Characterization of biochemical properties of an apurinic/apyrimidinic endonuclease from Helicobacter pylori

Apurinic/apyrimidinic (AP) endonucleases play critical roles in the repair of abasic sites and strand breaks in DNA. Complete genome sequences of Helicobacter pylori reveal that this bacterial specie has a single AP endonuclease. An H. pylori homolog of Xth (HpXth) is a member of exonuclease III family, which is represented by Escherichia coli Xth. Currently, it remains unknown whether this single AP endonuclease has DNA repair activities similar to those of its counterpart in E. coli and other bacteria. We report that HpXth possesses efficient AP site cleavage, 3’-repair phosphodiesterase, and 3’-phosphatase activities but not the nucleotide incision repair function. Optimal reaction conditions for HpXth’s AP endonuclease activity are low ionic strength, high Mg2+ concentration, pH in the range 7–8, and temperature 30 °C. The kinetic parameters measured under steady-state conditions showed that HpXth removes the AP site, 3’-blocking sugar-phosphate, and 3’-terminal phosphate in DNA strand breaks with good efficiency (kcat/KM = 1240, 44, and 5,4 μM–1·min–1, respectively), similar to that of E. coli Xth. As expected, the presence of HpXth protein in AP endonuclease—deficient E. coli xth nfo strain significantly reduced the sensitivity to an alkylating agent and H2O2. Mutation of active site residue D144 in HpXth predicted to be essential for catalysis resulted in a complete loss of enzyme activities. Several important structural features of HpXth were uncovered by homology modeling and phylogenetic analysis. Our data show the DNA substrate specificity of H. pylori AP endonuclease and suggest that HpXth counteracts the genotoxic effects of DNA damage generated by endogenous and host-imposed factors.


Introduction
(hereafter referred to as HpXth), homologous to E. coli Xth, encoded by ORF HP1526 [29][30]. Cell-free extracts from the mutant H. pylori strain with disrupted ORF HP1526 no longer have Mg 2+ -dependent AP site cleavage activity [31]. The H. pylori single xthA and double xthA mutY mutant strains show a four-and 37-fold increase in the spontaneous mutation rate in comparison with the wild-type control [32]. Taken together, these data strongly indicate that HpXth is involved in the repair of spontaneously occurring AP sites and other mutagenic DNA lesions in vivo. Nevertheless, at present, detailed biochemical characterization of this enzyme is lacking.
Since H. pylori seems to lack Nfo homologs, it was interesting to define the spectrum of DNA repair activities in its only AP endonuclease, HpXth. Here, we demonstrate that HpXth possesses AP site cleavage and 3'-repair phosphodiesterase functions. The kinetic parameters of HpXth-catalyzed repair activities were measured and compared to those of homologous AP endonucleases. The ability of HpXth to rescue an AP endonuclease-deficient E. coli strain after exposure to an alkylation agent and H 2 O 2 was studied. The evolution, mechanisms of DNA damage recognition, and biological roles of HpXth in the defence against spontaneous and host-induced damage to bacterial DNA are discussed.

Cloning and expression of H. pylori AP endonuclease in E. coli and purification of the HpXth protein
Genomic DNA of H. pylori was provided by the Republican Collection of Microorganisms (Astana, Kazakhstan). Reference gene sequence of xthA of H. pylori was retrieved from the published genome sequences: CP011330.1 (range: 1144587-1145339). The following PCR primers (shown in the 5'!3' direction) were applied to amplify the target HpXth cDNA: HpXth-NdeI, d(GGGAATTCCATATGAAACTGATTTCATGGAATGTGAAC) and HpXth-BamHI, d(CGCGGATCCGTTAAACTAATTCCAACCCTACCGG). The ORF of XthA from H. pylori was cloned into the pET28c(+) vector at the NdeI/BamHI sites, resulting in expression plasmid pET28c/HpXth. The encoded HpXth protein carries an N-terminal 6xHistag sequence. The DNA sequence of plasmid insert was verified against the sequence from the GenBank database in the Vector NTI software Advance™ v11.0 (Invitrogen, USA).
To express the recombinant proteins, E. coli Rosetta 2 (DE3) cells were transformed with plasmids, and the isolated Kn R transformants were grown in the Luria-Bertani (LB) broth with 50 μg/mL kanamycin. At optical density at 600 nm (OD 600 ) of 0.6, the cells were induced by 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG), and incubation was continued at room temperature with shaking (100 rpm) overnight. The induced cells were collected by centrifugation at 6000 × g for 7 min at 4˚C. The bacterial pellet containing HpXth was resuspended in a buffer consisting of 20 mM NaCl and 20 mM Tris-HCl (pH 8.0) and supplemented with the Complete Protease Inhibitor Cocktail (Roche Diagnostics, Switzerland). The cells were disrupted by a French press at 18,000 psi (1250 bar), and then sonicated on ice at 40% amplitude. The lysate was cleared by centrifugation at 40,000 × g for 60 min at 4 C. The supernatant was adjusted to 500 mM NaCl and 20 mM imidazole and loaded onto a 1 mL HiTrap Chelating HP column (GE Healthcare) charged with Ni 2+ . The bound proteins were eluted in a 20-500 mM imidazole gradient and the fractions containing a recombinant HpXth protein were pooled and loaded onto a 1 mL HiTrap Heparin column (GE Healthcare). The bound proteins were eluted with a 50-1000 mM NaCl gradient. Fractions containing the recombinant protein were stored at -20˚C in 50% glycerol.

