INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Helicobacter pylori is a gram-negative spiral bacterium that colonizes the gastric mucosa and is able to persist over decades if the infection is not treated. The chronic infection often occurs without symptoms, but some individuals develop severe features of upper gastrointestinal diseases such as peptic ulcer disease (PUD), gastric adenocarcinoma, and mucosa-associated lymphoid tissue (MALT) lymphoma (1, 13). The vacuolating cytotoxin is one of the putative virulence factors of H. pylori that might lead to ulcerations. It induces massive formation of acidic vacuoles in the cytoplasm of gastric epithelial cells in vivo (14) as well as in vitro in primary epithelial cells (9) or in permanent cell lines. It has also been demonstrated that oral administration of purified VacA to mice causes injury of the gastric mucosa (14). The gene coding for the cytotoxin exhibits a mosaic of different alleles, which can be separately detected by PCR (2, 3, 6). For the N-terminal region of vacA, coding for the signal peptide, three different families of sequences, termed s1a, s1b, and s2, can be differentiated. The midregion of vacA is represented by two different families of alleles, termed m1 and m2 (2).

The aim of this study was (i) to perform vacA typing of H. pylori isolated from German patients in order to test the applicability of this method to isolates from a population not yet examined, (ii) to evaluate the association of vacA genotypes with peptic ulceration and nonulcer dyspepsia (NUD), and (iii) to assess the association of vacA genotypes with the presence of the pathogenicity marker cagA.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Patients. Independent clinical isolates of H. pylori were obtained from 158 adults who underwent gastroduodenoscopy. Of these, 106 patients (mean age, 51 years) presenting with duodenal ulcerations (PUD) with a minimum ulcer size of 5 mm have been enrolled in a multicenter study (7) including 28 centers all over Germany. Biopsies of a control group consisting of 52 patients (mean age, 47 years) with NUD were taken by two gastroenterologists. Strains from patients with gastric ulceration, gastric cancer, or MALT lymphoma or who had taken antimicrobial agents 4 weeks prior to endoscopy were not included in the study.

Isolation and culture conditions. During endoscopy one antrum and one corpus mucosal biopsy were obtained from each patient. Each biopsy specimen was placed in a transport medium (Portagerm Pylori; Biomerieux) and sent to the laboratory within 24 h. The specimens were ground with a pellet pestle, spread on solid-agar plates, and incubated under microaerobic conditions at 37°C for 2 to 5 days. Yeast extract-cysteine blood agar base (Difco), supplemented with 0.0005% hemin, 0.007% potassium hydroxide, 0.001% vitamin K, 10% horse serum, 10% washed human erythrocytes, vancomycin (10 mg/liter), trimethoprim (5 mg/liter), and nystatin (1 mg/liter) after autoclaving, was used as the growth medium. Bacteria were identified as H. pylori by standard criteria (12). The primary cultures grown from antrum and corpus biopsy specimens were transferred to Microbank cryovials (Mast Diagnostica) and stored frozen at 70°C. These are referred to as stock cultures.

PCR-based typing. The DNA sequences of the primer oligonucleotides used for PCR and sequencing and the sizes of the corresponding PCR products are listed in Table 1. For PCR analysis, bacteria from frozen stock cultures were grown for 48 h in brucella broth supplemented with 10% fetal calf serum at 37°C with shaking under microaerobic conditions. The cells were harvested by centrifugation and resuspended in distilled water. A 1-µl aliquot from a 100-fold dilution of this suspension was analyzed in a PCR mixture (50 ml) containing 1 U of Taq DNA polymerase and 25 pmol of each primer in a buffer system described previously (4). Amplification was done in 30 cycles consisting of 94, 50, and 72°C (for 1 minute each). The PCR products were electrophoretically separated on a 2.5% agarose gel and stained with ethidium bromide.

DNA sequence analysis of the vacA midregion. To determine the nucleotide sequence of the vacA midregion, a 785-bp fragment was amplified by using the two primers VA6-F and VA5-R, which bind to conserved regions flanking the midregion of the vacA gene (Table 1). The PCR products were purified by using the QIAquik PCR purification kit from Qiagen and sequenced with the PRISM ready reaction dye cycle sequencing kit (Applied Biosystems), including fluorescence-labeled dideoxynucleotides. The fluorescent reaction products were separated on a denaturing polyacrylamide gel and analyzed with an Applied Biosystems 373A automated DNA-sequencing machine. The DNA sequence was analyzed with the HUSAR software package provided by the German Cancer Research Center in Heidelberg. The detailed alignments were done with GAP, CLUSTAL, or BESTFIT software. The dendrogram was constructed with pc/gene software, version 6.8, from IntelliGenetics Inc.

