Molecular Characterization of Benzimidazole Resistance in Helicobacter pylori

Abstract

A family of benzimidazole derivatives (BI) was shown to possess potent and selective activity against Helicobacter pylori, although the precise cellular target of the BIs is unknown. Spontaneous H. pylori mutants were isolated as resistant to a representative BI (compound A). Genomic DNA was isolated from a BI-resistant mutant, transformed into a BI-sensitive strain, and found to be sufficient to confer BI resistance. The resistance determinant was localized to a 17-kb clone after screening a lambda-based genomic library constructed from the BI-resistant strain. Upon sequencing and mapping onto the H. pylori strain J99 genome, the 17-kb clone was shown to contain the entire nuo operon (NADH:ubiquinone oxidoreductase). Further subcloning and DNA sequencing revealed that a single point mutation in nuoD was responsible for BI resistance. The mutation resulted in a G398S amino acid change at the C terminus of NuoD. Thirty-three additional spontaneous BI-resistant mutants were characterized. Sequencing of nuoD from 32 isolated mutants revealed three classes of missense mutation resulting in amino acid changes in NuoD: G398S, F404S, and V407M. One BI-resistant isolate did not have a mutation in nuoD. Instead, a T27A amino acid change was identified in NuoB. MIC testing of the wild-type H. pylori strain and four classes of nuo mutants revealed that all NuoD mutant classes were hypersensitive to rotenone, a known inhibitor of complex I (NADH:ubiquinone oxidoreductase) suggested to bind to NuoD. Further, a nuoD knockout verified that it is essential in H. pylori and may be the target of the BI compounds.

Helicobacter pylori is a microaerophilic gram-negative, spiral-shaped, motile bacterium that colonizes the gastric mucosa of humans (35). H. pylori is the leading cause of chronic gastritis and peptic ulcer disease and is associated with certain types of gastric cancer (3, 33). Currently, the most effective anti-H. pylori therapy consists of a proton pump inhibitor, such as omeprazole, plus two antimicrobials, commonly chosen from amoxicillin, metronidazole (MTZ), clarithromycin (CLR), and tetracycline (TET), and is taken two to four times daily for 7 to 14 days (21, 22). Several factors impact the success of a specific therapy. One such factor is the prevalence of drug-resistant H. pylori strains in different geographical areas. A study by Katelaris et al. revealed that 53% of the H. pylori strains evaluated were MTZ resistant, 8% were CLR resistant, and only 0.7% were TET resistant (22). Resistance continues to be of increasing concern, as it contributes to failed therapy. In addition, the current therapies rely on agents with broad-spectrum antimicrobial activities and thus disturb the normal balance of commensal microorganisms colonizing the gastrointestinal tract. This, in turn, leads to unpleasant side effects such as diarrhea and abdominal pain, which in many cases lead to poor patient compliance and premature termination of therapy. The Katelaris et al. study found that 8% of the total patient population took less than 90% of the prescribed medication (22). Reduced patient compliance may potentially lead to drug resistance, not only in the colonizing H. pylori but also in the normal flora and other opportunistic pathogens. Resistant clinical isolates of H. pylori have been identified for each of the major antibacterial classes used to treat infection, including amoxicillin (25), TET (8), CLR (16), and MTZ (24). There is currently a medical need for an effective anti-H. pylori therapy that relies on a single novel agent that selectively eradicates drug-resistant H. pylori and improves patient compliance.

Benzimidazoles (BIs) were previously synthesized that possess potent anti-H. pylori activities with MICs at which 90% of isolates were inhibited of ≤0.5 μg/ml for a diverse panel of 27 H. pylori strains, including several strains resistant to MTZ and CLR (6). In addition, many of these compounds have been characterized as being highly selective for H. pylori, with MICs at which 90% of isolates were inhibited of >64 μg/ml for a diverse collection of 25 aerobic bacterial strains and 18 anaerobic strains (6). These are attractive properties (e.g., potent and selective) for a prospective anti-H. pylori-specific agent to possess.

The cellular target of the BIs that is responsible for this highly potent and selective activity is currently unknown. The study of antimicrobial resistance is one means to gain an understanding of the mode of action and to identify the cellular target of an antimicrobial agent. In addition, resistance can be studied to examine the potential for resistance development throughout the research and development process. The intention of this study was to identify the mechanism of H. pylori resistance to BIs with the idea that it may help identify the molecular target for this structural class of compounds.

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture conditions.