Oligonucleotides
All the oligodeoxyribonucleotides with modifications and their regular complementary counterparts were purchased from Eurogentec (Seraing, Belgium) and are presented in the 5'!3' direction. They included the 30mer X-RT d(TGACTGCATAXGCATGTAGACGATGTGCAT) where X is tetrahydrofuran (THF, a synthetic analog of an abasic site), α-2'-deoxyadenosine (αdA), 5-hydroxycytosine (5ohC), or 5,6-dihydrouracil (DHU), and complementary 30mer oligonucleotides containing dA, dG, dC, or T opposite the lesion. DNA substrates were constructed as described previously [26,33]. Briefly, the complementary oligonucleotides were annealed; the resulting duplex oligonucleotides are referred to as X•C (G, A, T), respectively, where X is a modified residue. The DNA sequence contexts have previously been used to study the DNA substrate specificity of bacterial, yeast, and plant AP endonucleases [26,[34][35]. Chemical structures of DNA adducts used in this work are shown in S1 Fig. DNA end labeling were performed as described previously [26]. Briefly, oligonucleotides were either 5'-end labeled by T4 polynucleotide kinase (or by phosphatase minus mutant PNK) (New England Biolabs, OZYME, France) in the presence of γ/ 32 P]ATP (3000 CiÁmmol −-1 ) (PerkinElmer SAS, France) or 3'-end labeled by terminal transferase (New England Biolabs) in the presence of α[ 32 P]3'-dATP (cordycepin 5'-triphosphate, 5000 CiÁmmol -1 ) (PerkinElmer) as recommended by the manufacturers. The labeled oligonucleotides were annealed to their appropriate complementary oligonucleotides in a buffer consisting of 50 mM KCl and 20 mM HEPES-KOH (pH 7.5) at 65˚C for 3 min and cooled down to room temperature for 2 h. The resulting duplex oligonucleotides are referred to as X•C (G, A, T), respectively, where X is a modified residue.
The kinetic parameters of repair reactions were measured as described [37]. Briefly, 10-6000 nM duplex oligonucleotide substrate was incubated under the appropriate reaction conditions. The reaction products were quantified on a PhosphorImager after separation by denaturing polyacrylamide gel electrophoresis (PAGE). The kinetic constants were measured at the initial velocity, and K M and k cat values were determined by fitting the data to the one-site binding model in Prism 4 (GraphPad Software, USA). The kinetic parameters for AP endonuclease-catalyzed 3'!5' exonuclease reaction were determined by quantifying the reaction products expressed as a percentage of the total substrate. When an exonuclease generated multiple DNA products, the value obtained for each degradation fragment was multiplied by the number of catalytic events required for its formation, and the total exonuclease activity was calculated as the sum of these products.
The products of the reactions were analyzed as described [26]. Briefly, all reactions were stopped by adding 10 μL of a solution consisting of 0.5% SDS and 20 mM EDTA and then desalted by passing through homemade spin-down columns filled with Sephadex G25 (GE Healthcare) equilibrated in 7.5 M urea. The desalted reaction products were separated by electrophoresis in a denaturing 20% (w/v) polyacrylamide gel (7.5 M urea, 0.5× TBE). A Fuji Phosphor Screen was exposed to the gels and then scanned with Fuji FLA-9500 and analyzed in the Image Gauge v.4.0 software. At least three independent experiments were conducted for all quantitative and kinetic measurements and also gel images.

Preparation of anti-HpXth antibodies and Western blotting
To produce polyclonal antibodies, a six-month-old rabbit was immunized with the recombinant purified HpXth protein. On the first day of immunization, 0.3 mg of the HpXth protein diluted in phosphate buffer was injected subcutaneously in the back of the rabbit with complete Freund's adjuvant. After 7 days, a rabbit was immunized with 0.15 mg of the HpXth protein in the same buffer. The third, fourth and fifth immunizations with 0.15 mg of the antigen were carried out every 7 days. Two days later, the sixth immunization was carried out, and the next day the blood serum was collected and the production of anti-HpXth antibodies was checked by immunoblotting. Animal maintenance and experimental procedures were in accordance with the Kazakhstan national regulations based on the provisions of the Central Ethics Commission of Ministry of Health of the Republic of Kazakhstan (Astana, Kazakhstan) and the Legislation of the European Convention for the Protection of Vertebrate Animals Used for Experimental and other Scientific Purposes (ETS 123, Strasbourg, 1986). The protocol used in the present study was approved on January 5, 2016 by the Local Ethical Commission of the RSE "National Center for Biotechnology", Astana, Kazakhstan.

