Molecular Characterization of a Human DNA Kinase*

Human polydeoxyribonucleotide kinase is an enzyme that has the capacity to phosphorylate DNA at 5′-hydroxyl termini and dephosphorylate 3′-phosphate termini and, therefore, can be considered a putative DNA repair enzyme. The enzyme was purified from HeLa cells. Amino acid sequence was obtained for several tryptic fragments by mass spectrometry. The sequences were matched through the dbEST data base with an incomplete human cDNA clone, which was used as a probe to retrieve the 5′-end of the cDNA sequence from a separate cDNA library. The complete cDNA, which codes for a 521-amino acid protein (57.1 kDa), was expressed in Escherichia coli, and the recombinant protein was shown to possess the kinase and phosphatase activities. Comparison with other sequenced proteins identified a P-loop motif, indicative of an ATP-binding domain, and a second motif associated with several different phosphatases. There is reasonable sequence similarity to putative open reading frames in the genomes ofCaenorhabditis elegans and Schizosaccharomyces pombe, but similarity to bacteriophage T4 polynucleotide kinase is limited to the kinase and phosphatase domains noted above. Northern hybridization revealed a major transcript of approximately 2.3 kilobases and a minor transcript of approximately 7 kilobases. Pancreas, heart, and kidney appear to have higher levels of mRNA than brain, lung, or liver. Confocal microscopy of human A549 cells indicated that the kinase resides predominantly in the nucleus. The gene encoding the enzyme was mapped to chromosome band 19q13.4.

Transient DNA strand breaks and short gaps are frequently observed in cellular DNA. Many arise during regular cellular activity such as DNA replication, recombination, or differenti-ation. Others occur as a consequence of exposure to endogenous or exogenous DNA damaging agents. Repair of these strand interruptions is usually mediated by DNA ligases and polymerases. Both of these classes of enzymes require 3Ј-hydroxyl DNA termini, and the DNA ligases also require 5Ј-phosphate termini. However, the termini generated by nucleases, such as DNase II, and many produced by ionizing radiation bear 3Јphosphate and 5Ј-hydroxyl groups (1)(2)(3)(4), and therefore must be processed before they can be acted upon by DNA ligases or polymerases.
One enzyme that possesses the capacity to both phosphorylate 5Ј-hydroxyl termini and dephosphorylate 3Ј-phosphate termini is polynucleotide kinase (PNK). 1 The PNK from T4 phage has found widespread application in molecular biology, especially for radiolabeling DNA and oligonucleotides (5). It can act on DNA and RNA and even phosphorylate nucleoside 3Ј-monophosphates. However, the main cellular function of the T4 enzyme is not to repair DNA, but rather to counter the action of a phage endoribonuclease that cleaves tRNA (6). Eukaryotic PNKs fall into two categories depending on whether their preferred substrate is DNA or RNA (7). While both can phosphorylate 5Ј-termini, only the former have an associated 3Ј-phosphatase activity (8 -12).
Mammalian DNA kinases have been purified from a variety of sources including rat liver and testes and calf thymus (8 -19). The isolated enzymes share similar properties with regard to the kinase activity including an acidic pH (5.5-6.0) optimum (8 -18), and the minimum size of oligonucleotide that can be phosphorylated is in the range of 8 -12 nucleotides (11,15). The only significant discrepancy has been the molecular mass assigned to the polypeptides. Earlier reports regarding the PNK purified from rat organs indicated that the protein may be an 80-kDa homodimer composed of 40-kDa polypeptides (9,10,18), but PNK activity in tissue extracts detected on activity gels migrated as a 60-kDa polypeptide (20). Estimates for the size of calf thymus PNK have ranged from 56 to 70 kDa (16,17). We and others have recently purified the DNA kinases from calf thymus and rat liver to near homogeneity, making use of a broad spectrum of proteolysis inhibitors (11,12). The major protein band migrated as a 60-kDa peptide on polyacrylamide gels, but a minor band was observed at 40 kDa in the rat liver preparation.