Detection of cagA. The presence of the cagA gene was ascertained by PCR with primer oligonucleotides CAG-L and CAG-R (Table 1) under the cycling conditions described above.

Nucleotide sequence accession numbers. The vacA midregion of strains 003, 013, 026, 031, and 049 have been assigned EMBL database accession no. Y14740, Y14741, Y14742, Y14743, and Y14744, respectively.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Distribution of vacA alleles in patient groups. A total of 158 strains were investigated to determine their vacA genotypes. Of these, 106 were isolated from patients who suffered from duodenal ulceration (PUD) and 52 were isolated from patients with NUD. All three alleles of the vacA signal sequence could be detected and typed in all isolates as either s1a, s1b, or s2. Allele s1a was found in the PUD group with higher frequency (90%) than in the NUD group (58%). The allele s2, known to be associated with less-cytotoxic activity and with the absence of the pathogenicity-related gene cagA, was found in the NUD group at a much higher frequency (31%) than in the PUD group (4%). Both associations revealed statistical significance (P < 0.001 by the ² test). The midregion subtypes earlier defined as m1 and m2 (2, 3) were detectable in only 81.5% of the isolates, whereas in 18.5% no PCR product was obtained by using the primers specific for m1 and m2. The midregion subtype present in these strains was given the preliminary designation mx, which could be identified as a distinct subtype of m1 by further criteria (see below). With respect to this classification, the m1 and m2 alleles were equally distributed over the PUD and NUD groups. The data are summarized in Table 2.

Different vacA genotypes in strains from one patient. For each patient, the isolate from the gastric antrum and the isolate from the gastric corpus were analyzed separately. We had only one case in which H. pylori isolates with more than one vacA allele were obtained from a single patient. Strains 091A (isolated from the antrum) and 091C (isolated from the corpus) showed genotypes s1a m2 and s2 m2, respectively. This implied that the patient was infected with more than one strain. The results from this patient were not included in Table 2.

Further characterization of the midregion subtype mx. The 29 strains which gave no PCR product with the primers for typing m1 or m2 alleles were further investigated by DNA sequence analysis and by PCR with specifically designed primer oligonucleotides. For this purpose, the DNA sequences of 785-bp fragments including the divergent vacA midregions from five independent randomly selected untypeable isolates were determined. The analysis revealed six single-base-pair transitions within the binding site for the m1 antisense primer, VA3-R, whereas the binding site m1 sense primer, VA3-F, was completely conserved (Table 3). Binding sites for primers specific for the m2 allele were absent. In each of the five clinical isolates, which originated from patients living in different areas of Germany, the nucleotide exchanges were identical, with the exception that in isolate 026 an additional C-to-T transition was present (Table 3). In order to further investigate whether this variation was responsible for the insufficient amplification and to classify further mx strains, two new primers, VA3.1-R and VA3.2-R, were designed. The nucleotide sequence of primer VA3.1-R was identical with that of the altered binding site found in mx strains, and the primer VA3.2-R carried the sequence 5 bp upstream from the original annealing site (Tables 1 and 3). To determine whether these primers were suitable for investigation of the mx subtype, individual m1, m2, and mx strains were analyzed by PCR with primers VA3.1-R and VA3.2-R in combination with primer VA3-F (Fig. 1). This analysis revealed that primer VA3.1-R gave a strong amplification product with mx strains whereas primer VA3.2-R gave only weak amplification of the mx midregion sequence. Neither primer amplified the midregions of m1 or m2 strains, indicating that primer VA3.1-R, especially, could be useful to specifically detect the mx subtype. This was further confirmed by analysis of all 29 mx strains which could be successfully classified by this method (data not shown).

View larger version (23K): [in this window] [in a new window]   FIG. 1.   PCR analysis of the vacA midregion of defined m1, m2, and mx strains. Isolates 005C (m1), 004C (m2), 003C (mx), and 013C (mx) were analyzed with the four primer combinations which specifically detect the vacA subtypes m1 (VA3-F and -R) (lanes 1), m2 (VA4-F and -R) (lanes 2), and mx (VA3.1-R and VA3-F and VA3.2-R and VA3-F) (lanes 3 and 4, respectively). PCR products were separated on a 2.5% agarose gel and stained with ethidium bromide. The size of defined marker DNA fragments (1-kb DNA ladder; BRL) is indicated in base pairs to the right.

