During 2009, 206 volunteer subjects were recruited with symptoms of dyspepsia, 44.2% (91/206) men ranging in age from 18-68 years, in a population from Túquerres in the high Andes mountains of Colombia, characterized by a high prevalence of H. pylori and preneoplastic lesions[8,16]. Exclusion criteria included prior gastrectomy, chronic disease, intake of H₂-receptor antagonists, proton pump inhibitors, or antibiotic intake during the last four weeks prior to endoscopy. During upper gastrointestinal tract endoscopy, biopsies were obtained of gastric antral and body and embedded in paraffin for histopathological evaluation. Additional biopsies, two antral (greater and lesser curvature) and one gastric body (greater curvature) were taken for H. pylori culture and immediately frozen in thioglycolate with glycerol. The samples were delivered in liquid nitrogen to the Department of Pathology at Universidad del Valle (Cali, Colombia) for analysis. This research was approved by the Ethics Committee at Universidad del Valle. All the participants provided informed consent.

Histopathological diagnosis for each participant was independently evaluated by expert pathologists in antral and body gastric sections stained with hematoxylin and eosin, according to the Sydney classification system[17]. The categories were non-atrophic gastritis (NAG), multifocal atrophic gastritis without intestinal metaplasia (MAG), intestinal metaplasia (IM), and dysplasia. Cases with discordant diagnosis were again revised and consensus was reached.

Fragments of antral and body gastric mucosa were homogenized under sterile conditions in 200 µL of saline solution 0.89% using a homogenizer (Kimble-Kontes, Vinelan, NJ, United States). The homogenized was placed on Columbia agar plates (Oxoid, Basingstoke, Hampshire, England) with defibrinated sheep blood at 7% plus selective supplement for H. pylori (Dent) containing Vancomycin (10 mg/L), Sodium cefsulodin (5 mg/L), Trimethoprim lactate (5 mg/L), and Amphotericin B (5 mg/L) (Oxoid, Basingstoke, Hampshire, England). The agar plates were incubated under microaerophilic conditions (6% O₂, 6% CO₂, 88% N₂, using CampyPak Plus envelope, BBL, Nashville, TN, United States) at 37 °C from four to eight days until observing small gray translucent colonies. The typical colonies were sub-cultured and later identified as H. pylori through their morphology, Gram-stain, and biochemical analyses for oxidase, catalase, and urease activity.

Antimicrobial resistance was evaluated by agar dilution method according to guidelines from the Clinical and Laboratory Standards Institute[18] using Mueller-Hinton (MHA) agar (Merck KGaA, Darmstadt, Germany) supplemented with defibrinated sheep blood at 7%. Double serial clarithromycin and amoxicillin dilutions (Genfar laboratories, Bogotá, Cundinamarca, Colombia) of 0.25, 0.5, 1.0, 2.0, and 4.0 μg/mL were added to MHA plates. Bacterial isolates sub-cultured in Columbia agar for no more than 72 h were re-suspended in saline solution and turbidity was adjusted at a concentration equivalent to a McFarland 2 standard (containing 1 × 10⁷ or 1 × 10⁸ CFU/mL), 1 µL of the adjusted inoculum was placed directly onto each MHA agar plate with dilutions of the antibiotic. The minimum inhibitory concentration (MIC) of clarithromycin and amoxicillin was determined after 72 h of incubating the isolates under microaerophilic conditions at 37 °C, and it was recorded as the lowest concentration of the antibiotic that inhibits visible growth of the colonies[12]. An isolate was considered resistant to clarithromycin or amoxicillin when its MIC was ≥ 1.0 μg/mL. The H. pylori reference strain ATTC^® 43504 was used as quality control strain to monitor MIC precision when using the agar dilution method, considering MIC cutoff points < 0.015 µg/mL as sensitive and > 0.12 µg/mL as resistant to clarithromycin and amoxicillin.

