Helicobacter pylori Mutants Defective in RuvC Holliday Junction Resolvase Display Reduced Macrophage Survival and Spontaneous Clearance from the Murine Gastric Mucosa

Abstract

Homologous recombination contributes to the extraordinary genetic diversity of Helicobacter pylori and may be critical for surface antigen expression and adaptation to environmental challenges within the stomach. We generated isogenic, nonpolar H. pylori ruvC mutants to investigate the function of RuvC, a Holliday junction endonuclease that resolves recombinant joints into nicked duplex products. Inactivation of ruvC reduced the frequency of homologous recombination of H. pylori between 17- and 45-fold and increased sensitivity to DNA-damaging agents and the antimicrobial agents levofloxacin and metronidazole. The H. pylori ruvC mutants were more susceptible to oxidative stress and exhibited reduced survival within macrophages. Experiments with the H. pylori SS1 mouse model revealed that the 50% infective dose of the ruvC mutant was approximately 100-fold higher than that of the wild-type SS1 strain. Although the ruvC mutant was able to establish colonization with bacterial loads that were initially similar to those of the parental SS1 strain, infection was spontaneously cleared from the murine gastric mucosa over periods that varied from 36 to 67 days. These results demonstrate that, in this infection model, RuvC is essential for continued survival of H. pylori in vivo and raises the possibility that inactivation of ruvC might be of value in an attenuated vaccine strain.

Helicobacter pylori is a gram-negative, microaerobic, spiral bacterium that colonizes the stomachs of approximately half the world's population (13). Infection with H. pylori is associated with chronic gastritis and peptic ulceration and is also considered a risk factor for the development of gastric adenocarcinoma and mucosa-associated lymphoid tissue lymphoma (4, 37, 38). The bacterium is uniquely adapted for survival within the gastric mucosa, and colonization usually persists for years or even decades (13).

Genetic recombination is a potent force driving the evolution of microbial pathogens and facilitating the acquisition of antibiotic resistance, antigenic determinants, and virulence factors, which result in altered pathogenicity (30). Because H. pylori is naturally competent for transformation (35), new genetic material can be acquired by horizontal transfer, and it has the highest rate of recombination of any known bacterial species (1, 47). It exhibits a panmictic (free recombinational) population structure that results from frequent genetic recombination during mixed colonization by unrelated strains (1, 15, 17, 47). Gene transfer between H. pylori strains during colonization of an individual is common and can generate novel subtypes that exhibit changes in important virulence markers, such as the cag pathogenicity island (24). The ability to generate such an extraordinary degree of genetic diversity is likely to facilitate adaptation of H. pylori to environmental challenges within the stomach and contribute to the longevity of infection. As well as promoting genetic diversity, homologous recombination plays a critical role in conserving genetic identity by utilizing undamaged chromosomal copies as a template to replace missing or damaged nucleotides during DNA repair synthesis. Deficiency of DNA repair systems attenuates the virulence of other bacterial pathogens (6, 33), and efficient recombinational repair is likely to be important for H. pylori infection by facilitating survival from genotoxic products generated from the bacterium's own metabolism and the host inflammatory response.

The process of homologous recombination is initiated at ends or gaps in duplex DNA and involves exchange of single strands between homologous partners, leading to the formation of a four-way branched intermediate called a Holliday junction (for a review, see reference 29). Regression of stalled replication forks can also form Holliday junctions via annealing of nascent strands (43). Processing of junctions in this instance provides a means of restarting replication and thus avoiding potentially fatal blocks to chromosome transmission. In most bacteria, Holliday structures are processed by a tripartite molecular machine composed of a junction-targeting protein (RuvA), a helicase motor (RuvB), and a resolvase (RuvC) (44). Together, the RuvAB proteins catalyze the migration of the junction along DNA duplexes and consequently generate heteroduplex DNA (50). The RuvC endonuclease resolves the junction by dual-strand incision across the branch point to release nicked duplexes that can be sealed by DNA ligase (12, 20).

The potential importance of genetic recombination in promoting genetic diversity and ensuring efficient DNA repair prompted us to examine the role of Holliday junction processing in this important human pathogen. We constructed H. pylori ruvC mutants, characterized their phenotypes, and investigated the impact of these mutations in an established mouse model of infection (26). The results reveal that inactivation of ruvC confers reduced survival within macrophages and that RuvC is essential for the persistence of H. pylori infection.

MATERIALS AND METHODS

Bacterial strains, cell lines, and growth conditions.

Escherichia coli strain MC1061 (7) was used as a host for plasmid cloning experiments and was grown at 37°C on Luria-Bertani medium containing 100 μg of spectinomycin (Sigma Chemicals, Poole, United Kingdom)/ml or 25 μg of kanamycin (Sigma Chemicals)/ml as required.

