Dual Nuclease and Helicase Activities of Helicobacter pylori AddAB Are Required for DNA Repair, Recombination, and Mouse Infectivity*

Helicobacter pylori infection of the human stomach is associated with disease-causing inflammation that elicits DNA damage in both bacterial and host cells. Bacteria must repair their DNA to persist. The H. pylori AddAB helicase-exonuclease is required for DNA repair and efficient stomach colonization. To dissect the role of each activity in DNA repair and infectivity, we altered the AddA and AddB nuclease (NUC) domains and the AddA helicase (HEL) domain by site-directed mutagenesis. Extracts of Escherichia coli expressing H. pylori addANUCB or addABNUC mutants unwound DNA but had approximately half of the exonuclease activity of wild-type AddAB; the addANUCBNUC double mutant lacked detectable nuclease activity but retained helicase activity. Extracts with AddAHELB lacked detectable helicase and nuclease activity. H. pylori with the single nuclease domain mutations were somewhat less sensitive to the DNA-damaging agent ciprofloxacin than the corresponding deletion mutant, suggesting that residual nuclease activity promotes limited DNA repair. The addANUC and addAHEL mutants colonized the stomach less efficiently than the wild type; addBNUC showed partial attenuation. E. coli ΔrecBCD expressing H. pylori addAB was recombination-deficient unless H. pylori recA was also expressed, suggesting a species-specific interaction between AddAB and RecA and also that H. pylori AddAB participates in both DNA repair and recombination. These results support a role for both the AddAB nuclease and helicase in DNA repair and promoting infectivity.

Infection of the stomach with Helicobacter pylori causes a variety of diseases including gastritis, peptic ulcers, and gastric cancer (1). A central feature of the pathology of these conditions is the establishment of a chronic inflammatory response that acts both on the host and the infecting bacteria (2). Both epithelial (3,4) and lymphoid (5,6) cells in the gastric mucosa of infected individuals release DNA-damaging agents that can introduce double-stranded (ds) 2 breaks into the bacterial chromosome (7). The ds breaks must be repaired for the bacteria to survive and establish chronic colonization of the stomach. Homologous recombination is required for the faithful repair of DNA damage and bacterial survival. Alteration of the expression of one of a series of cell surface proteins on H. pylori occurs by an apparent gene conversion of babA, the frequency of which is reduced in repair-deficient strains (8,9). This change in the cell surface, which may allow H. pylori to evade the host immune response, is a second means by which recombination can promote efficient colonization of the stomach by H. pylori.
The initiation or presynaptic steps of recombination at dsDNA breaks in most bacteria involves the coordinated action of nuclease and helicase activities provided by one of two multisubunit enzymes, the AddAB and RecBCD enzymes (10). Escherichia coli recBCD null mutants have reduced cell viability, are hypersensitive to DNA-damaging agents, and are homologous recombination-deficient (11)(12)(13)(14). Similarly, H. pylori addA and addB null mutants are hypersensitive to DNA-damaging agents, have reduced frequencies of babA gene conversion, and colonize the stomach of mice less efficiently than wild-type strains (8).
The activities of RecBCD enzyme from E. coli (15)(16)(17)(18)(19) and AddAB from H. pylori (8) or Bacillus subtilis (20 -23) indicate some common general features of the presynaptic steps of DNA repair. In the case of E. coli, repair begins when the RecBCD enzyme binds to a dsDNA end and unwinds the DNA using its ATP-dependent helicase activities (17,24). Single-stranded (ss) DNA produced during unwinding, with or without accompanying nuclease, is coated with RecA protein (16,25). This recombinogenic substrate engages in strand exchange with a homologous intact duplex to form a joint molecule. Joint molecules are thought to be converted into intact, recombinant DNA either by replication or by cutting and ligation of exchanged strands (26).
