TEXT

Top Abstract Text References

Although most infections with Helicobacter pylori are asymptomatic, and some might even be beneficial for the host, the pathogen is usually eradicated with antibiotics such as amoxicillin and metronidazole in order to cure gastritis and peptic ulcer diseases (3, 4). Resistance against metronidazole is common among H. pylori strains, while there are only a few reports of amoxicillin-resistant H. pylori strains (6, 8; M. Guslandi, Letter, Lancet 353:241-242, 1999; A. A. van Zwet, C. M. Vandenbroucke-Grauls, J. C. E. Thijs, J. van der Wouden, M. M. Gerrits, J. G. Kusters, and C. M. Vandenbrouke-Grauls, Letter, Lancet 352:1595, 1998).

Metronidazole resistance of H. pylori was shown to be due to the mutational inactivation of rdxA (7, 11). In turn, a metronidazole-resistant strain (Mtz^r) was rendered sensitive when complemented with the rdxA gene of a metronidazole-sensitive strain (Mtz^s). It was concluded that RdxA functions as a metronidazole-reducing nitroreductase (11). Resistance against -lactam antibiotics like amoxicillin is generally due to hydrolysis by a -lactamase (5) or by mutational modification of the penicillin binding proteins (13).

To investigate the molecular basis for antibiotic resistance in H. pylori, we did an in vitro selection for amoxicillin and metronidazole resistance on the strains 503 (ATCC 43503) and 69A (69A and 888-0: clinical isolates obtained from R. Haas, Max-von Pettenkofer-Institut, Munich, Germany). In contrast to other investigators (12, 15), we performed the selection not on agar plates, but in liquid culture (brain heart infusion medium [BHI; Difco-BD Biosciences, Md.] plus 5% fetal calf serum [FCS; Eurobio, Les Ulis, France]) at 37°C with microaerobic incubation (5 to 6% O₂, 8 to 10% CO₂) in the presence of these antibiotics. The MIC was determined by inoculating logarithmically growing H. pylori cells with an optical density at 578 nm (OD₅₇₈) of 0.04 in 10 ml of BHI-5% FCS in 50-ml cell culture flasks (Greiner) on a shaker incubator at 90 rpm. The MIC of metronidazole was defined as no OD₅₇₈ increase in 10 to 14 days, and that of amoxicillin was defined as growth to an OD₅₇₈ lower than 0.4, with no increase after the first 24 h of incubation. We obtained stable metronidazole resistance after three serial passages over the course of 8 to 10 days with increasing metronidazole concentration in the growth medium (from 2 µg/ml to 25 µg/ml). The metronidazole MIC for nine independently selected 69A/Mtz^r and 503/Mtz^r strains was >25 µg/ml, while that for strains 503 and 69A was <5 µg/ml (data not shown). These results correspond to the observation of metronidazole resistance developing de novo during the course of a typical antibiotic therapy (1). In vitro selection of amoxicillin-resistant H. pylori strains was done similarly. Since the amoxicillin MIC for strains 503 and 69A was 0.02 to 0.05 µg/ml, we began selection at an amoxicillin concentration of 0.01 µg/ml. After 11 passages and 35 days under the permanent selective pressure of amoxicillin, the amoxicillin MIC for strain 69A reached 0.5 to 1 µg/ml, and after 35 passages and 89 days, an amoxicillin MIC of 15 µg/ml for the highly resistant strain 69A/Amx^r was obtained. Comparable selection results were obtained for the strain 503 (data not shown). In contrast to other Amx^r strains described in the literature (9), the resistance was stable after cultivation in the absence of antibiotics and storage at 80°C. The induction of resistance was specific for the antibiotic used for selection. Amoxicillin-resistant strains remained sensitive to metronidazole and vice versa.

