96-well Bio-Dot apparatus (Bio-Rad, Ivry-sur-Seine, France). DNA of the reference strain ATCC43504 and human DNA were used as positive and negative controls, respectively. The probes were generated from genomic DNA of ATCC43504 by PCR using the corresponding primer sets and purified with an Illustra GFX PCR DNA and Gel Band Purification Kit (Amersham Biosciences). With the use of the ECL Direct Nucleic Acid Labeling and Detection Systems (Amersham Biosciences), the probes were labeled with horseradish peroxidase, hybridized to the membranes at 42°C overnight, and, finally, exposed to Hyperfilm ECL.

A strain was defined as lacking a given gene if the results of PCR and dot-blotting were both negative, whereas the presence of a given gene required at least one positive result.

Nucleotide sequencing

The genomic regions of interest were initially amplified with primer set cagEmpty (Table 1) [27]. The amplicons were sequenced with a BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA) on an ABI Prism 310 Genetic Analyzer (Applied Biosystems). Sequence alignment was generated using the ClustalX program (http://clustalw.ddbj.nig.ac.jp/top-e.html).

Table 1 List of primers used in this study
Full size table

Analysis of H. pylori-induced IL-8 production in gastric cell lines

The experiments were performed twice independently, as described previously [8]. Briefly, the gastric epithelial cell line MKN45 was seeded into 24-well plates and grown overnight in RPMI 1640 medium supplemented with 10% FBS. H. pylori was harvested from the agar dishes and washed twice with PBS before being added to the culture wells with a bacterium-to-cell ratio of 50:1. The plates were incubated at 37°C for the indicated periods of time in an environment containing 5% CO2 and 95% air. The concentration of IL-8 in the cell culture supernatant was measured with the CXCL8/IL-8 ELISA Kit (R & D Systems, Minneapolis, MN, USA).

Analysis of H. pylori-induced hummingbird phenotype in AGS cells

The gastric epithelial cell line AGS was maintained and infected with clinical H. pylori strains as described above. After 36 h of co-culture, formation of the hummingbird phenotype was examined microscopically in five randomly chosen fields.

Quantification of H. pylori adhesion to gastric cell lines

Adhesion of H. pylori to gastric epithelial cells was measured by flow cytometry on a FACScalibur flow cytometer (Becton Dickinson, Franklin Lakes, NJ, USA), as described previously [28]. Briefly, H. pylori was added to culture wells containing MKN45 cells, as described above, and incubated for 3 h at 37°C. After vigorous washing three times with PBS to remove unbound bacteria, the cells were harvested and incubated with polyclonal rabbit anti-H. pylori antibody (DAKO, Glostrup, Denmark) (diluted 1:50) for 2 h at 4°C. After washing, the cells were incubated with fluorescein isothiocyanate (FITC)-conjugated goat anti-rabbit immunoglobulins (Invitrogen, Eugene, OR, USA) (diluted 1:200) for 2 h at 4°C. Finally, the cells were washed, resuspended in PBS, and immediately subjected to flow cytometry. The experiments for negative control samples were done exactly as described above, except that H. pylori was not used.

Flow cytometric data from 2 × 104 cells were collected and analyzed with WinMDI software (downloaded from http://facs.scripps.edu/software.html). Cellular debris, non-adherent bacteria, and cell clumps were excluded from analysis by gating. The results were expressed as mean fluorescence value ± standard error of the mean (SE) from four independent experiments.

Measurement of serum pepsinogen 1 (PG1), pepsinogen 2 (PG2), and gastrin

Serum PG1, PG2, and gastrin were quantified using ARCHITECT pepsinogen I.II (Abbott, Tokyo, Japan) and Gastrin RIA Kit (Kyowa, Tokyo, Japan), in accordance with the manufacturers’ instructions.

Statistical analysis

The quantitative data were expressed as mean ± SE. Fisher’s exact test, the Mann–Whitney rank sum test, independent t-test, one-way analysis of variance (ANOVA) test, and Spearman’s correlation were used. Differences at p < 0.05 were regarded as statistically significant.

