OUP user menu

Panmictic structure of Helicobacter pylori demonstrated by the comparative study of six genetic markers

Laurence Salaün, Céline Audibert, Geneviève Le Lay, Christophe Burucoa, Jean-Louis Fauchère, Bertrand Picard
DOI: http://dx.doi.org/10.1111/j.1574-6968.1998.tb12953.x 231-239 First published online: 1 April 1998

Abstract

We compared the classifications of strains obtained by analysis of several genetic markers to demonstrate the panmictic structure of Helicobacter pylori, previously suggested by the study of multilocus enzyme electrophoresis. A series of 39 strains, including 37 clinical isolates from patients with gastritis or ulcers from two regions of France, reference strain CIP 101260 and the Sydney strain (strain SS1), were used. They were studied by restriction fragment length polymorphism analysis of ribosomal DNA (ribotyping) using HindIII and HaeIII, by polymorphism analysis of the ureA-ureB and flaA genes by PCR-RFLP using HaeIII and MboI, by vacA genotyping and by the presence or absence of the cagA gene and of the insertion sequence IS605 detected by PCR. There was a high level of genetic polymorphism over the studied strains, with 38 ribotypes, 38 restriction profiles for the ureA-ureB gene, 19 restriction profiles for the flaA gene and five combinations of the signal and mid-region sequences of the vacA gene. Factorial analysis of correspondence and hierarchical clustering performed using each marker revealed that the different classifications of the strains were not correlated. This suggests there is much genetic recombination between strains and supports the hypothesis of a panmictic structure for the H. pylori species.

Key words
  • Helicobacter pylori
  • Population structure
  • Genetic diversity

1 Introduction

Helicobacter pylori is a pro-inflammatory gastrointestinal pathogen involved in the pathogenesis of chronic gastritis leading to peptic ulcer disease, gastric carcinoma or lymphoma [13]. Many efforts have been made to determine the molecular mechanisms of host/parasite interaction which may explain how H. pylori successfully colonizes gastric mucosa and escapes the host immune response. All the analyses of the genome of H. pylori have demonstrated a high level of genetic diversity in natural isolates of this pathogen [49]. However, relatively little is known about the population genetics of isolates classified as H. pylori, an area essential for understanding how the pathogen adapts to its particular niche.

Multilocus enzyme electrophoresis (MLEE) analysis [10] has suggested there has been much recombination in the H. pylori population. However, MLEE data should be interpreted with caution [11, 12] and drawing conclusions on the genetic structure of a bacterial population based on the study of a single marker is unwise [13]. The structure of a population can be assessed with more relevance by studying several DNA markers distributed over the entire genome and comparing the different classifications obtained; this method has already been used successfully for Escherichia coli population structure studies [13].

The present work is a comparison of the classifications of 39 strains by using six genetic markers. We studied ribosomal DNA polymorphism by the ribotyping method, and restriction polymorphism of the ureA-ureB and flaA genes by polymerase chain reaction-restriction fragment length polymorphis (PCR-RFLP). We determined the genotype of vacA and investigated whether the cagA gene and IS605 sequence were present. We also determined whether there was a correlation between the type of pathology and the place of the strains in the different classifications obtained.

2 Materials and methods

2.1 Bacterial strains

Thirty-seven strains of H. pylori from gastric biopsy samples of patients with gastritis or ulcers were collected from 1992 to 1997, in two geographic areas of France (Bretagne and Poitou). H. pylori CIP 101260 [14] (Institut Pasteur, Paris) and the Sydney strain SS1 [15] were included for reference purposes (Table 1). All strains were grown on Pylori agar (BioMérieux, Lyon, France). Cultures were incubated between 48 h and 10 days at 37°C under microaerobic conditions (Campypak Plus, Becton Dickinson, Cockeysville, MD). Cultures were stored in 10% v/v glycerol in trypticase soy broth (AES Laboratoire, Combourg, France) at −70°C.