At present, the cellular function of mammalian DNA kinases has not been elucidated. Clearly, one possibility is participation in the repair of strand breaks induced by DNA damaging agents, such as ionizing radiation or topoisomerase inhibitors (21,22). We have shown that, unlike T4 phage PNK, calf thymus PNK is able to efficiently phosphorylate the 5Ј-OH terminus at a nick and a one-nucleotide gap in a doublestranded DNA substrate (11). Furthermore, an in vitro system consisting of purified mammalian PNK, DNA polymerase ␤, and DNA ligase I was able to effect the complete repair of nicks and short gaps bounded by 3Ј-phosphate and 5Ј-OH termini (21). Alternatively, PNK could participate in a more regular function. For example, it has been observed that a proportion of Okazaki fragments have 5Ј-OH termini (23), which would have to be phosphorylated prior to ligation. As part of our ongoing study to address the question of the role of eukaryotic PNKs, this paper describes the molecular cloning, sequencing, cellular localization, and chromosomal mapping of human PNK.

EXPERIMENTAL PROCEDURES
Phosphorylation Assay-The DNA substrate containing 5Ј-OH termini was prepared by digestion of calf thymus DNA with micrococcal nuclease as described by Richardson (24). Each 5Ј-phosphorylation reaction mixture (20 l), containing 10 g of DNA substrate, 3 Ci of [␥-32 P]ATP (3000 Ci/mmol, Amersham Pharmacia Biotech), 500 nM unlabeled ATP, 80 mM succinic acid, pH 5.5, 10 mM MgCl 2 , 1 mM dithiothreitol, 1 mM EGTA, 2 g of bovine serum albumin, and protein fraction (typically 4 l), was incubated for 20 min at 37°C. The reaction was stopped and the DNA precipitated by addition of 200 l of 20% trichloroacetic acid and 100 l of 250 M sodium pyrophosphate containing 50 g of bovine serum albumin. Following centrifugation at 10,000 ϫ g for 10 min, the pellets were resuspended in 80 l of 0.1 M NaOH and reprecipitated by addition of 400 l of 10% trichloroacetic acid. This wash step was repeated once more before the radioactivity of the pellet was determined. As a control for kinase specificity (i.e. DNA versus protein), parallel reactions were carried out in the absence of the DNA substrate.
3Ј-Phosphatase Assay-The 3Ј-dephosphorylation of a 21-mer oligonucleotide (p21p) catalyzed by recombinant human PNK in Escherichia coli cell extracts was assayed by gel electrophoresis as described previously (21).
Partial Purification of Polydeoxyribonucleotide Kinase from HeLa Cells-A pellet of frozen HeLa S3 cells (3 ϫ 10 10 , approximately 50 ml packed cell volumes) was thawed in 200 ml of hypotonic buffer (10 mM Tris-HCl, pH 7.5, 2 mM MgCl 2 , 5 mM dithiothreitol, and 0.5 mM EDTA) containing a mixture of protease inhibitors (25 g/ml N ⑀ -p-tosyl-L-lysine chloromethyl ketone, 5 g/ml chymotrypsin, 1 g/ml aprotinin, 0.5 g/ml leupeptin, 0.5 g/ml pepstatin, and 1 mM ␣-toluenesulfonyl fluoride) and held for 20 min at 0°C before disruption in a Dounce glass homogenizer (15 strokes). Nuclei were collected by low speed centrifugation, and a protein extract was prepared in the presence of 0.3 M KCl as described previously (25). Sequential chromatography of the extract on a phosphocellulose P11 column (Whatman, Clifton, NJ) and an Ultrogel AcA34 gel filtration column (Sepracor/IBF, Marlborough, MA), and ammonium sulfate precipitation steps were carried out as described by Robins and Lindahl (26), except that the elution buffer for the first column contained 0.6 M KCl and the elution buffer for the second column contained 0.5 M NaCl. The active fractions in the second peak from the gel filtration column were