The key features of all relevant strains and plasmids are listed in Table 1. H. pylori strains were grown under microaerophilic conditions (5% [vol/vol] O₂, 10% [vol/vol] CO₂, 85% [vol/vol] N₂) at 37°C on blood agar plates (BAP) containing tryptic soy agar (Difco Laboratories, Detroit, Mich.) supplemented with 5% (vol/vol) defibrinated sheep blood (Becton Dickinson and Company, Sparks, Md.) or in brucella broth supplemented with 5% fetal bovine serum (Gibco BRL [Invitrogen; Carlsbad, Calif.]). Where appropriate, kanamycin was added to a final concentration of 25 μg/ml. H. pylori cultures were typically incubated for 48 h after routine subculture and for 3 to 5 days after plating for selection of transformants.

TABLE 1.

Strains and vectors

Strain or vector (clone)	Relevant genotype, phenotype, or sequence	Source or reference
H. pylori strains
AH244	Wild type	1
J99	Wild type	2
J99A.1	Compound A-resistant J99; NuoD G398S	This study
J99A.2	Compound A-resistant J99; NuoD F404S	This study
J99A.3	Compound A-resistant J99; NuoD V407M	This study
J99A.4	Compound A-resistant J99; NuoB T27A	This study
E. coli strains
ER1793	e14(mcrA6) Δ(mcrC-hsdRMS-mrr) 114::IS10	27; New England BioLabs
TOP10	F−, mcrA Δ(mrr-hsdRMS, mcrBC) φ80lacZΔM15 endA1 recA1	20; Invitrogen
ECO523	K-12 derivative; F−mcrA mcrB IN(rrnD-rrnE)1 lambda−	ATCC 27325 (W3110)
ECO524	ECO523 with Tn10 insertion in tolC
Vectors (clones)
LambdaGEM-11	Lambda-based cloning vector	Promega
Clone 17	Lambda GEM-11 containing 17,083-bp fragment from J99 genomic library containing compound A resistance gene	This study
pUC19	Cloning vector, Apr	36
pGEM-T	Cloning vector, Apr	Promega
pILL600	pBR322 containing C. coli aphA-3 gene, Kmr	23
pSM126	pUC19 containing 5,707-bp XbaI-SacI fragment from clone 17, Apr	This study
pSM127	pUC19 containing 4,032-bp SacI-XbaI fragment from clone 17, Apr	This study
pSM128	pUC19 containing 1,908-bp SacI fragment from clone 17, Apr	This study
pSM129	pUC19 containing 5,465-bp SacI fragment from clone 17, Apr	This study
pSM130	pUC19 containing 3,017-bp SacI-EcoRI fragment from pSM129, Apr	This study
pSM131	pUC19 containing 807-bp EcoRI fragment from pSM129, Apr	This study
pSM132	pUC19 containing 1,641-bp EcoRI-SacI fragment from pSM129, Apr	This study
pSM135	pSM132 containing 718-bp deletion in nuoD replaced by aphA-3 in the same transcriptional orientation	This study

Escherichia coli strains were grown aerobically at 37°C on Luria-Bertani (LB) agar or in broth supplemented with either ampicillin at 100 μg/ml or kanamycin at 50 μg/ml.

Antimicrobial compounds.

Ampicillin, kanamycin, piericidin A, and rotenone were obtained from Sigma (St. Louis, Mo.). The BI derivative, compound A, was prepared at AstraZeneca R&D Boston as previously described (Fig. 1) (6).

FIG. 1.

Structure of BI-containing compound A.

General DNA manipulations.

Standard molecular biology protocols were used for PCR, DNA cloning, agarose gel electrophoresis, and sequencing (30). Relevant oligodeoxynucleotide primers used in this study are listed in Table 2. Restriction endonucleases were purchased from New England BioLabs (Beverley, Mass.). The Rapid DNA ligation kit (Roche Diagnostics Corp., Indianapolis, Ind.) was used for DNA cloning. High-fidelity PCR Supermix (Invitrogen) was used to amplify DNA fragments for cloning, transformation, and sequence analysis. All PCR-generated clones and selected PCR-generated DNA fragments were sequenced (both strands) using an ABI Prism 3100 genetic analyzer after preparing ABI Prism BigDye Terminator cycle sequencing version 2.0 ready reactions (PE Biosystems, Foster City, Calif.). The resulting DNA sequence chromatographs were assembled and analyzed using Sequencher software version 4.0.5 (Gene Codes Corporation, Ann Arbor, Mich.). Oligodeoxynucleotide primers used for PCR and sequencing were synthesized by Invitrogen (Table 2). DNA was extracted from agarose gels with the QIAEX II kit (QIAGEN Inc., Valencia, Calif.).