Bacterial DNA was extracted from pure H. pylori cultures that showed resistance to clarithromycin and/or amoxicillin in each of the serial dilutions and from the isolates themselves in baseline (without antibiotic pressure). In all cases, single colonies were taken with swabs and washed in a 1.5 mL tube with 200 µL of sterile saline solution (0.89%). Tube contents were homogenized in vortex for 10 s and centrifuged in a 320R universal micro-centrifuge (Hettich Inc, Tuttlingen, Germany) at 13000 rpm for 2 min at 4 °C. Thereafter, the supernatant was discarded and the bacterial pellet was re-suspended in 300 µL of Lysis buffer (3 μL of Tris HCl 1 mol/L, pH 8.0 (Promega, Madison, WI, United States), 3 μL of EDTA 0.5 mol/L (Calbiochem, Gibbstown, NJ, United States), 15 μL of SDS 10% (Calbiochem, Gibbstown, NJ, United States), 3 μL of proteinase K (10 mg/mL) (Invitrogen, Carlsbad, CA, United States), and 276 μL of distilled water) and incubated in dry well (Labnet, Edison, NJ, United States) at 56 °C for 18 h. Then, the proteinase K was inactivated by heating at 70 °C for 10 min and 120 µL of NaCl 5 mol/L (Calbiochem, Gibbstown, NJ, United States) was added, homogenized by vortex and centrifuged for 5 min at 13000 rpm; the supernatant was transferred to another tube in which two volumes of absolute alcohol were added (Mallinckrodt, St.Louis, Mo, United States), again centrifuged at 13000 rpm for 20 min, then discarding the supernatant and adding 200 µL of ethanol at 70%. Once more, it was centrifuged for 5 min at 10000 rpm and then the supernatant was discarded; each tube was dried by inversion at room temperature. Finally, 50 µL of TE buffer were added (Tris-HCl 10 mmol/L, pH 8.0 (Promega, Madison, WI, United States), EDTA 1 mmol/L, pH 8.0 (Calbiochem, Gibbstown, NJ, United States) and the DNA was stored at -20 °C. The yield and purity of the DNA was determined by optical density at 260/280 nm in Gene Quant II^® spectrophotometer (Pharmacia Biotech, Piscataway, NJ, United States) according to manufacturer’s instructions.

To genotype the virulence of H. pylori isolates resistant in baseline to clarithromycin and amoxicillin, separate PCR reactions were carried out in a thermocycler (Swift MiniPro^TM, Esco Technologies, Hatboro, PA, United States) for vacA-s and vacA-m, cagA, cag empty-site. Four sets of primers were initially used in this study. To amplify the s region of the vacA gene, VA1F and VA1R primers were used, which amplify a fragment of 259 bp (allele s1) and 286 bp (allele s2). To detect the m region of the vacA gene, HPMGF and HPMGR primers were employed, which resulted in the amplification of a fragment of 401 bp (allele m1) and 476 bp (allele m2). The cagA gene was amplified to validate the ability of the EPIYA polymerase chain reaction (PCR) to detect the presence of the cag PAI, using cagAF and cagAR primers that amplify a fragment of 183 bp from the 5’ end of the cagA gene[19]. The negative isolates for the cagA gene were in all cases confirmed by cag empty-site PCR employing ES-F and Rnew-1R primers[7], upon detecting a fragment of 106 bp that indicated the loss of the cag PAI. Each experiment included a positive and negative reaction control.

To detect the presence of the cag PAI and characterize the number of EPIYA motifs present in the 3’cagA variable region in resistant isolates at each of the antibiotic dilutions (with antibiotic pressure) and in same isolates in baseline (without antibiotic pressure), a PCR reaction was carried out using cagA2530S and cagA3000AS primers[11]. The size of the amplicons expected varied in the range of 370 bp (2 EPIYA motifs), 470 bp (3 EPIYA motifs), 570 bp (4 EPIYA motifs), 670 bp (5 EPIYA motifs) ± 25 bp and were approximately 100 bp equidistant, indicating the presence of multiple repeated sequences. In cases where a clear band was not evidenced or the presence of amplified, a second PCR was carried out under the same conditions, employing 1 µL of the initial amplified, along with the respective controls to detect possible cross contamination. During the development of the trials, H. pylori strain 26695 (ATCC 700392) were used (number of access AE000511.1) as size control of the nucleotide sequence encoding for three EPIYA motifs-ABC (470 bp ± 25 bp), and H. pylori strain ATCC 43504 that presents an EPIYA-ABCCC motif (670 ± 25 bp); thus, predicting the number of repetitions of the EPIYA motifs was possible by direct comparison of the sizes corresponding to PCR amplified.