pooled (63 mg of protein in a total volume of 54 ml), dialyzed against buffer A (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM dithiothreitol, 1 mM potassium phosphate, and 10% glycerol). The specific activity at this stage of purification was approximately 0.06 units/mg of protein, where one unit of enzyme is the amount required to incorporate 1 nmol of phosphate from ATP into micrococcal nuclease-treated DNA in 30 min at 37°C (27). The pooled material was loaded onto a column (2.5 ϫ 5.0 cm) of Bio-Gel HT hydroxyapatite (Bio-Rad) pre-equilibrated with buffer A. The column was washed with five volumes of buffer A before eluting bound protein with a 200-ml linear gradient of 50 -500 mM potassium phosphate in buffer A collecting in 5-ml fractions. The active fractions, 28 -33, were pooled and dialyzed against buffer B (10 mM potassium phosphate, pH 6.8, 4 mM 2-mercaptoethanol, and 10% glycerol) containing 50 mM KCl. The material was loaded onto a 1-ml HiTrap SP column (Amersham Pharmacia Biotech), washed with 10 column volumes of buffer B and eluted with a 30-ml linear gradient of 50 -600 mM KCl in 30 1-ml fractions. A peak of kinase activity eluted at fractions 10 -12. The contents of fraction 11 were dialyzed against buffer C (50 mM Tris-HCl,

HeLa cells 8 [LI][LI]YPE[LI]PR
HeLa cells a Leucine and isoleucine [LI] cannot be distinguished by low energy collision-activated dissociation as they are isomers. pH 7.5, 1 mM dithiothreitol, 1 mM potassium phosphate, and 10% glycerol) containing 50 mM NaCl, and loaded onto a Mono S PC 1.6/5 column attached to a SMART micropurification chromatography system (Amersham Pharmacia Biotech). Protein was eluted with a 2-ml linear salt gradient of 50 -450 mM NaCl at a flow rate of 100 l/min in 20 100-l fractions. After assaying the fractions for DNA kinase activity, a small quantity of each fraction was examined by SDS-PAGE to determine which polypeptide correlated with activity. The remaining contents of the fraction with the peak of kinase activity (fraction 12) was further fractionated by gel electrophoresis and electroblotted onto polyvinylidene difluoride membrane.
Amino Acid Sequencing-The electroblotted HeLa protein was stained with sulforhodamine B (0.05% w/v in 30% v/v aqueous metha-nol, 0.1% v/v acetic acid) using a rapid-staining protocol (28). The dried, stained protein was then digested in situ on the polyvinylidene difluoride membrane with trypsin (Roche Molecular Biochemicals, modified) for 18 h at 30°C and the peptides extracted with 1:1 v/v formic acid/ ethanol (29). Aliquots were sampled and directly analyzed by matrixassisted laser desorption ionization (MALDI) time-of-flight mass spectrometry using a LaserMat 2000 mass spectrometer (Thermo Bioanalysis, UK) (30). Additional aliquots were quantitatively esterified using 1% v/v thionyl chloride in methanol and also analyzed by MALDI to provide acidic residue composition (31). Native and esterified peptide masses were then screened against the MOWSE peptide mass fingerprint data base (32). The remaining digested peptides (Ͼ90% of total digest) were then reacted with N-succinimidyl-2-morpholine ace- Amino acid residues 170 -176 are predicted to be associated with the phosphatase activity; residues 301-304 may be a nuclear localization signal; residues 372-380 constitute a P-loop ATP-binding domain; residues 402-464 may represent a DNA binding domain. Nucleotide sequences indicating the initiation and stop codons, the polyadenylation signal, and two 5Ј-UTR sequences with homology to complementary sequences in the 5Ј-UTR of DNase II are underlined.