TABLE 2.

Oligodeoxynucleotide primers

Primer name	Sequence
T7 pro	TAATACGACTCACTATAGGG
SP6 pro	ATTTAGGTGACACTATAGAATACT
JHP-1184-F	GTCATATGGCTCAAAATTTCACGAAACTCAACC
JHP-1184-R	CAAAGAATTGAGGAGCTCATGCC
CF-1	CATTTTTGAACATGACGATAACC
CF-2	CGATTTTACTAGAGCTTAACCG
CF-3	GAAAATGTGGGCGTTGTAACG
CF-4	GACGCAAAACTACGCCTTGATGC
CR-1	GGGATTTTCTTAAACGCATGC
CR-2	GTGGGCATGTATTCGTTATAGG
CR-3	CGCCCCCAATCCTTATAGCG
CR-4	CAATACCGATCATAACTATCGC
CR-5	GATGTGATAAAAGCTAGGGGC
Nuo-1F	ATTTTGCACTTGCATGGGGAATTG
Nuo-1R	GGGTAATTTTTAATCCATTGGGTGC
Nuo-2F	GCTCTTCAATTAGTTTGAAAGAAGCC
Nuo-2R	CATTTCTTCAAACTGCATGCCTGCG
Nuo-3F	CCTTATTTCTGACATCAACATTGGC
Nuo-3R	GATTAAAATCGCTATGCCCCCAGGG
Nuo-4F	GGTGTTGTTGAGCATCGCTTATGCC
Nuo-4R	CGCTTAAAAGGGTGTATAAAAAGGC
UM-nuo-F1	GCACTTGCATGGGGAATTG
UM-nuo-F2	CCAAACTCATAGAAATGG
UM-nuo-F3	GTTTTCATGTTCCCTTGG
UM-nuo-F4	CTGATGTGATGATCATCG
UM-nuo-F5	GAGGGTAGCGCTTATGAGG
UM-nuo-F6	CATGATTGGGTAGGCCACC
UM-nuo-F7	GTAGCCATGGGCAATTGC
UM-nuo-F8	GTTGGATTTGATGGAGG
UM-nuo-F9	GAAATTGATGAAAGCATTCG
UM-nuo-F10	GGTGGATAGATGAAACGC
UM-nuo-F11	GAAAAAAGCGATTTGTGCG
UM-nuo-R1	GTATAAAAAGGCGTTGTTAG
UM-nuo-R2	GTAATTGATCTTAAACCC
UM-nuo-R3	GCATCCGCTAAATATTGC
UM-nuo-R4	CTATCGCCATAATTGCCC
UM-nuo-R5	CAAAAACACGCTCATCGC
UM-nuo-R6	CATTTGGTTATCGTCATG
UM-nuo-R7	GCCAAGCATGTCATACGC
UM-nuo-R8	CTGGCGCGATTGTTTTTGC
UM-nuo-R9	CAAAATCAAACCTTGAACC
UM-nuo-R10	GACACCCTATTAGGCTGC
UM-nuo-R11	CAATGTAATAAATGAGGG

E. coli strain TOP10 competent cells (Invitrogen) were used for all plasmid-based DNA cloning and propagation, and E. coli strain ER1793 (New England BioLabs) was used for all lambda-based DNA cloning and propagation. Transformation of E. coli cells was performed as described by the manufacturer (Invitrogen).

Genomic DNA for PCR analysis and transformation was isolated using the Wizard genomic DNA purification kit, plasmid DNA was prepared by alkaline lysis using the Wizard Plus plasmid miniprep kit, and phage DNA was purified using the Wizard Lambda Preps DNA purification system according to the manufacturer's instructions (Promega, Madison, Wis.). Genomic DNA for library construction was prepared from H. pylori using a published modification of the guanidine lysis protocol of Pitcher and coworkers (26, 28). This protocol was used in place of the Wizard genomic DNA purification kit to reduce the amount of chromosomal shearing.

Bacteriological methods.