View this table:
Table 1

Origins and characteristics of the 39 H. pylori strains studied

No.StrainaAssociated
diseaseb,c
RFLP rDNAPCR-RFLP
ureA-ureB
PCR-RFLP
flaA
vacA typecagA
detection
IS605
detection
HindIIIHaeIIIbHaeIIIMboIHaeIIIMboIMid-region typebSignal sequence type
1REF1GRI1RA1UH1UM1FH1FM1m1s1b+
2B30GRI2n.d.UH2UM2FH2FM2n.d.s1a+
3B33URI3RA2UH3UM3FH2FM3m1s1+
4B46GRI4RA3UH4UM3FH2FM4m1s1a++
5B53URI5RA4UH3UM4FH2FM5m2s1a+
6B81URI6n.d.UH5UM5FH2FM6m2s1a+
7B101URI4RA5UH6UM6FH2FM7m1s1a+
8B102URI7RA6UH7UM3FH2FM8n.d.s1a+
9B104URI8RA7UH2UM3FH2FM9m2s2
10B109URI9RA8UH8UM7FH2FM1m1s1a+
11B110URI10RA9UH9UM3FH2FM10m1s1b+
12P1GRI8RA10UH10UM8FH3FM1m1s1b++
13P5GRI1RA11UH11UM9FH2FM11m1s1b+
14P6GRI8RA12UH12UM10FH2FM1m2s2+
15P7GRI8RA13UH13UM11FH2FM2m1s1a
16P14URI8RA14UH14UM12FH2FM12m1s1b+
17P15GRI8RA15UH7UM4FH2FM8m1s1a++
18P17GRI1RA16UH15UM13FH2FM13m2s2+
19P19GRI11RA17UH16UM4FH2FM2m2s2
20P20GRI12RA18UH5UM4FH2FM13m1s1a++
21P24URI13RA19UH17UM14FH2FM3m1s1b+
22P25GRI4RA20UH18UM1FH2FM1m2s2
23P26GRI14n.d.UH19UM1FH2FM1n.d.s1b++
24P34GRI15n.d.UH3UM3FH2FM2m1s1+
25P41GRI16RA21UH20UM15FH2FM2m2s2
26P45GRI17RA22UH1UM3FH2FM6m1s1b+
27P48GRI18RA23UH7UM16FH2FM1m2s1b+
28P54URI13n.d.UH21UM3FH2FM1m1s1b++
29P55GRI19RA24UH22UM3FH2FM1m2s1a++
30P91GRI8n.d.UH23UM17FH3FM13m2s1a+
31P98GRI20RA25UH11UM4FH2FM1m1s1b+
32P122GRI4RA26UH2UM4FH2FM1m2s1+
33P123URI21RA27UH24UM18FH2FM14m2s2+
34P124URI22RA28UH23UM19FH4FM15m2s1b
35P127URI23n.d.UH5UM3FH2FM8m1s1b++
36P130n.d.RI1n.d.UH23UM3FH2FM8m2s1b+
37P133URI8n.d.UH3UM2FH2FM16m1s1a+
38P136GRI24n.d.UH25UM4FH2FM13m1s1b++
39SYDNURI25n.d.UH26UM20FH2FM1m2s2+
  • aREF1 is the H. pylori reference strain CIP 101260 [14]; SYDN is the Sydney strain SS1 [15]; the geographic origin of the H. pylori strains is indicated as follows: B=Bretagne, P=Poitou.

  • bn.d.: not determined.

  • cG: gastritis; U: ulcer.

2.2 RFLP analysis of rDNA (ribotyping)

Bacterial DNA was isolated as previously described [6]. DNA (3 μg) was digested with HindIII and HaeIII (Boehringer Mannheim, Meylan, France) according to the manufacturer's instructions, transferred to Hybond N+ nylon membrane (Amersham, Les Ulis, France) by Southern blotting and probed with a chemiluminescent ribosomal probe, as previously described [16].