H. pylori strains G27 and SS1 were used in the experiments (8, 26). A total of 40 H. pylori strains from the United States, South America, Asia, and Europe were used to study the distribution of ruvC in clinical isolates of H. pylori. An rdxA mutant derivative of SS1, deficient in the production of an oxygen-insensitive NADPH nitroreductase, was used as a positive control for animal colonization studies (22). A chloramphenicol-resistant SS1-derived mutant deficient in the production of a type II restriction enzyme (HP0091; a kind gift from A. Labigne) was used as the source of genomic DNA for assessing competence for natural transformation. H. pylori strains were routinely cultured under microaerobic conditions at 37°C on a blood agar (BA) base 2 (Oxoid, Basingstoke, United Kingdom) plates supplemented with 10% horse blood and an antibiotic-fungicide mix consisting of 10 μg of vancomycin (Sigma Chemicals)/ml, 2.5 IU of polymyxin (Sigma Chemicals)/liter, 5 μg of trimethoprim (Sigma Chemicals)/ml, and 4 μg of amphotericin B (Sigma Chemicals)/ml. Liquid cultures of H. pylori were grown under identical conditions in brain heart infusion (BHI) broth containing 0.2% cyclodextrin and the antibiotic-fungicide mix. H. pylori transformants were selected on medium supplemented with 25 μg of kanamycin (Sigma Chemicals)/ml and 12.5 μg of chloramphenicol (Sigma Chemicals)/ml.

The macrophage cell line J774A.1 (American Type Culture Collection, Manassas, Va.) was grown in Dulbecco's Eagle medium (Invitrogen, Paisley, United Kingdom) containing 4 mM l-glutamine (Sigma Chemicals) and 10% fetal calf serum (Invitrogen) in a humid atmosphere at 37°C and 5% CO₂.

DNA techniques.

DNA manipulations were carried out by using standard techniques (42). Mini or midi Qiagen columns (Qiagen, Crawley, United Kingdom) and a QiaAmp DNA extraction kit (Qiagen) were used for plasmid and rapid chromosomal DNA preparations, respectively. PCR was carried out according to the manufacturer's recommendations with Taq DNA polymerase (Promega, Southampton, United Kingdom).

Plasmids.

To construct an H. pylori ruvC deletion, oligonucleotides (ruvC-1 and ruvC-2; Table 1) were used to amplify a 543-bp DNA fragment (fragment 1) containing the 5′ region of the ruvC gene flanked by ClaI and EcoRI restriction sites. A second pair of primers (ruvC-3 and ruvC-4; Table 1) were used to generate a 546-bp DNA fragment (fragment 2) containing the 3′ region of the ruvC gene flanked by BamHI and PstI restriction sites. Following PCR amplification, fragment 1 was restricted with ClaI and EcoRI and cloned into the plasmid vector pILL570 (25) cut with the same enzymes. The nonpolar kanamycin cassette liberated from pUC18K2 (32) by restriction with EcoRI and BamHI was inserted into the intermediate recombinant plasmid (containing fragment 1) that had been linearized with the same enzymes. Finally, the resultant plasmid (containing fragment 1 and the nonpolar cassette) was linearized with BamHI and PstI, and the BamHI- and PstI-restricted fragment 2 was inserted. The resulting construct, pJEN3, carries a 277-bp deletion of the ruvC gene replaced with the kanamycin cassette. Plasmid pCTB2-Cat was donated by J. Atherton and is a derivative of pCTB2 (9) containing the 5′ region of vacA and the cat_GC chloramphenicol resistance gene. Plasmid pHel2 is an E. coli-H. pylori shuttle vector (19) that carries the cat_GC chloramphenicol resistance gene.

TABLE 1.

Oligonucleotides used in this study

Oligo-nucleotide	Sequence (5′-3′)a
ruvC-1	ccatcgatATCTTCTCCCATTTAGTCGC
ruvC-2	ggaattcATATTAATGAACCCGGCCG
ruvC-3	gcggatccTTGCTTAACATCACAAGCG
ruvC-4	aaaactgcagTATGAGCTTGCCGCAAGCACCGG
ruvC-5	ATGCGTATTTTAGGAATAGATCCGGGC
ruvC-6	ATGGAGCTTTAAGCGTTGCGCATGCG

^aRestriction sites for BamHI (ruvC-3), ClaI (ruvC-1), EcoRI (ruvC-2), and PstI (ruvC-4) are underlined. Lowercase letters indicate nucleotides that were added at the 5′ end to create a restriction site.

Hybridization.

For Southern blot hybridization DNA was blotted onto nylon membranes (Roche Diagnostics Ltd., Lewes, United Kingdom) by capillary transfer in 20× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate). Hybridizations were performed with a digoxigenin labeling and detection system according to the manufacturer's instructions (Roche Diagnostics Ltd). Digoxigenin-labeled DNA probes were synthesized by random incorporation of digoxigenin-11-dUTP by PCR using oligonucleotide ruvC-5 and ruvC-6 primers specific to the corresponding gene (Table 1). Hybridizations were performed at 37°C in DIG EasyHyb (Roche Diagnostics Ltd), and the filters were washed twice with 2× SSC and 0.1% sodium dodecyl sulfate at room temperature and then with 0.1× SSC and 0.1% sodium dodecyl sulfate at 68°C. Membranes were developed with the CDP-Star chemiluminescent detection system (Roche Diagnostics Ltd.) according to the manufacturer's instructions.

Natural transformation of H. pylori.

