Direct Detection of Helicobacter pylori Mutations Associated with Macrolide Resistance in Gastric Biopsy Material Taken from Human Immunodeficiency Virus-Infected Subjects

Abstract

One hundred forty gastric biopsies were tested by microbiological methods and by amplifying a sequence of 23S rRNA and identifying mutations associated to clarithromycin resistance. Seventy-six specimens were positive for Helicobacter pylori. Mutational analysis revealed alterations in 18 (39.1%) of 46 and 2 (8.7%) of 23 samples from human immunodeficiency virus-seropositive and -seronegative persons, respectively. The results of the mutational analysis fully correlated with those of the susceptibility tests.

Helicobacter pylori is a gram-negative bacterium that plays a central role in the pathogenesis of chronic gastritis, peptic ulcer diseases, gastric adenocarcinoma, and mucosa-associated lymphoid tissue cancer (3, 7, 19). In most European countries, a drug combination including clarithromycin (CLA) is recommended for ulcer therapy (8) but macrolide resistance is an important factor in the clinical outcome of eradication therapy (16); indeed, CLA resistance has been found in more than 10% of the strains from some European countries (4, 17) and some Italian regions (11; M. P. Dore, B. Are, M. Carta, I. Mura, A. Maida, and G. Realdi, Abstract, Gastroenterology 114:G0445, 1998.). CLA susceptibility testing therefore seems to be necessary and cost-effective before initiation of therapy and is essential after the first treatment failure.

H. pylori CLA resistance is associated with single-base mutations within the peptidyltransferase-encoding region of the 23S rRNA gene (18, 23, 24). Three stable mutations have been described in which the adenine residues at positions 2143 and 2144 are replaced with guanine (A2143G and A2144G) or cytosine (A2143C) (10).

H. pylori is a delicate and slow-growing organism, which means that classic microbiological approaches are often cumbersome in determining a resistance profile. This problem may be overcome by PCR-based techniques that allow rapid analysis of a relatively small number of H. pylori cells in a gastric biopsy sample (2, 13-15, 20, 22).

The aim of our study was to evaluate the effectiveness of a genotypic strategy for the direct detection of CLA resistance in gastric biopsy material by means of double gradient-denaturing gradient gel electrophoresis (DG-DGGE) (6, 21).

Forty-two (23 CLA-sensitive and 19 CLA-resistant) H. pylori clinical isolates were used to establish the technique. Twenty-two of these strains were cultured from gastric biopsy materials obtained from the persons enrolled in the study.

One hundred forty gastric biopsy specimens were available for the study. One hundred two were consecutively obtained from 99 human immunodeficiency virus (HIV)-infected persons (23 females, 76 males) with dyspepsia. Eight samples (7.8%) were collected from persons who had received H. pylori antimicrobial therapy a mean of 18.6 (range, 1 to 41) months before. At the time of endoscopy, 33 persons belonged to Centers for Disease Control and Prevention group A, 31 belonged to group B, and 35 belonged to group C. Thirty-eight samples were collected from 38 persons without HIV infection (12 females, 26 males). Thirteen (34.2%) of these specimens were obtained from persons who received H. pylori eradication therapy a mean of 15 months before sample collection.

Endoscopy revealed 46 active or scarred gastric or duodenal ulcers, 54 gastric or duodenal erosions, and 40 cases of normal or hyperemic mucosa. During each endoscopic procedure, three antral mucosal biopsy specimens were taken, after patient informed consent was obtained, and processed for microbiological H. pylori culturing, histological examination, and DNA amplification by PCR. The H. pylori cultures were set up approximately 1 to 3 h after endoscopy.

The samples were routinely stained with hematoxylin and eosin for histopathological diagnosis and H. pylori morphological detection; when there was a doubt, they were also stained with Giemsa. The biopsy specimens were reviewed by a single pathologist.

The biopsy samples were cultured on Columbia agar with 5% sheep blood and Thayer Martin agar and incubated at 37°C under microaerophilic conditions for a maximum of 7 days. H. pylori was identified on the basis of colony morphology, microscopy, and positive oxidase and rapid urease activities. CLA sensitivities were determined by means of MIC measurements with E-test strips (AB Biodisk, Solna, Sweden) (9).

