Purification and Characterization of Human NTH1, a Homolog ofEscherichia coli Endonuclease III

The human endonuclease III (hNTH1), a homolog of the Escherichia coli enzyme (Nth), is a DNA glycosylase with abasic (apurinic/apyrimidinic (AP)) lyase activity and specifically cleaves oxidatively damaged pyrimidines in DNA. Its cDNA was cloned, and the full-length enzyme (304 amino acid residues) was expressed as a glutathione S-transferase fusion polypeptide in E. coli. Purified wild-type protein with two additional amino acid residues and a truncated protein with deletion of 22 residues at the NH2 terminus were equally active and had absorbance maxima at 280 and 410 nm, the latter due to the presence of a [4Fe-4S]cluster, as in E. coli Nth. The enzyme cleaved thymine glycol-containing form I plasmid DNA and a dihydrouracil (DHU)-containing oligonucleotide duplex. The protein had a molar extinction coefficient of 5.0 × 104 and a pI of 10. With the DHU-containing oligonucleotide duplex as substrate, theK m was 47 nm, andk cat was ∼0.6/min, independent of whether DHU paired with G or A. The enzyme carries out β-elimination and forms a Schiff base between the active site residue and the deoxyribose generated after base removal. The prediction of Lys-212 being the active site was confirmed by sequence analysis of the peptide-oligonucleotide adduct. Furthermore, replacing Lys-212 with Gln inactivated the enzyme. However, replacement with Arg-212 yielded an active enzyme with about 85-fold lower catalytic specificity than the wild-type protein. DNase I footprinting with hNTH1 showed protection of 10 nucleotides centered around the base lesion in the damaged strand and a stretch of 15 nucleotides (with the G opposite the lesion at the 5′-boundary) in the complementary strand. Immunological studies showed that HeLa cells contain a single hNTH species of the predicted size, localized in both the nucleus and the cytoplasm.

Reactive oxygen species are generated as by-products of oxidative phosphorylation or by ionizing radiation and induce extensive base damage that is mainly repaired via the base excision repair (BER) 1 pathway. This repair is initiated by removal of the damaged base, catalyzed by a DNA glycosylase (1)(2)(3). There are two classes of DNA glycosylases with distinct substrate specificities: the monofunctional simple glycosylase and the glycosylase with associated AP lyase activity. All oxidized base lesions are removed from DNA by DNA glycosylase/AP lyases, which not only catalyze removal of the base lesion but also cause strand cleavage via ␤-elimination. The Escherichia coli endonuclease III (Nth) recognizes a wide range of oxidized pyrimidine derivatives, including ring-saturated and ring-fragmented derivatives such as thymine glycol (Tg), 5-hydroxycytosine, 5,6-dihydrouracil (DHU), and urea (1,(3)(4)(5)(6). This enzyme has been well conserved from E. coli to the humans (7,8). On the other hand, oxidized purine lesions are also repaired by DNA glycosylase/AP lyases, i.e. Mut M (Fpg) of E. coli or OGG of eukaryotes (yeast and mammals), which do not share extensive sequence similarity. Furthermore, unlike these enzymes, the 23.4-kDa E. coli Nth contains a [4Fe-4S] cluster (9). Structural analysis of Nth by x-ray crystallography reveals that this repair enzyme consists of two ␣-helical domains, which contain a helix-hairpin-helix motif and [4Fe-4S] cluster loop (10,11). DNA binds to the cleft between two domains of the enzyme, while the catalytically important lysine and aspartic acid residues (Lys-120 and Asp-138) are positioned at the mouth of the pocket. A unified mechanism of DNA glycosylase/AP lyase activity proposed by Dodson et al. (12) suggests that Asp-138 of the protein deprotonates the Lys-120 residue, which then attacks the deoxyribose C-1 of the lesion, causing release of the base and formation of a covalent Schiff intermediate with DNA (11)(12)(13). The Schiff base intermediate undergoes several transformations resulting in strand cleavage via ␤-elimination (or successively ␦-elimination) to leave 5Ј-phosphate and 3Ј-␣,␤-unsaturated aldehyde (or 3Ј-phosphate for ␤␦-elimination) ends (14). Several eukaryotic homologs of Nth have been cloned from Saccharomyces cerevisiae (15,16), Schizosaccharomyces pombe (17), Caenorhabditis elegans (7,8,18), and mice. 2 Two DNA-binding motifs, as well as Lys and Asp residues corresponding to the catalytic Lys-120 and Asp-138 of E. coli enzyme, were well conserved among these homologs, except that S. cerevisiae Ntg1 does not have a [4Fe-4S] cluster (15).
Although the primary steps of the BER pathway are common to all organisms, the repair pathway in eukaryotes appears to be quite complex. For example, the mammalian cells appear to have at least two distinct BER pathways (19 -21). Although they appear to be homologs of E. coli Nth, eukaryotic enzymes, in particular hNTH, need to be characterized thoroughly because of many critical differences between the bacterial and eukaryotic BER. For example, the repair of Tg, presumably mediated by hNTH1, involves XPG protein, a component of the nucleotide excision repair pathway, which plays an unidentified structural role (22). Preliminary studies on the cloning of hNTH1 have recently been published by other laboratories (7,8). In this report, we have described characterization of the hNTH1 protein including a direct identification of the catalytic residue.

