ABSTRACT
We found that NCTC11637, the type strain of Helicobacter pylori, the causative agent of peptic ulcer disease and an early risk factor for gastric cancer, is metronidazole resistant. DNA transformation, PCR-based restriction analysis, and DNA sequencing collectively showed that the metronidazole resistance of this strain was due to mutation in rdxA (gene HP0954 in the full genome sequence of H. pylori 26695) and that resistance did not depend on mutation in any of the other genes that had previously been suggested: catalase (katA), ferredoxin (fdx), flavodoxin (fldA), pyruvate:flavodoxin oxidoreductase (porγδαβ), RecA (recA), or superoxide dismutase (sodB). This is in accord with another recent study that attributed metronidazole resistance to point mutations inrdxA. However, the mechanism of rdxAinactivation that we found in NCTC11637 is itself also novel: insertion of mini-IS605, one of the endogenous transposable elements of H. pylori, and deletion of adjacent DNA sequences including 462 bp of the 851-bp-long rdxA gene.
Helicobacter pylori is a microaerophilic gram-negative bacterium that chronically infects the gastric mucosa of more than half of all persons worldwide and is a major cause of chronic active gastritis and peptic ulcer disease (23) and an early risk factor for gastric cancer (17). Antimicrobial therapy that leads to eradication of infection and ulcer healing is often achieved with metronidazole, a nitroheterocyclic compound [1-(2-hydroxyethyl)-2-methyl-5-nitroimidazole], in combination with other drugs (30). Resistance to metronidazole is common, however, and is the major factor in many cases of failure to eradicateH. pylori and cure the disease (6, 20, 24).
For metronidazole to kill H. pylori, it must be taken up by an energized membrane (22) and then reduced to form a toxic metabolite. Several mechanisms of metronidazole resistance have been proposed in recent years, including enhanced scavenging of toxic oxygen radicals by an altered catalase or superoxide dismutase (7, 11, 26), a more efficient DNA damage repair mechanism (9, 27), and loss of function of a critical reductase (15, 16). Recent studies of fresh clinical isolates by several of us (12) indicated that metronidazole resistance often results from point mutation in rdxA (HP0954, in the fully sequenced genome of strain 26695 [28]), a gene that encodes an oxygen-insensitive NADPH nitroreductase. However, some researchers have questioned the generality of this interpretation (13, 21).
Here we report that NCTC11637, the type strain of H. pylorithat Barry Marshall had isolated in Australia and that we obtained from the American Type Culture Collection, displays a high level of resistance to metronidazole and that this resistance is attributable to mutational inactivation of rdxA. Our experiments further suggest that mutation in other genes, previously invoked to explain the metronidazole resistance phenomenon, does not contribute to the high-level metronidazole resistance of NCTC11637 (katA[catalase], fdx [ferredoxin], fldA[flavodoxin], porγδαβ [pyruvate:flavodoxin oxidoreductase], recA [RecA], and sodB[superoxide dismutase]).
MATERIALS AND METHODS
Bacterial strains and growth conditions. H. pyloristrains used in this study were NCTC11637 (MIC of >256 μg/ml) (obtained from the American Type Culture Collection as ATCC 43504), which was found here to be metronidazole resistant (Mtzr) (3), and the unrelated, metronidazole-susceptible (Mtzs) strain NCTC11638 (MIC of <0.032 μg/ml). Bacteria were routinely grown on Dent plates (Columbia agar plates supplemented with 7% lysed horse blood and H. pylori selective supplement [Oxoid, Basingstoke, United Kingdom]). Plates were incubated at 37°C in a microaerophilic atmosphere of 5% O2–10% CO2–85% N2.
Determination of MIC.MICs were determined with the Etest as described before (10).