The genomic differences between H. pylori isolates found before and after antibiotic pressure were evaluated by using random primers: 1254 (5’-CCGCAGCCAA-3’) and 1281 (5’-AACGCGCAAC-3’)[20]. These oligonucleotides were amplified in a final 12.5 µL reaction volume composed of 2.5 µL of PCR buffer (10 mmol/L of Tris-HCl pH 8.0 and 50 mmol/L of KCl-Promega, Madison, WI, United States), 3 mmol/L MgCl₂, 1 U of Go taq polymerase (Promega, Madison, WI, United States), 250 μmol/L of dNTPs (Promega, Madison, WI, United States), 25 pmol of each primer, and 1 µL of bacterial DNA. The amplification was carried out in a thermocycler (Swift MiniPro^TM, Esco Technologies, Hatboro, PA, United States), prior denaturalization for 5 min at 94 °C, followed by 45 cycles of: 94 °C for 1 min, 36 °C for 1.30 min, 72 °C for 2 min, followed by a final incubation at 72 °C during 10 min. Band size was estimated by using a 100 bp DNA Ladder (Fermentas International Inc., Vilnius, Lithuania). Each isolate was tested with the primers described under the same conditions at least twice and only evident and reproducible bands were evaluated.

All the amplicons obtained in each of the previously described PCR reactions were run on agarose gel (SeaKim, FMC Bioloabs) at 2%, stained with ethidium bromide (Invitrogen, Carlsbad, CA, United States) at 0.5 µg/mL, in an electrophoresis chamber (Fotodyne Inc., Hartland, WI, United States) at 75 V for 40 min provided by a EC-105 Compact Power Supply (Thermo Fisher Scientific Inc, Asheville, NC, United States).

Initially, univariate and bivariate analyses were conducted; the categorical variables were described as proportions. To determine the statistical significance of the differences in the proportions of cag PAI detection via cagA PCR and EPIYA PCR among baseline resistant isolates and in amoxicillin and clarithromycin dilutions, the McNemar test was employed for data related to the SPSS 15.0 statistical package (SPSS Inc., Chicago, IL, United States). Images of the amplicons obtained via RAPD PCR were analyzed through the Gel-Pro Analyzer 4.5 program for Windows (Media Cibernetics, Inc, Rockville, MD, United States). Molecular weights of the well-defined, single-pattern bands were analyzed with this program and a matrix of binary data was constructed based on the presence (1) and absence (0) of the bands observed (polymorphic) among the isolates. Ratios among isolates were established through cluster analysis carried out with the SPSS 15.0 statistical package (SPSS Inc., Chicago, IL, United States), where each of the fingerprint bands were denoted as variables. The dendrograms were designed by following Ward’s clustering method and the estimation of distances between each pair of H. pylori isolates was calculated with the squared Euclidean distance. In the clusters, fingerprints with distances less than or equal to five were considered related and distances above five were unrelated. The cluster analysis and the association to antimicrobial susceptibility were evaluated by using the χ² exact. The statistical significance was accepted with a P value ≤ 0.05.

Among the 176 participants diagnosed through histopathology, 140 (79.5%) presented non-atrophic chronic gastritis, four (2.3%) multifocal atrophic gastritis without intestinal metaplasia, and 32 (18.2%) were diagnosed with multifocal atrophic gastritis with intestinal metaplasia (data not shown).

The antimicrobial susceptibility of the 149 isolates is shown in Table 1; MIC₅₀ and MIC₉₀ values are indicated. A total of 136 (91.3%) of the bacterial isolates were sensitive to clarithromycin and amoxicillin, while six (4%) of the isolates were AmxR, and four (2.7%) were ClaR. Only three (2%) of the isolates were resistant to both antibiotics. Global resistance was at 8.7% (13/149). The histopathological diagnosis of the participants with isolates that showed resistance was as follows: nine (69.2%) presented non-atrophic chronic gastritis, two (15.4%) had multifocal atrophic gastritis without metaplasia and two (15.4%) had multifocal atrophic gastritis with intestinal metaplasia. It is interesting to find that 50% (2/4) participants diagnosed through histopathology with multifocal atrophic gastritis presented AmxR/ClaR double-resistant isolates (SV323 and SV399) (Table 2).

The genotyping results of H. pylori isolates resistant to amoxicillin and clarithromycin are shown in Table 2. In 23% (3/13) of the participants with isolates resistant to the antibiotics evaluated, colonization was documented with multiple H. pylori strains; in one of the cases (SV438) it was impossible to determine the infecting bacterial genotype. It was observed that in all isolates the cagA-positive genotype predominated. vacA s1m1 alleles were predominant, present in 84.6% (11/13), while s2m2 alleles were only observed in 7.7% (1/13) of the isolates. None of the resistant isolates obtained in this study presented genotypes s1m2 or s2m1.