tate (SMA) in order to enhance b-ion abundance and facilitate sequence analysis by tandem mass spectrometry (33). Dried peptide fractions were treated with 7 l of freshly prepared, ice-cold 1% w/v N-succinimidyl-2-morpholine acetate in 1.0 M HEPES (pH 7.8 with NaOH) containing 2% v/v acetonitrile. Following reaction for 20 min on ice, the reaction was terminated by the addition of 1 l of heptafluorobutyric acid and diluted with an equal volume of water. The solution was then injected in three 5-l aliquots onto a capillary reverse-phase column (300 m x 15 cm) packed with POROS R2/H material (Perseptive Biosystems, MA) equilibrated with 2% v/v methanol, 0.05% v/v trifluoroacetic acid running at 3 l/min. The adsorbed peptides were washed isocratically with 15% v/v methanol, 0.05% v/v trifluoroacetic acid for 30 min at 3 l/min to elute the excess reagent and HEPES buffer. Derivatized peptides were eluted with a single step gradient to 75% v/v methanol, 0.1% v/v formic acid and collected in two 3-l fractions. The derivatized peptides were then sequenced by low energy collision-activated dissociation using a Finnigan MAT TSQ7000 tandem triple quadrupole mass spectrometer and a Finnigan MAT LCQ ion-trap mass spectrometer, both instruments fitted with nanoelectrospray sources (34,35). Collision-activated dissociation was typically performed with collisional offset voltages between Ϫ18 and Ϫ30 V.
Two tryptic peptides from previously purified calf thymus PNK (11) were sequenced by the Harvard Microchemistry Facility (Cambridge, MA) using either an ABI 477A protein sequencer (Applied Biosystems, Foster City, CA) or an HP G1000A (Hewlett Packard, Palo Alto, CA). Confirmation of sequence was obtained by MALDI time-of-flight mass spectrometry on a LaserMat 2000 mass spectrometer.
Isolation and Sequencing of Polynucleotide Kinase cDNA-DNA sequences derived from the peptide sequences were used to screen the dbEST data base (NIH). A cDNA clone from infant brain (clone number 32798 inserted in lafmid BA) was identified and obtained from the I.M.A.G.E. Consortium. The cDNA insert (1548 bp) was fully sequenced, using an automated ABI Prism 377 DNA analysis system (Applied Biosystems), and confirmed the presence of the poly(A) tail, and a large open reading frame, but no clearly identifiable start codon. A 609-bp probe, prepared by digestion of clone 32798 with HindIII and PstI (New England Biolabs, Beverley, MA), was subsequently used to screen a gt11 HeLa cell 5Ј-STRETCH PLUS cDNA library (CLON-TECH, Palo Alto, CA) by a standard protocol (36). Ten positive clones were isolated, none of which contained a poly(A) tail. The largest insert (1.5 kilobase pairs) was amplified by PCR using the forward and reverse primers with Pfu DNA polymerase (Stratagene, La Jolla, CA), and then sequenced. Putative full-length cDNA was reconstituted as follows: (i) the PCR-amplified product was digested with SacI and shrimp alkaline phosphatase (Amersham Pharmacia Biotech), and the larger fragment (1.1 kilobase pairs) isolated by agarose gel electrophoresis, (ii) the DNA of clone 32798 was digested with SacI, (iii) the DNA molecules were ligated using phage T4 DNA ligase (Amersham Pharmacia Biotech), and (iv) the ligation product was digested with EcoRI.