Natural transformation.Metronidazole-resistant NCTC11638 transformants were created by natural transformation of NCTC11638 (Mtzs) with DNA from the NCTC11637 (Mtzr) strain, essentially as described by Wang et al. (31). As a control, bacteria were transformed with TE (10 mM Tris-HCl [pH 8.0], 1 mM EDTA) instead of DNA. Transformants (selected on Dent plates containing 32 μg of metronidazole per ml) were generated at a frequency of 10−5 per viable cell. From one such experiment, 12 individual colonies were selected at random, and the metronidazole MICs were determined.
PCR.Unless noted otherwise, standard techniques were used for all DNA manipulations (5). Primers (Table1) for PCR amplification were based on the full genome sequence of strain 26695 (28) plus other entries in the public database. PCR was performed in an automated thermal cycler (Mastercycler 5330; Eppendorf, Hamburg, Germany), in a final volume of 100 μl by using the Primezym DNA polymerase kit (Biometra, Göttingen, Germany), with approximately 2.5 ng of template genomic DNA and 100 pmol of each primer. Amplification conditions are listed in Table 1. The size of the PCR products was evaluated by electrophoresis on a 1.5% agarose–0.5× Tris-borate-EDTA gel containing 0.5 μg of ethidium bromide per ml.
Oligonucleotide primers used in this study
Cloning and sequencing of the PCR products.PCR products were ligated into the pGEM-T vector (Promega, Madison, Wis.) according to the manufacturer’s recommendations. The ligation products were then transformed into Escherichia coli ER1793 (18), and transformants were selected on Luria-Bertani plates supplemented with 100 μg of ampicillin per ml. Plasmid DNA was isolated from transformants with the QIAprep spin plasmid (Qiagen, Hilden, Germany). These plasmids were then used as templates for DNA sequencing with the Thermo-Sequenase premixed cycle sequencing kit (Amersham, Little Chalfont, Buckinghamshire, United Kingdom) with the M13 forward and reverse primers. Sequencing was performed on an Amersham Vistra 725 DNA sequencer, and data were analyzed with Lasergene software (DNASTAR, Madison, Wis.).
Restriction fragment length polymorphism (RFLP).Approximately 300 ng of PCR product was digested with restriction endonucleases (New England Biolabs, Beverly, Mass.). The endonucleases used for RFLP analysis of the various genes are indicated in Table 1. The restriction enzyme digestion profile was subsequently analyzed by electrophoresis on a 1.5% agarose gel.
RESULTS
Metronidazole-resistant transformant derivatives of NCTC11638.The type strain, NCTC11637, is resistant to metronidazole (MIC of >256 μg/ml). We used a DNA transformation and PCR product characterization strategy to determine the basis of its resistance. This was based on findings that Mtzr determinants can be transferred to Mtzs strains by transformation (31) and that the differences in distributions of restriction sites in a given gene from unrelated strains usually allow those strains to be distinguished (1). Mtzr (32 μg/ml) transformant derivatives of NCTC11638 were obtained at a frequency of approximately 10−5 per recipient cell with NCTC11637 genomic DNA, in contrast to a frequency of <10−7 per recipient cell in DNA-free mock-transformation controls. Each of the 12 transformants tested, although selected at 32 μg/ml, was resistant to more than 256 μg/ml, as was its Mtzr NCTC11637 parent.
RFLP analysis of the selected genes.A number of different genes (fdx [ferredoxin], fldA [flavodoxin],porγδαβ [pyruvate:flavodoxin oxidoreductase],recA [RecA], katA [catalase], andsodB [superoxide dismutase]) had been postulated by various other groups to contribute to high-level Mtzr. The possible involvement of any of them was tested by scoring whether the Mtzr transformants contained NCTC11637-type or NCTC11638-type alleles at these loci in an RFLP test. First, primers for amplification of these genes were designed (Table 1), based on the sequences in the public databases. PCR products of the expected size were obtained from the NCTC11637 donor DNA, from the NCTC11638 recipient, and from each transformant.
To maximize the power of RFLP analysis and to get a better sense of the sequence divergence of these two much-studied strains, we PCR amplified and sequenced each of the candidate genes and thereby identified restriction sites that differed usefully between the parental strains. The PCR product corresponding to each candidate gene from each of the 12 Mtzr transformants was then analyzed by restriction with appropriate enzymes (Table 1).
