Genotyping for cagA, vacA, iceA, and IS605

From each pool, chromosomal DNA was prepared as described previously [24]. DNA concentrations were measured by fluorometry (Dynaquant 200; Hoefer Pharmacia Biotech, Piscataway, NJ). The cagA [11], vacA [8], iceA [9], and IS605 [6] status for each population was determined by PCR and analyzed by gel electrophoresis, as described previously. In addition, the results of cagA, vacA, and iceA genotyping were confirmed by PCR followed by line probe assay [9]. For IS605, primers were complementary to orfA, homologous to the transposase of IS200, as described elsewhere [6]. Ten nanograms of DNA template was used for each reaction. Results were analyzed by electrophoresis in 1% agarose gels (Eastman Kodak, Rochester, NY).

RAPD-PCR

Forty nanograms of DNA template was used for RAPD-PCR with primers 1254, 1281, 8635, 9355, and d11344, as described elsewhere [25]. Results were analyzed by electrophoresis on 1% agarose gels. RAPD photographs obtained with primers 1254, 9355, and d11344 were scanned with a densitoscanner (Scanjet 4C; Hewlett-Packard, Bristol, UK), and the images were stored as TIFF (Tagged Image File Format) files with Deskscan II version 2.3 software (Hewlett-Packard).

Amplified fragment length polymorphism (AFLP)

AFLP analysis was done as described elsewhere [26, 27], with the following modifications. In a final volume of 20 μL, 20 ng of purified chromosomal DNA was digested with 1 U of EcoRI (Pharmacia, Uppsala, Sweden) and 1 U of MseI (New England Biolabs, Beverly, MA). The final volume then was increased to 30 μL by addition of 50 pmol each of the EcoRI adaptor and MseI adaptor, 1.2 U of T4 DNA ligase (Pharmacia), 1 mM ATP, and ligase buffer. The adaptors were allowed to ligate to the restriction fragments for 16 h at 16°C, after which the sample was diluted with distilled water in a final volume of 500 μL. A Texas Red fluorescent-labeled EcoO primer (5′-AGACTGCGTACCAATTC-3′; Isogen Bioscience, Maarssen, The Netherlands) was used for DNA amplification in 10 μL of a reaction mixture containing 0.3 ng of template DNA, 20 ng of labeled EcoO primer, 1 μL of 2 mM dNTP, 60 ng of unlabeled MseO primer (5′-GACGATGAGTCCTGAG-3′), 1 U of Taq polymerase (Perkin-Elmer, Norwalk, CT) in 10 mM Tris-HCl (pH 8.3), 50 mM KCl, and 1.5 mM MgCl₂. Amplification was done in a GeneAmp PCR system 9700 thermal cycler (Perkin-Elmer) with 35 cycles: denaturation (30 s at 94°C), annealing (30 s at 65–56°C), and DNA molecule extension (60 s at 72°C). In the first 12 cycles, the annealing temperature was lowered 0.7°C per cycle. After completion of the cycle program, 3 μL of loading buffer (Amersham Life Science, Cleveland) was added to the reaction mixtures, after which 2.5 μL of each sample was analyzed according to the manufacturer's instructions in a Vistra 725 automated DNA sequencer (Amersham Life Science). Fluorescent AFLP images were stored as TIFF files with the Vistra 2 TIFF software (Amersham).

Densitometric scanning and data processing

Computer-assisted analysis of the digitized RAPD images and the AFLP images was done by means of the GelCompar 4.0 software (Applied Maths, Kortrijk, Belgium). The obtained images were normalized for lane-to-lane differences in mobility by the alignment of patterns obtained on multiple specimens of standard DNA size markers (Amersham) included at regular intervals in each gel, and subsequently the background was subtracted by use of a curve-fitting algorithm. The levels of similarity of these digitized patterns then were calculated with the Pearson product moment correlation coefficient (r) of the GelCompar program, and the UPGMA method (unweighted-pair group method using arithmetic averages) was used to cluster the obtained patterns [28].

Sequencing of recA and IS605 PCR products

From each of the 26 pools, a 246-bp segment of H. pylori recA was amplified, as described elsewhere [29]. From each of the pools possessing IS605, a 371-bp segment from the IS200 homologous region also was PCR-amplified, as described elsewhere [6]. From each PCR run, 10 μL of product was analyzed by gel electrophoresis to exclude the presence of additional bands. The PCR product was purified by use of a Qiaquick PCR purification kit (Qiagen, Chatsworth, CA), and the DNA sequence was determined on both strands by use of an automated sequencer (ABI 377; Applied Biosystems, Foster City, CA). For the recA products from the strains from patient 3, AciI (New England Biolabs) digestion was done. Interstrain comparisons of 206-bp recA and 323-bp IS200 regions were made by use of the GCG Pileup program (Genetics Computer Group, Madison, WI) [30]. Statistical confidence in the trees was tested with 100 bootstrap replicates as implemented by PAUP (Smithsonian Institute, Washington, DC).