Although the AddAB and RecBCD enzymes appear to play similar roles in promoting recombination and DNA repair, they differ in several ways. RecBCD is a heterotrimer, composed of one copy of the RecB, RecC, and RecD gene products (27), whereas AddAB has two subunits, encoded by the addA and addB genes (21,28). The enzyme subunit(s) responsible for helicase activity can be inferred from the presence of conserved protein domains or the activity of purified proteins. AddA, RecB, and RecD are superfamily I helicases with six highly conserved helicase motifs, including the conserved Walker A box found in many enzymes that bind ATP (29 -32). A Walker A box is defined by the consensus sequence (G/A)XXGXGKT (X is any amino acid (29). RecBCD enzymes in which the con-served Lys in this motif is changed to Gln have a reduced affinity for ATP binding (33,34) and altered helicase activity (17,(35)(36)(37).
A nuclease domain with the conserved amino acid sequence LDYK is found in RecB, AddA, AddB, and many other nucleases (38). The conserved Asp plays a role in Mg 2ϩ binding at the active site; Mg 2ϩ is required for nuclease activity (39). The recB1080 mutation, which changes codon 1080 from the conserved Asp in this motif to Ala, eliminates nuclease activity (39).
We have recently shown that addA and addB deletion mutants are hypersensitive to DNA-damaging agents and impaired in colonization of the mouse stomach compared with wild-type strains (8). To determine the roles of the individual helicase and nuclease activities of H. pylori AddAB in DNA repair and infectivity, we used site-directed mutagenesis to inactivate the conserved nuclease domains of addA and addB and the conserved ATPase (helicase) domain of AddA. Here, we report that loss of the AddAB helicase is sufficient to impair H. pylori DNA repair and infectivity and, when the genes are expressed in E. coli, homologous recombination. AddAB retains partial activity in biochemical and genetic assays when either of the two nuclease domains is inactivated but loses all detectable nuclease activity when both domains are inactivated. Remarkably, H. pylori AddAB can produce recombinants in E. coli only in the presence of H. pylori RecA, suggesting a species-specific interaction in which AddAB facilitates the production of ssDNA-coated with RecA protein. Our results show that both the helicase and nuclease activities are required for the biological roles of H. pylori AddAB.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Growth Conditions-E. coli and H. pylori strains with their genotypes are listed in Table S1. Plasmids are listed in Table S2. Allele numbers and designations for polypeptides with nuclease (NUC) or helicase (HEL) mutations are indicated as superscripts for clarity. E. coli strains were constructed by CaCl 2 -mediated transformation; fresh transformants were made prior to each experiment. H. pylori strains were constructed by natural transformation. E. coli and H. pylori strains were cultured as described by Amundsen et al. (8).
Mutant Construction-Mutations in the nuclease (addA NUC and addB NUC ) and helicase motifs (addA HEL ) were introduced into pSA405 (pDuet-1 with addA and addB (8)) using the QuikChange II XL site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions. Oligonucleotides (Integrated DNA Technologies) carrying the desired mutations are listed in Table S3. Plasmids with the targeted mutations were identified by sequencing and introduced into E. coli strain V3060. Helicase and nuclease motif mutant alleles were introduced into H. pylori at the rdxA locus by natural transformation (40) of the relevant NSH57-derived deletion strain (8) with derivatives of the pRdxA vector (41) containing the appropriate point mutant gene subcloned from pSA405 derivatives described above. In all cases expression was driven from the endogenous promoter of the rdxA locus.
Genetic Assays-Recombination proficiency was measured in crosses as described previously (42) using E. coli Hfr donor strain V1306 (his ϩ rpsL ϩ ; PO44) and F Ϫ recipients (hisG4 rpsL31) carrying plasmids expressing E. coli recBCD, H. pylori addAB, H. pylori recA, or combinations of these and lacking E. coli chromosomal recA or recBCD null alleles. Selection was for His ϩ Str R .
Cell-free Extract Preparation and Assays-Cell-free extracts were prepared and assayed for helicase and nuclease activities as described (8).