EXPERIMENTAL PROCEDURES
Cloning of hNTH1 cDNA-To isolate hNTH1 cDNA, a human bone marrow cDNA library was screened using a mouse cDNA of the Nth homolog 2 under low stringency conditions described previously (23). One positive clone was obtained from 1 ϫ 10 6 independent plaques. The cDNA was subcloned into pUC19 (designated as phNTH1), and the sequence was determined.
Construction of Expression Plasmid of Glutathione S-Transferase (GST)-hNTH1 Fusion Protein-The DNA sequence encoding amino acids 1-304 of the open reading frame was amplified by polymerase chain reaction (20 cycles at 94°C for 1 min, 65°C for 1 min, and 72°C for 2 min) from the phNTH1 plasmid using Pfu DNA polymerase (Stratagene) and the following primers: Pmet1 (5Ј-CTT GGA TCC ATG ACC GCC TTG AGC GCG AGG-3Ј) and Pter (5Ј-CTC GAA TTC AGA GAC CCT GGG CGG CCG G-3Ј). The polymerase chain reaction product was subcloned into the pGEX-2T plasmid at BamHI/EcoRI sites. The recombinant plasmid was designated pGEX-hNTHmet1 and confirmed the original DNA sequence.
Expression and Purification of hNTH1 Protein-E. coli BL21(DE3)-pLysS carrying pGEX-hNTHmet1 was grown at 37°C until the absorbance at 600 nm reached 0.6. The culture was cooled to 28°C and, after the addition of IPTG to 0.1 mM, was grown at 28°C for 4 h, prior to chilling to 0°C. All subsequent procedures were carried out at 4°C. After the bacteria were harvested by centrifugation, they were resuspended in buffer A (50 mM Tris-HCl, pH 8.0, 500 mM NaCl, 2 mM EDTA, and 0.02% Triton X-100) and then sonicated (6 ϫ 30 s) on ice at full power using a Braun-Sonic U. After centrifugation of the cell lysate (30 min at 15,000 ϫ g), the supernatant was applied to a glutathione-Sepharose 4B (Amersham Pharmacia Biotech) column (5 ml) equilibrated with buffer A. After washing with buffer A, the GST fusion protein was eluted with buffer A containing 15 mM reduced glutathione (Sigma). The fractions that were a yellowish brown color were pooled and dialyzed against buffer B (20 mM Tris-HCl, pH 8.0, 300 mM NaCl, 2 mM EDTA, 0.02% Triton X-100, and 6 mM 2-mercaptoethanol). The GST-hNTH1 fusion protein was then digested with thrombin (0.28 units/mg fusion protein; Novagen) at 4°C for 16 h to cleave the GST adduct. After the reaction was stopped with phenylmethylsulfonyl fluoride (1 mM), the digest was passed through a glutathione-Sepharose column equilibrated with buffer C (20 mM Tris-HCl, pH 8.0, 6 mM 2-mercaptoethanol, and 10% glycerol) containing 300 mM NaCl, to remove GST and undigested fusion protein. The hNTH1 in the pooled flow-through fraction was then fractionated on an SP-Sepharose Fast Flow (Amersham Pharmacia Biotech) column with a linear gradient from 300 to 600 mM NaCl in buffer C. The fractions with the yellowish brown color were analyzed by SDS-PAGE. The full-length hNTH1 was eluted from the column later than the truncated hNTH1; these were dialyzed separately versus buffer D (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.2 mM DTT, 0.005% Triton X-100, and 50% glycerol) and stored at Ϫ20°C.
Oligonucleotide Substrates-A DHU-containing 55-mer oligonucleotide (DHU-55) with the sequence 5Ј-ATT ATG CTG AGT GAT ATC CCT CTG GCC TTC GAA CCC XAC CTC AAC CTC TGC CCA C-3Ј (where X represents DHU) was a gift from Dr. P. Doetsch (Emory University). A uracil (U)-containing 13-mer oligonucleotide (U-13) (5Ј-GCA CAG UCA GCC G-3Ј) was purchased from Life Technologies, Inc. A uracil-containing 50-mer (U-50) with the sequence 5Ј-TCG AGG ATC CTG AGC TCG  AGT CGA CGU TCG CGA ATT CTG CGG ATC CAA GC-3Ј and its complementary strand were synthesized by the NIEHS Center Core Lab at the University of Texas Medical Branch (UTMB). The oligonucleotides were gel-purified and labeled either at the 5Ј-end using T4 polynucleotide kinase and [␥-32 P]ATP, or at the 3Ј-end with Klenow DNA polymerase and [␣-32 P]dCTP as described previously (24,25). An oligonucleotide with a reduced AP site was prepared from a uracilcontaining oligonucleotide by treatment with uracil DNA glycosylase and NaBH 4 . Briefly, uracil-containing DNA (50 pmol) was incubated with 2 units of uracil DNA glycosylase (Life Technologies) in 14 mM Tris-HCl (pH 8.0), 1 mM DTT, 0.1 mg/ml bovine serum albumin, and 0.1 M NaBH 4 at 37°C for 30 min. After extraction with phenol/chloroform and desalting by gel filtration the DNA was lyophilized by vacuum centrifugation.
