PCR products amplified from the reference strains 26695, HPAG1 and J99 were identical to the published sequences sizes. Of the locus VNTR-2576 and VNTR-614, the PCR products sequencing were consistence with our electrophoresis results. The exact number of tandem repeats at each locus could be determined from the sizes of the PCR products.

In this study, 30 VNTR loci were candidated from the H. pylori database. And we finally identified 12 VNTR loci using analysis, which were also distributed throughout the H. pylori genome (Table 1). There's no variation in the other 18 loci, which were removed in the following study. The variation in repeat numbers is divergence at the 12 VNTR loci. The main characteristics of the 12 VNTR loci are listed in Table 2, including the diversity index of each locus.

A MLVA system to the molecular typing of H. pylori strains has been developed. On the basis of the 12 VNTR loci, the profiles of each isolate were obtained (Figure 1). The clinical H. pylori strains were divided into 127 MTs, which has not been described previously. According to cluster analysis, most strains from the same focus presented with the same or similar MTs (Figure 1). In addition, strains from the same focus were dispersed in the cluster tree. As shown in Figure 1, the 86.7% (13/15) of the Tokyo isolates had very similar MTs and could be clustered into Group A. One of the remaining Tokyo isolates belonged to the Group C, and the others were scattered distribution. Of the Southern and Eastern Chinese isolates, 74.4% (43/32) were clustered into group B, including B₁, B₂ and B₃ subgroups, and the rest strains were related to Group A, C and D. Of the isolates from Northern China, 60.7% were clustered into two major branches, groups C₁ (37.5%, 21/56) and C₂ (23.2%, 13/56), and other strains were scattered. Of the Western China isolates, 86.0% (37/43) were clustered into group D. The strains Tibet 1, 14, 23 and 43 were related to Group A, Tibet 37 and Tibet 35 were related to Group B₂ and C₂.

A minimum spanning tree was constructed on the basis of strains from different ethnic groups: 43 Tibetan, 33 Mongolian, 15 Yamato as well as 27 Han (Figure 2). There was a tendency to cluster into four main subgroups. However, there're still some exceptions, such as the Hangzhou-12 and 21, of Han strains (associated with gastritis and peptic ulcer), were related to the Tibetan strains group. Tibetan strains 1 and 43 (gastritis), were related to the Mongolian group, and Mongolian 16, (gastric cancer), was related to the Japanese group.

Among the 202 samples, 14.9%, 55.9%, 25.2% and 4.0% of patients presented with non-ulcer dyspepsia (NUD), gastritis (G), peptic ulcer (PU) and gastric cancer (GC), respectively. And in our study there's no significant relationship between the H. pylori MTs and the related diseases.

A total of 202 H. pylori strains were included in this study and the background information of the strains is listed in Table 3. The 187 clinical strains were isolated from various regions of China during 1998 and 2010; an additional 15 strains were presented as a gift by Institute of Medical Science University of Tokyo Japan in 2008. Patients ranged from 12 to 75 years old (mean age 44 years). All the patients reporting the symptoms of gastritis (G), peptic ulcer (PU) or gastric cancer (GC) underwent upper gastroendoscopy for both visual examination and biopsy collection. The strains were isolated from gastric biopsy gastrointestinal endoscopy of selected patients, who had not received non-steroidal anti-inflammatory drugs, proton pump inhibitors or other antibiotics during the last 2 months, revealed that out of 202 patients, 172 had either G, DU or GC and 30 had non-ulcer dyspepsia (NUD). Written consent was taken from all the patients before collection of the biopsy. The study was approved by the ethics review board at Third Military Medical University, and informed consent was obtained from all patients before participation.

Bacteria were separated and cultured in Skirrow medium with 5% fresh sheep blood at 37°C for 24 h~72 h in a micro-aerobic environment. H. pylori genomes were extracted using genomic DNA isolation kits (Omega Biotek Inc). Culture and identification of H. pylori were done by appropriate biochemical tests and amplification of 16S rDNA using species-specific primers

VNTR loci were selected from the MLVA database http://minisatellites.u-psud.fr/ASPSamp/base_ms/bact.php by estimating the size of PCR products on agarose gels. The repeat sequence of loci ≥ 10 bp, consistency of repeat unit ≥ 90% and a minimum of two alleles in three reference strains of H. pylori (26695, HPAG1, J99) were selected for this research. The locations, copy numbers, sizes of the loci and the gene(s) involved are also listed in Table 1.

A PCR reaction mixture (30 ml) containing 10 ng of DNA template, 0.5 mM of each primer, 1 unit of Taq DNA polymerase, 200 mM of dNTPs and 10 × PCR buffer (500 mM KCl, 100 mM TrisHCl (pH 8.3) 25 mM MgCl₂) was utilized. Amplification was carried out in a DNA thermocycler (MJ Research PTC-225) with denaturation at 94°C for 8 min, followed by 30 cycles of denaturation at 94°C for 45 s, annealing at 52°C for 45 s and elongation at 72°C for 1 min 26. A 10-min elongation at 72°C was performed after the last cycle to ensure complete extension of the amplicons.

Five μl of the PCR products were run on standard 3% agarose gels in 0.56TBE buffer at 8-10 V/cm. Gel lengths of 10 to 40 cm were used according to PCR product size and repeat unit size. Strains in which alleles had been precisely measured by re-sequencing or by direct comparison with a sequenced reference strain were used (In this study DNA from 26695, HPAG1 and J99 were used for this purpose). Multiple interspersed negative controls containing no DNA were included each time PCR was performed.

PCR products of 202 strains on VNTR-2576 and VNTR-614 sites were sequenced directly with a Taq Dye Deoxy Terminator Cycle Sequencing Kit on an ABI 377 sequencer (Applied Biosystems).

The number of repeat units in 12 VNTR loci were analyzed and inputted into BioNumerics version 5.1 software (Applied-Maths, Sint-Martens-Latem, Belgium), and gel images were obtained using the BioNumerics software package version 6.0 (Applied-Maths, Sint-Martens-Latem, Belgium) or using UVB gel image analysis. The number of repeat units in each locus was deduced by the amplicon size, flanking sequence length and repeat unit size. Data from agarose gel electrophoresis and UVB gel image analysis, obtained by capillary electrophoresis machines, were imported into BioNumerics by creating a virtual gel image. Gel image data were converted into characteristics data sets. Cluster analysis of Neighbor-joining tree (N-J) was carried out using the categorical similarity coefficient and the Ward method. A minimum spanning tree was inferred using characteristic data from cluster analysis. The polymorphism of each locus was represented by Nei's diversity index 27, calculated as DI = 1-∑(allelic frequency)².

Twenty clinical strain genomes from China and Japan were amplified and multiple DNA samples from each strain yielded PCR products with identical sizes at all loci. Chongqing26 and Tibet36 each yielded no product at one locus, possibly because of mutations or poor quality DNA samples.

The stabilities of the VNTR loci were investigated in a long-term experiment in which the 20 test H. pylori isolates used were sub-cultured into fresh Skirrow medium 30 times by serial passages at two or three day intervals. The DNA from the strains cultivated in each passage was extracted and subjected to MLVA analysis. The results of the VNTR analysis demonstrated no difference in tandem repeat numbers (data not shown).