Ciprofloxacin Sensitivity Testing-For H. pylori, ϳ3 ϫ 10 7 bacteria were inoculated on horse blood plates. For E. coli, ϳ1 ϫ 10 7 bacteria were inoculated on LB agar plates. A single E-test strip (AB Biodisk) was placed in the center of each plate, and plates were incubated at 37°C for 2-4 days (H. pylori) or 1 day (E. coli). The minimal inhibitory concentration (MIC) was determined according to the manufacturer's instructions.
Mouse Infections-Female C57BL/6 mice 24 -28 days old were obtained from Charles River Laboratories and certified free of endogenous Helicobacter infection by the vendor. The mice were housed in sterilized microisolator cages with irradiated PMI 5053 rodent chow and autoclaved corncob bedding, and acidified reverse osmosis-purified water was provided ad libitum. All studies were done under the practices and procedures of Animal Biosafety Level 2. The facility is fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International, and all activities were approved by the Fred Hutchinson Cancer Research Institutional Animal Care and Use Committee.
Groups of five mice were infected with H. pylori via oral gavage. After 1 week bacteria were recovered, and a competitive index was calculated as described (8).

Construction of addAB Nuclease and Helicase Mutants-To
assess the role of the nuclease and helicase activities of AddAB in DNA repair, recombination, and infectivity, we genetically inactivated one or another function. Nuclease and helicase domains were identified by comparing the amino acid sequences of H. pylori AddA and AddB with known motifs in E. coli RecB and RecD and in B. subtilis AddA and AddB (8,20,32,38,39,43,44). To inactivate the putative AddA helicase, we targeted the Walker A box responsible for ATP binding by changing codon 18 from Lys to Gln (K18Q), here designated AddA HEL . The nuclease domains of AddA and AddB were identified by the conserved RecB nuclease motif LDYK. To inactivate the nuclease domain, we changed the Asp in this motif to Ala in addA (D897A) or in addB (D667A), here designated AddA NUC or AddB NUC , respectively. Mutations were introduced into the addAB genes carried on an E. coli expression vector. For analysis in H. pylori, mutant alleles were integrated and expressed at the rdxA locus, and the endogenous locus was deleted (8,41).
Both AddA and AddB Have ATP-dependent dsDNA Exonuclease Activity-We assayed nuclease activity, as acid-soluble products of radioactive DNA, in extracts of E. coli cells with a chromosomal recBCD deletion (⌬recBCD) bearing a plasmid encoding the wild-type or mutant addAB alleles. Extracts from cells without AddAB (vector control), AddA HEL B, and the AddA NUC B NUC extract had very little ATP-dependent nuclease activity, indicating that inactivating both nuclease domains or the single helicase domain eliminates detectable exonuclease activity ( Table 1). The AddA HEL B enzyme likely fails to degrade DNA because, as for the RecBCD enzyme, the exonuclease acts almost exclusively during DNA unwinding (45)(46)(47)(48)(49). In contrast, inactivating the single nuclease motif in addA or in addB reduced the nuclease activity to about half that of the wild type (Table 1). This suggests that the individual nuclease domains of AddA and AddB can act independently during DNA unwinding.
We next measured the nuclease activity of mutant and wildtype AddAB enzymes in intact cells. In E. coli, the nuclease activity of RecBCD enzyme degrades the DNA of phage T4 lacking the protective gene 2 protein; this restriction greatly reduces the efficiency of plating (EOP) of phage T4 gene 2 mutants (50,51). In E. coli ⌬recBCD cells, expression of AddAB or AddA NUC B or AddAB NUC reduced the EOP to about 10 Ϫ5 of that with the vector control ( Table 1). The EOP was ϳ1 on cells with addA HEL or addA NUC B NUC , indicating that phage plating was not restricted in these cells, consistent with the absence of nuclease activity in assays of the corresponding extracts ( Table  1). The strong reduction of EOP on addA NUC B or addAB NUC cells indicates that each of the two nuclease domains has signif-icant nuclease activity, as observed in cell-free extracts (Table  1). Thus, both the intracellular and cell-free extract results suggest that the single nuclease domains can act independently but are dependent on the helicase for detectable nuclease activity.