DNA Glycosylase/AP Lyase Assay-End-labeled oligonucleotide substrates were incubated with appropriate amounts of the hNTH1 at 37°C for 5 min in a reaction mixture (10 l) containing 50 mM Tris-HCl, pH 8.0, 75 mM NaCl, and 1 mM DTT. After the reaction was stopped with 100 l of 0.1% SDS, the DNA was extracted with phenol/chloroform and ethanol-precipitated along with 20 g of glycogen (Boehringer Mannheim). The precipitate was dissolved in 10 l of loading solution (96% formamide, 20 mM EDTA, 0.05% bromphenol blue, and 0.05% xylene cyanol), heated at 95°C for 5 min, and then loaded onto a denaturing 20% polyacrylamide gel in 7 M urea and TBE buffer (89 mM Tris, 89 mM boric acid, and 2 mM EDTA). Radioactivity in the incised oligonucleotide was quantified by exposing the gel to a PhosphorImager (Molecular Dynamics). Another assay for DNA glycosylase/AP lyase, based on the rate of conversion of damaged form I DNA to form II, was also carried out using the Tg-containing pBluescript II SK(Ϫ), which was prepared by treating of the DNA with 0.04% OsO 4 at 70°C for 8 min (26). The reaction mixture (20 l) contained 50 mM Tris-HCl, pH 8.0, 75 mM NaCl, 1 mM DTT, 250 ng of damaged plasmid DNA, and an appropriate amount of the enzyme. After incubation for 5 min at 37°C, 2.2 l of stopping solution (40% glycerol, 0.2% bromphenol blue, and 1% SDS) was added, and the samples were electrophoresed on 0.8% agarose gels in TBE buffer. The intensities of the bands of form I and form II DNA were quantitated from an image of the gel captured using an Eagle Eye II video system and attached software (Stratagene). The average number of nicks (or Tg) per molecule was calculated by assuming a Poisson distribution of Tg formation in the plasmid (27).
Tg-DNA Glycosylase Assay-Plasmid pBluescript II SK(Ϫ) DNA was labeled with [methyl-3 H]thymidine 5Ј-triphosphate by the random priming method and treated with 0.4% OsO 4 at 85°C for 10 min. A reaction mixture (20 l) containing 50 mM Tris-HCl, pH 8.0, 75 mM NaCl, 1 mM DTT, and 300 ng of OsO 4 -treated DNA (36,000 cpm) was incubated at 37°C for 30 min with different concentrations of purified proteins. After incubation, 30 l of stopping solution (0.5 M sodium acetate and 0.67 mg/ml of calf thymus DNA) and 125 l of cold ethanol were added. The DNA was precipitated at Ϫ80°C and centrifuged at 15,000 ϫ g for 15 min, and the ethanol-soluble radioactive material was measured in supernatant by scintillation counting.
DNA Trapping Assay with NaCNBH 3 -The DNA-trapping reaction was performed in a 10-l mixture containing appropriate amounts of hNTH1 and substrate duplex oligonucleotide, 50 M Tris-HCl, pH 8.0, 1 mM EDTA, and 50 mM NaCNBH 3 (Aldrich). After incubation at 37°C for 10 min, the samples were boiled for 3 min in standard SDS-PAGE loading buffer and separated by electrophoresis on a 12.5% SDS-polyacrylamide gel (28). The gel was dried on DE81 paper (Whatman) and analyzed by autoradiography. Relative trapping efficiencies were determined by PhosphorImager analysis.
Identification of the Amino Acid Residue in hNTH1 That Forms Covalent Linkage to the Substrate-To make AP site-containing DNA, a 32 P-labeled U-13 oligonucleotide duplex (42 nmol) was incubated with 175 units of uracil DNA glycosylase in a reaction buffer containing 50 mM Tris-HCl (pH 8.0) and 1 mM EDTA. After the reaction at 37°C for 30 min, hNTH1 (about 60 nmol) was allowed to react with the AP site-containing DNA in the presence of 50 mM NaCNBH 3 at 37°C for 2 h. After the reaction mixture was loaded onto a 5-ml Econo-Pac Q cartridge (Bio-Rad) equilibrated with 20 mM Tris-HCl (pH 8.0), it was washed with equilibration buffer, and the trapped protein-DNA complex was eluted with a linear gradient of 0 -0.5 M NaCl in 20 mM Tris-HCl (pH 8.0). The fractions were assayed by SDS-PAGE and Coomassie Blue staining for the presence of the complex. The fractions from 0.15 to 0.25 M NaCl containing the complex (12 nmol) were pooled and dialyzed versus 50 mM NH 4 HCO 3 . After boiling for 5 min, the complex was hydrolyzed with a combination of 20 g of endoproteinase Glu-C and 20 g of trypsin (Promega, sequence grade) at 37°C for 24 h, and the treatment was repeated for an additional 48 h. The digest was lyophilized in a Speed Vac, dissolved in 0.1% trifluoroacetic acid, and applied to a Vydac C18 (300-Å pore, 5-mm diameter) reversed-phase HPLC column (1 ϫ 250 mm) equilibrated with 15% acetonitrile containing 0.1% trifluoroacetic acid. The column was eluted at a flow rate of 80 ml/min with a gradient (35 min) of 15-40% acetonitrile containing 0.1% trifluoroacetic acid.