Alkylation and oxidative DNA damage sensitivity
Drug treatment was performed as previously described [26]. In brief, overnight bacterial cultures were diluted 100-fold in LB broth containing 150 μg/mL ampicillin and 0.05 mM IPTG and incubated at 28˚C until the mid-exponential phase (OD 600 = 0.6). The cells were collected by centrifugation, washed once, and resuspended in phosphate-buffered saline (PBS). Tenfold serial dilutions were prepared in PBS as well. Alkylating agent MMS was added to 2.5 mL of molten soft agar (0.6% agar in LB) at 46˚C containing 1 mM IPTG, followed immediately by 0.25 mL of each cell dilution, and the mixture was poured onto the surface of a 1.5% solid LB agar (25 mL) plate. For an oxidizing-agent sensitivity test, each cell dilution was exposed to 10 mM H 2 O 2 for different periods, and then the cells were mixed with molten soft agar and poured onto LB plates. Colonies were scored after 1-2 days of incubation at 37˚C. Data used in survival curve construction and analysis are representative of at least three independent experiments.

Phylogenetic analysis and model building
The sequences of the conserved domains of E. coli Xth, human APE1 and APE2, Neisseria meningitidis Nape, and Methanothermobacter thermautotrophicus Mth212 were used to search the nonredundant NCBI database of protein sequences by means of BLAST [39]. One representative sequence from each family per taxonomic class was selected for analysis, unless the same species had two paralogs from the same family, in which case, both were selected. Structural model of HpXth was built by multiple iterative threading with searching of the full PDB database using I-TASSER [40]. All sequence alignments and phylogenetic trees were built in Clustal Omega [41]. The conservation analysis was carried out with the Taylor set of amino acid properties [42] and the hierarchical analysis approach outlined elsewhere [43], except that the conservation at any position was expressed as the mean of the number of crossed borders in the Venn diagrams covering all sequence pairs in the compared alignments and normalized to the number of pairs. Images were prepared in the PyMol software (Schrödinger, New York, NY). Phylogenetic trees were visualized using iTOL [44].

Liquid chromatography with tandem mass spectrometry (LC-MS/MS)
Electrophoretic separation of proteins was performed by SDS-PAGE in a 4-14% Bis-Tris gradient gel. The protein bands were excised from the gel and processed as described previously [45][46]. Next, 15 μL of 190 mM ammonium bicarbonate was added to 100 μL of 13.33 ng/μL trypsin (Promega, V5280) for activation of the enzyme, and the samples were incubated at 37˚C for 16 h. The peptide mixture was next purified in ZipTip C18 pipette tips (Millipore, 0.6 μL bed, ZTC18S096). The peptide-containing supernatants were dried at 35˚C in a Speed-Vac for 30 min, then resuspended in 10 μL of 0.1% trifluoroacetic acid for mass spectrometry experiments.
We used a trapping column setup (Acclaim PepMap100 C18 precolumn, 5 mm × 300 μm; 5 μm particles; Thermo Scientific) and a Dionex high-performance liquid chromatography pump (Ultimate 3000 RSLCnano System, Thermo Scientific). For this experiment, peptides were separated on a Acclaim Pep-Map RSLC column (15 cm × 75 μm, 2 μm particles; Thermo Scientific) with a 75 min multistep acetonitrile gradient (Buffer A: 0.1% formic acid; buffer B 90% acetonitrile/10% H 2 O in 0.1% formic acid) at a flow rate of 0.3 μL/min, according to the following time table: 0 min, 2% B; 10 min, 2% B; 58 min, 50% B; 59 min, 99% B; 69 min, 99% B; 70 min, 2.0% B; 75 min, 2.0% B. The unmodified CaptiveSpray ion source (Capillary 1300 V, dry gas 3.0 L/min, dry temperature 150˚C) was employed to interface the LC system with an Impact II (Bruker). For quantification, full-scan MS spectra were acquired at a spectral rate of 2.0 Hz followed by acquisition of one MS/MS spectrum. For data acquisition from the sample, the two most abundant precursor ions were selected for fragmentation, resulting in a total cycle time of 3 s. The mass range of the MS scan was set to extend from m/z 150 to 2200 in positive ion polarity mode.
We used Mascot software to perform searches against the SwissProt 2014_08 database (546,238 sequences; 194,363,168 residues). Search parameters were set as follows: variable modifications, oxidation (M), fragment ion mass tolerance, 20 ppm; parent ion tolerance, 20 ppm.