In each case, except for rdxA (see below), the results showed unambiguously that the Mtzr transformants contained alleles of the Mtzs NCTC11638 recipient at each of these loci. This is exemplified by the Sau3AI and MseI RFLP patterns of sodB shown in Fig.1. Thus, the ability of NCTC11637 genomic DNA to transform NCTC11638 to Mtzr did not depend on acquisition of special donor alleles of any of these previously suggested candidate genes.
Ethidium bromide-stained agarose gel displaying the restriction endonuclease digestion pattern of the sodBamplicon. (A) Digestion with Sau3AI; (B) digestion withMseI. Lane 1, H. pylori NCTC11638 (Mtzs); lane 2, H. pylori NCTC11637 (Mtzr); lanes 3 to 14, H. pylori NCTC11638 transformants (Mtzr). The positions of the molecular size markers (1-kb ladder; Stratagene) are indicated on the left.
In contrast to these negative results, PCR amplification from NCTC11637rdxA, the gene whose inactivation had been implicated recently in the Mtzr of fresh clinical isolates, yielded a product of 650 bp, 200 bp smaller than expected from the published sequences and as observed with NCTC11638 (Fig.2). Each of the 12 Mtzrtransformants of NCTC11638 also yielded a product of only 650 bp with the rdxA-specific primers. This result indicated that the Mtzr of NCTC11637 was due to mutation in rdxA. Independent confirmation for the involvement of rdxA in the Mtzr phenotype of strain NCTC11637 was obtained by transformation of wild-type NCTC11638 with the aberrant 650-bp PCR product obtained from NCTC11637. This resulted in high numbers of Mtzr transformants, while no such transformants were observed in the control transformation, where we used the 850-bprdxA PCR fragment obtained from strain NCTC11638.
Ethidium bromide-stained agarose gel of therdxA amplicon. Lane 1, H. pylori NCTC11638 (Mtzs); lane 2, H. pylori NCTC11637 (Mtzr); lanes 3 to 14, H. pylori NCTC11638 transformants (Mtzr). The positions of the molecular size markers (1-kb ladder; Stratagene) are indicated on the left.
Sequence analysis of the rdxA genes.The basis of the unexpected shortness of the NCTC11637 rdxA locus was determined by sequencing. In the donor strain and also in the transformants analyzed, we found an insertion of a novel 260-bp segment just before the start of HP0953 (the gene just downstream ofrdxA) and a deletion of a 462-bp segment including 177 bp from the 3′ end of rdxA (Fig.3). A BLAST search showed that 41 and 33 bp at the left and right ends, respectively, of this 260-bp element are closely matched to the corresponding left and right ends of the 2-kb insertion sequence (IS) IS650, whereas its central 186 bp were unrelated to sequences in full-length IS605(2). Rather, this segment corresponded to the mini-IS605 (Fig. 4) that was found at the right end of the cag pathogenicity island in one strain (NCTC11638, where it is referred to as is605, in contrast to IS605) (8). The genome sequence of the fully sequenced strain 26695 contains eight such mini-IS650 elements (Fig. 4), but none is in itscag pathogenicity island (28). Just after we submitted the manuscript, the complete genomic DNA sequence of H. pylori J99 was published (4), and this sequence reveals the presence of five partial IS605 elements (Fig. 4).
Alignment of the DNA sequences and the predicted encoded proteins of the rdxA amplicons. Mtzs, H. pylori NCTC11638; Mtzr, H. pylori NCTC11637 and the metronidazole-resistant transformants of NCTC11638. The sequence of the donor NCTC11637 amplicon is identical to that of all the metronidazole-resistant NCTC11638 transformants. The primers used to amplify the rdxA region (and amino acids derived therefrom) are indicated in boldface; the inverted repeats of the IS605 (7) are in boldface and underlined. The amino acids directly above the DNA sequence correspond to H. pylori NCTC11638; the ones directly below the DNA sequence correspond to NCTC11637 and the transformants. Gaps in the DNA sequence are marked with dashes, and identical nucleotides are indicated by dots.