Construction of Site-specific Mutants of hNTH1-The mutants K212Q and K212R were generated using Stratagene's Chameleon double-stranded site-directed mutagenesis kit. For this purpose, hNTH1 from pGEX-hNTHmet1 was cloned into the pBluescript vector between the BamHI and EcoRI sites. The primer used to mutate the ScaI site in the Amp r gene of the vector was 5Ј-CTG GTG AGT ATT CAA CCA AGT-3Ј. The mutagenic primers were 5Ј-G GTG GCG CTG CCC GGG GTT GGG CCT CAG ATG GCA CAC CTG G-3Ј for K212Q, and 5Ј-G GTG GCG CTG CCC GGG GTT GGG CCT AGG ATG GCA CAC CTG G-3Ј for K212R, where the mismatches are underlined. The entire hNTH1 amino acid coding region was sequenced to ensure the presence of only the desired mutation.
Immunological Analysis-Antibody was prepared by immunizing rabbits with purified hNTH1 protein (Alpha Diagnostics, San Antonio) and then affinity-purified by binding to hNTH1-GST protein coupled to NHS-activated Sepharose (Amersham Pharmacia Biotech). Western blot analysis was carried out with the affinity-purified antibody (1:250) and horseradish peroxidase-linked anti-rabbit second antibody (1:2000; Amersham Pharmacia Biotech) by using the enhanced chemiluminescence (ECL) technique (Amersham Pharmacia Biotech) according to the manufacturer's protocol. In an indirect immunofluorescence assay, HeLa S3 cells cultured on microscope coverslips were fixed in acetone/ methanol (1:1), treated with 50 mM NH 4 Cl for 10 min, and then permeabilized by a 3-min incubation in 0.1% Triton X-100. The cells were preincubated with human serum for 30 min, washed in PBS containing 0.5% bovine serum albumin and 0.1% Tween 20 (PBS-BT), and finally exposed to affinity-purified hNTH1 antibody (1:100) for 30 min at 37°C. After a 30-min wash with PBS-BT, fluorescein-conjugated anti-rabbit second antibody (Organon Teknika bv, Boxtel, Netherlands) was applied for 30 min at 37°C. The cells were then washed in PBS-BT for 30 min, mounted on a microscope slide in 50% glycerol in PBS, and photographed for fluorescence in a Zeiss UV microscope.
Other Methods-Proteins were quantitated by the dye-binding method (Bio-Rad) using bovine serum albumin as the standard. Measurement of pI by isoelectric focusing and the determination of the extinction coefficient of hNTH1 were carried out as described previously (29). Automated NH 2 -terminal amino acid sequence analysis was performed in the UTMB Protein Chemistry Laboratory, using a Perkin-Elmer/Applied Biosystems Procise protein/peptide sequencer. DNase I footprinting was performed essentially as described previously (25).

RESULTS
Overexpression of Recombinant hNTH1 in E. coli and Purification of the Enzyme-The sequence of the hNTH1 cDNA independently cloned and used in this study is identical to the published sequence of Hilbert et al. (7), except that our clone has two additional nucleotides at the 5Ј-end. The sequence of another hNTH1 cDNA cloned by Aspinwall et al. (8) is different from ours by three nucleotide substitutions within the coding region, as well as a 33-nucleotide addition at the 5Ј-end. The number of amino acid residues of hNTH1 cited here is based on our sequence and that of Hilbert et al. (7).
The DNA sequence encoding amino acids (1-304) of the hNTH1 was amplified by polymerase chain reaction, and the resulting product was inserted into a pGEX-2T vector to over-express the full-length hNTH1 fused to the C terminus of the GST protein. SDS-PAGE of the proteins from bacteria induced or uninduced by IPTG indicated that the 60-kDa protein was overexpressed from the plasmid (data not shown); this molecular mass agreed well with the expected size (59.8 kDa) of the GST-hNTH1 fusion protein. We used proteinase-deficient E. coli to overexpress the enzyme. The majority of the fusion protein was expressed in the soluble form by incubating the cells at 28°C after the addition of IPTG and was readily purified by affinity chromatography (Fig. 1A).