Results

cagPAI status in H. pylori strains from Vietnam

Based on the results of PCR and dot-blotting, we found that 91 out of 103 strains had all of the cagPAI genes (regarded as intact cagPAI), 12 contained only part of cagPAI (regarded as partial cagPAI), and none were cagPAI-negative (Fig. 1b).

In the 12 strains possessing partial cagPAI, four patterns of deletion were observed: deletion from cagδ through cagA (five strains), deletion from cagδ to the left half of cagA (four strains), deletion from the right half of cagβ to the left half of cagA (two strains), and deletion from the right half of cagY through cagA (one strain) (Fig. 1b).

By sequencing, we clarified the endpoints of cagPAI deletion in two representative strains, Del-61 and Del-146. Alignment with the sequence of H. pylori strain 26695 (GenBank accession number AE000511) revealed that the deletion in strain Del-61 spanned the stretch from nucleotide 556291 (corresponding to nucleotide 2716 of cagY) to nucleotide 584478 (located in the intergenic region between hp0548 and hp0549). In strain Del-146, the deletion extended from nucleotide 550933 (corresponding to nucleotide 717 of cagβ) to nucleotide 578714 (corresponding to nucleotide 1794 of cagA) (Fig. 1c).

cagPAI status and clinical outcomes

Twenty-five (27.5%) of 91 patients infected with intact cagPAI strains developed peptic ulcer, whereas none of the 12 patients infected with partial cagPAI strains had such disease. This difference was statistically significant (p < 0.05) (Table 2).

Table 2 Association between cagPAI status and clinical outcomes
Full size table

We analyzed the relationship between cagPAI status and the histological scores in 78 gastritis patients. Except for mononuclear cell infiltration and H. pylori density in the antrum, other histological scores were significantly higher in patients infected with intact cagPAI strains than in those infected with partial cagPAI strains, irrespective of the biopsy site (p < 0.05) (Fig. 2). Intestinal metaplasia was rare, and was, therefore, excluded from the analysis.

Fig. 2
figure 2

cagPAI status and chronic gastritis. Generally, the histological scores in patients infected with intact cagPAI strains were higher than in patients infected with partial cagPAI strains, irrespective of the biopsy site (antrum, corpus, and upper lesser curvature). *p < 0.05, Mann–Whitney rank sum test; N.I neutrophil infiltration; M.I mononuclear cell infiltration; Atr atrophy; Hp.D H. pylori density

Full size image

Taken together, these data suggested that H. pylori strains containing partial cagPAI were associated with less severe disease and milder gastritis than those with intact cagPAI.

Association of serum PG1 and PG2 with cagPAI status

Patients infected with intact cagPAI strains had significantly higher serum concentrations of PG1 and PG2 than those infected with partial cagPAI strains (77.3 ± 5.5 vs. 54.6 ± 4.2 ng/ml and 19.7 ± 1.6 vs. 10.9 ± 1.0 ng/ml, p < 0.05), whereas there was no difference in serum gastrin between the two groups (143.4 ± 8.3 vs. 120.8 ± 14.5 pg/ml). Interestingly, we also observed correlations between PG1 level and neutrophil infiltration score in the corpus (p = 0.018) and between PG2 level and neutrophil infiltration score in both the antrum and the corpus (p = 0.023 and p < 0.001, respectively).

As expected, the PG1/PG2 ratio, which reflects gastric mucosal atrophy, was significantly lower in patients infected with intact cagPAI strains than in those infected with partial cagPAI strains (4.2 ± 0.14 vs. 5.1 ± 0.28, p < 0.05).

