Isolation and Characterization of the cDNA Encoding Bovine Poly(ADP-ribose) Glycohydrolase*

The synthesis and rapid turnover of ADP-ribose poly- mers is an immediate cellular response to DNA damage. We report here the isolation and characterization of cDNA encoding poly(ADP-ribose) glycohydrolase (PARG), the enzyme responsible for polymer turnover. PARG was isolated from bovine thymus, yielding a pro- tein of approximately 59 kDa. Based on the sequence of oligopeptides derived from the enzyme, polymerase chain reaction products and partial cDNA clones were isolated and used to construct a putative full-length cDNA. The cDNA of approximately 4.1 kilobase pairs predicted expression of a protein of approximately 111 kDa, nearly twice the size of the isolated protein. A single transcript of approximately 4.3 kilobase pairs was detected in bovine kidney poly(A) (cid:49) RNA, consistent with expression of a protein of 111 kDa. Expression of the cDNA in Escherichia coli resulted in an enzymatically active protein of 111 kDa and an active fragment of 59 kDa. Analysis of restriction endonuclease fragments from bovine DNA by Southern hybridization indicated that PARG is encoded by a single copy gene. Taken together, the results indicate that previous reports of multiple PARGs can be explained by proteolysis of an 111-kDa enzyme. The deduced amino acid sequence of the bovine PARG shares little or no homology with other known proteins. However, it contains a putative bipar- tite nuclear location signal as would be predicted for a nuclear protein. The availability of cDNA clones for PARG should

The biological consequences of genomic damage include recovery of normal cell function, cellular survival leading to malignant transformation, or cell death by necrosis or apoptosis (1). Among the many variables that can affect the ultimate biological consequence of DNA damage to a particular cell are (i) the amount, type, and location of the DNA damage and (ii) the cellular response elicited by the damage. An immediate cellular response to DNA damage is the activation of poly(ADP-ribose) polymerase (PARP) 1 by DNA strand breaks (2). PARP catalyzes the conversion of NAD to multibranched polymers containing up to 200 ADP-ribose residues (3). Increases in polymer levels of more than 100-fold may occur within minutes (4). PARP is a major acceptor for ADP-ribose polymers in an automodification reaction (5)(6)(7), while histones and other DNA binding proteins also are modified to a lesser extent (8,9). Once synthesized, polymers are rapidly turned over (10,11), being converted to free ADP-ribose by the action of poly(ADP-ribose) glycohydrolase (PARG) (12,13). An ADP-ribosyl protein lyase has been proposed to catalyze removal of protein-proximal ADP-ribose monomers (14).
While the changes in chromatin structure that accompany the rapid synthesis and turnover of this large polyanion are still poorly understood, ADP-ribose polymer metabolism has been linked to the enhancement of DNA repair (15)(16)(17)(18), limitation of malignant transformation (19 -21), enhancement of necrotic cell death (22), and involvement in programmed cell death (23,24). To date, studies of the structure and function of the enzymes of ADP-ribose polymer metabolism have been mainly limited to PARP (25). PARG has been isolated to apparent homogeneity from bovine thymus (26,27), guinea pig liver (28), and human placenta (29), and basic enzymatic features have been established (26). However, the structure of the protein has not been elucidated. We report here the isolation and characterization of the cDNA encoding bovine PARG. The availability of cDNA clones for PARG should facilitate detailed studies of the enzyme and the involvement of ADP-ribose polymer metabolism in cellular responses to DNA damage.

EXPERIMENTAL PROCEDURES
Purification of PARG-PARG was purified from bovine thymus tissue (Pel-Freez) by modifications of previously published procedures (26,27). The enzyme was isolated up to the polyethylene glycol (PEG)-6,000 fractionation step as described previously (27). However, DNA-agarose and heparin-Sepharose chromatographic steps used previously were omitted, and the PEG-6,000 fraction was applied directly to an affinity matrix of poly(ADP-ribose)-dihydroxyboronyl-Sepharose (PADPR DHB-Sepharose). The active fractions eluted from PADPR DHB-Sepharose (25 ml) were pooled, placed in dialysis tubing, concentrated against dry PEG-20,000 to approximately 12 ml, and dialyzed against 2 liters of 20 mM potassium phosphate buffer, pH 8.0, 0.1% Triton X-100, 5 mM ␤-mercaptoethanol, 0.1 mM thioglycolic acid, 0.4 M KCl (buffer A). The sample was loaded onto a 1.0 ϫ 11-cm Toyopearl AF-Red (Supelco) column, and PARG was eluted with an 80-ml linear gradient of 0.4 -2 M KCl in buffer A. The active fractions, eluting at approximately 1.25 M KCl, were pooled, placed in dialysis tubing, concentrated against solid sucrose to approximately 9 ml, and dialyzed against 20 mM potassium phosphate buffer, pH 7.2, 0.75 M KCl, 0.1% Triton X-100, 10% glycerol, 5 mM ␤-mercaptoethanol, 0.1 mM thioglycolic acid. PARG activity was determined as described by Ménard and Poirier (30), and protein content was determined by the method of Bradford (31). The final preparation was quantified by SDS-PAGE (32) and Coomassie Blue staining to compare the intensity of the protein band with a known amount of bovine serum albumin (33).