Helicase Activity of addAB Mutants-The AddAB and RecBCD enzymes unwind linear dsDNA in an ATP-dependent reaction. We measured helicase activity under reaction conditions that minimize nuclease activity, with excess ATP relative to Mg 2ϩ (45). Cell-free extracts were reacted briefly (2 min) with the 32 P end-labeled substrate, and the products were separated by agarose gel electrophoresis and visualized by autoradiography. Extracts containing AddAB, AddA NUC B, AddAB NUC , or Add-A NUC B NUC unwound dsDNA, producing similar amounts of ssDNA products (Fig. 1). As expected, mutation of one or both of the nuclease motifs had no apparent effect on the helicase activity. In contrast, inactivation of the AddA ATP-binding domain (AddA HEL B) abolished helicase activity, and no ssDNA was detected even with more than 10 times the amount of extract protein that gave detectable unwinding with the other strains ( Fig. 1).
Collectively, the experiments described above demonstrate that site-directed mutations in the nuclease and helicase domains of AddA or AddB altered the expected target activities (Table 1 and Fig. 1). We next determined how loss of these activities affected DNA repair, recombination, and infectivity.
Sensitivity of H. pylori addAB Mutants to DNA-damaging Agents-AddAB plays a central role in DNA repair, and addAB null mutants are sensitive to DNA-damaging agents (8,32). We wished to determine in H. pylori whether inactivation of the helicase domain or one or both of the nuclease domains of AddAB increased the sensitivity of cells to the DNA gyrase inhibitor ciprofloxacin. Ciprofloxacin induces ds breaks in the DNA, a target for AddAB-mediated repair (52)(53)(54). H. pylori strains with chromosomal addA or addB deletions had a lower MIC of ciprofloxacin than the wild type (Fig. 2).
To compare the phenotype of wild type and nuclease or helicase mutants, addA or addB alleles (wild type or mutant) were inserted into the chromosomal rdx locus, at which inserted genes are often overexpressed compared with the normal chromosomal location (8,43). H. pylori strains with a mutation in the helicase motif (addA HEL ) or the nuclease motif of either addA or addB (addA NUC or addB NUC ) had a lower MIC of ciprofloxacin than wild-type addAB at rdx (Fig. 2). These data  suggest that AddAB helicase and nuclease are required for repair of ciprofloxacin-induced damage.
Homologous Recombination in E. coli Requires Both Wildtype H. pylori AddAB and RecA-DNA damage induced by ciprofloxacin is thought to be repaired by homologous recombination between the damaged DNA and an intact duplex. Recombination proficiency can be measured quantitatively in E. coli following conjugation with an Hfr strain, which introduces broken DNA into the recipient cells. Because both AddAB and RecBCD have helicase and nuclease activities ( Table 1 and Fig. 1), we tested the ability of wild-type and mutant AddAB to promote recombination in E. coli recipients lacking RecBCD, an enzyme essential for recombination in E. coli Hfr crosses (11,12,14).
To measure conjugation, we expressed H. pylori AddAB from plasmids in E. coli hisG rpsL (His Ϫ Str R ) strains with a chromosomal recBCD deletion. These strains were mated with an rpsL ϩ Hfr strain that transfers hisG ϩ early, and His ϩ [Str R ] recombinants were selected (42). The recombinant frequency in the recB21 null mutant was reduced more than a thousandfold compared with that in a recBCD ϩ strain ( Table 2, compare  crosses 1 and 2). Surprisingly, strains expressing wild-type AddAB were as recombination-deficient as recB21 (Table 2, cross 5). Strains expressing AddA or AddB with altered nuclease or helicase domains were also recombination-deficient (data not shown). This suggests that the nuclease and helicase activities provided by wild-type H. pylori AddAB are not able to produce viable recombinants in E. coli.