Analysis of H. pylori-induced IL-8 production and hummingbird phenotype in gastric cell lines

We randomly selected nine intact cagPAI strains and ten partial cagPAI strains and compared their abilities to induce IL-8 secretion and the hummingbird phenotype in gastric epithelial cells. As shown in Fig. 3a, b, the intact cagPAI strains always induced much higher production of IL-8 and a higher rate of the hummingbird phenotype than the partial cagPAI strains or mock at any time point (p < 0.01). In contrast, there was no significant change in hummingbird phenotype formation or IL-8 secretion from cells infected with partial cagPAI strains as compared to mock, indicating that the cagPAI of these strains was nonfunctional.

Fig. 3
figure 3

cagPAI status and H. pylori-induced IL-8 production and hummingbird phenotype in gastric epithelial cells. The intact cagPAI strains induced much higher production of IL-8 at 6 h, 12 h, and 24 h (a) and higher rate of the hummingbird phenotype at 36 h (b) compared to the partial cagPAI strains or mock, but there was no significant difference between the partial cagPAI strains and mock (a and b). *p < 0.01, independent t-test

Full size image

H. pylori adhesion to gastric epithelial cells is associated with cagPAI status

Because cagPAI encodes the type IV secretion system, which targets the α5β1 receptor of the host cell [29], we examined whether internal deletions within this locus affect the ability of H. pylori to bind to the gastric epithelium. Flow cytometry experiments with the 19 clinical isolates of H. pylori mentioned above showed that the intact cagPAI strains bound to gastric epithelial cells more strongly than the partial cagPAI strains (Fig. 4a). To confirm this observation, we repeated the experiments with H. pylori TN2 and its various mutant strains. We found that deletion of the cagA gene did not significantly affect the binding ability of H. pylori. In contrast, the adhesion of TN2ΔcagG to gastric epithelial cells was significantly lower than that of TN2, but was still higher in comparison with TN2ΔcagE, TN2ΔcagY, or TN2ΔcagPAI. Knock-out of the cagE or cagY gene reduced adhesion to the level of TN2ΔcagPAI. However, all of the mutant strains still retained considerable adhesion ability compared to the negative control, indicating the involvement of other factors in the binding activity of H. pylori. In fact, as expected, the deletion of oipA and babA, which are well-known adhesion factors, also decreased the H. pylori binding ability significantly (Fig. 4b).

Fig. 4
figure 4

cagPAI status and H. pylori adhesion to gastric epithelial cells. a In 19 clinical isolates selected randomly, the adhesion of partial cagPAI strains to gastric epithelial cells was significantly lower than that of intact cagPAI strains. *p < 0.05, Mann–Whitney rank sum test. b Except for TN2ΔA, other mutant strains showed decreased adhesion compared with the wild type. *, **, ***, ****p < 0.05 compared with TN2 wt, TN2ΔA, TN2ΔG, and TN2ΔE or TN2ΔY, respectively, one-way analysis of variance (ANOVA) and independent t-test

Full size image

Discussion

The aim of this study was to characterize the cagPAI structure of Vietnamese H. pylori isolates, which have never been investigated before. To overcome the weaknesses of previous studies examining only a limited number of cagPAI genes, we used up to 32 probes covering all 30 genes located in this locus. Moreover, the combination of PCR and dot-blotting allowed us to avoid any false-negative results of PCR due to variations in the primer annealing sites. We found that cagPAI was present in all of the Vietnamese H. pylori isolates, the majority of which (91/103) had a complete set of cagPAI genes, while the remaining 12 possessed only part of the locus, indicating that cagPAI was obviously not a uniform, highly conserved structure. Our results suggest that cagA should not be regarded as an absolute marker for intact cagPAI in Vietnam because the presence of cagA predicted an intact cagPAI status correctly in only 88% of cases (91/103). Moreover, deletion within the cag island spanned quite a long DNA stretch containing several genes, suggesting that cagPAI should be investigated using several sets of primers spanning the locus, rather than focusing on certain genes.