Successful amplification took place of the 3’ variable region of the cagA encoding for EPIYA phosphorylation motifs of the CagA protein in 92.3% (12/13) of the baseline resistant isolates (without antibiotic pressure) and in 60% (36/60) of the resistant isolates growing at each of the serial amoxicillin and clarithromycin dilutions (with antibiotic pressure) (Table 2). This permitted confirming the presence of the cag PAI previously detected by cagA-specific PCR, as well as characterizing the number of EPIYA motifs present. Amplicon size was in the range of 470-670 bp (± 25 bp), according to that suggested by Panayotopoulou et al[11] except for some cases where strong bands were obtained -reproducible with unexpected molecular weights close to 170 and 270 bp (± 25 bp) (Figure 1A and Table 2). A single PCR product was observed in 49.3% (36/73) of the isolates obtained between baseline resistant samples and antibiotic dilutions and more than one PCR product with sizes corresponding to different numbers of EPIYA repetitions in 16.43% (12/73) of the isolates. All isolates negative for EPIYA PCR (n = 25) were confirmed by cag empty-site PCR, finding only four isolates as true cagA-negative. When comparing amplification proportions of EPIYA PCR vs cagA PCR in detecting cag PAI in isolates growing in the amoxicillin and clarithromycin serial dilutions, significant differences were found (McNemar test P = 0.006 and P = 0.031, respectively), contrasting with the baseline resistant isolates where no significant differences were found between both PCR methods (P > 0.05) (Table 3).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Electrophoretic analysis. A: EPIYA polymerase chain reaction (PCR) products from DNA of the Helicobacter pylori (H. pylori) isolates SV301 and SV316 that grew in the Amoxicillin serial dilutions, the PCR products were analyzed in agarose gel at 2%. Lines a and b positive controls; line a: Clinical H. pylori isolate (PZ5085) with EPIYA ABCC motif (570 ± 25 bp) confirmed by sequencing; Line b: A mix of DNA from isolate PZ5085 and from the H. pylori strain 26695 with EPIYA-ABC motif (470 ± 25 bp). M: 100 bp DNA ladder. Distribution of different molecular weights among isolates is an indication of the presence of multiple EPIYA repetitions; B: Random amplified polymorphic DNA fingerprints generated with primer 1254 in H. pylori isolates resistant in baseline (without antibiotic pressure) and resistant isolates growing in each of the serial antibiotic dilutions (with antibiotic pressure); line N: negative reaction control; M: 100 bp ladder.

Primers 1281 and 1254 generated a reproducible RAPD fingerprints; they were capable of discriminating 73 different profiles. The number and size of the bands obtained with primer 1281 varied from 3 to 20 and from 76 to 1111 bp, respectively. It was determined if antibiotic pressure was associated to RAPD clusters. Most of the baseline resistant isolates (12/13; 92.3%) and under antibiotic pressure of amoxicillin (18/32; 56.3%) and clarithromycin (18/28; 64.3%) were included in cluster I in comparison to the distribution of these isolates in clusters II and III (P = 0.003).

With primer 1254, well-defined patterns were generated of two to 19 fragments in a range from 49 to 1950 bp (Figure 1B). The dendrogram based on RAPD profiles obtained with this primer includes the antimicrobial susceptibility for each isolate (Figure 2). Cluster analyses for this primer showed three main clusters in a parsimonious solution. In none of the clusters was the segregation of the baseline isolates and under antibiotic pressure significant (P = 0.704).

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Dendrogram of the profile of random amplified polymorphic DNAs generated with primer 1254 in Helicobacter pylori isolates resistant in baseline and resistant isolates growing at each of the serial dilutions of amoxicillin and clarithromycin. Three separate clusters (I, II and III) are indicated. In none of the clusters was segregation of sensitive and resistant isolates significant (P = 0.704). Cluster analyses were designed by following Ward’s hierarchical clustering method and estimation of distances between each pair of Helicobacter pylori (H. pylori) isolates were calculated with the squared Euclidean distance. Distances between isolates are given in a 25-point standardized scale that reflects with sufficient precision the existing proportionality between the original distances. The fusions produced near the origin of the scale (left) indicate that the cluster formed is quite homogeneous. Conversely, fusions produced on the final zone of the scale (right) indicate the cluster formed is quite heterogeneous. Atb: Antibiotic, Dil: Dilution.