Mutation frequencies were determined as follows: H. pylori was grown in broth culture to mid-exponential phase (optical density at 600 nm [OD₆₀₀], ∼0.8), centrifuged, and suspended in brucella broth to a concentration of 100 OD₆₀₀, of which 10 μl (∼10⁸ cells) was spotted onto BAP for overnight incubation. The following morning, each spot containing approximately 10¹⁰ cells was suspended in 1 ml of brucella broth and dilution plated onto BAP containing no compound (viable counts) or compound A at 2 times the MIC (resistant counts). The mutation frequency was calculated by comparing the number of viable colonies on BAP with the number of colonies observed on compound A-containing plates. MICs were determined by plating 5 μl of brucella broth containing 10⁶ H. pylori cells, grown as described above, onto media containing twofold-increasing concentrations of compound A. The minimum concentration defining the threshold limit for growth was reported as the MIC.

Transformation of H. pylori.

H. pylori was transformed with DNA as previously described (34). The source of DNA (sheared genomic, restriction endonuclease digested, PCR generated, or plasmid) did not impact the method of transformation. Briefly, H. pylori was grown to an OD₆₀₀ of 0.5 to 0.8. The cells were then concentrated by centrifugation and suspended to a final concentration of OD₆₀₀ = 100 in brucella broth. Twenty microliters was spotted onto BAP and incubated for 2 h at 37°C under microaerophilic conditions. Next, approximately 1 μg of DNA was applied directly onto the spotted H. pylori cells (sterile water was used as a negative control) and incubated overnight. The spotted cells were suspended into 1 ml of brucella broth and spread directly onto BAP containing the appropriate antimicrobial compound for selection of transformants. Colonies would generally appear within 3 to 5 days.

Construction and characterization of an H. pylori genomic library.

BglI-digested H. pylori genomic DNA (9 to 23 kb) was ligated into LambdaGEM-11 BamHI arms and packaged with Packagene extract (Promega). Phage buffer (20 mM Tris-HCl [pH 7.4], 100 mM NaCl, 10 mM MgSO₄) and chloroform were added to the packaged DNA library and gently mixed by inverting the tube six times. The chloroform was then allowed to settle to the bottom of the tube. The resulting phage lysate, minus the chloroform, was subsequently used for transduction experiments.

Preparation of the plating bacteria was done according to the method of Sambrook et al. (30). Briefly, E. coli strain ER1793 was prepared for transduction by first growing it in 50 ml of sterile LB supplemented with 0.2% maltose (to induce the λ receptor) overnight at 37°C on a rotary shaker (250 rpm). The cells were concentrated by centrifugation (4,000 × g; 10 min) and suspended to an OD₆₀₀ of ∼2 in 10 mM MgSO₄. Cells could be stored at 4°C from several days to 3 weeks.

The packaged phage was titrated to determine the appropriate dilution to generate well-separated plaques on a lawn of ER1793. For each plate of plaques, 100 μl of the appropriately diluted phage was added to 100 μl of the prepared ER1793 cells described above. The phage were allowed to adsorb to the cells for 30 min at 37°C. Three milliliters of molten (47°C) LB top agar (autoclaved 1.0 g of Bacto-tryptone, 0.8 g of NaCl, 0.6 g of agarose in 100 ml of H₂O) was added to a 5-ml tube, gently vortexed, and immediately poured onto 1-day-old LB agar plates prewarmed to 37°C. Once the top agar hardened, the plates were incubated at 37°C overnight (8 to 10 h). The following morning, well-separated plaques were picked and plated onto fresh lawns of ER1793 using a 36-square grid. Ten such 36-grid plates were prepared to comprise a 360-clone H. pylori library (10 pools of 36). To prepare lysate from each of the 10 pools, the plates were overlaid with 5 ml of SM buffer (sterile 0.01% gelatin, 50 mM Tris-HCl [pH 7.5], 100 mM NaCl, 8 mM MgSO₄). Following this, the top agarose was scraped with a glass spreader (avoiding the bottom agar) and transferred to a conical 50-ml centrifuge tube. The agarose was broken up with a spatula and incubated at room temperature for 30 min with periodic shaking. The tube was then centrifuged at 10,000 × g for 15 min at 4°C, and the supernatant was transferred to a fresh tube and gently mixed with chloroform to 0.3% (vol/vol) by inverting the tube six times. One additional centrifugation step was added to settle out the remaining agarose and the chloroform. The recombinant phage DNA was purified from the lysate medium using Promega's Wizard Lambda Preps DNA purification system. The purified phage DNA was used for restriction analysis, transformation of H. pylori, and DNA sequencing.

RESULTS

Genetic basis of H. pylori BI resistance.