In E. coli, RecBCD enzyme appears to interact with RecA protein during the presynaptic stages of homologous recombination, loading RecA onto DNA to produce a recombinogenic ssDNA filament (16); this activity is required to produce recombinants (55). We tested the possibility that AddAB did not support recombination in E. coli because it was unable to load E. coli RecA protein onto DNA. RecA protein is required for Hfr conjugation (56); the recombinant frequency in recBCD ϩ E. coli was reduced more than a thousand-fold in the presence of recA56, a recA null allele ( Because repair of DNA damage can occur by homologous recombination, we tested whether co-expression of H.pylori addAB and recA could reduce the sensitivity of an E. coli recBCD deletion strain to UV light or to the DNA gyrase inhibitor, ciprofloxacin. Both agents lead to the production of dsDNA breaks. The strains expressing H. pylori addAB alone FIGURE 2. Both AddAB nuclease and helicase activities are required for H. pylori wild-type resistance to ciprofloxacin. The minimal inhibitory concentration of ciprofloxacin for each strain was determined using E-test strips; the mean Ϯ S.E. is indicated (n ϭ 3-5). Strains are derivatives of H. pylori strain NSH57. The wild-type or mutant NUC or HEL addAB allele was integrated at the rdx locus, and the corresponding endogenous locus was deleted.  Table S2). b Strains have the indicated E. coli chromosomal recA alleles with or without plasmid pSA502 expressing the H. pylori recA ϩ gene. c Data are the mean Ϯ S.E. from 3 to 6 independent matings, except where data from two independent experiments are shown. For cross 9, p Ͻ 0.0001 compared with cross 5.
For crosses 10 and 11, p Ͻ 0.005 compared with cross 5.
were about as sensitive to UV light as a strain with the recB null allele (recB21) (Fig. 3). The E. coli strain with both H. pylori AddAB and H. pylori RecA showed greater resistance to UV light, approaching the resistance of the E. coli wild type. A similar pattern was observed in the sensitivity to ciprofloxacin (Fig.  4). The MIC for the E. coli strain expressing H. pylori addAB and recA was similar to that for recBCD ϩ E. coli. Strains expressing H. pylori addAB or recA alone were about as sensitive to ciprofloxacin as a recBCD null mutant. These data further support a species-specific interaction between AddAB and RecA in DNA repair. We next tested the role of the AddAB helicase and nuclease activities in promoting Hfr recombination in E. coli expressing H. pylori recA. The recombinant frequency in strains with a single nuclease domain alteration (AddA NUC B or AddAB NUC ) was reduced about 10-fold compared with that with the fully wild-type combination (AddAB with H. pylori RecA) but was significantly greater than that in the crosses without H. pylori RecA (p Ͻ 0.005; Table 2, crosses 10 and 11). AddA NUC B NUC and AddA HEL B were as recombination-deficient as a null mutant even in the presence of H. pylori RecA ( Table 2, crosses 12 and 13). Taken together, these data indicate that both the nuclease and helicase activities of AddAB are required for DNA repair and homologous recombination but that a single nuclease domain in AddAB can provide sufficient activity to produce recombinants, although at a reduced level.
H. pylori AddAB Nuclease and Helicase Are Required for Efficient Colonization of the Mouse Stomach-Deletion of recA, addA, or addB reduces the efficiency of H. pylori stomach colonization in mice (8), suggesting a role for DNA repair during infection. To further explore the importance of AddAB during infection, we determined the contribution of distinct AddAB biochemical activities to infectivity.