cDNA cloning and purification of the HpXth protein
The cDNA coding for AP endonuclease HpXth was prepared by PCR from the DNA of H. pylori, and oligonucleotide primers as described in Materials and Methods. The gel-purified PCR 753 bp fragment was inserted into pET28c(+) to create pET28c/HpXth expression plasmid. Competent E. coli DH5α cells were transformed with the ligation mixture, and plasmids containing inserts were isolated from the transformant colonies. The primary structures of inserts in the plasmids were verified by sequencing. A comparison of the cloned cDNA with the respective reference sequences present in GenBank revealed polymorphism in HpXth, manifested in the form of amino acid substitution: G181D. The calculated molecular mass of His-tagged HpXth is 31.6 kDa.
To characterize the DNA repair activities of the recombinant HpXth protein, we affinitypurified the bacterial AP endonuclease from the E. coli Rosetta 2 (DE3) strain expressing the Nterminal His-tagged form of HpXth as described in Materials and Methods. After that, the homogeneity of protein preparations was assessed by SDS-PAGE in a 12% gel, revealing >95% purity of the HpXth preparations (S2 Fig). To confirm the identity of the purified protein, the bands were excised from the gel and in-gel digested with trypsin. The resulting peptide mixtures were subjected to matrix-assisted laser desorption ionization-time of flight MS and electrospray ionization tandem mass spectrometry, respectively, and individual peptide sequences were identified based on their molecular weights. Most of the identified peptides matched the expected amino acid sequences of the HpXth protein, with~80% sequence coverage of the protein. These results indicated that the purified protein is indeed the AP endonuclease from H. pylori.

Characterization of AP site cleavage activity of the H. pylori AP endonuclease in the presence of metal cofactors
Previous study has shown Mg 2+ -dependent AP site cleavage activity in a cell-free extract of H. pylori dependent on the function of gene XthA coding for a homolog of E. coli Xth [31]. Nonetheless, it remained unclear whether the HpXth protein, the product of XthA, has efficient AP endonuclease and other DNA repair activities. To clarify this issue, we measured incision activities of the purified HpXth protein on the 5'-[ 32 P]labeled 30mer oligonucleotide duplex THF•T, containing THF, an abasic site analog, at position 11, in the presence of 5 mM MgCl 2 . As shown in Fig 1, incubation of THF•T with HpXth generated a fast-migrating 10mer cleavage product (lanes 2-11) indicating the presence of AP endonuclease activity. HpXth exhibited a very efficient AP site cleavage activity because 0.5 nM enzyme was able to cleave >95% of 10 nM oligonucleotide substrate after only 5 min incubation under our experimental conditions (lanes 4 and 11). Incubation of THF•T with 1 nM HpXth generated the second minor cleavage product migrating similarly to the 9mer fragment (lane 5) suggesting degradation of the initial 10mer cleavage product by the 3'!5' exonuclease activity of the enzyme.
To characterize the AP site cleavage activity of H. pylori AP endonuclease in more detail, we measured the HpXth protein-mediated cleavage of THF•T in a buffer supplemented with various concentrations of divalent metal chlorides and sulphate, including MgCl 2 , MnCl 2 , CaCl 2 , CoCl 2 , NiCl 2 , CuSO 4 , ZnCl 2 , or FeCl 2 , (Fig 2 and S3 Fig). As shown in Fig 2A, in the presence of 1 or 5 mM Mg 2+ or Mn 2+ , the purified HpXth protein exerted an efficient AP site cleavage activity by generating fast-migrating 10mer cleavage fragments (lanes 6 and 7 and 10 and 11) which migrated similarly to the 10mer product generated by the control enzyme, human APE1 (lane 1). Nonetheless, higher concentrations of Mg 2+ or Mn 2+ (10-20 mM) resulted in a strong decrease in the HpXth-catalyzed AP site cleavage (Fig 2A, lanes 9 and 13, and Fig 2B). The presence of 1 or 5 mM Ca 2+ caused much less efficient stimulation of HpXth's AP site cleavage activity as compared to Mg 2+ or Mn 2+ (Fig 2A, lanes 14 and 15). Again, high concentrations (10-20 mM) of Ca 2+ resulted in strong inhibition of the AP endonuclease activity (lanes 16 and 17). It is noteworthy that no cleavage of AP site by HpXth was observed in the absence of any metal cations or in the presence of EDTA (lanes 3-5), indicating the absolute necessity of divalent metal ions for enzymatic catalysis. Taken together, these findings mean that HpXth has a nonlinear dependence on the presence of various concentrations of Mg 2+ and Mn 2+ : its AP site cleavage activity increased in the presence of 1-5 mM and then decreased in the presence of 10-20 mM MgCl 2 and MnCl 2 (Fig 2).
Further characterization of HpXth regarding the divalent-metal requirement showed that at a very low (0.01 mM) concentration, Zn 2+ , Ni 2+ , Fe 2+ , or Cu 2+ stimulated the AP endonuclease activity, while the same metal cations at 0.1-1.0 mM strongly inhibited AP site cleavage activity of the bacterial enzyme (S3 Fig). To further substantiate inhibitory effects of the above metals, we measured HpXth activity toward THF•T in a buffer supplemented with various concentrations of both metals MgCl 2 and ZnCl 2 . The results revealed that even low concentrations of ZnCl 2 (0.1-1 mM) in the reaction mixture already containing 1-5 mM MgCl 2 , strongly inhibited AP site cleavage activity of HpXth (S4 Fig), suggesting that Zn 2+ and heavier metal cations can efficiently compete with Mg 2+ for the enzyme's metal-binding site. Finally, all these results suggest that among metal cations heavier than Mg, only Mn 2+ and Ca 2+ can stimulate HpXth AP endonuclease activity in the concentration range 1-5 mM, whereas all other cations, except Co 2+ , stimulate this enzymatic activity only at very low concentrations.