Southern hybridization

Two micrograms of DNA derived from IS605-positive isolates was digested with HindIII and electrophoresed on a 1.2% Tris-acetate/EDTA agarose gel (0.04 M Tris-acetate, 0.001 M EDTA). DNA was transferred to a nylon membrane by capillary transfer [31]. The probe for IS200 was made by PCR, as described elsewhere [6], and labeled by nick translation and hybridization by means of the Renaissance chemiluminescent kit (DuPont NEN Research Products, Boston), following the manufacturer's instructions.

Lewis phenotype analysis

Bacteria were harvested from 2 TSA plates after 48 h of growth for ELISA analysis of Lewis-x and -y expression, as described elsewhere [32]. In brief, microtiter plates were coated with whole H. pylori cells and incubated with mouse monoclonal antibodies to Lewis-x or -y, goat anti-mouse antibodies coupled to alkaline phosphatase, and p-nitrophenylphosphate as substrate. The results were expressed as optical density (OD) at 410 nm × 1000 U for H. pylori coating concentrations corresponding to 1 μg of protein per well. In addition to controls without primary or secondary antibodies, OD values of wells coated with equal concentrations of Escherichia coli HB101 cells were subtracted from the values of the wells coated with H. pylori. As controls, aliquots of the same preparation of cell suspensions of reference H. pylori strains previously determined to strongly express Lewis-x or Lewis-y, or to express neither (C1, MO19, and 88-22), were included; their OD values varied within a range of ± 10% between different assays [33].

Antibiotic susceptibility assays

Bacteria were harvested from two TSA plates after 48 h of growth and were assayed for susceptibility to amoxicillin, clarithromycin, tetracycline, and metronidazole by inoculation onto fresh TSA plates containing E-strips (AB Biodisk, Solna, Sweden) with single antibiotics in increasing concentrations. After 72 h of growth under microaerobic conditions, the MIC was defined as the boundary of the contact zone between bacterial growth and the strip.

Southern hybridization

To further differentiate the IS605-positive pairs among the 5 patients possessing them and to determine the degree of similarity within the pairs, we performed Southern hybridization by use of a probe within orfA, the region homologous to IS200. Each patient's initial IS605-positive strain showed its own unique hybridization pattern, with the number of hybridizing fragments within the chromosomal DNA varying from 1 to 5 (results not shown). In each case, the patterns for the initial and follow-up pools were identical, indicating IS605 stability within the H. pylori genome for a total of 503 person-months.

Sequence analysis of recA and IS605 PCR products

To provide an estimate of the rate and nature of point mutations, we determined both the nucleotide and deduced amino acid sequences for a 206-bp recA PCR product from each of the 26 pools and for a 323-bp PCR product of the IS605 region from the pools possessing this insertion element. For patients 3–13, for whom the above-described genotyping studies did not reveal any differences between the paired pools, the 206-bp recA products from the initial and follow-up samples were 100% identical. In contrast, there was 96.7% and 98.0% identity, respectively, within pairs 1 and 2 (figure 1), for which the above-described studies revealed differences with respect to cagA, vacA, iceA, or IS605 genotypes. As such, the probability of identity of the baseline and follow-up pools of pairs 3–13 was significantly higher than that of the isolates from pairs 1 and 2 (P = .0009; t test) and significantly higher than any out-of-pair comparison (mean ± SE identity, 96.8% ± 0.06%; P = .0002; t test). For pair 3, repeated direct sequencing of the PCR products from both pools showed either a C or a T at position 39. We then isolated DNA from multiple single colonies of both the initial and follow-up pools of this strain, performed recA PCR, and did AciI restriction digestion on each product. By agarose gel electrophoresis, AciI digestion yields 3 bands if C is at position 39 or 2 bands in case of a T. Of 16 colonies from the initial pool, 15 (94%) had a T and 1 (6%) a C at this position. Of 19 colonies of the follow-up pool, 17 (89%) had a T and 2 (11%) a C. Thus, the C/T polymorphism at position 39, which was synonymous, existed within the H. pylori population at both baseline and follow-up in patient 3 and did not represent a sequencing artifact. For the 5 patient pairs (3, 4, 6, 8, and 13) that contained IS605 (table 1), the intrapair identity of a 323-bp PCR sequence was 100%, whereas the out-of-pair identity was 95.9% ± 0.03% (P < .001; t test; figure 2). These studies confirm the high degree of relatedness of the paired pools from each person. The clustering of in-pair sequences in the dendrograms for patients 3–13 was supported in a majority of bootstrap replicates and therefore was statistically significant. No out-of-pair sequences were grouped in a majority of bootstraps.