Transformation of H. pylori was carried out essentially as described previously (21). Briefly, bacteria were inoculated as 1-cm patches and grown for 5 h under microaerobic conditions at 37°C before the addition of 1 μg of chromosomal (isolated from a chloramphenicol-resistant SS1-derived HP0091 mutant) or circular plasmid DNA (pJEN3, pCTB2-Cat, or pHel2). After further incubation for 18 h, the bacteria from each individual patch were suspended in 500 μl of BHI supplemented with 0.2% cyclodextrin, serially diluted, and plated directly on selective and nonselective plates for enumeration. After incubation for 72 h, the colonies were counted, and the transformation frequency was calculated as the number of kanamycin- and chloramphenicol-resistant colonies per microgram of DNA per recipient CFU (mean of at least three experiments).

Sensitivity to mitomycin C and antimicrobial agents.

Susceptibility to mitomycin C (Sigma) and metronidazole (Sigma) was assessed by agar dilution determination of the MIC. Inoculates yielding 10⁴ CFU/spot were inoculated onto plates of Iso-Sensitest agar (Oxoid) enriched with 10% horse blood containing doubling dilutions of metronidazole and mitomycin C. The MIC was defined as the lowest concentration of antibiotic inhibiting growth when the plates were read after 72 h incubation under microaerobic conditions (generated as described above) at 37°C. Susceptibility to amoxicillin, clarithromycin, and levofloxacin was assessed by the Epsilometer test (E-test; AB Biodisk, Solna, Sweden) according to the manufacturer's instructions.

UV irradiation sensitivity assays.

For UV sensitivity assays, bacterial cell suspensions were serially diluted and 10 μl of each dilution was spotted on BHI plates supplemented with 0.2% cyclodextrin. Cells were irradiated with 0, 5, 10, 20, 30, 45, and 60 J of 254-nm UV light generated by a UV lamp (Harovia Lamps, Slough, United Kingdom)/m² at a distance of 58 cm. Plates were immediately wrapped in aluminum foil to prevent photoreactivation and colonies were enumerated after 72 h incubation under microaerobic conditions at 37°C in the dark. Data are reported as the mean numbers of CFU per milliliter ± standard errors of the means from three independent determinations and analyzed with the t test.

Oxidative stress and acid survival assays.

Oxidative stress resistance was determined by disk inhibition assay. Sterile 6-mm-diameter paper disks were applied to BA plates that had been streaked for confluent growth. Samples (10 μl) of 30% hydrogen peroxide (Sigma Chemicals), 80% cumene hydroperoxide (Sigma Chemicals), and 1.9% methyl viologen (Sigma Chemicals) were applied to the disks, and the zone of inhibition of growth was measured after the plates had been incubated for 48 h at 37°C. Data are reported as mean zone diameters ± standard errors of the means from three independent determinations and analyzed with the t test. A P value of ≤0.05 was considered significant.

To determine the ability of bacterial cells to survive under acid conditions, bacteria were harvested from 48-h culture plates and suspended in BHI broth supplemented with 0.2% cyclodextrin to yield a final suspension of ∼10⁷ CFU/ml. The suspensions were adjusted to pH 4.0 and 7.0 and incubated under microaerobic conditions at 37°C with shaking (250 rpm) in either the presence or absence of 10 mM urea (Sigma Chemicals). Samples were removed at 0, 1, and 24 h, serially diluted, and plated directly on BA plates for enumeration. Results were obtained from three independent experiments.

Macrophage killing assay.

The survival of H. pylori within macrophages was investigated as described by Odenbreit et al. (36), with minor modifications. J774A.1 cells were seeded in 24-well plates to a density of 2 × 10⁵ cells per well in cell culture medium. After 24 h at 37°C and 5% CO₂, the medium was replaced by fresh cell culture medium with or without 1 mM cytochalasin D and incubated for 1 h at 37°C and 5% CO₂. The medium was again replaced by fresh cell culture medium with or without 1 mM cytochalasin D containing bacteria from a liquid culture at a multiplicity of infection of 50, and infection was synchronized by a short centrifugation step (3 min at 600 × g). After nonadherent bacteria had been removed by three washes with phosphate-buffered saline (PBS), the cells were incubated for 1 h at 37°C and 5% CO₂. To kill extracellular bacteria, gentamicin was added to the cells to a final concentration of 100 μg/ml, mixed, and incubated for a further 60 min at 37°C and 5% CO₂. After five washes with PBS and another incubation step in fresh medium with or without cytochalasin D for 2 h, cells were lysed for 5 min with ice-cold PBS-0.1% saponin. Appropriate dilutions of the supernatant were plated on BA plates and incubated for 2 to 3 days under microaerobic conditions at 37°C to count the surviving bacteria. Data are reported as mean numbers of CFU per milliliter ± standard errors of the means of inoculating and surviving bacteria from three independent determinations and analyzed with the t test.

Animal colonization.