Expression of PNK cDNA in E. coli-The cDNA was amplified by PCR using Pfu DNA polymerase and primers with tails that provided cleavage sites for NdeI (5Ј-TTTGAATTCCCATATGGGCGAGGTGGAG-CCCCCGGGC-3Ј) and BamHI (5Ј-CGCGGATCCTCAGCCCTCGGAGA-ACTGGCAG-3Ј) and then subcloned into the expression plasmid pET-16b (Novagen Inc., Madison, WI). The new plasmid (pPNK-His), which codes for a His-tagged derivative of PNK, was transfected into host E. coli bacterial strain BLR(DE3) (Novagen). The bacteria were grown at 37°C to an OD 600 of 0.6 in 100 ml of LB medium containing 50 g/ml ampicillin and 12.5 g/ml tetracycline. Zinc chloride was then added to the medium to a final concentration of 0.015 mM, and PNK expression was induced at 30°C for 3 h by addition of 0.4 mM (final concentration) isopropyl-1-thio-␤-D-galactopyranoside (Sigma). After harvesting the cells by centrifugation at 5000 ϫ g at 4°C for 5 min, they were resuspended in 10 ml of extraction buffer (50 mM Tris-HCl, pH 7.5, 0.015 mM ZnCl 2 , 6 mM mercaptoethanol). Lysozyme was added to a final concentration of 100 g/ml together with Triton X-100 (final concentration, 0.1%), and, after incubation at 30°C for 15 min, the bacteria were disrupted by sonication. The soluble fraction was separated from the insoluble fraction by centrifugation at 12,000 ϫ g for 15 min at 4°C. The insoluble fraction was resuspended in 1 ml of extraction buffer.
Northern (RNA) Hybridization Analysis-The 609-bp HindIII/PstI fragment used to screen the HeLa cDNA library was also used to probe a human multiple tissue Northern blot (CLONTECH) containing 2 g (per lane) of polyadenylated RNA isolated from eight different human tissues. Hybridization was performed at 68°C for 1 h under conditions described by the manufacturer. As a control for the amounts of mRNA in each lane, the membrane was reprobed with a sequence of ␤-actin cDNA provided by CLONTECH.
Antibodies and Confocal Microscopy-A synthetic peptide antigen was prepared commercially (SSPEQ, Quebec) from the first 17 amino acids of peptide sequence 1 (Table I)  b, DNA kinase activity in the soluble fraction recovered from E. coli transfected with pET-16b and pPNK-His. The DNA phosphorylation assay was carried out with (ϩ) and without (Ϫ) DNA as described under "Experimental Procedures" using 10 g of cell extract. c, 3Ј-phosphatase activity in the insoluble fraction recovered from E. coli transfected with pET-16b and pPNK-His. The extracts (10 g of total protein) were tested for their capacity to remove the 3Ј-phosphate from a 5Ј-32 Plabeled 21-mer oligonucleotide (4 pmol) in 20 min as described previously (21). grown as a monolayer on glass microscope slides to 80% confluence. Following rinsing in PBS, the cells were fixed in 95% ethanol at Ϫ20°C for 15 min. The slides were allowed to dry, and were incubated for 1 h at room temperature with 1% skim milk powder in PBS to minimize nonspecific binding of the immunoreagents. Following extensive PBS rinsing, the slides were incubated overnight at 4°C in the rabbit polyclonal antiserum (diluted 1/30 in PBS), in a humidified atmosphere. The cells were then rinsed extensively with PBS, and rhodamineconjugated goat anti-rabbit IgG (HϩL, Cappel Laboratories, Durham, NC) was applied at a dilution of 1/30 in PBS for a 1-h incubation at 37°C in a water-saturated atmosphere. The unbound fluorescent antibody was removed by extensive washing in PBS, and the slides were covered with coverslips for confocal microscopy using PBS/glycerol, 1:1 as a mounting medium. The instrumentation and the procedures for the confocal laser scanning microscopy have been described previously (38).
Fluorescence in Situ Chromosomal Hybridization-Fluorescence in situ hybridization was performed as described previously (39). Human metaphase cells were prepared from phytohemagglutinin-stimulated peripheral blood lymphocytes. Biotin-labeled probes were prepared by nick translation using Bio-16-dUTP (Enzo Diagnostics, Farmingdale, NY). One PNK probe was clone 32798 (including the plasmid vector). A second probe, which provided a 440-bp sequence stretching from the 5Ј-untranslated region into the 5Ј-end of the translated sequence, was generated by PCR amplification of the HeLa cDNA clone using the gt11 forward primer and a reverse primer, 5Ј-GTGGAGGCCATTGAC-CAAATA-3Ј. The two clones were labeled and co-hybridized to the chromosome preparations. Hybridization was detected with fluorescein-conjugated avidin (Vector Laboratories, Burlingame, CA), and chromosomes were identified by staining with 4,6-diamidino-2-phenylindole-dihydrochloride.