The GST-hNTH1 fusion protein was subsequently cleaved by thrombin to isolate the wild-type hNTH1. We observed two polypeptides of 36 and 34 kDa on SDS-PAGE in the thrombin digest in addition to the 26-kDa band (GST), indicating that thrombin cleaved at two sites in the fusion protein (Fig. 1B). NH 2 -terminal amino acid sequencing of product b showed that the protein was hNTH1 with two additional amino-terminal amino acids (Gly-Ser) derived from the vector. The NH 2 -termi- nal sequence analysis of band c indicated that this band was a truncated form of hNTH1 lacking 22 residues at the NH 2 terminus. The conditions for thrombin digestion were optimized for maximum generation of the full-length protein. After removal of uncleaved protein and GST by chromatography on a glutathione affinity column, two hNTH1 polypeptides with different sizes were finally separated from each other by SP Sepharose chromatography, and each was judged to be apparently homogeneous by SDS-PAGE (Fig. 1C, lanes 5 and 6).
Some physicochemical properties of the full-length hNTH1 protein are summarized in Table I DNA Glycosylase/AP Lyase Activity of Purified hNTH1-The activity of the purified hNTH1 protein was measured using a Tg-containing plasmid DNA and a DHU-containing oligonucleotide. Full-length and truncated hNTH1 nicked the form I Tg plasmid, but not the normal plasmid, at about the same rate to convert it into form II (Fig. 2A). The GST-hNTH1 fusion protein also had Tg-DNA-specific nicking activity, as previously reported by Hilbert et al. (7), but the specific activity of the protein was about half that of the non-fusion enzymes. The hNTH1 incised the DHU-containing oligonucleotide at the DHU site to generate a 3Ј ␣,␤-unsaturated aldehyde (Fig. 2B,  lane 2), which could subsequently be converted to a 3Ј-OH terminus by treatment with E. coli endonuclease IV (Fig. 2B,  lane 3). This showed that hNTH1 had an AP lyase activity that cleaved damaged DNA via ␤-elimination. The product of ␤␦elimination, with a 3Ј-phosphate terminus, was not observed.
The pH optimum of hNTH1 was established to be pH 8 with the Tg-containing plasmid substrate (Fig. 3A). The maximal activity of hNTH1 was observed in 75 mM NaCl or KCl, and higher salt concentrations inhibited the enzyme (Fig. 3B). EDTA was not inhibitory even at 10 mM concentration (Fig.  3C). Some divalent cations (Mg 2ϩ , Mn 2ϩ , and Ca 2ϩ ) inhibited hNTH, and about 50% of the normal activity was observed in the presence of 5-10 mM concentrations of these ions (Fig. 3C).
We measured the reaction rate of hNTH1 with DHU-containing oligonucleotide duplexes having different bases opposite the DHU (data not shown). DHU is a derivative of cytosine, so its opposite base should be guanine, but DHU is able to pair to adenine during DNA replication. That hNTH1 cleaved both duplex oligonucleotides at nearly the same rate indicated that the enzyme activity was not significantly affected by the presence of A or G opposite DHU.
The apparent K m of hNTH1 for DHU-55 was calculated to be about 47 nM from the Lineweaver-Burk plot of the DHU-55 cleavage reaction (data not shown). The enzyme had a k cat of ϳ0.6/min for DHU-55. The value of k cat was corrected by the fraction (25%) of active molecules in the enzyme preparation. Nearly identical kinetic parameters (K m of 36 nM; k cat of ϳ0.7/ min) for hNTH1 were obtained with an AP site-containing 50-mer made from a U-50 oligonucleotide.
Footprinting Analysis of hNTH-bound Substrate-The DNase I footprint for hNTH1 bound to a DHU-containing oligonucleotide is shown in Fig. 4. When the DHU-containing strand was labeled at the 5Ј-end, four nucleotides on the 5Ј-side of DHU were protected from DNase I, but a footprint on the 3Ј-side was difficult to see because of the presence of cleavage product (Fig. 4B). However, using 3Ј-end-labeled DHU-containing oligonucleotide, five nucleotides on the 3Ј-side of DHU and four nucleotides on the 5Ј-side were shown to be protected from  DNase I (Fig. 4C). When the complementary strand was labeled, protection was observed spanning a zone of 15-nucleotide with a hypersensitive site region in between (Fig. 4D). No protection was observed when unmodified oligonucleotide was used (Fig. 4A). Schematic representations of the protected region or hypersensitive sites are depicted in Fig. 5. Similar footprints were obtained using oligonucleotides containing a reduced AP site or tetrahydrofuran at the site of DHU (data not shown).
Cross-linking of the Reaction Intermediate by NaCNBH 3 -DNA glycosylases with associated AP lyase activity normally use an amino group as a nucleophile, resulting in a transient covalent imino enzyme-DNA substrate Schiff base intermediate (12,30). This intermediate can be stabilized by reduction with NaCNBH 3 and monitored by DNA band shifts on SDSpolyacrylamide gels (DNA trapping assay). The assay using a DHU-55 oligonucleotide duplex showed that hNTH1 can be trapped with the DHU-containing oligonucleotide (Fig. 6A). That a dihydro-U:A oligonucleotide was trapped with the same efficiency as the dihydro-U:G oligonucleotide (data not shown) confirmed that the reaction with hNTH1 was independent of the complementary strand. The AP site-containing oligonucleotide was also a good substrate for trapping by hNTH1 (Fig. 6B).