The purification procedure for the bovine thymus PARG summarized in Table I is typical for results obtained from six separate preparations of the enzyme. Purification from 500 g of bovine thymus achieved approximately 50,000-fold purification and yielded approximately 20 g of purified protein. Analysis of the final preparation by SDS-PAGE revealed that more than 95% of the protein migrated at an apparent molecular mass of approximately 59 kDa (Fig. 1).
Peptide Sequencing-Prior to proteolytic fragmentation, the purified PARG (40 g in 100 l of 0.4 M ammonium bicarbonate buffer, pH 8.0, 8 M urea) was incubated in a final concentration of 2.2 mM dithiothreitol at 56°C for 15 min. Iodoacetamide was added to a final concentration of 2.0 mM, and the sample was incubated at 25°C for 15 min. After dilution with an equal volume of water, 1.5 units of immobilized L-1tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin (Pierce) was added, and the sample was incubated at 37°C for 18 h with gentle rotary shaking. Finally, the mixture was subjected to centrifugation at 16,000 ϫ g for 5 min to separate the tryptic fragments from the immobilized trypsin. The tryptic fragments were adjusted to 0.05% in trifluoroacetic acid and separated on a 4.6 mm ϫ 25 cm, Microsorb MV, C 4 reversed-phase HPLC column (Rainin) eluted with an 80-min linear gradient from 4 to 44% acetonitrile in 0.05% trifluoroacetic acid. Four oligopeptide fractions, with approximate elution times of 61, 63, 68, and 75 min, were selected for peptide sequence analysis by the Edman degradation method (34). Amino acid sequence data of four oligopeptides, designated by their approximate HPLC elution times from the reversed-phase column, are shown in Table II. cDNA Cloning-To obtain cDNA clones encoding bovine PARG, PCR amplification experiments were followed by the screening of two different bovine cDNA libraries. Fig. 2 depicts two PCR products and eight cDNA clones that were isolated to provide a putative full-length cDNA clone encoding bovine PARG. For each of the cDNA inserts characterized, the sequence of both strands was determined by the dideoxynucleotide chain termination method (35) using Sequenase from U.S. Biochemical Corp.
The first step leading to the isolation of cDNA clones was to synthesize two multidegenerate 17-mer primers, GAYCAYAAYGARTGYYT and CKRTANGTYTCNGCRTA (where Y represents T/C, R is A/G, K is T/G, and N is A/T/C/G), based on two regions of the oligopeptide 68 sequence (DHNECL and YATEYR, Table II). Using the multidegenerate primers and an oligo(dT)-primed bovine thymus cDNA gt11 library BL1019b (CLONTECH), PCR amplification generated a 74-bp DNA fragment with a deduced amino acid sequence identical to the corresponding region of oligopeptide 68. Next, two specific 24-mer oligonucleotide primers (ATCATCACAGGTACTGAGCAGTAC and GCCTGT-GTATTCACTGTACTGCTC) based on the sequence of this 74-bp DNA were used in combination with gt11 forward and reverse primers to amplify PCR products 1 and 2 from the bovine thymus library. PCR product 1 contained 231 bp of sequence including the region encoding the N-terminal region of oligopeptide 68 and the entire sequence of oligopeptide 61. PCR product 2 contained 757 bp, which included sequence encoding the C-terminal region of oligopeptide 68 and the entire sequence of oligopeptide 63. The sequence information obtained from PCR products 1 and 2 was used to isolate cDNA clones obtained by the screening of bovine thymus and bovine kidney cDNA libraries. A 518-bp EcoRI-HindIII fragment from PCR product 2 was used as a probe to screen approximately 1 ϫ 10 6 independent clones from the bovine thymus library. Two positive cDNA clones (clones 1 and 2) were isolated, which overlapped PCR products 1 and 2. However, attempts to obtain clones from the bovine thymus library that contained sequence 5Ј to clone 2 were unsuccessful. Thus, a 231-bp EcoRI-KpnI fragment from clone 2 was used as a probe to screen approximately 5 ϫ 10 5 independent