We first tested the ability of H. pylori AddAB nuclease or helicase mutants to colonize mice following co-infection with wild-type bacteria. After 1 week, we recovered bacteria from the stomach and calculated a competitive index (CI), the ratio of mutant to wild-type bacteria recovered. A CI of less than 1 for a mutant indicates a colonization defect relative to wild-type. The addA NUC B and addA HEL B strains had a CI of 0.1 or less in all mice tested, indicating a deficiency comparable with that of addAB null mutants (Fig. 5) (8). In fact, no mutant bacteria were recovered from infections with the addA NUC B strain; mutant bacteria were recovered in only 1 of 10 mice with the addA HEL B mutant. The addAB NUC strain demonstrated considerable mouse-to-mouse variation. In four mice the CI was less than 1; two mice yielded no mutant bacteria, but in one mouse the mutant colonized to a higher level than wild-type. These data   . AddAB nuclease and helicase activities are required for wildtype H. pylori stomach colonization. Mice were orally infected with a 1:1 mixture of the indicated H. pylori strains. After 1 week the bacteria colonizing the stomach were harvested and the competitive index determined (ratio of nuclease or helicase mutant to deletion or wild type (WT) in the output corrected for the input ratio). Each datum is the CI from one mouse. Open symbols indicate mice from which only wild-type or deletion mutant bacteria were recovered and represent an upper limit on the competitive index with 95% confidence (Poisson distribution). Similarly, black symbols indicate animals in which only the nuclease or helicase mutant was recovered and represent a lower limit on the competitive index with 95% confidence. Gray symbols indicate animals where both strains in the inoculum were recovered at 1 week. The results of infections with two different addA HEL clones are indicated by circles and squares. CI Ͻ 1 indicates a colonization defect, and CI Ͼ 1 indicates out-competition.
indicate that in most cases, full AddAB nuclease activity is required to maintain H. pylori viability during mouse colonization; the activity provided by a single nuclease domain is not sufficient. The addA HEL B strain was as deficient as the addA null mutant (8) in colonization activity, further demonstrating the role of helicase and nuclease activities in supporting infection (Table 1 and Fig. 1).
We next compared the infectivity of H. pylori addA NUC B, addAB NUC or addA HEL B to ⌬addA or ⌬addB; the deletion strains lack nuclease activity and are deficient in DNA repair and colonization (8). Similarly, extracts of H. pylori expressing the AddA or AddB subunit alone lack ATP-dependent dsDNA exonuclease activity, suggesting that both subunits are required for nuclease activity (8). As expected, the partial nuclease activity of H. pylori addA NUC B or addAB NUC allowed more efficient colonization than ⌬addA or ⌬addB strains (Fig. 5).
Remarkably, the CI for addA HEL B relative to ⌬addA in 10 mice was 0.1 or less, indicating that the helicase mutant was more deficient in colonization than a strain that expressed only the AddB subunit (Fig. 5). This suggests that AddA HEL B is stably expressed during infection and that there is an activity remaining in the enzyme that reduces the viability of H. pylori during colonization. Under the "Discussion" we have suggested an explanation for this observation. Collectively, these results indicate that the nuclease and helicase activities of AddAB are required to maintain H. pylori viability for efficient colonization.

DISCUSSION
Repair of DNA damage is required to maintain viability in all organisms. In pathogenic bacteria such as H. pylori, DNA repair is required to reverse host-inflicted damage that occurs as a consequence of the immune response, which releases agents, including reactive oxygen species, that introduce ds breaks into the bacterial chromosome (7). Failure to repair such damage limits bacterial colonization of the host. A central component and initiating agent of the repair pathway of many bacteria is the AddAB or closely related RecBCD enzyme. These enzymes use a complex set of activities to repair DNA damage by homologous recombination. We reported previously that H. pylori AddAB promotes DNA repair and is required for efficient colonization of the stomach (8). Here, we used site-directed mutagenesis of functional enzyme motifs to dissect the role of the ATP-dependent ds exonuclease and helicase activities of H. pylori AddAB. We found that both activities were required for repair and efficient colonization of the stomach in a mouse model.