Genomic DNA from the H. pylori isolates and biopsies was prepared with the QIAamp Tissue Kit (Qiagen, Hilden, Germany). Aliquots of DNA (1 μg) were used for time release PCR (1) of a 268-bp fragment of the 23S rRNA gene (GenBank accession no. U27270) by means of primers HP1 (5′ AACGGTCCTAAGGTAGCG) and HP2 (5′ CATCAAGGGTGGTATCTCA). The former oligonucleotide carries, at its 5′ end, an additional 30 bp of a GC-rich sequence. The annealing temperatures were 58°C for the first 20 cycles and 56°C for the following 30 cycles. A negative control was tested every three samples. The heteroduplex molecules were generated by mixing the PCR products obtained from the sample with those from a wild-type CLA-sensitive strain. (6, 21). DG-DGGE was carried out as previously described (21). Briefly, the PCR products were electrophoresed in a colinearly increasing double gradient of 40 to 55% denaturant and 6 to 10% polyacrylamide gels in 1× TAE buffer (20× TAE buffer is 0.8 M Tris base-0.4 M sodium acetate-0.02 M EDTA, pH 7.4) at a constant temperature of 60°C at 50 V overnight. The gels were stained with a 1:5,000 dilution of Sybr Green II (FMC Bio Products, Rockland, Maine). The 23S rRNA PCR products were sequenced with the same primers and an automated DNA sequencer (model 377; Applied Biosystems, Foster City, Calif.).

Optimization of the mutational analysis protocol was achieved by processing of 23 CLA-sensitive and 19 CLA-resistant H. pylori strains. All of the PCR products obtained from the resistant strains contained mutations, and the DG-DGGE analysis also revealed altered homoduplex molecules and/or the formation of heteroduplex molecules. Four different band patterns were generated (Fig. 1A), and direct DNA sequencing confirmed these results and showed the following mutations: A2143C (one strain), A2143G (nine strains), A2144G (eight strains), and A2144G and T2183C (two strains). The possibility of a mixture of two or three bacterial clones was observed in five strains. Four of these showed the presence of a mutated clone and a wild-type clone, since heteroduplex formation was observed without mixing with the wild type (Fig. 1A, lanes 17, 18, 22, and 23); the fifth showed a mixture of two mutated clones and one wild-type clone (Fig. 1A, lanes 15 and 16). In two of the CLA-resistant strains, a point substitution (T2183C) located outside the drug binding site was associated with a mutation causing CLA resistance.

FIG. 1.

DG-DGGE analysis of the 268-bp region of the H. pylori 23S rRNA gene. Photographs of DG-DGGE patterns show the electrophoretic mobility of PCR fragments. Both the homoduplex and heteroduplex patterns are shown for each sample carrying a mutation. The heteroduplex molecules were prepared by mixing them with the PCR products obtained from a CLA-susceptible H. pylori strain without mutations in the 268-bp 23S rRNA gene region. Wt, PCR products from the CLA-susceptible H. pylori strain. (A) Patterns obtained from bacterial suspensions. (B) Patterns obtained from the amplification products of gastric biopsy materials. See the text for detailed explanations.

The results of the conventional and 23S rRNA PCR examinations for the presence of H. pylori in the gastric biopsy specimens are shown in Table 1. A total of 76 gastric biopsy specimens were therefore positive for H. pylori, as revealed by conventional methods and/or 23S rRNA PCR. Assuming the ideal case with no false-negative results, the sensitivities of histology, culture, and 23S rRNA PCR, respectively, were 98.7, 72.4, and 90.8%.

TABLE 1.

Comparison of 23S rRNA PCR and microbiological methods for detection of H. pylori in gastric biopsy samples

No. of samples	Result obtained or no. (%) of samples
	Conventional methods		23S rRNA PCR
	Histology-microscopy	Culture
64	−	−	−
54	+	+	+
14	+	−	+
6	+	−	−
1	+	+	−
1	−	−	+
140 (total)	75 (53.6)	55 (39.3)	69 (49.3)

Table 2 summarizes the results of susceptibility testing and the genotypic detection of mutations in the 23S rRNA gene of H. pylori by processing of gastric biopsy specimens. DG-DGGE detected mutations in 20 (43.5%) of 46 PCR products obtained from samples collected from HIV-seropositive persons and 5 (21.7%) of the 23 obtained from samples collected from HIV-seronegative persons. Four different DG-DGGE band patterns were obtained from these samples, which consisted of sets of four bands when both heteroduplexes and homoduplexes were distinguishable or of three bands when the homoduplexes showed the same migration (Fig. 1B). DNA sequencing confirmed the presence of the following mutations in all of the samples with altered DG-DGGE patterns: A2143G (6 samples), A2144G (11 samples), A2144G and T2183C (3 samples), and T2183C (5 samples). In seven samples, the simultaneous presence of mutated and wild-type DNAs was observed (Fig. 1B, lanes 3, 4, 5, and 6). Base substitutions known to cause CLA resistance (at positions 2143 and 2144) were detected in 20 of the 25 samples containing mutations. A single-base substitution at position 2143 or 2144 generated DG-DGGE band patterns that were clearly distinguishable from those obtained with the T2183C substitution alone or in association with a mutation at position 2144 (Fig. 1B). The DG-DGGE patterns generated by T2183C alone and in association with A2144G were similar but could be differentiated on the basis of heteroduplex formation by mixing the two PCR products (Fig. 1B, lane 17). After exclusion of the samples carrying the single T2183C substitution, DG-DGGE analysis revealed mutations associated with CLA resistance in 18 (39.1%) of 46 samples collected from HIV-seropositive persons and in 2 (8.7%) of 23 samples collected from HIV-seronegative persons (Table 2).