clones of the bovine kidney 5Ј stretch plus cDNA gt11 library BL3001b (CLONTECH). Three positive cDNA clones (clones [3][4][5] were obtained, all of which contained sequence 5Ј to clone 2. Each of these clones also contained sequence encoding oligopeptide 75. Clones 1-5 provided multiple overlapping sequences in the 3Ј-terminal portion of a consensus cDNA, but additional clones were sought to obtain overlapping sequences for the 5Ј-terminal region. Thus, a 436-bp EcoRI-KpnI fragment located at the 5Ј end of clone 3 was used as a probe to screen approximately 6 ϫ 10 5 independent clones of the bovine kidney library. Clones 6 -8 provided overlapping sequences for the 5Ј-terminal region. The full-length cDNA was constructed by ligating a 3.9-kb XbaI-NsiI fragment from pWL11 (clone 1 cDNA insert in pTZ18R (36)) and a 3.0-kb NsiI-XbaI fragment from pWL13 (clone 4 cDNA insert in pTZ18R). The resulting plasmid, termed pWL30, contained the 4,070-bp full-length cDNA.
Expression of PARG in Escherichia coli-PARG was expressed using two different bacterial expression systems, the pTrcHis Xpress System (Invitrogen), in which the expressed protein contains a leader polyhistidine sequence, and the glutathione S-transferase (GST) gene fusion system (Pharmacia Biotech Inc.). For expression in the pTrcHis Xpress system, three different DNA fragments were amplified and inserted into the pTrcHis expression plasmid. Construct A, containing the cDNA sequence Ϫ3 to 2,946, was prepared by subcloning a 2.9-kb XhoI-EcoRI DNA fragment amplified from pWL30 with primers WIN34 (GCT-GCGGGTCTCGAGCATGAGTGCGGGC) and WIN15 (GCGTCTAGAA-  For expression of PARG as a GST fusion protein, an insert containing the cDNA sequence from position 1138 to 2946 was prepared by subcloning a 1.8-kb EcoRI-EcoRI fragment amplified from pWL30 with the oligonucleotide CCAATTTGAAGGAGGAATTCCCGCCGCCACCATG-AATGATGTGAATGCCAAACGACCTGGA and WIN15 as primers. The resulting DNA fragment was inserted into the EcoRI site of the pGEX-2T expression vector, and the plasmid was used to transform E. coli NM522 cells. For expression experiments, bacterial cultures were grown at 37°C in 1% Bacto-tryptone, 0.5% yeast extract, and 0.5% NaCl to approximately 0.6 A 600 /ml and were induced with 1 mM isopropyl-␤-D-thiogalactoside (IPTG). Cells were collected by centrifugation, and crude extracts were prepared by sonication (10 A 600 /ml) in 10 mM sodium phosphate buffer, pH 7.2, 150 mM NaCl, 0.5 mg/ml lysozyme, 0.1 mg/ml phenylmethylsulfonyl fluoride, 1 mM EDTA, 0.7 g/ml pepstatin A, 0.5 g/ml leupeptin, and 1 g/ml aprotinin. Cell extracts were subjected to centrifugation, and the supernatant fraction was used for assay. PARG assay conditions were as described previously (30). Following incubations, portions of reaction mixture were analyzed by thin layer chromatography (30) or subjected to anion exchange HPLC. Anion exchange HPLC utilized a Whatman Partisil SAX column equilibrated with 7 mM potassium phosphate buffer, pH 4.0, at a flow rate of 1 ml/min. The sample was diluted in the same buffer, applied to the column, and eluted with a 30-min linear gradient from 7 mM potassium phosphate buffer, pH 4.0, to 250 mM potassium phosphate buffer, 0.5 M KCl, pH 4.0. Activity gel assays for PARG were done by casting polyacrylamide gels with automodified PARP containing [ 32 P]ADP-ribose polymers as described previously (37). Following electrophoresis, PARG was renatured by incubating the gels at 25°C in 5 volumes of 50 mM sodium phosphate buffer, pH 7.5, 50 mM NaCl, 10% glycerol, 1% Triton X-100, 10 mM ␤-mercaptoethanol, changing the buffer every 3 h for a total of five changes. After an additional incubation at 37°C for 3 h, gels were dried, and PARG activity was detected following autoradiography as a clear band on a black background. Cell extracts containing PARG fused to GST were examined for binding to glutathione-Sepharose 4B (GSH-Sepharose) (Pharmacia Biotech Inc.) according to the specifications of the manufacturer.
