Search for Novel Candidate Mutations for Metronidazole Resistance in Helicobacter pylori Using Next-Generation Sequencing

Abstract

Metronidazole resistance is a key factor associated with Helicobacter pylori treatment failure. Although this resistance is mainly associated with mutations in the rdxA and frxA genes, the question of whether metronidazole resistance is caused by the inactivation of frxA alone is still debated. Furthermore, it is unclear whether there are other mutations involved in addition to the two genes that are associated with resistance. A metronidazole-resistant strain was cultured from the metronidazole-susceptible H. pylori strain 26695-1 by exposure to low concentrations of metronidazole. The genome sequences of both susceptible and resistant H. pylori strains were determined by Illumina next-generation sequencing, from which putative candidate resistance mutations were identified. Natural transformation was used to introduce PCR products containing candidate mutations into the susceptible parent strain 26695-1, and the metronidazole MIC was determined for each strain. Mutations in frxA (hp0642), rdxA (hp0954), and rpsU (hp0562) were confirmed by the Sanger method. The mutated sequence in rdxA was successfully transformed into strain 26695-1, and the transformants showed resistance to metronidazole. The transformants containing a single mutation in rdxA showed a low MIC (16 mg/liter), while those containing mutations in both rdxA and frxA showed a higher MIC (48 mg/liter). No transformants containing a single mutation in frxA or rpsU were obtained. Next-generation sequencing was used to identify mutations related to drug resistance. We confirmed that the mutations in rdxA are mainly associated with metronidazole resistance, and mutations in frxA are able to enhance H. pylori resistance only in the presence of rdxA mutations. Moreover, mutations in rpsU may play a role in metronidazole resistance.

INTRODUCTION

Helicobacter pylori is a spiral-shaped Gram-negative bacterium that infects more than half of the world's population and is a major cause of chronic gastritis, peptic ulcer diseases, gastric cancer, and mucosa-associated lymphoid tissue lymphoma (1, 2). The eradication of H. pylori not only improves peptic ulcer healing but also prevents its recurrence and reduces the risk of developing gastric cancer (3–5). Furthermore, other H. pylori-related disorders, such as mucosa-associated lymphoid tissue lymphoma, atrophic gastritis, and intestinal metaplasia, have been shown to regress after antimicrobial therapy (6–8). Metronidazole has been used widely in combination therapies, such as metronidazole-based triple therapy, concomitant therapy, and bismuth-containing quadruple therapy, to eradicate this bacterium (5, 9, 10). Although treatment success depends on several factors, such as smoking status and patient compliance, antibiotic resistance is the major cause of treatment failure (11–13). However, along with clarithromycin resistance, resistance to metronidazole has arisen independently and is becoming increasingly common (14, 15).

Resistance to metronidazole was described previously and is predominantly associated with mutations in rdxA (hp0954), a gene encoding an oxygen-insensitive NAD(P)H nitroreductase (16–19). Mutations in two additional genes, frxA (hp0642) and frxB (hp1508, encoding a ferredoxin-like enzyme), both of which encode NAD(P)H-flavin oxidoreductases, have been shown to enhance H. pylori resistance when found along with rdxA gene mutations (20–23). However, the precise mechanism of metronidazole resistance is still debated, given that metronidazole resistance may also arise in H. pylori with mutations in the frxA gene only (22, 24). In addition, it is unclear whether other mutations in genes outside rdxA or frxA are associated with metronidazole resistance.

Recently, next-generation sequencing (NGS) has been applied to clarify the evolution and pathogenicity of H. pylori, as well as to identify its novel virulence factors (25–31). Another interesting practical application is the detection of genomic changes related to drug resistance through a comparison of the genomes of wild-type strains and of those that survive antibiotic treatment (32). Using next-generation sequencing, we are able to detect potential mutations throughout the genome of H. pylori and therefore identify novel mutations if they exist. In this study, we used next-generation sequencing to characterize the genomic changes associated with metronidazole resistance in H. pylori, where we confirmed mutations in the frxA and rdxA genes that are known to be associated with metronidazole resistance in H. pylori, and we identified a putative novel mutation in the rpsU gene that likely plays a role in metronidazole resistance.