Effects of pH, temperature, ionic strength, and substitution of a conserved amino acid on H. pylori AP endonuclease activity
To confirm the biochemical properties of the purified HpXth protein, we measured cleavage of the 30mer THF•T duplex at various MgCl 2 concentrations, ionic-strength levels, pH levels, and temperatures. As shown in Fig 3, MgCl 2 (panel A), pH (panel B), and temperature (panel C) dependence of the HpXth-catalyzed AP site cleavage was clearly bell-shaped. HpXth manifested the highest AP endonuclease activity at 5 mM MgCl 2 , pH 8.0, and 30˚C. Notably, the H. pylori AP endonuclease was sensitive to elevated temperature and ionic strength and was inhibited at 37˚C and 50 mM KCl; furthermore 10-fold inhibition was observed at 150 mM KCl (Fig 3D). These results suggest that the requirements for reaction conditions of HpXth are different from those of E. coli Xth and human APE1 AP endonucleases. To ensure that the observed AP site cleavage activity is not due to trace contamination via expression of host endonucleases, we constructed mutant of HpXth by site-directed mutagenesis, then purified it by the same scheme as that for the wild-type enzyme. We introduced the mutation D144N into HpXth because in the Xth family enzymes, the residue homologous to D144 of HpXth forms a coordination bond with the catalytic metal ion. In the human APE1 protein, the corresponding mutation D210N reduces the enzymatic activity~10,000-fold [47]. The purified HpXth-D144N mutant protein was incubated with the 5'-[ 32 P]labeled THF•T and recessed Exo20•RexT Rec duplexes to measure AP site cleavage and 3'!5' exonuclease activities, respectively. Note that the recessed Exo20•RexT Rec duplex oligonucleotide was composed of a long regular 40mer RexT template fragment and a short complementary regular 5'-[ 32 P]labeled 20mer Exo20 fragment. As shown in Fig 4, mutant H. pylori AP endonuclease, even when present in excess, did not show any detectable AP site cleavage activity (lanes 5-7) as compared to wild-type HpXth (lanes 2-4). Of note, HpXth D144N mutant concomitantly lost its nonspecific 3'!5' exonuclease activity toward the Exo20•RexT Rec duplex, whereas HpXth-WT exerted a DNA-degrading activity under the same reaction conditions (Fig 4,  lanes 12-14 versus lanes 9-11). Altogether, these results indicated that D144 of HpXth is

Characterization of 3'-repair and NIR activities of pure H. pylori AP endonuclease
Hydrolytic AP endonucleases, in addition to their classic AP site cleavage activity, catalyze the removal of 3'-blocking sugar-phosphates and 3'-terminal phosphate remnants from DNA strand breaks generated by ROS or DNA glycosylases/AP lyases. To test whether HpXth has the 3'-repair phosphodiesterase activities, we prepared nicked Exo20 THF •RexT Nick and Exo20 P •RexT Nick duplexes composed of a long 40mer RexT template fragment and two short complementary DNA fragments: the regular 19mer and 20mer Exo20 fragment, containing a 3'-terminal THF and 3'-phosphate residue, respectively. Of note, Exo20 THF •RexT Nick and Exo20•RexT Rec duplexes with 5'-[ 32 P]labeled Exo20 were used to measure the 3'-repair phosphodiesterase and 3'!5' exonuclease activities, respectively. The Exo20 P •RexT Nick duplex with 3'-[ 32 P]labeled Exo20 was employed to measure the 3'-phosphatase activity.
As shown in Fig 5A, HpXth cleaves the nicked 5'-[ 32 P]labeled Exo20 THF •RexT Nick duplex generating a 20mer DNA fragment that migrates faster than the 20mer Exo20 THF substrate, indicating that the enzyme removes the 3'-terminal THF residue, thus leaving a Exo20 fragment with 3'-OH (lanes 2-6 and 10-12). Even at a low (0.05 nM) enzyme concentration, HpXth removed the 3'-THF residue in more than 90% of Exo20 THF •RexT Nick duplex  bottom of the gel (Fig 5B, lanes 6-9), similar to the product generated by calf intestinal alkaline phosphatase (CIP; lane 1), indicating that the bacterial AP endonuclease removes the 3'-terminal [ 32 P] residue leaving an unlabeled Exo20 fragment. Of note, a higher protein concentration of HpXth (5 nM) was required to remove more than 50% of phosphate residues from the Exo20 P •RexT Nick duplex (lane 7) as compared to the Exo20 THF substrate.
We further examined the DNA substrate specificity of H. pylori enzyme by means of 3'-[ 32 P]labeled 30mer duplex oligonucleotides containing a damaged nucleotide: αdA, DHU, or 5ohC. In control experiments, APE1 cleaved αdA•T, DHU•G, and 5ohC•G duplexes (Fig 6,  lanes 1, 6, and 11). No cleavage 5' to the lesion site of DNA duplexes was observed in the presence of HpXth (lanes 3-5, 8-10, and 13-15), suggesting that H. pylori AP endonuclease does not possess a NIR activity under the experimental conditions tested.