Mutations Inactivating the H. pylori AddAB Nucleases and Helicase-The RecB, AddA, and AddB proteins of E. coli, B. subtilis, and H. pylori share the RecB nuclease motif involved in binding Mg 2ϩ to the nuclease active site (8,20,38,57,58). A mutation changing the aspartic acid to alanine in AddA (D897A) and AddB (D667A) together eliminated nuclease activity (AddA NUC B NUC ; Table 1). Interestingly, the same codon change in only one or the other subunit reduced, by about half, the specific activity of the corresponding mutant enzymes (Table 1). This indicates that the nuclease domains of AddA and AddB can act independently during DNA unwind-ing. Yeeles and Dillingham (20) have shown that each subunit of B. subtilis AddAB degrades one strand of a DNA duplex. This dual nuclease mechanism is consistent with the data reported here. The observation that an H. pylori addA or addB deletion strain lacks ATP-dependent nuclease activity (8) indicates, however, that the subunits cannot act alone. We infer that there is a complex interaction of AddAB subunits as there is for the subunits of the E. coli RecBCD enzyme (55, 59 -65).
The helicase activity of AddAB converts dsDNA to ssDNA in an ATP-dependent reaction. The ssDNA intermediate is thought to invade a homologous duplex following the initiating steps of homologous recombination. Mutations that change a lysine to glutamine in the ATP-binding domain reduce the affinity of ATP binding and alter the activity of other helicases (17, 32-34, 36, 37, 66). Cell-free extracts containing AddA HEL B enzyme failed to unwind DNA, although those containing AddA NUC B, AddAB NUC , or AddA NUC B NUC produced similar amounts of ssDNA as extracts containing AddAB enzyme (Fig.  1). The importance of helicase activity to the overall mechanism of enzyme action is demonstrated by the absence of dsDNA exonuclease activity in AddA HEL B extracts (Table 1). This suggests, as with RecBCD enzyme, that the exonucleolytic degradation occurs during or after DNA unwinding.
Evidence for Species-specific Interactions of RecA and RecBCD or AddAB-Genes encoding a representative of the AddAB and RecBCD classes of enzymes are found in nearly all of the bacterial genomes sequenced (67,68). 3 This suggests a common need for ATP-dependent exonucleases and helicases in DNA metabolism and raises the possibility that the gene products from different organisms might be interchangeable (10). Clearly, H. pylori AddAB has nuclease activity in E. coli (Table 1 and Ref. 8), but expression of addAB failed to restore recombination proficiency in the same strain (Table 2, cross 5). Partial recombination proficiency in the recBCD deletion strain was restored only when H. pylori AddAB and RecA were expressed together ( Table 2, cross 9). Similarly, recombinants were produced in the E. coli recA56 null mutant only when H. pylori RecA and AddAB were co-expressed ( Table 2, cross  8). This indicates a species-specific interaction between AddAB and RecA in two different genetic backgrounds. This result may stem from H. pylori AddAB having a third activity demonstrated for RecBCD enzyme: loading RecA protein onto ssDNA as RecBCD unwinds the DNA (16).
The RecA loading activity of RecBCD enzyme plays an essential role in creating the active form of an ssDNA end that undergoes strand exchange with homologous duplex DNA; mutants that fail to load RecA are recombination-deficient (55). Analysis of complexes formed between RecA and the C terminus of RecB protein identified a region that interacts with RecA (69). Similar experiments using RecB and non-cognate DNA strand exchange proteins from B. subtilis, Saccharomyces cerevisiae, and phage T4 indicate that the efficiency of the interaction compared with that with E. coli RecA is reduced (69). The observation that AddAB from B. subtilis restores viability and recombination proficiency to an E. coli recBCD deletion mutant (70) suggests that E. coli RecA interacts to at least a limited degree with B. subtilis AddAB. The interaction of B. subtilis RecA with E. coli RecB outside of cells may parallel the ability to produce recombinants inside cells, although to date there is no reported evidence that B. subtilis AddAB actively loads RecA (20).
Additional genetic evidence for a species-specific interaction has been inferred from differences in recombination proficiency observed in E. coli when RecBCD and RecA are provided by different species of Enterobacteriaceae (71,72). Although the relative recombination proficiencies supported by cognate pairs of RecA and RecBCD from E. coli, Serratia marcescens, and Proteus mirabilis in an E. coli recBCD-recA deletion mutant varied, the frequency of recombinants was reduced severalfold in the presence of non-cognate pairs (71,72). Additional work is needed to define the nature of the RecBCD-or AddAB-RecA interaction.