2.3 ureA-ureB RFLP patterns and flaA RFLP patterns

The oligonucleotides used for the PCR amplification of specific segments were: 5′-AGGAGAATGAGATGA-3′ and 5′-ACTTTATTGGCTGGT-3′ for ureA-ureB[8]; and 5′-ATGGCTTTTCAGGTCAATAC-3′ and 5′-GCTTAAGATATTTTGTTGAACG-3′ for flaA[17]. 10 ng of phenol-extracted DNA was added to 100 μl of reaction mixture containing PCR buffer (Eurobio, Les Ulis, France), 3 mM MgCl2, 0.25 mM of each deoxynucleotide and 20 pmol of each primer. Taq polymerase (2.5 U; Eurobio, Les Ulis, France) was added, and the reaction mixture was overlaid with 60 μl of mineral oil. PCR was performed with a Techne PHC-3 thermal cycler. The amplification cycle consisted of an initial denaturation at 95°C for 5 min followed by denaturation at 94°C for 1 min, primer annealing at 50°C (ureA-ureB) or 55°C (flaA) for 1 min, and extension at 72°C for 2 min. The final cycle included extension for 5 min at 72°C. Samples were amplified through 30 consecutive cycles [18]. Aliquots (10 μl) of amplification products were subjected to electrophoresis in a 1% agarose gel (w/v), stained with ethidium bromide and viewed with a UV transilluminator. DNA (90 μl) was precipitated by adding 1/10 volume of 3 M sodium acetate and 2 volumes of ethanol 99.9% (v/v). The samples were incubated at −20°C for 2 h and centrifuged at 12 000×g for 30 min. The pellet was dried and dissolved in 25 μl of 6 mM Tris-HCl, pH 7.5/6 mM NaCl/0.1 mM EDTA. 10 μl of the concentrated DNA solution (1–5 μg) was incubated with 10 U of HaeIII or MboI restriction enzyme (Boehringer Mannheim, Meylan, France) for 16 h, and subjected to electrophoresis in a 2% agarose gel (w/v) as described above.

2.4 vacA genotyping

The vacA signal sequence type and mid-region nucleotide sequence type of each isolate were determined by PCR with extracted genomic DNA using the primers described previously [7]. PCR was performed as described for PCR-RFLP with annealing at 52°C to amplify the mid-region and at 55°C to amplify the signal sequence of vacA. Amplification products were digested, subjected to electrophoresis and detected as described for PCR-RFLP.

2.5 Detection of cagA gene and of IS605

PCR amplification of cagA and of IS605 was performed using the previously described primers 5′-GATAACAGGCAAGCTTTTGAGG-3′ and 5′-CTGCAAAAGATTGTTTGCGAGA-3′ for cagA[19], and 5′-CGCCTTGATCGTTTCAGGATTAGC-3′ and 5′-CAACCAACCGAAGCAAGCATAATC-3′ for IS605 (C. Burucoa, personal communication). PCR was carried out as described for vacA genotyping with annealing at 57°C. Amplification products were detected as described above.

2.6 Statistical analysis

The typing data were summarized as six two-way tables: (i) ribotyping with HindIII, (ii) ureA-ureB polymorphism with HaeIII, (iii) ureA-ureB polymorphism with MboI, (iv) flaA polymorphism with HaeIII, (v) flaA polymorphism with MboI and (vi) vacA genotyping and detection of cagA and IS605. Each table had 39 rows, one for each strain, and a column for each rDNA fragment (for ribotyping) or DNA fragment (for ureA-ureB and flaA RFLP patterns), vacA genotype and detection of the cagA gene and IS605. The presence or absence of each character was binarily coded, present=2 or absent=1. Factorial analysis of correspondence (FAC) [2022] and hierarchical clustering [23] were performed from each table and the results compared. The analysis was performed with SPAD.N software (CISIA, Saint-Mandé, France) on a PC.

3 Results

3.1 Ribotyping

The restriction endonucleases HindIII and HaeIII were used to digest H. pylori chromosomal DNA. HindIII digestion resulted in 25 pattern types (RI1–RI25) (Table 1). The RI8 pattern was shared by eight strains (20.5%). The RFLP patterns consisted of 2–5 bands from 1400 bp to 9500 bp in size, except for strains P41 and P98 which had 13 and 9 bands respectively. The DNA samples of 11 strains (28%) were not cleaved by HaeIII and there were 28 pattern types among the remaining 28 strains (RA1–RA28). These patterns consisted of 4–10 bands from 890 bp to 9500 bp in size. A total of 38 ribotypes, resulting from the combination of HindIII and HaeIII RFLP patterns, were identified among the strains (Table 1).