Spontaneous mutation frequencies were determined for two H. pylori strains exposed to 1 μg of compound A/ml (2 times the MIC) (Fig. 1). The mutation frequencies were 3.1 × 10⁻⁸ (standard deviation, 2.7 × 10⁻⁸) for strain AH244 and 1.3 × 10⁻¹⁰ (standard deviation, 4.9 × 10⁻⁹) for strain J99.

Genomic DNA was isolated from a compound A-resistant mutant of strain J99 and from its isogenic parent (wild type [WT] J99). The isolated DNAs were transformed into a highly competent H. pylori laboratory strain, AH244. A negative (water) control was included in this experiment to compare the DNA treatments to the spontaneous rate of resistance. Greater than 1,000 H. pylori colonies were observed following transformation with 1 μg of DNA isolated from the resistant strain, while only 3 and 5 colonies were observed after transformation of an equal number of cells with 1 μg of DNA from WT J99 and from the negative control, respectively. The latter treatments that resulted in only a few colonies reflected the spontaneous resistance rate determined for H. pylori strain AH244. The results from this transformation experiment clearly indicated that there was a transformable genetic basis to compound A resistance in H. pylori.

Construction of a genomic library of DNA isolated from a BI-resistant mutant of H. pylori strain J99.

Given the results described above, a DNA library was constructed to screen for the compound A resistance gene(s). BglII was chosen to cleave the DNA into larger fragments (the J99 genome is predicted to have 231 BglII sites) for library construction. Three fractions from the BglII-restricted DNA, corresponding to 3.0 to 3.5 kb (F1), 6.5 to 9.5 kb (F2), and 9.5 to 23 kb (F3), were purified after electrophoretic separation on an agarose gel. Approximately 0.5 μg of each DNA fraction was used to transform WT AH244 cells. This was followed by selection for resistance on BAP containing 1 μg of compound A/ml. The results from this transformation experiment indicated that fraction F3 (9.5 to 23 kb) contained the resistance gene, since it was the only fraction that generated compound A-resistant colonies above the spontaneous resistance rate.

λGEM-11 (lambda-based vector) was chosen as the vehicle of choice to construct the genomic library, since it accepts inserts between 9.5 and 23 kb in size and the BamHI-generated cloning arms of λGEM-11 are compatible with the BglII-generated inserts in fraction F3. The BamHI arms were ligated with F3 DNA, packaged, and titrated onto lawns of ER1793 to determine the most appropriate working dilution of the packaged ligation reaction mixture. Phadnis and coworkers showed previously that ER1793 was by far the best E. coli host (out of eight tested) for cloning H. pylori DNA into λGEM-11 (27).

The formula of Clarke and Carbon, N = ln(1 − P)/ln(1 − 1/n), where N = the number of individual recombinants, P = the probability of including any DNA sequence of a random library, and n = the size of the genome relative to a single cloned fragment, was used to determine how many clones should be screened to be >99% confident that the constructed library was complete (7). In this study, there were 62 BglII fragments in the sequenced H. pylori strain J99 genome between 8,869 and 23,373 bp in size. These 62 fragments were predicted to comprise 868 kb, or roughly 52% of the genome. Using the above formula, 99% confidence would require 283 clones. Our strategy included 360 clones, giving a confidence level of 99.7%.

Initially, 10 pools each comprising 36 individually amplified clones were prepared and analyzed by agarose gel electrophoresis. The pooled DNA was digested with SacI, which cuts the λGEM-11 arms away from the insert. As expected, the arms were liberated and observed as heavily staining bands at approximately 9 and 20 kb. The cloned DNA inserts from each pool were observed as unique profiles of smaller SacI fragments ranging in size from approximately 2 to 20 kb. This suggested that the library was random and contained many unique DNA inserts.

Screening for the compound A resistance gene(s).

In the first round of screening the 10 pools of 36 clones, 2 pools (numbers 4 and 7) were identified as positive after transformation of WT AH244. Pool 7 was chosen for further analysis. Pool 4 is discussed below in the section describing the sequencing of additional spontaneous mutants. Pool 7 was further divided into six pools of six clones, and the DNA was transformed into WT AH244 for selection on 1 μg of compound A/ml. Pool 3 contained a positive clone in this experiment, and it was therefore divided into the six individual clones for transformation. A single clone, 17, was identified as containing the genetic determinant(s) responsible for resistance.

Characterization of clone 17.