Northern and Southern Blot Analysis-Total cytoplasmic RNA and poly(A) ϩ RNA were isolated from bovine kidney MDBK cells (ATCC CCL22) using TRIzol reagent (Life Technologies, Inc.) following the manufacturer's recommendations. RNA was fractionated on denaturing agarose gels (38), transferred to nylon membranes, and hybridized with clone 4 (Fig. 2) radiolabeled by a random hexamer priming method (39). Total genomic DNA was prepared from bovine thymus tissue as described previously (38), and DNA (10 g) was digested with EcoRI, BglII, XbaI or PstI, fractionated on a 1% agarose gel, transferred to a nylon membrane (Hybond Nϩ, Amersham), and hybridized using an 828-bp HindIII fragment of clone 1 radiolabeled as described for clone 4 above (39). Prehybridizations and hybridizations were carried out at 42°C in 50% formamide, 0.25 M sodium phosphate buffer, pH 7.2, 0.25 M NaCl, 7% SDS, 1 mM EDTA.

Isolation of cDNA Clones Encoding Bovine PARG-
The combined nucleotide sequence of clones 1-8 (Fig. 2) predicted a full-length cDNA clone of 4,070 bp containing 257 bp of 5Јnoncoding sequence, a single open reading frame of 2,931 bp, and a 3Ј-noncoding region of 882 bp. Fig. 3 shows the complete nucleotide sequence (GenBank TM accession number U78975) and the deduced amino acid sequence, which predicts a protein of 977 amino acids and a molecular mass of 110.8 kDa.
Northern Blot Analysis of Bovine Kidney RNA-An unexpected feature of the consensus full-length cDNA clone was that it predicted the expression of a protein of approximately 111 kDa (Figs. 2 and 3), while the enzymatically active PARG from thymus had a molecular mass of approximately 59 kDa (Fig. 1). To determine the size of the RNA transcript for PARG, total RNA and poly(A) ϩ RNA were isolated from MDBK cells and annealed using clone 4 as the hybridization probe. A single transcript of approximately 4.3 kb was detected in the poly(A) ϩ RNA (Fig. 4B, lane 2). Thus, the transcript size was consistent with the expression of a 111-kDa PARG protein.
Expression of PARG in E. coli-To determine whether the isolated cDNA encoded PARG, three different constructs containing specific regions of the cDNA were inserted into the pTrcHisB expression vector. Constructs A and B contained the entire open reading frame of 110.8 kDa, which together with the fusion partner predicted a protein of approximately 115 kDa. Construct B also contained the 3Ј-untranslated region of the clone. Since the isolated PARG of approximately 59 kDa contained enzymatic activity, construct C contained only the 75-kDa carboxyl-terminal region of the PARG, which predicted a fusion protein of approximately 79 kDa. Following transformation of cells with the three expression constructs, cell extracts were examined for PARG activity using three different assay methods. Using a thin layer chromatography assay that measures release of [ 32 P]ADP-ribose from [ 32 P]ADP-ribose polymers (30), PARG activity was detected in extracts from cells transformed by each of the constructs. Fig. 5A shows results obtained with constructs B and C. No activity was detected in cells transformed with the empty vector, but activity was detectable without induction by IPTG, indicating a leaky lac promoter. The addition of IPTG resulted in a timedependent increase of up to approximately 4.5-fold in enzymatic activity. Fig. 5A also shows that the enzymatic activity was strongly inhibited by the presence of ADP-hydroxymethylpyrrolidine diol, a specific inhibitor of PARG (40, 41). The The cDNA Encoding Bovine Poly(ADP-ribose) Glycohydrolase material released from ADP-ribose polymers was shown to be exclusively ADP-ribose by strong anion exchange HPLC (Fig.  5B), demonstrating that the cell extracts did not contain any other ADP-ribose polymer-degrading enzymes such as phosphodiesterase, which catalyzes the formation of AMP and phosphoribosyl-AMP (42).