Alignment of the mini-IS605 sequence present in the rdxA gene from strain NCTC11637 (see Fig. 3); thecag pathogenicity island (cag PAI) of strain NCTC11638 (7), GenBank accession no. U60176 bp 22476 to 22735; one of the eight partial copies of IS605 present in strain 26695 (27), GenBank accession no. HPAE000537 bp 2867 to 2606; and one of the five partial copies present in strain J99 (4), GenBank accession no. HPAE000537 bp 2633 to 2891. The inverted repeats of IS605 (7) are indicated in boldface and underlined. For definitions of the dots and dashes, please see the legend to Fig. 3.
DISCUSSION
The results of our study of the type strain NCTC11637 confirm the conclusions of Goodwin et al. (12) that inactivation ofrdxA is responsible for much or all of the high-level metronidazole resistance so often seen with H. pyloriclinical isolates. There was no evidence that mutation insodB, katA, ferredoxin, flavodoxin,porγδαβ, or recA genes made any contribution to the high-level metronidazole resistance of NCTC11637. Although the PCR products for the above loci did not always cover the complete gene, it is generally assumed (as was confirmed by our analysis of the rdxA gene sequences) that natural transformation results in the exchange of a large fragment of DNA. Hence, not finding any indications for gene exchange in the central portion of these six genes in 12 independent Mtz-resistant transformants provides strong evidence against the involvement of these genes and the loci adjacent to them. However, our data do not exclude the possibility that, for the low-level resistance that is observed with some H. pylori strains (13), these genes may still play a role. Only determination of enzyme activities can provide the definitive proof for the relative contribution of each gene in the Mtzr of an individual strain.
All Mtzr mutations in rdxA that Goodwin et al. had observed were small point mutations (base substitutions), whereas the loss of RdxA function in NCTC11637 was caused by a deletion of much of the rdxA gene, associated with insertion of mini-IS605. This is a novel transposable element that does not contain any significant open reading frame and that accordingly, we conjecture, may depend on another, larger element such as full-length (2 kb) IS605 for its movement. Associated with mini-IS605 in this strain was a deletion that removed 462 bp of rdxA. Several different IS elements that are not related to IS605 or mini-IS605 had been shown in other species to generate deletions adjacent to their sites of insertion (14, 25, 29). In two cases at least (IS1 and IS50) (25, 32), deletion is independent ofrecA function, and we propose that an equivalent IS element-mediated adjacent-deletion process may have operated here. In one scenario, mini-IS605 was inserted betweenrdxA and the adjacent HP0953 gene, leaving a strain that was still metronidazole susceptible but highly prone to mutate to Mtzr by adjacent deletion. An alternative possibility of insertion and concomitant deletion has not been ruled out.
This is the first report of a transposon-based spontaneous mutation associated with antibiotic resistance in H. pylori. However, some one-third of strains from Europe and the Americas carry full-length IS605, a similar fraction carry the distantly related IS606 element, and more than half of the strains seem to carry mini-IS605 (whether full-length IS605 is present or not) (19). Endogenous IS elements are well known in other species as agents of spontaneous mutation and genome rearrangements and evolution. We propose that they may function similarly in H. pylori, perhaps often contributing to drug resistance in the face of mounting antibiotic usage and other adaptive changes of considerable evolutionary significance.
ACKNOWLEDGMENTS
This work was supported by the Netherlands Digestive Disease Foundation, the Medical Research Council of Canada grant R-14292 and Astra Canada, and the National Institutes of Health (DK48029, AI38166, and HG00820).
We thank A. B. Brinkman for his technical support.
FOOTNOTES
- Received 19 January 1999.
- Returned for modification 19 April 1999.
- Accepted 27 August 1999.
- Copyright © 1999 American Society for Microbiology