Consistent with previous reports [30, 31], we found that H. pylori strains with partial cagPAI were less likely to cause peptic ulcer diseases, and were associated with milder gastritis, indicating that these strains were less virulent than those possessing intact cagPAI. These findings were in agreement with the in vitro experiment results showing that partial cagPAI strains lacked almost all ability to induce IL-8 secretion and the hummingbird phenotype in gastric cell lines. Their adhesion to epithelial cells was also significantly decreased. The intactness of cagPAI is a prerequisite for assembly of the type IV secretion system, by which H. pylori sticks to the host cell surface, disrupts cell junctions by translocating CagA into the cytoplasm, and induces the secretion of inflammatory cytokines [6, 11, 12, 14, 29, 32]. These various events allow H. pylori to avoid mechanical clearance, escape the acidic environment of the stomach lumen, gain nutrients, and, finally, to survive, replicate, and effectively colonize the gastric mucosa [33–36]. In fact, our histological scoring data showed that the density of H. pylori strains with intact cagPAI in biopsy specimens was generally higher, suggesting that these strains had a better capacity to colonize the gastric mucosa, compared with those possessing partial cagPAI.

The adhesion of H. pylori to epithelial cells is reported to be mediated by several bacterial factors, especially outer membrane proteins such as BabA, OipA, and SabA [25, 37–39]. Our data support the role of oipA and babA in adhesion, and also indicate the significance of cagPAI in the adhesion of H. pylori to gastric epithelial cells. It has been reported that the pilus of H. pylori TFSS is able to bind to the integrin α5β1 receptor on gastric epithelial cells via the RGD motif of CagL protein [29]. Because TFSS is encoded by several genes in cagPAI, deletions inside this locus are likely to affect the TFSS structure, leading to a reduction in binding activity. As both cagE and cagY are essential for the formation of TFSS [14], the deletion of these genes abolishes the cagPAI-dependent binding activity, as our data showed. However, the deletion of cagG appeared to have only a minor effect on the adhesion ability of H. pylori. Interestingly, cagG has been reported to be unnecessary for the induction of IL-8 secretion from host cells [14]. This result, together with ours, suggests that the CagG protein may play only a supplementary role in the activity of the TFSS. However, it should be noted that these in vitro data may not actually reflect what happens in the in vivo condition because H. pylori can adapt well to the gastric environment. Further in vivo binding studies are required to address this issue.

The pepsinogens are digestive proenzymes that are specifically secreted by the gastric mucosa. Pepsinogens are divided into two distinct types, PG1 and PG2, which have different immunological and biochemical properties. PG1 is secreted by chief and mucous neck cells of the corpus [40, 41]. PG2 is produced also by these cells and others such as the cells in the cardiac and pyloric glands, as well as Brunner’s glands in the duodenum [40, 41]. The PG1/PG2 ratio has long been used as a non-invasive method for the evaluation of gastric mucosal atrophy [40, 41]. Consistent with the histological scores, the PG1/PG2 ratio in patients infected with intact cagPAI strains was lower, suggesting that the gastric mucosal atrophy in these patients was more advanced than in patients infected with partial cagPAI strains. Moreover, we found that the serum concentration of PG1 and PG2 in H. pylori-positive patients was associated with cagPAI status, and this is the first time that such a phenomenon has been reported. In in vitro experiments, H. pylori was shown to directly stimulate pepsinogen secretion from gastric cells independently of cagPAI [42], a finding that seemed to be contradictory to ours. However, it should be noted that H. pylori might be able to affect pepsinogen secretion indirectly by inducing gastric mucosal inflammation, which cannot be reflected in an in vitro experiment. Correlations between the serum concentration of PG1 and PG2 and neutrophil infiltration in the antrum and/or corpus were observed in our study, as well as in others [43, 44]. Therefore, it can be speculated that intact cagPAI strains, with their stronger ability to induce gastric mucosal inflammation, are likely to stimulate more secretion of PG1 and PG2 in vivo than partial cagPAI strains.

In conclusion, our study has found intact and partial cagPAI in about 88 and 12% of Vietnamese H. pylori, respectively. Partial cagPAI strains are biologically less virulent and, thus, less likely to cause a severe clinical outcome than intact cagPAI strains.