Clone 17 DNA was prepared for sequencing and subcloning. The DNA sequence was determined for the junctions of the cloned insert by using primers specific for the T7 and SP6 promoters at the vector arm junctions (Table 2). The sequenced junctions demonstrated that the clone spanned from nucleotides 1310501 to 1327589 on the J99 chromosome (2; http://scriabin.astrazeneca-boston.com/hpylori/). The genes contained on clone 17 were as follows: frr (ribosome releasing factor), pyrE (orotate phosphoribosyltransferase), JHP1179 (H. pylori-specific gene), JHP1180 (gene of unknown function), and the entire nuo operon of 14 open reading frames (JHP1181 to -1194) encoding NADH oxidoreductase (Fig. 2A).

FIG. 2.

Characterization of clone 17. (A) Map of clone 17 containing the compound A resistance gene(s). The map contains the restriction endonuclease sites used for subcloning into smaller fragments represented by plasmids pSM126, pSM127, and pSM129. (B) Map of pSM129 and subclones of pSM129, represented by pSM130, pSM131, and pSM132, with the corresponding restriction endonuclease sites used for cloning.

Clone 17 was subcloned into pUC19 using the restriction endonuclease sites SacI and XbaI (Fig. 2A). The resulting clones (pSM126, pSM127, pSM128, and pSM129) were then transformed into WT AH244 and WT J99 to identify the positive subcloned DNA. pSM129 was the only clone capable of transforming H. pylori strains J99 and AH244 to compound A resistance. The DNA insert in pSM129 was further subcloned in pUC19 by using the existing restriction endonuclease sites EcoRI and SacI (Fig. 2B). These subclones (pSM130, pSM131, and pSM132) were transformed into WT AH244 and WT J99 to identify pSM132 as the only positive clone. Based on the H. pylori genome sequence, pSM132 contained the entire nuoD gene and the first 171 nucleotides of the 5′ end of nuoE (Fig. 2B). The 1,643-bp DNA insert in pSM132 was sequenced in its entirety, revealing only a single nucleotide change relative to the WT J99 sequence. This change was a G-to-A nucleotide transition, resulting in a glycine-to-serine missense mutation at residue 398 of NuoD.

Sequencing of nuoD from additional independent spontaneous compound A-resistant mutants.

Thirty-three independent H. pylori strain J99 resistant mutants were isolated on BAP containing 1 μg of compound A/ml. Genomic DNA was prepared from each of the 33 resistant mutants, and specific primers (JHP1184-F and -R) were used to PCR amplify the entire nuoD gene from each of the mutants. DNA sequencing of each PCR product was completed using primers CF-1 thru CR-5 (Table 2) to determine whether nuoD carried the mutation, as was shown for the first isolated compound A-resistant mutant in clone 17. Figure 3 summarizes the sequencing results for all 33 mutants aligned with the WT J99 NuoD sequence. All but one of the mutants had a single change at the C terminus of NuoD (Fig. 3). Twenty-eight of the mutants had the glycine-to-serine change at residue 398 (G-to-A transition; J99A.1), three had a phenylalanine-to-serine change at residue 404 (T-to-C transition; J99A.2), one had a valine-to-methionine change at residue 407 (G-to-A transition; J99A.3), and one did not have a detectable mutation in nuoD (J99A.4). The other positive pool (number 4) identified during the initial library screening process was subjected to PCR and sequence analysis as above. The positive clone in this pool also contained nuoD with a single G-to-A transition at the same position as observed for the J99A.1 class of mutants (Fig. 3).

FIG. 3.

Sequencing of nuoD from 33 independent spontaneous compound A-resistant mutants. The sequences show the alignment of the C terminus of NuoD (amino acids 391 to 409) from four classes of mutants. The changed amino acid is shown in gray. The numbers in parentheses refers to the number of isolates representing each of the mutant classes. The numbers above the WT sequence refer to the amino acid position in NuoD. The diamonds indicate the amino acid residues previously shown to be involved in resistance to rotenone in R. capsulatus (10, 13, 29).

Characterization of the unmapped mutant J99A.4.