MATERIALS AND METHODS

In vitro selection of a metronidazole-resistant H. pylori strain.

The wild-type H. pylori strain 26695 (denoted 26695-1 in our previous study [32] and susceptible to metronidazole) was obtained from a subculture of the original 26695 strain purchased from the American Type Culture Collection (ATCC). A metronidazole-resistant strain was cultured from strain 26695-1 after exposure to low concentrations of metronidazole, as described previously (24, 32, 33). Briefly, a single colony of 26695-1 was inoculated onto Mueller-Hinton II agar (Becton Dickinson, Sparks, MD, USA) supplemented with 10% defibrinated horse blood (Nippon Biotest Lab, Tokyo, Japan) without antibiotics. The plate was incubated at 37°C under microaerophilic conditions (5% O₂, 10% CO₂, and 85% N₂) for 72 h. The colonies were harvested and inoculated into brucella broth (Becton Dickinson) containing 10% horse serum. First, the culture medium was exposed to serially doubling concentrations of metronidazole (0.5, 1.0, 2.0, 4.0, 8.0, 16.0, and 32.0 mg/liter) via an agar dilution method. The colonies exposed to the maximum concentrations of metronidazole were obtained and subcultured at the same concentration 5 times before being exposed to higher concentrations. Briefly, the cultures were allowed to reach a density of a McFarland opacity standard of approximately 1.0 to 2.0, and the culture medium containing bacterial cells was directly replicated onto the metronidazole-containing dilution agar plates. The plates were incubated at 37°C under microaerophilic conditions. After being confirmed resistant to metronidazole using an Epsilometer test (Etest) (AB Biodisk, Solna, Sweden) and agar dilution method, the isolates were transferred to metronidazole-free blood agar plates ≥5 times, followed by a redetermination of the final MICs to assess the stability of this resistance.

Antibiotic susceptibility testing.

Etest and an agar dilution method were used to determine the MICs of metronidazole for the strain used in this study. Briefly, Mueller-Hinton II agar medium supplemented with 10% defibrinated horse blood was used as the culture medium, and the turbidity of the culture suspension, which was adjusted to be equivalent to a McFarland opacity standard of 1.0 to 2.0, was used for inoculation. The Etest strip of metronidazole was placed on the plate and incubated for 3 to 5 days at 37°C under microaerophilic conditions. The agar dilution MIC tests were performed according to the standard method recommended by the National Committee for Clinical Laboratory Standards (NCCLS) (34, 35). The plates contained 1.5-fold dilutions (to achieve small scales that agree with Etest) of metronidazole, with concentrations ranging from 1.0 to 128 mg/liter. The MIC was defined by the point of intersection of the inhibition ellipse zone with the graded strip for the Etest and the lowest concentration of metronidazole that completely inhibited visible growth for the agar dilution method. Strains were considered resistant when the MIC value was ≥8 mg/liter for metronidazole (34, 36). The MIC tests were performed for each putative resistant strain independently at least three times using both the Etest and agar dilution methods.

Determination of candidate mutations.