Steady-state kinetic parameters of the DNA repair reactions catalyzed by H. pylori AP endonuclease
To further substantiate the DNA substrate specificity of the purified HpXth protein, we determined steady-state kinetic parameters of the repair activities and calculated K M , k cat , and k cat / K M values for the cleavage of various DNA substrates. The 30mer THF•T and 40mer Exo20 THF •RexT Nick , Exo20 P •RexT Nick , and Exo20•RexT Rec duplex oligonucleotides with an AP site residue and a 3'-THF, 3'-P, and 3'-OH group were applied to measure AP endonuclease, 3'-phosphodiesterase, 3'-phosphatase, and 3'!5' exonuclease activities, respectively. For comparison of the kinetic parameters, E. coli Xth and human APE1 data are presented. As shown in Table 1, steady-state kinetic parameters of HpXth-catalyzed repair reactions revealed that the H. pylori enzyme has a strong substrate preference for AP sites in duplex DNA. The k cat /K M value of HpXth for AP site cleavage is 28-and 9-fold higher than that of its 3'phosphodiesterase and 3'!5' exonuclease activities. It is noteworthy that the k cat /K M values for AP site cleavage and 3'-phosphodiesterase activities of HpXth toward the THF•T and Exo20 THF •RexT Nick duplexes were quite similar to those of E. coli Xth, but nearly 10-fold lower than those of human AP endonuclease 1 (APE1). Similar to human APE1, HpXth has a weak 3'-phosphatase activity with~220-fold lower k cat /K M as compared to that of its AP endonuclease function. Overall, these kinetic parameters suggested that H. pylori AP endonuclease is similar to E. coli Xth and can be considered a major cellular enzyme that processes AP sites and DNA strand breaks containing 3'-blocking groups in vivo.

Homology modeling and phylogenetic analysis of HpXth
AP endonucleases belonging to the Endonuclease-Exonuclease-Phosphatase (EEP) superfamily are classified into five families based on their sequence homology: ExoIII-like (with E. coli Xth being the prototypical member), NApe-like (similar to Neisseria NApe endonuclease), Mth212-like (similar to Methanothermobacter Mth212 endonuclease), Ape1-like and Ape2-like (similar to human APE1 and APE2, respectively). Unlike its E. coli homolog, HpXth is classified as an Ape1-like endonuclease by the NCBI Conserved Domain Database [48]. We re-examined the phylogenetic relations among the EEP superfamily AP endonucleases using a set of 284 sequences evenly spread over all domains of life (Fig 7A and S1 File). HpXth, together with several other bacterial and archaeal sequences, formed a separate clade among the Ape1-like family members (labeled as Archaea+Bacteria in Fig 7A). Bacillus subtilis ExoA protein, an efficient AP endonuclease with a low NIR activity [49], also fell into this clade. A closer look at Epsilonproteobacteria, the class to which H. pylori belongs, revealed that basal Epsilonproteobacteria possess AP endonucleases from ExoIII-like, Mth212-like, and Ape1-like families, but some of those were lost during the evolution: in Helicobacteraceae, only Ape1-like sequences remain, while its sister group, Campylobacteraceae, retained only Mth212-like AP endonucleases.
https://doi.org/10.1371/journal.pone.0202232.g006 Each substrate was used to measure a specific DNA repair function under the optimal reaction conditions: the THF•T duplex for the AP endonuclease activity; the Exo20 THF •RexT Nick nicked DNA duplex for the 3'-repair phosphodiesterase activity; the Exo20 P •RexT Nick nicked DNA duplex for the 3'-phosphatase activity; and the Exo20•RexT Rec recessed DNA duplex for the 3'!5' exonuclease activity (see Materials and methods). b Kinetic parameters of the E. coli Xth-catalyzed AP site cleavage and 3'-phosphodiesterase reaction were taken from other studies [26,59]. c Kinetic parameters of human APE1-catalyzed DNA repair reactions were taken from ref. [37].
https://doi.org/10.1371/journal.pone.0202232.t001 One exception is Asn9 of HpXth, which corresponds to Asp70 in the human protein and forms part of a catalytic Mg 2+ -binding site (Fig 7C). However, Ape1-like enzymes vary at this position, containing Asp, Asn, or small amino acids (Gly/Ala). Despite influencing the metal preference and/or the balance of endonuclease, exonuclease, and 3'-phosphodiesterase activities [33], both Asn and Asp are likely suitable for the biologically relevant function of EEP superfamily AP endonucleases.