As described above, a single compound A-resistant mutant was isolated without a detectable mutation in nuoD. We were interested to explore the basis of resistance in this isolate. Our first inclination was to look elsewhere in the 14-kb nuo operon for a mutation. In order to efficiently probe the operon without cloning and sequencing the entire stretch of DNA, a strategy was developed to more quickly identify a smaller region of DNA responsible for resistance. Primers were designed so that four overlapping fragments (approximately 4 kb each) covering the entire operon would be obtained by PCR. The PCR products (nuo1, nuo2, nuo3, and nuo4) were individually transformed into WT AH244 to determine which fragment may be responsible for resistance. The transformation results indicated that the resistance gene was located within the DNA encompassed by nuo1 that spanned from upstream of nuoA to within the protein coding sequence of nuoF. The entire nuo1 PCR product was sequenced using primers F1 thru F11 and R1 thru R11 (Table 2). The sequencing results identified only one difference relative to the WT J99 sequence. This difference was an A-to-G nucleotide transition in nuoB resulting in a threonine-to-alanine amino acid change at residue 27. The nuo1 PCR product was prepared a second time and sequenced to confirm the presence of this single mutation that supports resistance to compound A.

Directed knockout of nuoD to determine essentiality.

It was of interest to determine whether nuoD is essential in H. pylori, especially since it has been reported to be nonessential in E. coli (4, 5). The nuoD gene was cloned into pGEM-T after PCR amplification from WT J99 genomic DNA using primers JHP1184-F and -R (Table 2). A 718-bp internal deletion was created in nuoD by PCR using primers CR-2 and CF-4, followed by ligation of a Campylobacter coli kanamycin resistance gene (aphA-3 [23]) in place of the deleted DNA. The constructed knockout plasmid (pSM135) was transformed into WT H. pylori J99 and AH244, and the cells were plated onto selective medium. No kanamycin-resistant colonies were recovered for the J99 transformation, and 36 kanamycin-resistant colonies were recovered for the AH244 transformation. Eight of the 36 AH244 colonies were subcultured, and the genomic DNA was isolated to assess integration of the knockout construct into the WT nuoD gene by PCR analysis. PCR analysis was performed using primers CF-1 and CR-5 to amplify the region on either side of the aphA-3 gene. All eight kanamycin-resistant strains contained an intact WT nuoD (data not shown). These results suggested that nonspecific integration had occurred. Further, PCR analysis of the transformation mixture, prior to plating onto selective media, revealed that site-specific integration into nuoD had occurred in both lots of cells (J99 and AH244), but no viable colonies resulted with this genotype. This provided additional evidence that this recombination event was lethal. The conclusion drawn from the above results was that nuoD (or the nuo operon) is essential in H. pylori. Since nuoD is contained within an operon, we cannot rule out possible polar effects within the operon.

MIC determinations for H. pylori strain J99, the four classes of compound A-resistant mutants, and E. coli strains to compound A, piericidin A, and rotenone.

MICs were determined for WT H. pylori strain J99, each of its four classes of compound A-resistant mutants, and E. coli strains ECO523 and ECO524 (ECO523 with TolC⁻ mutation) using compound A, piericidin A, and rotenone. Rotenone and piericidin A are known complex I inhibitors that have been suggested to interact with NuoD in Rhodobacter capsulatus (12, 13, 18). The MICs for the four mutant classes treated with compound A were consistent at 2 μg/ml, while E. coli strain ECO523 and the isogenic TolC⁻ strain were insensitive to the highest concentration used (4 μg/ml) (Table 3). The TolC⁻ strain was used to address whether efflux may play a role in E. coli resistance. Several additional BI-containing compounds were screened against WT J99 and the four classes of resistant mutants, and cross-resistance was consistently observed (data not shown). In the case of rotenone, the three classes of nuoD mutants were hypersensitive, with MICs of <0.125 μg/ml, while the WT and nuoB mutant had the same MIC of 1 μg/ml (Table 3). Both E. coli strains were insensitive to rotenone at 4 μg/ml (Table 3). WT H. pylori strain J99 and the nuoB mutant were sensitive to piericidin A, with an MIC of 0.008 μg/ml, while the three nuoD mutants were two- to fourfold less sensitive (Table 3). Both E. coli strains were insensitive to piericidin A at 4 μg/ml, the highest concentration tested (Table 3). Growing E. coli under microaerophilic conditions in the presence of compound A, piericidin A, and rotenone yielded the same results (data not shown).

TABLE 3.