H. pylori genomic DNA was extracted using the QIAamp DNA minikit (Qiagen, Valencia, CA), according to the manufacturer's instructions. Genome resequencing was performed on the susceptible 26695-1 strain and one metronidazole-resistant strain obtained by in vitro selection using Illumina next-generation sequencing (HiSeq 2000; Illumina, Inc., San Diego, CA, USA). Raw sequencing reads (90-bp paired-end; mean insert size, 500 bp) were used to reconstruct the whole-genome sequences of both strains by mapping to the reference genome sequence of strain 26695 (GenBank accession no. NC_000915) using CLC Genomics Workbench version 4.0 (CLC bio, Aarhus, Denmark). Candidate mutations were obtained by comparing the reconstructed genomes of the two strains. The candidate mutations were confirmed by PCR using the primers listed in Table S1 in the supplemental material, followed by Sanger sequencing. The PCR in this study was carried out in 25-μl volumes containing 10× PCR buffer, 10 pmol of each primer, 2 mmol/liter MgCl₂, 200 μM each deoxyribonucleoside triphosphate (dNTP), 1 U of Ex Taq DNA polymerase (TaKaRa Bio, Inc., Otsu, Japan), and 20 to 50 ng of DNA. The PCR conditions were initial denaturation for 5 min at 94°C, 35 amplification steps (94°C for 30 s, 55°C for 30 s, and 72°C for 30 s), and a final extension cycle of 7 min at 72°C. The amplified PCR products were purified using the QIAquick purification kit (Qiagen, Inc.), according to the manufacturer's instructions, and the purified amplicons were sequenced with the BigDye Terminator version 3.1 cycle sequencing kit (Applied Biosystems, Foster City, CA, USA) using an ABI Prism 3130 genetic analyzer (Applied Biosystems), according to the manufacturer's instructions. The sequences were then aligned using Molecular Evolutionary Genetics Analysis (MEGA) 6.0 (Tempe, AZ, USA) to the reference sequence of strain 26695 deposited in GenBank (accession no. NC_000915).

Natural transformation of the candidate mutations.

The amplified PCR products containing either wild-type sequences or candidate mutations were separately introduced into metronidazole-susceptible H. pylori 26695-1 through natural transformation, as described previously (32, 37, 38). Briefly, recipient cells were inoculated onto Mueller-Hinton II agar plates and were grown for 5 h, after which 1.0 μg of PCR fragments diluted in TE (10 mM Tris-HCl [pH 8.0] and 1 mM EDTA) was added directly onto the bacterial lawn. After incubation for 24 h under microaerophilic conditions, the transformed cells were streaked onto Mueller-Hinton II agar plates containing metronidazole (1.0, 2.0, 4.0, 8.0, 16, 32, 64, and 128 mg/liter), and several single colonies were separately collected from the lowest to the highest concentrations on the metronidazole-containing plates where they were seen and spread onto metronidazole-free horse blood agar plates. The bacterial cells from each colony were harvested and diluted in brucella broth after incubation for 3 to 4 days. The culture medium was further inoculated onto metronidazole-free horse blood agar plates at least three times before being used to evaluate metronidazole susceptibility by Etest and the agar dilution method. Successful transformations and mutations were confirmed with PCR, followed by DNA sequencing analysis. For double mutation induction, amplified PCR products containing a two-base-pair insertion at position 571 (-571TA) in the frxA gene were introduced into transformants that contained a mutated rdxA PCR product. PCR fragments containing mutations in the rdxA gene obtained from a clinical metronidazole-resistant strain were used as positive-control PCR products. Amplified PCR products containing no mutations in the rdxA gene and the parental wild-type strain 26695-1 were used as a negative control. Each natural transformation was performed independently at least three times.

Nucleotide sequence accession numbers.

We deposited the genome sequences of wild-type 26695-1 and metronidazole-resistant strain 26695-1MET at GenBank under accession no. CP010435 and CP010436, respectively.

RESULTS

Establishment of resistant strain.

One metronidazole-resistant strain, denoted 26695-1MET, derived from the wild-type metronidazole-susceptible strain 26695-1, was obtained via exposure to low concentrations of metronidazole (up to 16.0 mg/liter) (Table 1). The final MICs were 128 mg/liter for 26695-1MET and 4 mg/liter for 26695-1.

TABLE 1.

Three mutations related to metronidazole resistance confirmed in 3 genes by PCR-based sequencing

Genea	Position	Mutation type	Wild typeb	Mutationc	Mutation in straind:
					26695-1MET	26695-1
frxA (hp0642)	571	Indel		TA	+	−
rdxA (hp0954)	3	SNPe	G	A	+	−
rpsU (hp0562)	37	SNP	G	T	+	−

^afrxA encodes a NAD(P)H-flavin oxidoreductase; rdxA encodes an oxygen-insensitive NAD(P)H nitroreductase; rpsU encodes the 30S ribosomal protein S21.

^bThe wild type column shows the nucleotide or segment of nucleotide in the wild-type strain.

^cThe mutation column shows the nucleotide mutation or insertion mutations found in the mutant strain.