Partial Purification and Peptide Sequencing of Human
PNK-Fractions of a crude extract of HeLa cells that was passed down an AcA 34 Ultragel size exclusion column in the presence of 0.5 M NaCl were shown to contain DNA kinase activity (Fig. 1). Two peaks of activity were apparent, the first migrating with the bulk of the higher molecular weight protein, which may suggest that PNK is bound in a complex to other proteins, and the second eluting with proteins in the range of 40 -100 kDa. Initial steps in the purification were carried out by conventional chromatography using gel filtration, hydroxyapatite, and cation exchange media. For the final step, the protein was applied on a SMART system precision column and eluted in 20 100-l fractions with a 2-ml salt gradient (50 -450 mM NaCl). The kinase assay revealed a peak of activity centering on fraction 12 ( Fig. 2A). Correlation of the intensities of the protein bands in fractions 10 -14 (Fig. 2B) with kinase activity FIG. 5. Confocal microscopy of PNK in human A549 cells. a, cells incubated with preimmune rabbit serum. There is faint cytoplasmic background labeling, but a lack of staining in the nucleus. b, cells incubated with rabbit polyclonal antibodies raised against a PNK-derived peptide (see "Experimental Procedures"). Pronounced nuclear staining is evident.
FIG. 6. PNK mRNA levels in human tissues. The polyadenylated RNA isolated from several tissues (CLONTECH) was probed with a 32 P-labeled 609-bp (HindIII-PstI) fragment from clone 32798 (containing nucleotides 205-809 of the PNK sequence). A cDNA probe to ␤-actin was used as a control for mRNA content.

FIG. 7.
In situ hybridization of biotin-labeled PNK probes to human metaphase cells from phytohemagglutinin-stimulated peripheral blood lymphocytes. The chromosome 19 homologues are identified with arrows; specific labeling was observed at 19q13.4. The inset shows partial karyotypes of two chromosome 19 homologues illustrating specific labeling at 19q13.4 (arrowheads). Images were obtained using a Zeiss Axiophot microscope coupled to a cooled charge-coupled device camera. Separate images of 4,6-diamidino-2-phenylindole-dihydrochloride-stained chromosomes and the hybridization signal were merged using image analysis software (NU200 and Image 1.57). strongly suggested that the ϳ60-kDa band (topmost of the three major bands in fraction 12, marked by an arrow) was responsible for the PNK activity. Accordingly, this band was chosen for amino acid sequencing.
Initial screening of the peptide mass fingerprint against the MOWSE protein sequence data base (approximately 210,000 entries) revealed no significant matches. Six tryptic peptides were then sequenced de novo by low energy collision-activated dissociation using both triple quadrupole and ion-trap mass spectrometry (peptides 3-8, Table I). The use of the SMA reagent permitted full sequences to be obtained for each peptide from a single collision spectrum. This was particularly important when sequencing by collision-activated dissociation using the ion-trap, where the SMA reagent typically yielded complete y-ion coverage. In addition, two peptide sequences (peptides 1 and 2, Table I) from our previously purified preparations of calf thymus PNK (11) were obtained by conventional amino acid sequencing.