3.2 ureA-ureB and flaA RFLP patterns

The 2.4-kb ureA-ureb segment was successfully amplified from the 39 H. pylori strains. HaeIII restriction analysis of the PCR product obtained for each strain yielded profiles of 3–6 fragments of 140–2180 bp in size. A total of 24 fragments were detected from the strains. There were 26 distinct pattern types (UH1–UH26) in the 39 strains (Table 1). MboI restriction analysis of the PCR product yielded patterns of 4–8 fragments between 110 bp and 890 bp in size. Sixteen fragments of various sizes were detected and there were 20 distinct pattern types (UM1–UM20) among the 39 strains (Table 1). A total of 38 pattern types, resulting from the combination of HaeIII and MboI RFLP patterns, were identified among the strains (Table 1).

The 1.5-kb flaA segment was amplified from all 39 H. pylori strains. HaeIII restriction analysis of the PCR product yielded patterns of two or three fragments with sizes between 400 bp and 890 bp. Six different size fragments were detected and four distinct pattern types (FH1–FH4) were identified among the strains (Table 1). MboI restriction analysis of the PCR product yielded patterns of 2–5 fragments with sizes between 200 bp and 890 bp. Eleven fragments were detected and 16 pattern types (FM1–FM16) were identified (Table 1). Twelve isolates (30.5%) had the FM1 pattern whereas 10 isolates (25.5%) had a unique pattern. A total of 19 pattern types, resulting from the combination of HaeIII and MboI RFLP patterns, were identified among the strains (Table 1).

3.3 vacA genotyping

The distribution of vacA m1 and m2 alleles in the 39 H. pylori isolates was determined (Table 1). Twenty (51.5%) of the 36 strains (92.5%) with DNA amplified by one of the two primer sets were of subtype m1 and 16 (41%) were of subtype m2. There was no amplification for the remaining three strains (7.5%).

We differentiated between type s1 and s2 vacA signal sequences. Thirty-one (79.5%) strains had s1 and eight (20.5%) strains s2 sequences (Table 1). The s1a and s1b subtypes were differentiated and the presence of the type s2 sequence was demonstrated for the nine strains. Thirteen type s1 strains (33.5%) were classified as subtype s1a and 15 (38.5%) as subtype s1b. The subtypes s1a or s1b were not determined for three strains (7.5%).

Five combinations of the vacA regions were identified among the 39 strains, the most common being m1s1b, in 11 strains (28%).

3.4 Detection of the cagA gene and of IS605

cagA was detected in 27 of the 39 H. pylori isolates (69%) by PCR (Table 1).

IS605 was detected in 15 of the H. pylori isolates (38.5%) (Table 1). PCR amplified similar DNA fragments of the expected size in strains containing IS605.

3.5 Statistical analysis

FAC and hierarchical clustering analysis were performed for each table and for each genetic marker. The classifications of the strains derived from these analysis were compared. No correlation was identified from these comparisons. The dendrogram based on the FAC for virulence factors typing data (Fig. 1A), at a genetic distance of 0.10, contains four virulence groups (I–IV). When these data are transferred to the dendrograms for ureA-ureB and flaA PCR-RFLP data (Fig. 1B) and HindIII RFLP rDNA data (data not shown), the virulence groups are randomly distributed on the two dendrograms. There were no correlations when the groupings from the dendrogram derived from the ribotyping data were compared with those obtained from ureA-ureB or flaA data (data not shown). There was also no correlation between pathology and either marker when epidemiological data were compared to the classifications of strains on the dendrograms.

Figure 1

Dendrograms based on the FAC using (A) vacA genotyping, cagA and IS605 detection and (B) ureA-ureB and flaA PCR-RFLP data. aThe four groups (I, II, III and IV) delineated by the dendrogram of the virulence factors in A (see Section 3) are indicated on the dendrogram in B.