Upon observing the distribution of the number of EPIYA repetitions in each of the serial dilutions of the antibiotics with respect to the baseline isolates, change at the number of EPIYA repetitions was found in 61.5% (8/13) of the baseline isolates under antibiotic pressure. In six of these isolates (SV301, SV328, SV440, SV323, SV399 y SV433), the change consisted in loss of EPIYA repetitions evidenced by the low molecular weight of the amplified bands. In contrast, two isolates (SV316 and SV333) revealed gain of EPIYA repetitions. The gain and loss of EPIYA motifs resulted in a diversity of H. pylori subclones after bacterial adjustment to changing conditions product of antibiotic pressure. In all cases, the genotyping results and the presence of a single band in the EPIYA PCR for the baseline isolate permitted discarding the presence of more than one infecting cagA-positive isolate, except for isolate SV328 that presented in baseline three CagA species that differed in the number of EPIYA repetitions; after antibiotic pressure, alteration of the number of EPIYA repetitions was evidenced in two of the CagA species present (MIC 0.25 µg/mL) and its subsequent loss with increased concentration of the antimicrobial (MIC 0.5 µg/mL). Also, in three isolates no change was noted in the 3’ variable region of the cagA and in the two remaining cases (SV338 and SV438) change could not be documented due to absence of EPIYA PCR amplification. We used RAPD PCR to evaluate the clonal relationship of the isolates and verify that these strains had the same origin. Specifically, the RAPD fingerprints generated by both primers reflected a close clonal relationship between baseline isolates (without pressure) and isolates growing in antibiotic dilutions (under pressure).

Helicobacter pylori (H. pylori) is a prominent risk factor for gastric cancer, higher risk in individuals infected with cagA-positive strains. Its eradication with antibiotics constitutes a chemoprevention strategy; however, treatments are not completely effective and seem to contribute to increasing the prevalence of less virulent strains.

Virulence-associate genotypes of H. pylori are increasingly recognized as determinants of the clinical outcome of the infection. The fact that virulence factors are linked to disease implies that they are a fixed characteristic, but this is not the case and genotype variations can occur through genetic rearrangements that either eliminates particular immunostimulatory genetic regions (cag pathogenicity island) or causes variations in potential immunostimulatory molecules (number of Glu-Pro-Ile-Tyr-Ala motifs in CagA) probably reflecting the local selection of H. pylori particular phenotypes as an escape resource to the immune response of the host induced by environmental pressures.

Some studies indicate that failure to eradicate H. pylori from the stomach may result in an increase in the prevalence of less virulent genotypes. It is unknown if this phenomenon is due to the positive selection of certain genotypes (cagA-negative^/vacA s2m2) in cases of multiple infection or loss of specific virulence factors (cag PAI) induced by antibiotic pressure. This study shows that antibiotic pressure does not induce loss of the cag PAI, but it can lead to genetic rearrangements within of the cagA variable region that can affect the level of tyrosine phosphorylation impacting on its cellular effects and lead to divergence of cagA-positive subclones, which as a set could alter the pathogenic process of H. pylori in cases with treatment failure.

These in vitro findings are specially important because of their possible implications for the treatment and evolution of gastro-duodenal diseases caused by H. pylori, suggesting that acquisition or deletion of EPIYA motifs product of the antibiotic effect can affect the phosphorylation level of tyrosine and leading to the divergence of cagA-positive H. pylori subclones. Further studies are needed aimed at evaluating the effect of anti-H. pylori treatment on the cagA variable region in H. pylori isolates from patients with treatment failure and deciphering the complex molecular mechanisms through which these genomic alterations are produced. Upon proving its effect on cagA, this methodological approach could constitute a useful prognosis tool in clinical practice, when detecting the presence of multiple subclones and determining the number and type of EPIYA motifs, permitting the prediction of pathogenic activity of the bacteria in unsuccessful treatments.

The cag pathogenicity island is a genomic fragment of approximately 40 kb that contains 30 genes that encode a type IV secretion apparatus, which export proteins from bacterial cytoplasm to host epithelial cells. The terminal gene in the island, cagA, is commonly used as a marker for the entire cag locus.

This manuscript deals with the evaluation of the in vitro effect of amoxicillin and clarithromycin on the cag PAI. The authors found that antibiotic pressure does not induce loss of the cag PAI and leads in most cases to genetic rearrangements within the 3’ region of cagA of the bacteria that lead to divergence of cagA-positive subclones. This study is interesting and original. Methods are appropriate, results are clearly presented and conclusions are corroborated by the results.