Isolation and Sequencing of Human PNK cDNA-Screening of the dbEST data base with peptides 1-6 revealed several human and murine cDNA clones with DNA sequences coding directly for the peptide sequences or with minor variations. Of these clones, clone 32798 (from the I.M.A.G.E. Consortium) contained the largest insert (1548 bp) of human DNA. Sequencing of clone 32798 (bases 208 -1636, Fig. 3) showed the presence of a poly(A) tail, a consensus -AATAAA-polyadenylation signal (bases 1589 -1594) and a large open reading frame, with a TGA stop codon at bases 1564 -1566, containing in-frame sequences coding for peptides 2-5 and 8. The first valine in peptide 2 of the calf thymus enzyme is replaced by an isoleucine in human PNK. The insert was not long enough to account for the size of the protein determined by SDS-PAGE analysis, so a probe was prepared from the 5Ј-end of the insert of clone 32798 and used to screen a gt11 HeLa cell cDNA library. Although 10 positives were obtained from this library, none contained a poly(A) tail. The largest insert (bases Ϫ90 to 1317; Fig. 3), however, extended the open reading frame to a putative start codon (bases 1-3) and included in-frame sequences coding for peptides 1, 6, and 7. Again there is a minor difference between the amino acid sequence between the bovine and human peptide 1. The 1100-bp sequences in common between clone 32798 and the HeLa clone are absolutely identical. The complete open reading frame codes for a 521-amino acid protein with a predicted molecular mass of 57,102 Da.
Bacterial Expression of Human PNK-After splicing together the full sequence shown in Fig. 3, the translated region was subcloned into an expression vector (pET-16b), to generate the plasmid pPNK-His, and expressed in a strain of E. coli as a His-tagged protein. After sonication of the bacteria to release protein, a strong band at ϳ60 kDa was observed in both the soluble and insoluble fractions (Fig. 4a). The 60-kDa band is the major protein in the insoluble fraction. Both the soluble and insoluble fractions showed clearly detectable levels of DNA kinase activity. (Data for the soluble fraction are shown in Fig.  4b. This is relatively straightforward because E. coli itself does not possess a DNA kinase activity that would interfere with the assay.) 3Ј-Phosphatase activity was measured by monitoring the removal of a 3Ј-phosphate group from a synthetic oligonucleotide (21). Fig. 4c shows the conversion of 5Ј-labeled p21p to p21 by protein from the bacteria harboring pPNK-His. In this case we made use of the protein in the inclusion bodies because nonspecific E. coli phosphatases were present in the soluble fraction.
Expression of PNK in Human Tissues-Northern blot analysis of RNA isolated from a number of human tissues (Fig. 5) indicated a major transcript of approximately 2.3 kilobases, although in some tissues a second less abundant but considerably larger (7.5 kilobases) transcript was observed. There were also notable differences in the levels of mRNA expression in the tissues examined, in particular, pancreas and, to a lesser extent, heart appeared to have elevated levels of the message.
Cellular Localization of PNK-Rabbit polyclonal antibodies were raised against a synthetic peptide composed of the first 17 amino acids in peptide 1 (Table I). These antibodies were used to visualize PNK in human A549 lung carcinoma cells by fluorescence confocal microscopy. The results, shown in Fig. 6, indicate that the protein accumulates in the cell nuclei.
Chromosomal Location of Human PNK-To localize the PNK gene, we performed fluorescence in situ hybridization of biotinlabeled PNK cDNA probes to normal human metaphase chromosomes. Co-hybridization of two probes, EST clone 32798 and a 440-bp sequence stretching from the 5Ј-untranslated region into the 5Ј-end of the translated sequence, resulted in specific labeling only of chromosome 19 (Fig. 7). Specific labeling of 19q13.3-13.4 was observed on four (2 cells), three (9 cells), two (9 cells), or one (5 cells the two cDNA probes hybridized individually. These results suggest that the PNK gene is localized to chromosome 19, band q13.4.

DISCUSSION
Gel electrophoresis of the partially purified DNA kinase activity from HeLa cells indicated that the human enzyme is a ϳ60-kDa polypeptide, and thus has a molecular mass similar to that of the proteins isolated from rat liver and calf thymus (11,12,20). The molecular mass of the 521-amino acid polypeptide encoded by the sequenced cDNA is 57,102 Da. The same computer program (ExPASy, Swiss Institute of Bioinformatics) also predicted a pI for the protein of 8.7, which is very close to the pI measured for the rat liver and calf thymus proteins of 8.6 and 8.5, respectively (10,11). Expression of the protein in E. coli, and the demonstration of its kinase and phosphatase activities (Fig. 4), confirmed that we had indeed cloned the cDNA for human PNK.