Six- to eight-week-old specific-pathogen-free CD1 mice (Charles River, Margate, United Kingdom) were housed in polycarbonate cages in isolators and fed a commercial pellet diet with water ad libitum. All animal experimentation was performed in accordance with Home Office license 40/2340 and institutional guidelines. Aliquots of 100 μl, containing 10⁸ to 10⁹ CFU of H. pylori in BHI supplemented with 0.2% cyclodextrin, were administered orogastrically to mice as described elsewhere (23). For all experiments, inoculating suspensions of SS1 and two independently constructed ruvC mutants were prepared from identical, low-subculture stocks (between 8 and 12 in vitro passages). Mice were killed at various time points up to 67 days after inoculation, and colonization with H. pylori was assessed by quantitative culture, serology, and histology as described previously (23). Briefly, the stomach of each mouse was removed, washed in physiological buffered saline, and divided longitudinally into tissue fragments so that each fragment contained cardia, body, and antrum. For each stomach, one fragment was placed in BHI supplemented with 0.2% cyclodextrin and another was placed in formalin. For the performance of quantitative bacterial cultures of stomach samples, tissue fragments were homogenized with disposable plastic grinders and tubes (PolyLabo, Strasbourg, France). The homogenates were serially diluted and plated directly onto BA plates for enumeration. To determine the 50% infective dose (ID₅₀), mice were infected with serial dilutions of bacteria as described above. The inoculum was serially diluted and plated to determine the actual bacterial dose. The mice were sacrificed, and the stomachs were cultured as described above to determine the number of animals infected at each dose. The Reed-Muench calculation was used to determine the ID₅₀ (40).

Serum samples were tested for H. pylori antigen-specific immunoglobulin G (IgG) antibody by a previously described enzyme-linked immunosorbent assay technique (23). Ninety-six-well Maxisorb plates (Nunc, Kamstrup, Denmark) were coated with 25 μg of a sonicated whole-cell extract of H. pylori SS1. Serum samples were diluted 1:100 and were added in 100-μl aliquots to coated microtiter wells. To allow for nonspecific antibody binding, samples were also added to uncoated wells. Bound H. pylori-specific antibodies were detected by using biotinylated goat anti-mouse immunoglobulin and streptavidin-peroxidase conjugate (Amersham, Little Chalfont, United Kingdom). The readings for uncoated wells were subtracted from those of the respective test samples. A cutoff value was determined from the mean optical density value ±2 standard deviations for the corresponding samples from naive uninfected mice. Samples with optical density readings greater than this cutoff value were considered positive for H. pylori-specific antibodies.

Gastric tissue fragments stored in formalin were cut in longitudinal sections (4 μm) and stained by the hematoxylin-eosin technique. Examination of the tissue sections for histopathological lesions was performed blind, and the presence of inflammatory cell infiltrates (polymorphonuclear and mononuclear cells), erosive lesions, edema, hyperplasia, and lymphoid follicle formation were graded in the range from “none” to “severe” according to the Sydney system (11).

RESULTS

Features and distribution of H. pylori ruvC and construction of an isogenic mutant.

In the sequenced H. pylori strains 26695 and J99 (2, 48), ruvC (HP0877 and JHP0811, respectively) encodes a polypeptide of 157 amino acids with a predicted molecular mass of 17.4 kDa and a pI of 9.26. The open reading frame is preceded by a putative ribosomal binding site located seven nucleotides upstream of the translational start codon (Fig. 1A). A putative promoter, displaying homology to the consensus sequence of an E. coli σ70 promoter, is located 44 bp from the start of ruvC (Fig. 1A). The ruvC gene is flanked by an iron-regulated outer membrane protein and a protein of unknown function. These flanking genes are divergently transcribed, suggesting that ruvC forms a single transcriptional unit. An orthologue of ruvA (HP0833; JHP0815) is positioned five genes upstream of ruvC (2, 48).

FIG. 1.

H. pylori ruvC promoter structure and conserved elements of the ruvC gene product. (A) Sequence of the putative ruvC promoter. Potential transcription and translation signals are indicated. RBS, ribosome binding site. (B) Amino acid sequence alignment of RuvC Holliday junction endonucleases of H. pylori (Hpy) and E. coli (Eco). Identical residues are marked with asterisks, and the conserved acidic residues known to be essential for E. coli RuvC catalysis are in bold.

Sequence comparisons of H. pylori RuvC revealed that it is most closely related to its paralog from Campylobacter jejuni (65.0% identity) and shares 34.7% identity with E. coli RuvC (Fig. 1B). The four acidic residues (Asp-7, -138, and -141 and Glu-66) known to be essential for junction resolution activity are conserved in H. pylori RuvC (41). Additional residues implicated in base-stacking interactions (51) and in sequence specificity (18) are also present. However, H. pylori RuvC lacks an equivalent of the last 16 residues present in the E. coli polypeptide. In E. coli, this region may form contacts with the RuvB ring helicase and act to stabilize binding of RuvC in the RuvABC junction complex.

We used the PCR to confirm the presence and distribution of ruvC in a collection of 40 clinical strains of H. pylori isolated from patients in different geographical locations who had undergone upper gastroduodenal endoscopy for gastric cancer, peptic ulceration, and nonulcer dyspepsia. Using the oligonucleotides ruvC-5 and ruvC-6 (Table 1), we successfully amplified an approximately 470-bp product from each of these strains (data not shown). To determine whether ruvC was present in other species of Helicobacter, chromosomal DNA isolated from Helicobacter acinonychis (NCTC 12686), Helicobacter felis (ATCC 51211), and Helicobacter cinaedi (NCTC 12423) was analyzed by dot blot hybridization using a probe corresponding to the ruvC nucleotide sequence. Genomic DNA isolated from H. pylori was used as a positive control, and genomic DNA isolated from a C57BL/6 mouse and pILL570 plasmid DNA were used as negative controls. A positive signal was obtained for all the Helicobacter species (data not shown). These data show that the gene encoding RuvC is present in all strains of H. pylori examined and is conserved in both gastric and enterohepatic members of the genus Helicobacter.