View larger version (12K): [in this window] [in a new window]   FIG. 2.   Dendrogram showing the relationships among vacA midregion sequences of subtypes m1, m2, and m1a. The midregion sequences from 11 German H. pylori isolates were aligned. The DNA regions analyzed correspond to nucleotides 2629 to 3078 in the vacA gene HP05676 deposited in the GenBank and EMBL databases, which served as a reference (6).

Relationship between vacA genotype and CagA status. The presence of the pathogenicity marker gene cagA was highly correlated with the vacA signal sequence type s1a (P < 0.001 by the ² test). This link between cagA and allele s1a was not exclusive, as indicated by the detection of six cagA-negative isolates exhibiting the vacA genotype s1a and, conversely, three cagA-positive isolates exhibiting the vacA genotype s2.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The present study revealed that the vacA allele s1a of H. pylori is strongly associated with PUD. On the other hand, the vacA allele s2 was predominantly found in isolates from patients with NUD and was very rare in strains isolated from ulcer patients. Therefore, the vacA allele s2 is not suspected to contribute to PUD.

It was reported in previous studies (2, 3, 6) that the observed cytotoxic activity was higher in s1a m1 than in s1a m2 strains and that cytotoxicity seemed to be completely absent in s2 m2 strains. It was proposed, as the most likely explanation for the high prevalence of the s1a genotype in the ulcer group, that the elevated toxicity of s1a m1 strains might contribute to the development of ulcerations. However, this explanation is not consistent with the findings of the present study, which indicate that, in contrast to the genotype of the signal sequence, the genotype of the midregion showed no correlation with the outcome of disease. If it is true that the ulcerative effect is exclusively dependent on cytotoxic activity, one would expect a higher frequency of m1 genotypes in the ulcer group and correspondingly more m2 genotypes in the NUD group. One explanation for these contradictory observations could be that the toxin assay with HeLa cells might not accurately reflect the in vivo effect occurring in the duodenal epithelial cells. It is supposed that the cytotoxin enters the cell by receptor-mediated endocytosis (8), and if this is true, the presence of specific receptor molecules on the surface of the target cell should influence the efficacy of the cytotoxin. Another explanation might be that factors other than VacA influence the development of ulcers. As shown before, and also demonstrated in this study, the vacA genotype s1a is strongly associated with the presence of the pathogenicity marker cagA (2, 3), whereas strains with vacA genotype s2 mostly lacked the cagA gene. This may imply that vacA genotype s1a is a marker for elevated virulence. Interestingly, the vacA midregion allele m2 was correlated with a negative cagA status (P < 0.001 by the ² test) (Table 5).

It was one of the objectives of this study to evaluate whether the vacA genotype might be a predictor of the duodenal ulcer risk of an infected person. However, our conclusions are limited by the fact that the NUD control group, in contrast to the PUD group, was heterogeneous and included patients with erosive gastritis as well as persons who might have suffered from ulcers before or might suffer from ulcers in the future. This potential selection bias could in part explain the relatively high prevalence of genotype s1a in the NUD group.

Nearly one-fifth of the H. pylori strains which were isolated harbored a vacA genotype that could not be classified by the proposed scheme (2, 3) and which we gave the preliminary name mx. The genotypes of all 29 of these isolates have been confirmed by allele-specific PCR, and 5 were additionally analyzed by DNA sequencing. Because the mx type investigated in this study highly resembled the midregion type m1, we would propose the term "m1a," analogous to the terms s1a and -b used for the signal sequence subtypes (2, 3). The results of the present study also provided evidence that all m1a isolates are of clonal origin and represent a distinct midregion genotype. To see such clear-cut differences from the results of vacA typing of strains of North American origin was surprising (2). To date, there is one other published study about H. pylori isolates from Japan, which reported that 84 of 87 clinical isolates were untypeable with respect to the midregion of the vacA gene (11). It would be of interest to evaluate in further studies if the m1a genotype is restricted exclusively to Western Europe or if it is also found in other geographic regions. We presume that geographical clustering of vacA genotypes in different populations exists. The primer we designed for detecting genotype m1a might be a useful tool in such studies.