To determine the size of the expressed enzymatic activity, an activity gel assay (37) was used. With construct A, activity was observed at approximately 115 and 59 kDa (Fig. 5C). No bands were produced from extracts from the IPTG-induced pTrcHisB vector that did not contain an insert. Extracts from cells transformed with construct B showed bands at approximately 115 and 59 kDa, and extracts from cells transformed with construct C showed bands at approximately 79 and 59 kDa (data not shown). During storage at 4°C, cell extracts lost activity migrating at the higher molecular weight, while the activity at approximately 59 kDa increased (data not shown). Expression of PARG in the pTrcHisB expression vector did not result in detectable amounts of protein by staining with Coomassie Blue. Thus, another construction was designed to overexpress a 69-kDa carboxyl-terminal region of the PARG as a fusion with GST, which allows convenient protein purification by affinity chromatography on a GSH-Sepharose column. Two hours after induction with IPTG, strong expression of a protein migrating at approximately 90 kDa was observed (Fig. 5D). This protein bound to GSH-Sepharose and was eluted by GSH. The construct contained a thrombin cleavage site between the GST and the 69-kDa region of PARG, and treatment of the material bound to GSH-Sepharose with thrombin resulted in the release of a protein that migrated at approximately 59 kDa. This result suggests that the protein purified from the bovine thymus may be larger than suggested by its migration on SDS-PAGE. Attempts to obtain high level expression of the full-length protein were unsuccessful. Southern Blot Analysis of PARG Genomic Complexity-Pre-  , lanes 1A and 1B) and poly(A) ϩ RNA (4 g, lanes 2A and 2B) were separated on a denaturing agarose gel. Panel A shows the ethidium bromide-stained gel, and panel B shows the autoradiogram of a Northern blot analysis using a random primed, 32 P-labeled DNA probe constructed from clone 4 (Fig. 2).
vious studies have reported that PARG isolated from nuclear fractions had a molecular mass of approximately 75 kDa (28,29), while PARG isolated from whole cell homogenates or postnuclear supernatant fractions had a molecular mass of approximately 59 kDa (26,27,43). These results suggest either that two or more genes may code for PARG or that proteolysis generates lower molecular weight forms from higher molecular weight forms. The cDNA isolated encoded a protein considerably larger than any PARG proteins previously described, consistent with the possibility that the different forms of PARG are derived from a single form by proteolytic cleavage. To test the hypothesis that PARG is encoded by a single copy gene, the genomic complexity of the PARG gene was analyzed by a Southern hybridization experiment. Genomic DNA was digested with four different restriction enzymes, which did not contain restriction sites within the carboxyl-terminal region of the PARG cDNA. Following electrophoresis, the restriction digests were subjected to hybridization with a probe that corresponded to the carboxyl-terminal region of the PARG cDNA. The analysis displayed in Fig. 6 shows that, in each restriction rolidine diol (40,41). Panel B, analysis of material released from ADPribose polymers by anion exchange HPLC. Extracts from a strain containing construct B were incubated with [ 32 P]ADP-ribose polymers (30), and a portion was analyzed by anion exchange HPLC as described under "Experimental Procedures." The elution times for AMP, ADPR, and PR-AMP are indicated by arrows. Panel C, analysis of PARG activity following SDS-PAGE. Crude extracts (1 l each) of cells transformed by pTrcHis without an insert (lane A) or containing construct A (lane B) were analyzed by an activity gel assay in which the gel was incubated following electrophoresis in 50 mM sodium phosphate buffer, pH 7.5, 50 mM NaCl, 10% glycerol, 1% Triton X-100, 10 mM ␤-mercaptoethanol to allow renaturation of PARG activity (37). The positions of molecular weight marker proteins are shown. Panel D, expression of a carboxyl-terminal region of PARG cDNA as a fusion protein with GST. A construct corresponding to a 69-kDa carboxyl-terminal region of the PARG cDNA was prepared and used to transform E. coli NM522 cells as described under "Experimental Procedures." Cell extracts were subjected to SDS-PAGE, and proteins were stained by Coomassie Blue. to electrophoresis on a 1% agarose gel, and blotted to nylon filters. The blot was annealed with a 32 P-labeled DNA probe corresponding to the carboxyl-terminal region of the PARG protein as described under "Experimental Procedures." digest, the probe hybridized primarily with a single restriction fragment. The fainter signals probably reflect the presence of introns in the PARG gene. This result indicates that PARG is encoded by a single copy gene in the bovine genome. DISCUSSION The rapid synthesis of ADP-ribose polymers that occurs in response to DNA strand breaks is accompanied by very rapid polymer turnover (4,10,11), indicating that PARP and PARG activities are closely coordinated as cells respond to DNA damage. While PARP has been widely studied, information concerning structure and function relationships of PARG is much more limited.