Isogenic mutants of H. pylori ruvC were generated from a plasmid construct (pJEN3) carrying a central 277-bp deletion of the H. pylori ruvC gene coupled with insertion of a nonpolar cassette composed of the aphA3 kanamycin resistance gene (49) lacking its promoter and terminator regions (32). H. pylori ruvC mutants were obtained by allelic exchange following natural transformation with the recombinant plasmid. Since H. pylori is genetically diverse, mutations were constructed in two strains, G27 and SS1. The genotypes of the constructed mutants were verified by PCR using oligonucleotides flanking ruvC (data not shown). The kanamycin-resistant ruvC mutants grew normally in vitro and had growth rates similar to those of the parent strains (data not shown).

Reduction of transformation frequency in ruvC mutants.

To examine whether disruption of ruvC adversely affected natural transformation, we compared the competence for transformation by chromosomal and plasmid DNA of the ruvC mutants with that of their parent strains (Table 2). The parental strain G27 had a transformation frequency of approximately 10⁻⁵, with chromosomal and plasmid DNA conferring chloramphenicol resistance, while the transformation frequency of strain SS1 was approximately 20-fold lower (Table 2). In both parental backgrounds, inactivation of ruvC resulted in ∼30- and ∼20-fold reductions in transformation frequency for chromosomal and plasmid (pCTB2-Cat) DNA, respectively (Table 2). While transformation of chromosomal and suicide plasmid DNA depends on recombinational allelic exchange, transformation with self-replicating plasmids is usually independent of recombination. The mutants were tested for transformation with the self-replicating plasmid pHel2. No chloramphenicol-resistant colonies were obtained when SS1 wild-type or ruvC mutant strains were transformed with pHel2. Chloramphenicol-resistant colonies were obtained in the G27 background, and no difference in transformation frequency was observed between the wild-type and ruvC mutant strains (Table 2). A plasmid of the expected size was recovered from a random sample of transformants, indicating that pHel2 was maintained in a plasmid form and that the chloramphenicol resistance obtained was not due to integration of cat_GC into the host chromosome.

TABLE 2.

Transformation frequency of wild-type and ruvC mutant strains with chromosomal and plasmid DNA

Strain	Transformation frequencya (% of parent frequency)
	Chromosomal DNA	Plasmid pCTB2Cat	Plasmid pHel2
G27	956 ± 660 (100)	745 ± 78 (100)	970 ± 400 (100)
G27ruvC::Km	21 ± 18 (2)	43 ± 38 (6)	733 ± 321 (76)
SS1	16 ± 8 (100)	58 ± 10 (100)	0
SS1ruvC::Km	0.6 ± 0.4 (4)	3 ± 1 (5)	0

^aNumber of resistant colonies (10⁻⁸) per microgram of DNA per recipient CFU. Data are means ± standard deviations for three experiments.

Sensitivity of ruvC mutants to DNA damage and antimicrobial agents.

Since E. coli ruvC mutants are defective in recombinational repair (28), we examined the sensitivity of ruvC mutants in G27 and SS1 backgrounds to DNA-damaging agents. Both mutants showed increased sensitivity to the DNA interstrand cross-linking agent mitomycin C compared to their respective wild-type strains (Table 3). They were also more sensitive to irradiation with 254-nm UV light (Fig. 2). The wild-type strains showed similar dose-dependent killing patterns, characterized by roughly 3 log units of killing over 60 s of UV irradiation. The ruvC mutant strains were significantly more sensitive to killing by UV light, with a maximal loss in viability at 60 s (P < 0.01) (Fig. 2).

TABLE 3.

Susceptibility of wild-type and ruvC mutant strains to mitomycin C and antimicrobial agents

Strain	MIC (μg/ml)a
	Mitomycin C	Amoxicillin	Clari- thromycin	Levo- floxacin	Metro- nidazole
G27	32.00	0.047	0.064	0.125	4.00
G27ruvC::Km	4.00	0.047	0.064	0.064	0.25
SS1	12.5	0.25	<0.016	0.094	0.5
SS1ruvC::Km	3.125	0.25	<0.016	0.032	0.016

^aResults are the means of two independent experiments.

FIG. 2.

Susceptibility of wild-type and ruvC mutant strains to UV irradiation. Survival of strains following exposure to UV radiation at 254 nm is shown. Data are means ± standard errors of the means from three independent determinations. Symbols: ▴, H. pylori G27; ▵, H. pylori G27ruvC::Km; ▪, H. pylori SS1; □, H. pylori SS1ruvC::Km.

We also assessed susceptibility to a variety of antimicrobial agents (Table 3). The ruvC mutant strains displayed increased sensitivity to metronidazole (which induces DNA degradation and strand breakage [34]) and levofloxacin (which interferes with replication by inhibiting the activity of topoisomerases [45]). There was no difference in susceptibilities of the wild-type and mutant strains to amoxicillin (which inhibits synthesis of the cell wall) or clarithromycin (which inhibits protein synthesis).

Susceptibility of ruvC mutants to oxidative and acid stress.