Drug sensitivity of an E. coli DNA repair-deficient strain expressing the H. pylori AP endonuclease
To examine the biological roles of HpXth's activities, we employed the phenotype rescue assay of E. coli mutant. We measured the sensitivity of AP endonuclease-deficient E. coli xthA nfo BH110 (DE3) strain harboring plasmid pBW21 coding for the E. coli Nfo and plasmid pBluescript SK(+) containing inserts coding for wild-type or mutant pathogen's AP endonuclease HpXth or HpXth-D144N, respectively, to exposure to MMS and H 2 O 2 . Western blotting using rabbit anti-HpXth polyclonal antibodies confirms the presence of the HpXth proteins in E. coli BH110 (DE3) cells harboring the plasmids that encode the AP endonuclease from H. pylori (S6 Fig). MMS methylates purines in cellular DNA, subsequently methylated purines are excised by DNA glycosylases in the BER pathway leading to the appearance of AP sites in DNA [56]. Exposure to H 2 O 2 oxidizes DNA bases and also generates single-strand DNA breaks with blocked 3' termini [57]. The AP endonuclease-deficient E. coli BH110 (DE3) strain lacking both Xth and Nfo is very sensitive to both agents [58]. As expected, the plasmid that encodes for E.coli Nfo rescues BH110 (DE3) cells after exposure to MMS and H 2 O 2 (Fig 8). Consistent with our biochemical results, the pBSK-HpXth plasmid encoding HpXth restored the resistance to MMS to the level close to that of plasmid-harboring E. coli Nfo, showing a 10-to 10 4fold increase in the cell survival at different concentration of MMS as compared to the control empty plasmid (Fig 8A). Plasmid pBSK-HpXth partially restored the resistance to H 2 O 2 showing a 5-to 100-fold increase in the cell survival at different periods of exposure to H 2 O 2 as compared to the control empty plasmid (Fig 8B). As expected, the plasmid harboring mutant HpXth-D144N did not confer resistance to MMS and H 2 O 2 , as was the case for the empty control plasmid, suggesting that this enzyme's catalytic activities are essential for their biological functions (Fig 8).
Remarkably, the E. coli BH110 (DE3) cells expressing HpXth-D144N were more sensitive to MMS as compared to cells carrying the empty control vector, suggesting that the mutant H. pylori protein may inhibit repair of AP sites by other back-up DNA repair pathways. Taken together, these results suggested that HpXth can efficiently repair AP sites and only to some extent 3'-blocking groups in E. coli.

Discussion
Unlike many bacteria, including E. coli, which possess two structurally unrelated AP endonucleases, the Helicobacter lineage has retained only a homolog of exonuclease III. Previously, it has been demonstrated that the mutants of H. pylori lacking HpXth (nucT HP1526 and xthA:: cat) have undetectable Mg 2+ -dependent AP site cleavage activity and show a 4-fold increase in the spontaneous mutation frequency over the wild-type strain, suggesting that HpXth is a functional homolog of E. coli Xth [31][32]. Nevertheless, HpXth has not been biochemically characterized to date. In this study, we have cloned cDNA encoding the HpXth, overexpressed and purified it from E. coli, and characterized the DNA substrate specificity of the enzyme. Our results suggest that HpXth protein possesses AP site cleavage, 3'-repair phosphodiesterase, 3'-phosphatase, and 3'-exonuclease functions. Furthermore, HpXth, just as E. coli Xth, but in contrast to their human homolog APE1, has no damage-specific nucleotide incision activity.

Enzymatic activities of HpXth
In line with the well-characterized E. coli Xth, our data indicate that HpXth is a divalent metal ion-dependent enzyme requiring Mg 2+ or Mn 2+ for the full activity (Figs 2 and 3). Of note, HpXth was considerably inhibited in the presence of 0.1-1.0 mM ZnCl 2 (S3 and S4 Figs), suggesting that Zn 2+ ions could thwart AP site repair in H. pylori. The detailed characterization of the reaction conditions showed that HpXth-catalyzed AP site cleavage is optimal at a high Mg 2+ concentration (5 mM), neutral pH, low ionic strength, and 30˚C (Fig 3). These reaction conditions with low salt and relatively high Mg 2+ more closely resemble the optimum of E. coli Xth [59] than that of human APE1 [13]. The Mg 2+ concentration, ionic strength, pH, and even temperature dependence curves of HpXth-catalyzed AP endonuclease activity are all bellshaped. These observations may imply that the catalytically active protein conformation is sensitive to changes in the reaction conditions; perhaps in vivo, the HpXth-catalyzed reactions are tightly regulated.
The preparation of recombinant HpXth displayed nonspecific 3'!5' exonuclease activity degrading nicked DNA after the AP site cleavage and undamaged recessed DNA duplexes (Figs 1 and 4). To confirm that the exonuclease activity is an intrinsic property of this AP endonuclease, we constructed and characterized the HpXth containing a D144N mutation that should disrupt the metal-binding shell. The mutant completely lost both the ability to cleave AP sites and the 3'!5' exonuclease activity (Fig 4). Hence, the the 3'!5' exonuclease function is intrinsically present in HpXth and may play an important physiological role.
To get an idea about comparative importance of the observed enzymatic activities of HpXth, we have measured the steady-state kinetic parameters of its AP site cleavage, 3'-repair phosphodiesterase, 3'-phosphatase, and 3'!5' exonuclease activities. The analysis of kinetic constants revealed that HpXth possesses efficient AP endonuclease and 3'-repair phosphodiesterase activities with k cat and k cat /K M values very similar to those of E. coli Xth (Table 1). For both bacterial enzymes, these constants were 2.5-10-fold lower in comparison with the human counterpart, APE1. Nevertheless, the kinetic parameters suggest that HpXth can efficiently protect H. pylori from endogenous and induced AP sites and strand breaks in genomic DNA.