Quantitation of the covalent complex formed in the DNA trapping assay established the fraction of active molecules in the hNTH1 preparation. The hNTH1 protein (1 pmol) was incubated with an increasing concentration of DHU-containing oligonucleotide in the standard buffer in the presence of NaC-NBH 3 (Fig. 6C). After incubation at 37°C, the reactions were analyzed for the number of hNTH1 molecule covalently trapped on DNA. The maximum complex formation (calculated by extrapolation from a double reciprocal plot of the data in Fig.  6C) was 0.25 pmol, indicating that 25% of the hNTH1 molecules were active.
Identification of the Residue That Forms Covalent Linkage to the Substrate DNA-The sequence homology of hNTH1 to E. coli Nth around the putative catalytic site suggested that the Lys-212 of hNTH1 was the nucleophilic residue that formed a covalent DNA complex. To confirm this possibility, we sequenced an HPLC-purified radiolabeled peptide complex obtained by proteolysis of a borohydride-trapped protein-DNA complex prepared using a radiolabeled 13-mer oligonucleotide substrate. After digestion of the complex with endoproteinase Glu-C and trypsin, the peptide fragment bearing the radiolabeled oligonucleotide eluted from a C18 reversed-phase HPLC column at about 30% acetonitrile (Fig. 7). The results of microsequence analysis of the HPLC fraction containing the peptide complex are shown in Table II. Two sequences, one major and one minor, were obtained. The major sequence was identical to the reported sequence of hNTH1 in the region representing Leu-203 to Met-218, which spanned the putative catalytic residue Lys-212. Sequence analysis results of cycle 10 (Lys-212) showed a blank for the major peak, consistent with a lysyl ⑀-amino group adducted with an oligonucleotide. Highly charged 2-anilino-5-thiazolinone amino acid residue complexes formed in sequencing reactions typically are not solvent-extractable from the reaction cartridge of the sequencer and thus give a negative result at that position run during sequence analysis. The minor sequence was established to be a combination tryptic and autocatalytic peptide cleavage product of endoproteinase Glu-C representing the sequence region Thr-191 to Lys-206. Cycle 10 of the endoproteinase Glu-C peptide sequence gave the expected yield for Thr-200, which served as an internal control and demonstrated the absence of any instrument malfunction at this cycle. Therefore, we concluded that Lys-212 was covalently cross-linked to DNA in the trapped hNTH1-substrate complex.
Site-Directed Mutagenesis of a Catalytic Residue Lys-212-To further confirm the involvement of this lysine residue in the enzyme's catalytic mechanism, Lys-212 was replaced with Gln or Arg by site-directed mutagenesis. The mutant proteins (K212Q and K212R) were expressed in a soluble form as GST fusion proteins in E. coli. The specific absorbance of each mutant protein at 410 nm was the same as that of the wild type protein, indicating that the mutant proteins each has an intact [4Fe-4S] cluster (data not shown). The K212Q mutant could not form a covalent complex with the DHU-containing oligonucleotide, and had no DNA glycosylase and AP lyase activity (Fig. 8, A-C). On the other hand, the K212R mutant could trap the DNA substrate and had DNA glycosylase/AP lyase activity, but at a lower level than the wild-type protein.
Kinetic studies showed that K212R had about a 2.5-fold higher K m and a 35-fold lower k cat than the wild type (Table III).  6. Formation of DNA-trapped complex. A, purified hNTH1 was incubated with a 5Ј-32 P DHU-oligonucleotide duplex at 37°C for 10 min with or without 50 mM NaCNBH 3 and analyzed by SDS-PAGE. B, an AP site-containing oligonucleotide duplex was prepared from U-50 oligonucleotide by treatment with uracil-DNA glycosylase (UDG) at 37°C for 10 min and then incubated with hNTH1 with or without NaCNBH 3 . C, quantitation of covalent complex formation. The hNTH1 (1 pmol) was allowed to react with various amounts of DHU-oligonucleotide in the presence of 50 mM NaCNBH 3 at 37°C for 30 min, and the maximum complex formation was calculated by extrapolation from a double reciprocal plot.
FIG. 7. Separation of radioactive peptides by HPLC. The hNTH ( 32 P-labeled)-oligonucleotide covalent complex was digested with trypsin and endoproteinase Glu-C and separated by HPLC as described under "Experimental Procedures." Peak R* contained the radioactivity, which was pooled, dried, and used for NH 2 -terminal sequencing.
b This peptide was a tryptic and autocatalytic cleavage product of endoproteinase Glu-C, which coeluted with the radiolabeled peptide during HPLC. c Expected lysyl residue was not evident in this cycle (see "Results").