Bacterial pathogens deficient in the production of recombination enzymes are more prone to DNA damage caused by reactive oxygen species (5). The susceptibilities of the wild-type and ruvC mutant strains to oxidative stress inducers were measured by a disk inhibition assay. Similarly derived mutants deficient in the production of oxygen-insensitive NADPH nitroreductase, RdxA (22), served as a control to exclude possible effects of the presence of the aphA3 gene, which confers kanamycin resistance in the mutant strain. Hydrogen peroxide, cumene hydroperoxide, and methyl viologen were added to filter paper disks applied to plates streaked for confluent growth, and susceptibilities were measured as zones of inhibition around the disks (Table 4). The H. pylori ruvC mutants were more sensitive to hydrogen peroxide, cumene hydroperoxide, and methyl viologen, with significantly greater zones of inhibition than the wild-type strains (Table 4). In contrast, the H. pylori rdxA mutants were not more sensitive to these oxidative stress inducers, as they had zones of inhibition similar to those of the wild-type strains (data not shown).

TABLE 4.

Susceptibility of wild-type and ruvC mutant strains to oxidative stress

Strain	Zone of inhibition (mm) witha:
	30% hydrogen peroxide	80% cumene hydroperoxide	1.9% methyl viologen
G27	12.6 ± 0.4	37.3 ± 0.5	37.0 ± 0.6
G27ruvC::Km	15.3 ± 0.4	43.7 ± 0.4	46.0 ± 0.6
SS1	11.0 ± 0.7	34.7 ± 0.6	45.0 ± 0.9
SS1ruvC::Km	18.7 ± 0.7	40.3 ± 0.7	71.7 ± 0.6

^aZones of inhibition were measured around filter paper disks saturated with 10 μl of the indicated compounds. Water added as a control did not yield any zones of growth inhibition. Results are the means ± standard errors of the means from three independent experiments. All ruvC mutant values were statistically significantly different (P < 0.05) from the wild-type strain controls.

To investigate whether mutation of ruvC had an effect on H. pylori viability at low pH, acid survival studies were conducted with the wild-type strain and the ruvC mutant. The experiments were performed in both the presence and absence of urea to differentiate between urease-dependent and urease-independent mechanisms. There was no significant decrease in viability of the wild-type and mutant strains after 1 or 24 h of exposure to pH 7 in either the presence or absence of urea. Likewise, both the wild-type and mutant strains survived incubation for 1 and 24 h at pH 4 in the presence of urea. In the absence of urea, approximately 10⁵ and 10² CFU of both the wild-type and mutant strains per ml survived after 1 and 24 h of exposure to pH 4.0, respectively. These experiments demonstrate that loss of ruvC expression does not significantly affect acid survival of H. pylori.

Sensitivity of H. pylori ruvC mutants to macrophage killing.

In other bacterial pathogens, mutations in recombination enzymes (encoded by recA and recBC) are known to confer susceptibility to the oxidative burst of macrophages (5). Given the sensitivity of H. pylori ruvC mutants to oxidative damage, we investigated whether RuvC contributed to the survival of H. pylori within macrophages. We employed a gentamicin killing assay in which J774A.1 macrophages were infected with either the mouse-adapted H. pylori strain SS1 or the SS1 ruvC mutant. H. pylori cells that were not taken up by macrophages were killed by gentamicin, while intracellular bacteria were rescued by lysis of the phagocytes and enumerated by plating on BA plates after 2 h. To monitor the efficiency of ingestion and extracellular killing of H. pylori by gentamicin, a cytochalasin D control was used. Cytochalasin D blocks the phagocytic uptake of H. pylori (36, 39) and resulted in a >10⁴-fold reduction in survival of H. pylori SS1 and ruvC mutant strains. Exposure of H. pylori in the absence of phagocytic cells resulted in significant reduction in bacterial survival after 1 h of gentamicin treatment (>6 log units). Without cytochalasin D, both the wild type and the ruvC mutant could be recovered after gentamicin treatment, with a 10²-fold reduction in survival of H. pylori SS1, compared to a 10⁴-fold reduction in survival of the ruvC mutant (Fig. 3). The ruvC mutant therefore survived significantly less well in J7774A.1 macrophages than the wild-type strain (P < 0.05) (Fig. 3). These results support a role for recombinational processes in survival of H. pylori within macrophages.

FIG. 3.

Survival of H. pylori SS1 (black bars) and H. pylori SS1 ruvC (gray bars) in J774A.1 macrophages after 2 h, determined with the gentamicin killing assay. Absolute numbers of surviving H. pylori are shown in comparison with the number of infecting bacteria. Data are means ± standard errors of the means from three independent determinations.

Mouse colonization by the ruvC mutant.

To assess the function of RuvC in colonization of the stomach, we investigated the potential of wild-type and mutant bacteria to infect the SS1 H. pylori mouse model. Six experiments were conducted in which two independently constructed ruvC mutants, as well as the parental SS1 strain, were used to orogastrically inoculate mice (10⁸ to 10⁹ CFU per mouse). Colonization by SS1 and SS1-derived mutants was assessed by quantitative culture of the stomachs (23). In an initial experiment, in which animals were sacrificed at 21 days, all 10 mice inoculated with the SS1 parental strain were colonized by H. pylori, with bacterial loads that ranged from 2.4 × 10⁵ to 6.7 × 10⁶ CFU per g of stomach (Fig. 4). In the control group, none of the 10 mice inoculated with BHI were infected with H. pylori. An SS1-derived mutant deficient in the production of oxygen-insensitive NADPH nitroreductase, RdxA (22), served as a control to exclude possible effects of the presence of the aphA3 gene, which confers kanamycin resistance in the mutant strain. The degree of colonization in mice (n = 10) inoculated with the rdxA mutant was similar to that in mice with the wild-type SS1 strain (P > 0.1). In contrast, only 5 of 10 mice inoculated with the RuvC-deficient H. pylori bacteria were colonized with H. pylori; the bacterial loads in these mice were between 5.4 × 10³ and 1.28 × 10⁶ CFU/g and were significantly lower than those in the mice colonized with the wild-type SS1 strain (P < 0.001) (Fig. 4).