To investigate the molecular mechanisms for resistance, we sequenced putative resistance genes from 69A strains: five independently selected 69A/Mtz^r strains and one 69A/Amx^r strain. The rdxA gene of each 69A/Mtz^r strain had at least one mutation compared to the copy of the wild-type 69A strain (Fig. 1A). In three cases, the mutations caused stop codons in the rdxA open reading frame, which is also common for metronidazole-resistant clinical isolates (Fig. 1B). In the remaining two strains, the mutations caused either one or two amino acid changes at positions conserved in metronidazole-sensitive strains (Fig. 1B). The 69A/Amx^r strain had no rdxA mutation (Fig. 1A), but had four pbp1 mutations (S414R, Y484C, T541I, and P600T) and one pbp2 mutation (T498I). All of these mutations cause amino acid changes at positions conserved in the Amx^s strains 26695 (16), J99 (2), and 69A (GenBank accession no. AF315503 and AF315504) and were located in the putative transpeptidase domains of the proteins (10).

View larger version (76K): [in this window] [in a new window]   FIG. 1.   Alignment of RdxA proteins from H. pylori 69A strains selected for antibiotic resistance (A) and those of H. pylori wild-type strains (B). (A) The strains 69A and 69A/Amxr were metronidazole sensitive, and the 69A/Mtzr1 to -5 strains were metronidazole resistant. (B) H. pylori strains 26695, 503, 69A, and 500 were metronidazole sensitive, and H. pylori strains 439 and 504 (ATCC 43504) were metronidazole resistant. A black background shows the conservation of amino acid residues. H. pylori amino acid sequences are given as follows: 26695, reference 16; strains 500 and 439, reference 11; strain 503, GenBank accession no. AF316109; strain 504, accession no. AF315501; and strain 69A, accession no. AF315502.

To prove that these mutations were indeed responsible for antibiotic resistance, we amplified these genes by PCR, purified the DNA by gel electrophoresis, and transformed it into antibiotic-sensitive strains. To do this, we used a simplified protocol for transformation of H. pylori, in which we added linear PCR fragments without an additional resistance marker to logarithmically growing H. pylori and then selected for transformants with amoxicillin or metronidazole, respectively. After transformation with the mutated rdxA genes from four 69A/Mtz^r strains, bacteria of three metronidazole-sensitive H. pylori strains (26695, 69A, and 888-0) were rendered resistant (Table 1). Transformation with the mutated pbp1 gene from the 69A/Amx^r strain rendered bacteria of these strains (26695, 69A, and 888-0) moderately amoxicillin resistant (MIC of 0.5 to 1 µg/ml). Transformation with the pbp2 gene from 69A/Amx^r caused no amoxicillin resistance. The cotransformation of pbp1 and pbp2 did not show increased resistance compared to transformation with pbp1 alone (Table 1).

To exclude the possibility that antibiotic resistance after transformation was due to a different spontaneous mutation, we sequenced the rdxA genes from six transformed metronidazole-resistant strains and the pbp1 genes from two transformed amoxicillin-resistant strains. In all but one case, we found the same mutations as in the respective donor strain (Table 2). The only exception was observed for transformation with the rdxA gene of 69A/Mtz^r4: The rdxA gene from 69A/Mtz^r4 contained two mutations (C19Y and C184F), while H. pylori transformed with this gene acquired only the C19Y mutation (Table 2). We therefore do not know if the C184F mutation is involved in metronidazole resistance.

Our findings independently confirm previous results (7, 11, 14) and expand their data in the sense that not only stop codons and extensive deletions in the rdxA open reading frame cause metronidazole resistance, but so do single amino acid changes. We have proven that two alterations of the RdxA protein, C19Y and T49K, have the same effect on the phenotype of H. pylori (Mtz^sMtz^r) as rdxA null mutations and conclude that they severely affect RdxA function. The same is true for deletion of the C-terminal 14 amino acids of RdxA. By systematically transforming rdxA genes with single mutations into metronidazole-sensitive H. pylori strains by the simplified protocol, we have illustrated how it would be possible to identify additional residues essential for RdxA function. We have also shown that pbp1 mutations can affect amoxicillin resistance, but are not sufficient for the high-level amoxicillin resistance of 69A/Amx^r. This indicates that mutations in more than one gene are probably required to render H. pylori amoxicillin resistant. This could explain the many cycles necessary for the in vitro selection of amoxicillin-resistant H. pylori strains and their low occurrence in vivo.