4 Discussion

Many studies have reported a high level of diversity in the H. pylori species, but only two works have studied the genetic structure of this pathogen, both by MLEE [10, 24]. A panmictic structure for the H. pylori population was suggested by Go et al. [10]. However, drawing conclusions about the structure of a population based on the study of a single marker can be misleading. Moreover, the MLEE method underestimates polymorphism because the net charge calculated does not always correlate well with electrophoretic mobility [25] and most enzyme polymorphisms are selectively neutral, or nearly so. Therefore, we analyzed several genes and compared the classifications obtained to assess the genetic structure of H. pylori. This method has already been used to assess the structure of E. coli populations [13] and the high level of correlation between the classifications obtained with this species demonstrated its clonal structure as previously demonstrated by Selander et al. [26] on the basis of MLEE. Conversely, in Pseudomonas aeruginosa species, the lack of correlation between the classifications obtained has demonstrated a high rate of recombination [27].

We studied the polymorphism of total DNA using the ribotyping method. Ribotyping seems to be a highly sensitive method for H. pylori strain identification [28, 29] even if only two copies of the rDNA operon are present [30]. We identified 38 ribotypes among the 39 strains. This high degree of diversity in rRNA gene patterns and the lack of a significant association between ribotype and clinical symptoms are consistent with data published before [9, 29].

The classifications obtained by PCR-RFLP analysis of the ureA-ureB and flaA genes did not correlate between each other, nor with ribotype or with vacA, cagA or IS605 markers.

The polymorphism of the pathogenicity determinant genes also supports the existence of an extensive recombination structure in the H. pylori species, because the classification of strains based on the polymorphism of the virulence genes was not correlated with ribotype and ureA-ureB or flaA gene classifications (Fig. 1). However, we found a strong association between the presence of cagA and vacA signal sequence type s1 (χ2=9.22, P<0.01), as previously noted by Atherton et al. [7]. Nevertheless, neither a particular vacA genotype nor the presence of cagA was correlated with one of the clinical traits studied (gastritis or ulcer). Only 62.5% of the strains isolated from ulcers were cagA+. The cagA gene is part of a 40-kb pathogenicity island (PAI) (the cag region) which can also contain the insertion sequence IS605. This sequence can cause large chromosome rearrangements that split the cag region in half [31]. The presence of IS elements may also cause rearrangements leading to partial or total deletions of the PAI [32]. Therefore, the presence of IS605 does not necessarily imply the presence of the cagA gene. This was demonstrated in our study because cagA and IS605 detections were not significantly associated. These particular genetic elements were previously demonstrated to be able to spread among the bacterial populations of other distantly related species by horizontal transfer [33]. IS605 is similar to an IS element found in the PAIs of Dichelobacter nodosus[33] and five of the 30 open reading frames within the cagA region have sequences similar to those of the plt gene of Bordetella pertussis and the vir gene of Agrobacterium tumefaciens[32, 34]. Moreover, the observations that many H. pylori isolates contain plasmids [4, 5], that some have phages [35] and that genomic rearrangements may occur [36] support the hypothesis of extensive recombination within this species. The mosaic structure of vacA is also consistent with this hypothesis. Several bacterial genes with similar mosaic structures have been described, including IgA proteases from Neisseria gonorrhoeae and penicillin binding proteins from N. meningitidis[7]. Such structures arise by horizontal transfer of DNA in species known to be naturally competent, a property shared with H. pylori[37]. These genetic transfers may occur when individuals are simultaneously infected with multiple H. pylori strains, as suggested by Atherton et al. [7]. However, even if transient multiple infection is more common than previously recognized, even in developed countries [38], H. pylori strains grow at many discrete foci in the gastric mucosa, with relatively little mixing [39], and so the opportunity for DNA transfer between different strains of H. pylori may be limited. Heterogeneity of the H. pylori species could give a selective advantage to this pathogen, allowing it to escape the defences of the host.

Acknowledgements

This study was financially supported by the Ministère de l'Education Nationale, de l'Enseignement Supérieur et de la Recherche, by the Université de Poitiers and by the Conseil Général du Poitou-Charentes.

References

View Abstract