MICs for H. pylori strain J99 WT and its four classes of compound A-resistant mutants and E. coli strains ECO523 and ECO524

Strain	MIC (μg/ml)
	Compound A	Rotenone	Piericidin A
J99 WT	0.5	1	0.008
J99 A.1	2	<0.125	0.016
J99 A.2	2	<0.125	0.032
J99 A.3	2	<0.125	0.032
J99 A.4	2	1	0.008
ECO523 (WT)	>4	>4	>4
ECO524 (TolC−)	>4	>4	>4

DISCUSSION

Complex I is a multisubunit protein structure located in the plasma membrane of bacteria and the inner membrane of mitochondria (11, 14, 17). It is a proton-pumping NADH:ubiquinone oxidoreductase (EC 1.6.5.3) that catalyzes the first step of oxidative phosphorylation (respiration) by transferring two electrons from NADH to ubiquinone (17). In E. coli, complex I is made up of 13 polypeptides encoded by genes of the nuo operon (nuoA to nuoN). In this operon, nuoC and nuoD are fused together to encode a single polypeptide. In contrast, the H. pylori nuo operon is made up of 14 genes where nuoC and nuoD are predicted to encode separate proteins (32).

Dupuis et al. have established R. capsulatus complex I as a model system for studying this multisubunit enzyme (14). Relevant to the present study, they have isolated mutants of R. capsulatus resistant to piericidin A (a potent inhibitor of complex I that binds close to or at the ubiquinone-binding site) that have a single missense mutation in nuoD resulting in an amino acid change (V407M) at the C terminus of the protein (9, 10). The V407M change maps to the same region that the mutations in H. pylori NuoD were observed in the present study (Fig. 3). The piericidin A-resistant mutants were also resistant to rotenone, a commonly used pesticide that has been shown to bind and inhibit complex I at the ubiquinone-binding site (9, 10, 12). In a follow up study, Prieur et al. constructed additional nuoD mutations by site-directed mutagenesis (29). One additional mutant was resistant to rotenone and piericidin A (D412E) (Fig. 3). In contrast, mutations within the analogous region of H. pylori NuoD resulted in hypersensitivity to rotenone but moderate resistance to piericidin A and compound A (Table 3). Based on the experiments in R. capsulatus, the authors proposed that the quinone polar head binds to the NuoD subunit (13, 29). In addition, they suggested that the binding site for the polar head of ubiquinone is located at the interface of NuoD and NuoB. This was, in part, based on photoaffinity labeling of NuoB from membranes of Paracoccus nitrificans and Thermus thermophilus with a ³H-labeled pyridaben analog (31). In the same study, both rotenone and piericidin A prevented binding of the ³H-labeled pyridaben analog (another pesticide) (31). This is compatible with the results of the present study, where compound A-resistant mutants had missense mutations in either the C terminus of nuoD or in nuoB. We speculate that NuoD and NuoB may provide a binding interface for compound A.

Studies with inhibitors have shown that E. coli complex I is extremely insensitive to rotenone, while complex I from R. capsulatus was notably sensitive (12, 18). These findings are consistent with our results showing that two strains of E. coli were insensitive to rotenone (and compound A), while H. pylori was sensitive to both of these compounds (Table 3). It may be that BI selectivity for H. pylori is due to an insensitivity of complex I from most other bacteria to the BI-containing compounds, much like E. coli complex I is insensitive to rotenone. A second possibility to explain selectivity is that complex I is not essential in most other bacteria; therefore, even if inhibited, the bacteria continue to grow. There is significant evidence that complex I is not essential in several organisms, such as E. coli (37), Salmonella enterica serovar Typhimurium (38), and R. capsulatus (15). In these studies, nuo was not essential, likely because there is an additional NADH dehydrogenase (Ndh II) that can compensate for inhibition of complex I. Ndh II and nuo dual knockouts have been demonstrated to be viable but more limited in growth rate (4, 5). In the present study, nuoD (or the nuo operon) was confirmed to be essential for the viability of H. pylori strains AH244 and J99 by knockout mutagenesis. There is no homolog to Ndh II in the H. pylori genome, suggesting that H. pylori is more limited by a lack of alternative cellular machinery (32).

In the present study, we found the mutation frequencies for H. pylori strains J99 and AH244 challenged with compound A to be low, ranging from 3.1 × 10⁻⁸ to 1.3 × 10⁻¹⁰. The compound A MICs for the resistant mutants were only moderately higher (4 times higher) than that observed for the WT J99 strain (Table 3). The hypothesis that the BIs may bind at the same site as ubiquinone and known complex I inhibitors, such as rotenone, is attractive based on the results of this study. There are several parallels to be drawn from the literature, especially regarding the R. capsulatus model, including the involvement of NuoD and NuoB. Alternatively, complex I may serve to alter the BIs into active anti-H. pylori agents, much like RdxA has been shown to be involved in converting metronidazole into an active anti-H. pylori drug (19). Complementary biochemical studies are needed to ascertain whether complex I is the target of the BIs in addition to being involved in resistance.