^dStrain 26695-1MET is a metronidazole-resistant strain; 26695-1 is a wild-type strain (26695). +, nucleotide mutation or insertion mutations occurred; −, no nucleotide mutation or insertion mutations occurred.

^eSNP, single-nucleotide polymorphism.

Detection of mutations in resistant strains using next-generation sequencing.

A genome-wide analysis of mutations that differentiated the two H. pylori strains 26695-1 and 26695-1MET was undertaken using Illumina next-generation sequencing. We mapped the short-read sequences of H. pylori 26695-1 and 26695-1MET to the 26695 genome, with coverage depths of 747× and 680×, respectively. A comparison of the two strains with the reference sequence 26695 identified 36 variants (19 single-nucleotide polymorphisms [SNPs] and 17 indels), of which 13 SNPs and 9 indels were shared by the two strains and were regarded as strain-specific variants that existed before the acquisition of drug resistance. However, 14 variants (6 SNPs and 8 indels) were found exclusively in the 26695-1MET strain within 7 genes, hp0413, cag7 (hp0527), frxA (hp0642), rdxA (hp0954), rpsU (hp0562), rrnA16S (hpr04), and rrnB16S (hpr07) (see Table S2 in the supplemental material). Among them, mutations in 3 genes (frxA, rdxA, and rpsU) were confirmed by PCR and sequencing. The mutation G3A in rdxA, which is a missense mutation, changes the start codon ATG (methionine) to ATA (isoleucine). A two-base-pair frameshift insertion at position 571 (-571TA) in frxA created a premature stop codon (TAA). The G38T mutation in the rpsU gene did not create a premature stop codon; however, it was responsible for an amino acid change from an aspartic acid (D) to tyrosine (Y) at the 13th amino acid of the rpsU protein sequence.

Confirmation that the mutations are involved in resistance using natural transformation.

To determine whether the three mutations in the three genes frxA, rdxA, and rpsU were necessary and sufficient to mediate metronidazole resistance, the mutated PCR products were transformed into metronidazole-susceptible H. pylori strain 26695-1 using natural transformation. The transformed cells were selected on Mueller-Hinton II agar plates supplemented with serial concentrations of metronidazole by using an agar dilution method (from 1.0 to 128 mg/liter by double dilution). Three candidate mutations were separately introduced into wild-type strain 26695-1 (Table 2).

TABLE 2.

Three candidate mutations introduced into strain 26695-1 via natural transformation using the agar dilution method under metronidazole selection

PCR products for genes containing mutationsa	Transformants recovered at maximum metronidazole concn (mg/liter) of:
frxA(-571TA)	2.0
rdxA(G3A)	8.0
rdxA(C46T, G238A, G352A)	32
rpsU(G37T)	2.0
rdxA (G3G)	2.0
Strain 26695-1	2.0

^aBold type indicates that the transformant showed resistance to metronidazole. PCR products containing three mutations C46T, G238A, and G352A in rdxA obtained from the clinical sample were introduced into the wild-type strain 26695-1 used as a positive control. PCR products without any mutations in rdxA were introduced into the wild-type strain 26695-1 and used as a negative control. Wild-type strain 26695-1 was also used as a negative control.