However, both mutant proteins formed a gel-shifted complex with a duplex containing a reduced AP site, indicating that the mutation of Lys-212 did not affect the specific DNA binding activity of hNTH1 (Fig. 8D). Immunological Characterization of hNTH1-A polyclonal antibody raised against hNTH1 used in immunoblotting experiments with HeLa S3 cell extract (Fig. 9A); a single band of 36 kDa with the same molecular mass as that of the recombinant protein was detected by SDS-PAGE. This result showed that the endogenous NTH was not only identical to the cloned enzyme but also may not have extensive post-translational modification. Finally, indirect immunofluorescence studies indicated that hNTH1 was distributed in both the nucleus and the cytoplasm (Fig. 9B, b). However, in some cells, the enzyme was concentrated in the nucleus Fig. 9B, c). DISCUSSION We have independently cloned the cDNA of human NTH and expressed the recombinant protein in E. coli in soluble form. Our clone is nearly identical to the hNTH1 cDNA reported by Hilbert et al. (7) but significantly different in sequence from the second hNTH1 cDNA reported by Aspinwall et al. (8). Although some preliminary properties of the enzyme were described in those earlier papers, we have provided a more comprehensive characterization of the enzyme in this report. The full-length enzyme with 304 amino acid residues was expressed as a GST fusion polypeptide whose cleavage with thrombin resulted in the production of the wild type protein with two additional NH 2 -terminal amino acid residues and a truncated protein with a deletion of 22 residues at the amino terminus (Fig. 1B). Kinetic analysis for Tg-DNA glycosylase/AP lyase activity showed that the two polypeptides (36 and 34 kDa) have identical enzymatic activity ( Fig. 2A) and thus indicated that the NH 2 -terminal amino acid residues of hNTH1 are not essential for its enzymatic activity. Indeed, the homology between hNTH1 and E. coli Nth starts from amino acid residue 95 of the former and residue 3 of the latter and continues to near the carboxyl termini of the proteins. The extended NH 2 -terminal segment of hNTH1 has putative nuclear and mitochondrial targeting signals and may also play a role in interactions with other proteins. The endogenous protein in HeLa cells has the same size as a recombinant protein in SDS-PAGE by Western blotting and is apparently localized in both the nucleus and cytoplasm as indicated by an immunofluorescence assay. However, in some cells, the enzyme is predominantly present in the nucleus. Furthermore, the cytoplasmic enzyme may actually be localized in the mitochondria.
Purified hNTH1 is yellowish brown in color and has an absorbance spectrum with a peak at 280 nm and a second peak at 410 nm due to the presence of a [4Fe-4S] cluster (Table I).
Four Cys residues of this domain are well conserved among eukaryotic Nth homologs (7,8) and are suggested to form a pocket with a domain of the helix-hairpin-helix motif important FIG. 8. Activity of mutant hNTH1. A, DNA trapping assay of the mutant proteins. Wild type (WT) or mutant proteins (2 pmol) were allowed to react with 32 P-labeled DHU-55 oligonucleotide duplex at 37°C for 10 min in the presence of 50 mM NaCNBH 3 . The reaction products were analyzed as described for Fig. 7. B, DNA glycosylase/AP lyase activities of mutant proteins. Various amounts of each enzyme were incubated with 250 ng of Tg-containing plasmid at 37°C for 30 min, and the resulting nicks were quantified. C, Tg-DNA glycosylase activity of mutant proteins. Various amounts of each enzyme were incubated with 300 ng of 3 H-labeled Tg-containing DNA at 37°C for 30 min, and the ethanol-soluble radioactivity was measured. D, electrophoretic mobility shift assays of mutant proteins. U-13 oligonucleotide and reduced AP (rAP) site-containing oligonucleotide derived from U-13 were used as probes.  9. Immunological analysis. A, Western blot detection of hNTH1 in HeLa cells. HeLa S3 cells were lysed in a buffer containing 10 mM Tris-HCl (pH 7.4), 150 mM NaCl, and 0.5% Nonidet P-40, and proteins were fractionated by SDS-PAGE (12.5% polyacrylamide). The hNTH1 band (lane 1) was visualized by affinity-purified anti-hNTH1 antibody. Lane 2, purified full-length hNTH1 (20 pg). B, intracellular localization of hNTH1 in HeLa S3 cells. Cells on coverslips were fixed and exposed to anti-hNTH1 antibody and then exposed to fluorescein isothiocyanate-labeled secondary antibody under a Zeiss UV microscope. a, preimmune serum; b and c, anti-hNTH antibody. Other details are given under "Experimental Procedures." for DNA binding (11). The absorbance ratio of 410/280 nm is useful as an index of the intactness of the [4Fe-4S] cluster (11). The value (0.23) for hNTH1 is smaller than that of E. coli Nth (0.38 -0.40), which may reflect the difference in the specific absorbance of these proteins at 280 nm (Table I). Titration of the recombinant protein with an oligonucleotide substrate indicated that only about 25% of the enzyme molecules in a typical preparation were active, probably due to misfolding of the eukaryotic protein during expression in E. coli.