TABLE 2.

Comparison of CLA susceptibility of H. pylori strains and genotypic detection of mutations associated with CLA resistance by DG-DGGE analysis of gastric biopsy samples obtained from HIV-seropositive and -seronegative patients with dyspeptic symptoms

Patients	No. of samples	No. histology positive	Phenotype analysis			Genotype analysis
			No. culture positive	No. (%) of samples		No. PCR positive	No. (%) positive by DG-DGGE analysis
				CLA susceptible	CLA resistant		Wild type	All mutationsa	Mutations at 2143 and 2144
HIV seropositive	102	49	37	23 (62.2)	14 (37.8)	46	26 (56.5)	20 (43.5)	18 (39.1)
HIV seronegative	38	26	18	17 (94.4)	1 (5.6)	23	18 (78.3)	5 (21.7)	2 (8.7)
Total	140	75	55	40 (72.7)	15 (27.3)	69	44 (63.8)	25 (36.2)	20 (29)

^aAll of the samples with electrophoretic mobility different from that of the CLA-susceptible reference strain are included here.

Our results show that determination of CLA susceptibility by genotypic analysis was possible in 69 (90.8%) of the 76 samples found to be H. pylori positive by conventional and/or PCR methods; only 55 culture-positive samples (72.4%) were phenotypically analyzed. The possible presence of CLA-associated mutations was ascertained in 14 culture-negative but histology-positive samples. In comparison with that of culturing methods, the sensitivity of our amplification protocol seems to be better than that described in the earliest studies by Sevin et al. (22) and Björkholm et al. (2) and at least similar to that reported recently by Maeda et al. (14).

DG-DGGE mutational analysis detected all three of the mutations (A2143C, A2143G, and A2144G) known to cause CLA resistance. In particular, although it is very rare, as it has been reported in only about 7% of resistant strains (24) and is undetectable by restriction fragment length polymorphism analysis, mutation A2143C generated clearly visible bands due to the formation of heteroduplex molecules. We found that the A2144G mutation was the most frequently observed single-base substitution, being found in 70% of the mutated amplification products. Although this report is too limited to reflect regional or national trends, a similar frequency has been reported in strains cultured in Japan (14). The single-base T2183C substitution was observed in eight (11.6%) of the PCR-positive samples. This mutation is located outside the drug binding site and is probably a simple polymorphism unrelated to CLA resistance.

There was a 100% correlation between the genotypic and phenotypic methods of CLA resistance detection. In seven (35%) of the mutation-containing samples, DG-DGGE analysis detected the simultaneous presence of altered and wild-type DNAs. CLA-resistant strains were cultured from all of these samples. There are two possible explanations for this. Either two bacterial populations, one of which carries a mutated 23S rRNA gene, are present, or H. pylori possesses two copies of the rRNA operon (5, 12). It has also been suggested that a heterozygous condition may exist in CLA-resistant H. pylori (24).

We observed a significantly higher incidence (37.8%) of CLA-resistant strains cultured from our HIV-infected persons than that observed in the general Italian and European populations (4, 11, 17; Dore et al., abstract). This may be related to the greater administration of macrolide compounds to HIV-infected persons. A review of the clinical charts of the persons enrolled in this study showed that only eight (8.1%) had received previous CLA-containing H. pylori eradicant therapy, but 29 (29.3%) had received a total of 44 treatments with azithromycin or CLA for upper respiratory or disseminated Mycobacterium avium complex infection in the 18 months preceding gastroscopy.

In conclusion, our time-saving and relatively simple mutational analysis efficiently detected the H. pylori 23S rRNA gene mutations associated with CLA resistance. This assay may be particularly useful when increasing macrolide use and the consequently increased risk of bacterial resistance require rapid pretreatment methods to facilitate optimal drug choices.