The colonies from a candidate mutation in rdxA [named rdxA(G3A)]were successfully obtained from plates containing 4.0 to 8.0 mg/liter metronidazole (Table 2). At least 8 colonies from the plates with the lowest to the highest metronidazole concentrations were obtained for further evaluation of MICs, and the final MICs were 16 mg/liter for rdxA(G3A) (Table 3). The corresponding mutation in each transformant was confirmed with PCR-based sequencing. No mutations in the frxA and rpsU genes were confirmed in these selected transformants. We did not obtain any transformants for the candidate mutations in frxA and rpsU, even on the plates containing 2.0 mg/liter metronidazole. PCR fragments containing 3 mutations, C46T, G238A, and G352A, in the rdxA gene obtained from a clinical metronidazole-resistant strain (MIC, 48 mg/liter) from our previous study (39) were used as positive-control PCR products. Several colonies (named rdxA[C46T, G238A, and G352A]) were observed on the plates beginning with metronidazole concentrations from 4.0 to 32 mg/liter (Table 2). At least 8 colonies from these metronidazole-containing plates were obtained for further evaluation of MICs and confirmation of the mutation. The final MICs of all selected rdxA(C46T, G238A, and G352A) colonies were 48 mg/liter (Table 3). None of three candidate mutations in rdxA, frxA, and rpsU were observed in all positive-control transformants with PCR-based sequencing. To control for spontaneous mutation, the PCR products that were amplified with the same primers for the rdxA gene but without any mutations were introduced into the metronidazole-susceptible strain 26695-1 as a negative control. We did not observe any colonies in repeated experiments, even on the plates containing 4.0 mg/liter metronidazole (colonies were observed only on the plates containing 2.0 mg/liter metronidazole). We also used the metronidazole-susceptible parental strain 26695-1 without transformation and showed that the colonies were obtained only on the plates containing 2.0 mg/liter metronidazole.

TABLE 3.

PCR-based sequencing results and final MICs of successful transformants

Transformants with mutationsa	rdxAb	frxAc	rpsU	Final MICs (mg/liter)
rdxA(G3A)	G3A	WT	WT	16
rdxA(G3A) + frxA(-571TA)	G3A	-571TA	WT	64
rdxA(C46T, G238A, G352A)	C46T, G238A, G352A	WT	WT	48
Strain 26695-1	WT	WT	WT	4.0

^aTransformants that carried mutations in the rdxA gene obtained from clinical samples were used as a positive control. Wild-type strain 26695-1 was used as a negative control.

^bG3A, the substitution of guanine to adenine at position 3 in the rdxA gene.

^cWT, wild type; -571TA, two-base-pair insertion at position 571 in the frxA gene.

C46T, G238A, and G352A, the substitution of cytosine to thymine, guanine to adenine, and guanine to adenine at position 46, 238, and 352 in rdxA gene, respectively.

To evaluate the synergic effects of the candidate mutations, further natural transformations were performed with the same method as that used for single transformation. Amplicons containing -571TA in frxA obtained from 26695-1MET were introduced into the rdxA(G3A) transformants, and we obtained transformants containing double mutations in the frxA and rdxA genes, which were confirmed with PCR-based sequencing. These transformants showed higher MICs (64 mg/ml) than those with single mutations.

DISCUSSION

Metronidazole has been widely used in combination therapies with other antimicrobials to eradicate H. pylori and other anaerobic bacterial infections (5, 9, 10). Resistance to metronidazole, a significant cause of H. pylori treatment failure (11–13), is becoming increasingly common in many countries worldwide (14, 15). It is now accepted that metronidazole resistance is predominantly caused by mutations in rdxA (16–19) and that mutations in frxA enhance the resistance of rdxA gene mutations (20–23). However, metronidazole-resistant strains without mutations in the rdxA and frxA genes have been reported, indicating that additional genes are involved in metronidazole resistance (40–44). In this study, we successfully cultured a single metronidazole-resistant strain under the selection of low metronidazole concentrations from a susceptible strain and identified seven genes with a number of genetic variants associated with metronidazole resistance using next-generation sequencing. However, mutations in the 4 genes, including hp0413, cag7, rrnA16S, and rrnB16S, were not confirmed by PCR sequencing. This probably is because these mutations are mostly located in the repeated region of these gene sequences where next-generation sequencing (NGS) may not read well. Surprisingly, only a single sequence in addition to that in the two well-known rdxA and frxA genes related to metronidazole resistance was confirmed to contain a variant associated with resistance. Finally, we confirmed mutations in the rdxA and frxA genes known to be associated with metronidazole resistance through natural transformation experiments.