The amino acid sequence (Fig. 3) indicates that this is a novel protein. However, as shown in Fig. 8, predicted proteins of Caenorhabditis elegans and Schizosaccharomyces pombe have ϳ30% similarity to human PNK. There are several large blocks of high similarity. Two in particular are probably associated with the two known activities of PNK. Other proteins possess these consensus sequences. The first, residues 372-380, conforms to the sequence pattern (A/G)-X 4 -G-K-(S/T) of the P-loop consensus sequence (40), which is an ATP/GTP binding domain found in many kinases, including nucleoside kinases such as adenylate and uridylate kinases (Table II). The second, residues 170 -176, is a sequence found in several phosphatases (Table II), including carbohydrate-phosphate phosphatases like phosphoglycolate phosphatase and glycerol-3-phosphatase. This might be anticipated for a DNA phosphatase i.e. that dephosphorylates a deoxyribose phosphate. The homologous sequence in T4 PNK is part of a motif predicted by Jilani et al. (12) to be associated with the phosphatase activity of the T4 enzyme. Their conclusion was partly based on the observation that the first Asp in the homologous sequence of human phosphomannomutase 1 (Table II) has recently been shown to form an acyl-phosphate when incubated with its substrate (41). It is reasonable to suggest that the homologous Asp in human PNK may be a phosphate acceptor in the course of the enzyme's phosphatase activity. Both the P-loop and the phosphatase consensus sequences are found in phage T4 polynucleotide kinase and a putative PNK/RNA ligase from the nuclear polyhedrosis virus. There is, however, little other homology between the human kinase and the phage T4 PNK.
Two other potential functional regions were identified. The short motif, RKKK, residues 301-304, could be a nuclear localization signal belonging to a class of such signals consisting of four or more Arg or Lys residues or any combination of the two amino acids. That PNK has such a motif is consistent with the results of the confocal microscopy ( Fig. 6), indicating the nuclear localization of the protein. The large domain between residues 402 and 464 (Fig. 3) is suggested by the program MacPattern to be a DNA binding domain. It is a relatively cysteine-rich sequence, and these amino acids may be involved in structuring such a domain.
Within the 5Ј-untranslated region (5Ј-UTR) of PNK are two closely spaced sequences (nucleotides Ϫ89 to Ϫ79 and Ϫ77 to Ϫ69) that have a high level of homology to sequences in the complementary strand of the 5Ј-UTR of DNase II (42) (Table  II). This is of interest because DNase II, which has recently been implicated in apoptosis and cell differentiation (43)(44)(45)(46), generates DNA strand breaks with 3Ј-phosphate and 5Ј-OH termini. If these nicks are repaired, it would presumably require the action of a polynucleotide kinase and, therefore, the regulation of the two enzymes may be coordinated.
The Northern blot (Fig. 5) suggested high expression of the protein in human pancreas. However, it must be borne in mind that this tissue is isolated from cadavers up to 3 h post-mortem. It is possible that the high level of PNK expression in pancreas reflects the high level of DNA degradation (presumably by nucleases such as DNase II) seen in post-mortem pancreatic tissue (47).
We have mapped the location of the gene for PNK to chromosome 19q13.4. Among the other genes that have been mapped to this locus are DNA polymerase ␦, the apoptosis regulator gene BAX, protein kinase C, and several zinc finger proteins. The DNA repair enzymes, DNA ligase I and ERCC1, are located at 19q13.3 and DNase II at 19q13.2. Note Added in Proof-The accompanying article by Jilani et al. (48) describes the independent identification and characterization of the same human DNA kinase, which they term PNKP.