The frequency of recombinants produced by H.pylori AddAB and RecA was reduced compared with that of fully wild-type E. coli (Table 2; compare cross 8 and 9 to cross 1). This may reflect the level of expression of the H. pylori genes in E. coli, protein stability, or another factor. One factor that limited RecA protein activity in our experiments may be the reported post-translational modification of H. pylori RecA by glycosylation in H. pylori but not E. coli (73,74). Expression of H. pylori RecA failed to complement the UV sensitivity of an E. coli recA mutant (74). Our results ( Table 2 and Fig. 3) suggest that expression of H. pylori addAB and recA together would partially restore the UV resistance of an E. coli recA mutant.
Roles of AddAB in H. pylori DNA Damage Repair and Infection-DNA damage from the environment or from cellular activities such as replication or transcription must be repaired for H. pylori to infect and successfully colonize the host. The role of AddAB nuclease and helicase activities in DNA repair was tested by determining the sensitivity of H. pylori wild-type and mutant strains to the topoisomerase inhibitor ciprofloxacin. DNA lesions resulting from ciprofloxacin treatment include ds breaks, putative entry sites for AddAB. The addA HEL B strain was as sensitive to ciprofloxacin as addA or addB deletion strains (Fig. 2). This observation is consistent with the lack of nuclease and helicase activities in the corresponding strains or enzymes (Table 1 and Ref. 8). Inactivation of a single nuclease domain increased sensitivity to ciprofloxacin relative to wild type (Fig. 2), indicating that the helicase and partial nuclease activity remaining in addA NUC B and addAB NUC strains are insufficient to repair DNA damage and maintain cell viability. These data support the importance of AddAB nuclease and helicase activity in maintaining H. pylori viability by facilitating DNA repair following host-inflicted DNA damage.
Successful colonization of the mouse stomach by H. pylori requires resistance to the host immune response and maintenance of bacterial viability. The importance of AddAB nuclease and helicase activity in this process is indicated by the observation that addA and addB deletion mutants colonize less efficiently than wild-type (8), addA NUC B, or addAB NUC strains (Fig. 5). The helicase and partial nuclease activities remaining in strains with a single nuclease domain mutation were sufficient to promote more efficient colonization than addA or addB deletion strains which lack both activities.
Interestingly, H. pylori addA HEL B demonstrated a greater colonization defect than the addA deletion strain (Fig. 5). The mutant AddAB enzymes in both strains lacked detectable nuclease and helicase activity. This suggests that the AddA HEL B mutant enzyme might act upon damaged DNA in such a way that an AddAB-independent repair pathway cannot be used for repair. This pathway would be functional in addA deletion strains and contribute to colonization. The B. subtilis AddA HEL B mutant enzyme binds tightly to dsDNA ends (20); we infer that the same is true for the H. pylori mutant enzyme. AddA HEL B enzyme likely binds to a dsDNA end, blocking accessibility of other repair enzymes. A RecN homolog in H. pylori has been described recently (75); its function may be to join dsDNA ends together for repair (76) in an AddAB-independent pathway. Other pathways may also be involved in repair.
It is clear that DNA repair and homologous recombination contribute to the pathogenicity of bacteria in significant ways. DNA damage inflicted by reactive oxygen species generated as part of the immune response or as a consequence of oxidative stress produced by antibiotic treatment (77) must be repaired for bacteria to establish infection. In many organisms, DNA damage induces the SOS response, the expression of a group of genes that promote survival (78). Expression of low fidelity polymerases results in mutagenic DNA repair that can produce mutations that affect virulence or increase resistance to antimicrobial agents (79). SOS induction requires RecBCD in E. coli (80,81), and AddAB may function in the same capacity in other organisms. Establishing the role of the AddAB helicase and nuclease activities in DNA repair, colonization, and the response to a changing host environment will contribute to our understanding of how H. pylori efficiently colonizes the stomach and perhaps suggest a target for antimicrobial intervention.