Using natural transformation experiments, we confirmed that transformants containing three mutations, C46T, G238A, and G352A, in the rdxA gene used as a positive control showed moderate resistance to metronidazole (MIC, 48 mg/liter). Mutation G3A in rdxA was successfully transformed into metronidazole-susceptible strain 26695-1, which had a low metronidazole MIC (16 mg/liter). Although the methodology has been established and used not only for experiments on metronidazole resistance (38, 45–48) but also for clarithromycin (32), tetracycline (49), fluoroquinolones (50, 51), and amoxicillin (52, 53), spontaneous mutations may have occurred under the metronidazole selection introduced during our experiment. However, no transformants were obtained with the plates containing 2.0 mg/liter metronidazole when the metronidazole-susceptible strain 26695-1 was transformed with PCR products without any mutations, which suggests that spontaneous mutations are unlikely to account for the resistant phenotype observed.

Transformants containing two mutations of -571TA in frxA and G3A in rdxA showed resistance to metronidazole with higher MICs than those with the transformants containing only G3A in rdxA (48 mg/liter versus 16 mg/liter, respectively). These results clearly demonstrate the presence of synergistic effects and that mutations in frxA contribute to the achievement of higher MICs in the presence of rdxA gene mutations. These findings support the hypothesis that mutations in the rdxA gene are the main mechanism of metronidazole resistance in H. pylori and that mutations in the frxA gene can increase the degree of resistance in the presence of mutations in the rdxA gene, as described in previous studies (16–23).

We were unable to obtain transformants containing a mutation at -571TA in frxA alone, although this mutation created a premature stop codon (TAA), which suggests that it may have caused metronidazole resistance; therefore, we hypothesize that the inactivation of frxA itself cannot induce metronidazole resistance. In contrast to other studies that concluded that mutations in frxA without mutations in rdxA might cause metronidazole resistance (22, 24, 40), we suggested that this may be due to other additional but uncharacterized mechanisms of metronidazole resistance that coexist with the mutations in the frxA gene (24).

We have identified, for the first time, the rpsU gene to be associated with metronidazole resistance. rpsU encodes a 30S ribosomal protein, S21, which is involved in protein synthesis (54–56), and was previously identified with antibiotic-resistant recombinants in Escherichia coli (57), which suggests that it may play a role in metronidazole resistance in H. pylori via an unknown mechanism. We found 2 metronidazole-resistant clinical isolates from Vietnam that had two different missense mutations in rpsU but not metronidazole-susceptible clinical isolates, in which one strain had mutations in rdxA gene and the other one did not have a mutation in either rdxA or frxA (positions G100A and G123T; T. T. Binh and Y. Yamaoka, unpublished observation). However, we were unable to obtain transformants using PCR products containing mutations in this gene. We do not know whether the transformants did not have a strong enough metronidazole-resistant phenotype to be selected under metronidazole selection when this mutation was introduced into parental strain 26695-1, because we used simple PCR products with a mutation for natural transformation, indicating the absence of selection with antibiotic cassettes (e.g., chloramphenicol selection). We are now trying to transform this candidate mutation into 26695-1 via selection by using chloramphenicol cassettes to confirm whether it can play roles in metronidazole resistance. Furthermore, our study has several limitations. We did not obtain metronidazole-resistant strains without mutations in the frxA and rdxA genes in order to confirm the presence of other mutations outside these two genes that are associated with metronidazole resistance; therefore, further work is required to identify the role of mutations in addition to those known in the frxA and rdxA genes. On the other hand, it is well known that next-generation sequencing alone cannot read the whole genome, as one contig and some sequences of the genome may not be read completely, especially in the repeated regions of the DNA sequences (58, 59). Therefore, we may have missed some other mutations in other genes that may be related to metronidazole resistance. Nonetheless, we did confirm that next-generation sequencing technology can be a useful tool for screening mutations related to drug resistance.

In conclusion, we first analyzed the genome profile for metronidazole resistance in H. pylori using next-generation sequencing, showing that this technology is useful for identifying mutations that differ between a susceptible and resistant strain, and it offers a significant advantage over candidate gene approaches that examine only a fraction of the genome at any one time. Our study confirms that mutations in the rdxA gene are mainly associated with metronidazole, whereas mutations in the frxA gene enhance H. pylori resistance exclusively in the presence of rdxA mutations. Finally, the ribosomal gene rpsU may be an additional candidate associated with metronidazole resistance.