The recombinant human homolog of E. coli Nth has been shown earlier to be nearly identical to Nth in enzyme activity (7,8). The enzyme releases Tg, urea, and other residues from the damaged DNA and has an AP lyase activity that cleaves DNA strand via ␤-elimination; however, no detailed kinetic studies were reported. In our studies, we carried out kinetic analyses using Tg-containing form I plasmid DNA and DHUcontaining duplex oligonucleotide substrates. We found that the enzyme had a pH optimum of about 8 and optimal activity in 75 mM NaCl. Unlike a typical DNA glycosylase/AP lyase, the enzyme was not inhibited with a high concentration of EDTA (10 mM) but was unexpectedly inhibited (ϳ50%) by 5-10 mM divalent cations. The AP lyase activity of hNTH1 involves almost exclusively ␤-elimination, as evidenced by the production of 3Ј ␣,␤-unsaturated aldehyde termini. A DHU residue is derived from cytosine by deamination and ring saturation, so its paired base is usually guanine. However, DHU will pair with adenine during DNA replication. That hNTH1 enzyme had a similar reaction rate with either dihydro-U:G or dihydro-U:A pair-containing oligonucleotide substrate indicates that its activity was not dependent on the opposite base. Thus, the mechanism for substrate recognition appears to be different from that of OGG1 and OGG2, the eukaryotic 8-oxoguanine DNA glycosylase/AP lyases, whose activities strictly depend on the base opposite 8-oxoguanine (31)(32)(33). 3 At the same time, both OGG and NTH require duplex DNA as the substrate. The apparent K m for a DHU-containing substrate, 47 nM, was similar to those of E. coli Nth (91 nM) and S. cerevisiae Scr2 (109 nM) for a 37-mer DHU-containing duplex oligonucleotide (16) and that of E. coli Nth (61 nM) for an AP site-containing duplex oligonucleotide (11). The enzyme had a k cat of ϳ0.6/min for the DHU-containing DNA; this value is surprisingly much lower than 1.1 s Ϫ1 for AP site-containing DNA observed for E. coli Nth (11). A similar difference was observed between mammalian N-methylpurine-DNA glycosylase and its E. coli functional homolog, AlkA (29). This observation suggests that the mammalian enzymes in the BER system, including DNA glycosylases, may interact in vivo with other cellular components that increase their turnover. Indeed, the DNA deoxyribophosphodiesterase activity of polymerase ␤ via ␤-elimination is enhanced by interaction with AP endonuclease (35), and polymerase ␤ also interacts with DNA ligase III via XRCC1 protein (36). Another line of evidence that Tg in H 2 O 2 -treated XP-G/CS mutants was removed at about half the initial rate of the normal cells implies an interaction between the DNA glycosylase and XP-G protein (22).
Specific interaction of DNA glycosylase with the substrate can be observed by DNase I footprinting analysis of the Michaelis complex, which was trapped because of the extremely low k cat of the enzyme (25). The specific binding of hNTH1 to the DHU-containing duplex oligonucleotide covered 10 nucleotides centered near DHU in the damaged strand and 15 nucleotides with G opposite DHU in the complementary strand. When E. coli Nth was bound to a reduced AP site-containing oligonu-cleotide, it protected 9 -11 nucleotides on each strand from cleavage by DNase (37).
Based on extensive sequence conservation among Nth proteins of various bacteria and eukaryotes, Lys-212 was predicted to be the active site nucleophile in hNTH1. However, while this requirement was suggested, no direct evidence was shown for the corresponding active site lysine of Nth (Lys-120 in E. coli) or of any other organism. In the case of E. coli Nth, identification of this residue was suggested from the x-ray crystallographic structure and site-directed mutagenesis studies (10,11). In this report, we have shown for the first time that Lys-212 is indeed the catalytic residue in hNTH1 that forms a covalent complex with DNA, by determining the sequence surrounding this residue in a peptide derived from the trapped complex. Although a similar result was obtained earlier with hOGG1 (38), it was important to demonstrate that the conserved Lys residue in endonuclease III, the prototype DNA glycosylase/AP lyase, is indeed the nucleophile (11). Site-specific substitution of Lys-212 in hNTH1 with Gln inactivated the enzyme's catalytic activity, but not the DNA-binding activity specific for damaged DNA, as was shown with an analogous mutant of E. coli Nth (11). However, the K212R mutant was active although with a much lower k cat and about 85-fold less catalytic specificity than the wild-type protein. A similar situation was observed with the K241R mutant of yeast OGG1 (39). It is not clear whether Arg can functionally substitute for Lys as a nucleophile in Schiff base formation or whether another Lys residue closed to the catalytic pocket can also serve as the donor in a slightly changed conformation. Importantly, the recombinant wild type and mutant hNTH1 proteins will be useful for structure-function studies of NTH and determining its precise role as a critical component of the mammalian BER system.