Phylogenetic distribution and structural features of EEP activities
In addition to the AP site and 3 0 -end processing activities, certain EEP superfamily enzymes possess the NIR activity that nicks DNA at some sites containing damaged bases. Notably, the NIR activity is distributed unevenly in the phylogeny: it has been reported for human APE1 [13], Arabidopsis ARP [33], and Methanothermobacter Mth212 [60], but is lacking or negligible in ExoIII-like enzymes [13,26], bacterial/archaeal Ape1-like enzymes [ [49] and this work] and plant Ape1L [35]. Ape2-like enzymes, in view of their low AP endonuclease activity and the predominant 3'!5'-exonuclease activity [61], are also likely to be free of the NIR activity. The identification of NIR-deficient HpXth, grouped with low-NIR B. subtilis ExoA, allowed us to revisit the requirements for efficient NIR among the EEP endonucleases.
The phylogenetic analysis of the EEP superfamily AP endonucleases suggests that either the NIR function was present in an ancestral EEP endonuclease and then was lost in certain lineages (such as ExoIII-like, bacterial/archaeal Ape1-like, and plant Ape1L), or was independently acquired by APE1, ARP, and Mth212-like homologs. Because the loss of a function is more probable than independent emergence of one in several lineages, we examined the conservation of residues in different EEP groups to find which positions differentiate NIR-proficient enzymes from NIR-deficient ones, assuming the Ape1-like and Mth212-like enzymes to be NIR-proficient unless clustering with B. subtilis ExoA/HpXth or Ape1L. We identified three positions that could be related to the NIR proficiency; of note, all of them are involved in forming contacts with DNA away from the active site (Fig 7B, carbons colored red). The first two corresponded to Cys99 and Gly127 in human APE1. Gly127 was replaced by Lys/Arg in all Ape1L enzymes, whereas the homologs of Cys99 were found to have small side chains in most NIR-proficient enzymes and aliphatic bulky ones in NIR-deficient enzymes (e.g., Met in HpXth or Val in E. coli Xth). In human APE1, Cys99 and Gly127 form a part of the structure that is inserted into the DNA minor groove 5' to the damaged nucleotide [55]. At the same side but closer to the lesion, Gly176 in APE1 corresponds to small-side chain residues in the major groove in most NIR-proficient enzymes, whereas NIR-deficient ones possess a charged or polar group at the equivalent position (Glu in E. coli Xth, Gln in HpXth). These differences could lead to different modes of DNA kinking upon binding to the enzyme, and these conditions might be crucial for the correct entry of base-containing nucleotides into the active site of AP endonucleases. A comparison of Ape1-like NIR-proficient enzymes with NIR-deficient ones reveals that the former group has Phe/Tyr in the active site pressing against the everted abasic nucleotide (Phe266 in human APE1), whereas all NIR-deficient enzymes have a bulkier Trp side chain at this position, possibly interfering with the correct eversion of the damagedbase-containing nucleotide during NIR.
Many bacterial enzymes, including AP endonucleases, are different from their eukaryotic counterparts by having no or little regulatory additions to the catalytic domains. Comparing human APE1 with bacterial Xth, it is clear that the human protein carries an additional N-terminal tail, where all regulatory Lys residues are located. Sequence alignment of human APE1 and E. coli and H. pylori Xth clearly indicate that there is no counterparts of regulatory Lys [62] and redox-active Cys residues [63] in the bacterial enzymes (S7 Fig). This observation most likely reflects the differences in the intracellular distribution of the human and bacterial enzymes and in their involvement in the transcription regulation.

Biological role of a single H. pylori AP endonuclease
Efficient AP endonuclease and 3'-end repair functions of HpXth suggest that H. pylori AP endonuclease plays an important role in the repair of AP sites and single-strand DNA breaks with blocked 3' termini generated by alkylating agents, ROS, or DNA glycosylases. Consistent with DNA substrate specificities characterized in vitro, the expression of HpXth in E. coli AP endonuclease-deficient BH110(DE3) strain conferred resistance to MMS comparable to that provided by a plasmid encoding E. coli Nfo (Fig 8A). In addition, HpXth imparted resistance to H 2 O 2 but to a lesser degree than E. coli Nfo did ( Fig 8B); this finding may be consistent with less efficient 3'-end repair functions as compared to the AP site cleavage activity (Table 1). Also, the absence of the NIR function in HpXth found here is suggestive of possibly increased sensitivity of H. pylori to oxidizing agents such as bleomycin and organic peroxides, that induce specific lesions that are subject to NIR [58,64]. In conclusion, our findings together with earlier reports unequivocally identify HpXth as the principal AP endonuclease and 3'-repair phosphodiesterase in H. pylori, protecting this important human pathogen from the genotoxic effects of endogenous and induced AP sites and DNA strand breaks.
Supporting information S1 File. Full-length sequences of 284 Endonuclease-Exonuclease-Phosphatase superfamily AP endonucleases, whose highlighted domains were subjected to the phylogenetic analysis. Domains and their families were assigned according to the Conserved Domain Database [48].