Open in a separate window

Figure 1

Sequences of 206-bp polymerase chain reaction product derived from Helicobacter pylori recA for 26 paired H. pylori pools. For each pair shown in the dendrogram, original and follow-up pools are designated A and B, respectively. Sequences were identical for initial and follow-up pools of pairs 3–13, each pair having its unique sequence. In contrast, there was 96.7% and 98.0% identity within pairs 1 and 2, respectively. Pools 3A and 3B (*) consisted of mixed populations with silent C/T polymorphism at position 39. In-pair clusters each were supported in majority of bootstrap replicates.

Open in a separate window

Figure 2

Sequences of 323-bp polymerase chain reaction product derived from Helicobacter pylori IS605 insertion element, for those isolates containing this element (pairs 3, 4, 6, 8, and 13 and second pool of pair 1). Pool designations in this dendrogram are identical to those in figure 1. Intrapair identity was 100%, whereas out-of-pair identity was 95.9% ± 0.03%. In-pair clusters each were supported in majority of bootstrap replicates.

Lewis antigen expression

To determine whether prolonged colonization was associated with phenotypic differences, we examined expression of Lewis antigens, which represent the terminal structures of the O antigens of the H. pylori lipopolysaccharide molecules [32, 35–37]. For the 11 related pairs (3–13), no significant changes were noted in Lewis-x expression over time (mean ± SE, baseline vs. follow-up, 870 ± 240 vs. 737 ± 248; P = .70; paired t test). However, for these same pairs, the expression of Lewis-y decreased over time from 737 ± 196 to 452 ± 197 (P = .05; paired t test). These results suggest that there is in vivo selection for particular phenotypes (e.g., Lewis-y expression).

Other phenotypes

All pools were susceptible to amoxicillin, tetracycline, and clarithromycin (MIC, <0.016, <1.5, and <2 μg/mL, respectively). Pool 1B was resistant to metronidazole (MIC, >8 μg/mL), as were pools 4A, 4B, 11A, and 13B (table 1). Because cagA has a variable number of 3′ repeats that affect the size of the cagA protein, we examined immunoblots of cells from pairs 3–13. For 8 of the 11 pairs, as expected, cagA products were present (table 1). No size shifts were noted for any of these pairs.

RAPD-PCR

The sequence and Southern analyses examine only a small part of the H. pylori genome. We next used more comprehensive methods for assessing genetic drift without focus on specific genomic loci. We performed RAPD-PCR, because this technique yields strain-specific profiles based on the presence of particular sequences, which vary with different primers, within the entire bacterial genome [25]. As expected, and confirming the sensitivity of this method, the pools in pairs 1 and 2 showed clearly dissimilar RAPD patterns with each of the primers used (figure 3). For the remaining 11 pairs, the RAPD profiles of the initial and follow-up pools were similar. The profiles for each pair differed from those of other pairs. However, within each pair, differences were noted in the number and locations of amplimers generated by 1–4 primers. In an analysis of the profiles obtained with primers 1254, 9355, and d11344 and pairs 3–13, the range of interpair similarity was 27%–69% (mean ± SD, 44% ± 14%), whereas, as expected, the intrapair similarity, with a range of 65%–95% (mean ± SD, 82% ± 9%), was substantially higher (figure 3). The RAPD profiles of each pool and intrapair difference were reproducible on repetition of RAPD testing with the same or newly isolated DNA. In addition, for 10 pools (5 pairs), these profiles also were examined after prolonged culture on TSA plates with passage of strains every 48 h for 20 times. In each case, the initial and follow-up isolates had identical profiles. The differences also were reproducible when the process of culture, DNA isolation, and RAPD testing was repeated in another laboratory by another operator. These data suggest that intrapair differences in RAPD patterns were not due to laboratory- or operator-related conditions, nor to in vitro handling of the strains, but represent genetic changes that occurred in vivo within the individual pools over time.

Open in a separate window

Figure 3

Analysis of random arbitrarily primed DNA (RAPD) polymerase chain reaction profiles of 26 paired Helicobacter pylori pools obtained with primers 1254, 9355, and d11344. Pool designations are identical to those in figure 1. RAPD photographs were digitized, and levels of similarity between patterns were calculated (GelCompar 4.0 software; Applied Maths, Kortrijk, Belgium). The range of intrapair similarity for pairs 3–13 was 65%–95%, but for pairs 1 and 2, similarities were only 46% each.

Financial support: Stichting van Helten of the Royal Dutch Academy of Sciences; NIH (DK-53707); the Medical Research Service of the Department of Veterans Affairs; and the Iris and Homer Akers Fellowship in Infectious Disease.