FIG. 4.

Assessment of mouse colonization 21 days after inoculation with H. pylori SS1, SS1 ruvC, SS1 rdxA, or BHI. Each point represents one mouse.

To further characterize the apparent colonization defect of the ruvC mutant, a time course experiment was performed in which a group of 40 mice were infected with the wild-type SS1 strain and another group of 40 mice were infected with the ruvC mutant (Fig. 5). An inoculum containing 10⁸ CFU of the appropriate bacteria was used for each mouse. Five mice from each group were sacrificed on each of days 0 (5 h postinoculation), 1, 2, 4, 7, 14, 23, and 36, and the viable counts of each strain were determined. Five hours after infection, mice inoculated with SS1 were colonized with approximately 10³ CFU/g, while the stomachs of mice inoculated with the ruvC mutant contained approximately 10⁴ CFU/g (Fig. 4). By day 4, the bacterial load for the mice infected with SS1 reached approximately 10⁶ CFU/g and remained at this level for the remaining time points. Although the bacterial loads recovered from mice infected with the ruvC mutant did reach 10⁶ CFU/g by day 7, the counts were significantly lower than those from the SS1-infected mice for the 7-, 14-, 23-, and 36-day time points (Fig. 4). The density of colonization of the ruvC mutant declined after day 7, and no ruvC mutant bacteria were recovered from any of the five mice at the 36-day time point. In contrast, the SS1-infected mice were still colonized at day 36, with bacterial counts between 8.12 × 10⁵ and 3.65 × 10⁶ CFU/g. To confirm that these results were due to clearance of the ruvC mutant, further kinetic experiments were carried out with two independently constructed ruvC mutants. In these experiments, ruvC mutant bacteria could not be recovered by day 49 and day 67 (data not shown). H. pylori infection in these animals was also assessed by serology and histological analysis. No H. pylori antigen-specific IgG antibody was detected in animals inoculated with BHI supplemented with 0.2% cyclodextrin. Although anti-H. pylori IgG was detected in all mice infected with either the wild type or the mutant strain, there was no significant difference in the magnitudes of the serological responses mounted by these groups of animals (data not shown). No significant cell infiltration was seen in uninfected mice during the 9-week observation. Mice infected with the wild type and the ruvC mutant strain showed no or mild inflammation at all time points up to 9 weeks postinoculation. There was no difference in the grade of inflammation between mice infected with the wild type and those with the ruvC mutant, and pathological lesions such as microerosions, hyperplasia, gland atrophy, and edema were absent from all animals.

FIG. 5.

Kinetics of mouse colonization with H. pylori strain SS1 or SS1 ruvC. A group of 40 mice were infected with H. pylori SS1, and another group of 40 mice were infected with the ruvC mutant. Five mice from each group were sacrificed on each of days 0 (5 h postinoculation), 1, 2, 4, 7, 14, 23, and 36, and the viable counts of each strain were determined. Each point represents the mean log CFU per gram of tissue ± the standard error of the mean for each subgroup of five mice. Symbols: ⧫, H. pylori SS1; •, H. pylori SS1ruvC::Km.

Finally, the ID₅₀s of the wild-type and mutant strains were determined by infecting mice with 10-fold serially diluted bacteria and determining the number of animals colonized at each dose. The Reed-Muench calculation was used to determine the number of bacteria required to obtain colonization of 50% of the animals (40). Table 5 shows that while the ID₅₀ for the wild-type SS1 strain was 1.4 × 10⁴ bacteria, the ID₅₀ of the ruvC mutant was approximately 100-fold higher (1.7 × 10⁶ bacteria). These results indicate that recombinational processes play a vital role in the establishment and long-term survival of H. pylori in the mouse gastric mucosa.

TABLE 5.

ID₅₀ determination for wild-type and ruvC mutant strains in mice^a

Strain anddilution	No. (%) of mice infected (n = 5)
SS1
2.5 × 108	5 (100)
2.5 × 107	5 (100)
2.5 × 106	5 (100)
2.5 × 105	5 (100)
2.5 × 104	3 (60)
2.5 × 103	1 (20)
SS1ruvC::km
9.6 × 108	5 (100)
9.6 × 107	5 (100)
9.6 × 106	4 (80)
9.6 × 105	2 (40)
9.6 × 104	2 (40)
9.6 × 103	1 (20)

^aMice were inoculated with wild-type SS1 or the ruvC mutant at the indicated dilutions. Reed-Muench ID₅₀ calculations were 1.4 × 10⁴ CFU/ml for the wild type and 1.7 × 10⁶ CFU/ml for the ruvC mutant.

DISCUSSION

The panmictic population structure and natural competence of H. pylori strongly suggest that recombinational exchange confers a selective advantage in colonization of the stomach and host adaptation. However, our knowledge of the molecular mechanisms that mediate genetic exchange and the precise contribution of this process to the virulence of this important human gastric pathogen remains rudimentary. The series of experiments described in the present study confirm that recombinational processes are important in transformation and DNA repair in H. pylori and show that they are also involved in resistance to oxidative stress and macrophage survival. Our finding that a mutant defective for RuvC production is attenuated for colonization of a mouse model of infection provides evidence that genetic recombination plays a crucial role in establishing infection of the gastric mucosa. Furthermore, our observation that infection by the ruvC mutant was spontaneously cleared indicates that this process is essential for the long-term survival of H. pylori within the stomach and raises the possibility that inactivation of ruvC might be of value in an attenuated vaccine strain.

Phylogenomic studies have shown that H. pylori carries all three components of the RuvABC system required for processing and resolving Holliday junctions (3, 14). In addition to finding the ruvC gene in all of the clinical isolates of H. pylori examined, we also showed that it was present in gastric and enterohepatic Helicobacter species, providing evidence for evolutionary conservation of RuvC across this genus. Although H. pylori RuvC contains all the key residues known to be required by E. coli RuvC for binding and resolution (18, 51), it lacks the C-terminal 16-amino-acid extension that may be crucial for interaction with RuvB and stabilization of binding of RuvC on the RuvAB junction complex. This suggests that the functional interactions between the components of the resolvasome structure may be significantly different in H. pylori and E. coli.

In E. coli, mutational inactivation of ruvA, ruvB, or ruvC results in a modest reduction in genetic recombination (two- to threefold) but a significant defect in DNA repair (27). Inactivation of ruvC in H. pylori resulted in DNA repair defect similar to that observed in E. coli but a far greater reduction in the frequency of recombination. This disparity may reflect differences in the repertoire of recombination enzymes available in H. pylori or may be due to unique features of DNA processing during natural transformation. A similarly severe defect in transformation has also been observed with Streptococcus pneumoniae recG mutants, while the corresponding mutants in E. coli are relatively proficient at recombination (31). Given that RecG also participates in processing branched intermediates and displays a significant functional overlap with RuvABC (27), examination of the relative contribution of both pathways to H. pylori transformation and DNA repair merits investigation.

Although H. pylori infection is associated with the recruitment of lymphocytes, macrophages, and polymorphonuclear cells, the bacterium appears to have evolved effective mechanisms to resist this strong inflammatory response and establish persistent gastric infection (16). In particular, although H. pylori can be internalized by phagocytes, it appears able to survive for prolonged periods within these cells and presumably is able to resist damage by free radicals derived from the phagocytic respiratory burst (36, 46). The ability to repair DNA damage resulting from the oxidative burst of macrophages is known to be essential for full virulence and for survival of Salmonella within macrophages (5). Our finding that recombination-deficient mutants of H. pylori are also more sensitive to oxidative stress and intracellular killing by macrophages provides evidence that maintaining the integrity of its DNA is vital for viability within phagocytic cells. The ability to escape the functions of professional phagocytic cells present at the site of infection may help explain how the bacteria are able to survive and chronically colonize the gastric mucosa.

Although a number of bacterial factors have been shown to be essential for establishing colonization, little is known of the mechanisms that allow H. pylori to maintain a chronic infection. It is important to identify such factors, since it is persistence of the bacteria that ultimately leads to the induction of mucosal damage and carcinoma. Our in vivo experiments provide the first evidence that recombinational processes are important for establishing and maintaining long-term H. pylori infection. Although mutations in the ruvC gene of H. pylori strain SS1 resulted in bacteria that were able to colonize the murine gastric mucosa, the ID₅₀ of the ruvC mutant was 100-fold higher than that of the wild-type strain. The most striking finding was that infection was spontaneously cleared from the murine gastric mucosa by periods that varied from 36 to 67 days. This is the first time that such an attenuated strain of H. pylori has been described. This colonization defect is unlikely to be due to reduced viability of the ruvC mutant, since the growth characteristics of the mutant were similar to those of the wild-type strain. Increased susceptibility of the ruvC mutant to oxidative stress and macrophage killing may in part explain the reduced capability of the bacteria to survive and persistently colonize the gastric mucosa. During the time course experiments, we attempted to determine whether the elimination of the ruvC mutant coincided with the appearance of host inflammatory cell infiltration of the murine gastric mucosa. However, it is recognized that H. pylori SS1 induces relatively mild inflammatory changes in the murine gastric epithelium (10), and the lack of significant gastritis in our mice made it impossible to test this hypothesis.

By facilitating the acquisition of new genes from diverse sources and the rearrangement of existing genes, homologous recombination may be critical for adapting surface antigen expression in response to environmental challenges present within the stomach. Our data lead us to speculate that recombinational exchange may make a particularly important contribution to the chronicity of H. pylori infection by substituting alternate versions of surface-expressed adhesins and antigens, thereby facilitating immune evasion. Although our preliminary investigation indicated that the serological response mounted by ruvC mutant-infected mice was of a magnitude similar to that of the response in mice infected with the wild-type strain, we aim to further characterize the humoral and cellular responses in response to infection with this strain. These studies will indicate whether inactivation of ruvC might be of value in the development of an attenuated vaccine strain.