The isolation of partial PARG cDNA clones (Fig. 2) has allowed the construction of a cDNA clone that codes for the sequence of all four oligopeptide sequences present in the isolated protein (Table II, Fig. 3) and contains PARG activity when expressed in E. coli (Fig. 5). The cDNA clone (Fig. 3) has features typical of cDNAs that code for mammalian proteins. These include (i) an oligo A (putative poly(A) ϩ ) sequence at the 3Ј end, (ii) a polyadenylation signal (AATAAA) (44) 12 bp upstream from the oligo A sequence, (iii) a sequence of ATTTA in the 3Ј-untranslated region thought to play a role in selective mRNA degradation in mammalian cells (45), (iv) a single open reading frame, and (v) a nucleotide sequence around the first start codon commonly found at known sites of initiation of translation (46). The likelihood that the cDNA clone constructed represents a full-length or nearly full-length clone for PARG is supported by the observation that hybridization of poly(A) ϩ RNA from bovine kidney cells with the cDNA showed a single band of hybridization of approximately the same size as the cDNA (Fig. 4).
The nucleotide sequence encoding bovine PARG indicates that PARG shares little or no homology with other known sequences. A search of sequence data banks has failed to reveal significant homology with any sequences coding for known proteins. A strong sequence homology has been observed with human and rat cDNA clones that likely represent partial clones for PARG from these species. Examination of protein sequence data bases also has shown that the deduced amino acid sequence of PARG lacks any sequence homology with known proteins. However, the amino acid sequence shares a significant homology with a protein sequence from Caenorhabditis elegans that may represent the PARG protein from this organism (47). The deduced amino acid sequence of PARG has been examined for a number of structural motifs that can be predicted from the primary amino acid sequence. We have noted that the expressed PARG protein can form dimers stable to SDS-PAGE conditions. In that regard, residues 871-907 show significant homologies to known leucine zipper dimerization sequences (48). Another motif identified is a putative bipartite nuclear location signal (NLS) (49). It is interesting that PARP also contains a bipartite NLS (50). Fig. 7 compares deduced amino acid sequences in the NLS region of the bovine PARG and homologous regions of putative PARG sequences from human, mouse, and C. elegans with the NLS region of PARP from seven different organisms. The putative NLS of PARG fulfills the criteria for bipartite NLS in that it contains conserved acidic and basic amino acid residues at two different locations each within the region of homology to the NLS of PARP (50).
An intriguing finding was that the PARG cDNA clone codes for a protein of approximately 111 kDa, which is nearly twice the size of the PARG protein isolated from bovine thymus (Fig.  1). It seems likely that PARG contains a protease sensitive site that, following proteolysis, yields a protein fragment of approximately 59 kDa that still retains enzymatic activity. Several pieces of evidence favor this possibility. (i) Expression of the carboxyl-terminal portion of the cDNA resulted in enzymatic activity (Fig. 5). (ii) All of the oligopeptides sequenced were located in the carboxyl-terminal half of the protein (Figs. 2 and 3). (iii) The only protein, other than the 59-kDa protein detected in the thymus preparation was approximately 111 kDa ( Fig. 1). (iv) The PARG activity expressed in bacteria was sensitive to proteolysis, yielding a protein of approximately 59 kDa (Fig. 5). (v) The cleavage site in PARG is in the region of the putative NLS, and the PARP NLS is located in a proteasesensitive site (24). Taken together with the data suggesting that bovine PARG appears to be coded for by a single copy gene (Fig. 6), proteolysis seems likely to explain the presence of PARG activity with a molecular mass of approximately 74 and 59 kDa in bovine thymus preparations (51). Likewise, a similar mechanism could explain previous reports of a PARG of 74 kDa isolated from nuclear fractions of guinea pig liver and human placenta (28,29) and a PARG of 59 kDa isolated from postnuclear fractions of guinea pig liver (43).
While proteolysis of a larger protein to yield smaller proteins retaining PARG activity seems likely to explain the size heterogeneity of PARG previously reported, it remains to be determined if proteolysis normally occurs in vivo or whether it occurs during purification of the enzyme. While our results show that a full-length protein can be expressed containing PARG activity (Fig. 5C), the molecular size of PARG in vivo also remains to be determined. If PARG occurs as a larger protein, an interesting possibility is that the amino-terminal region may be involved in the regulation of enzymatic activity.
The availability of cDNA clones coding for PARG should facilitate studies of structural and functional aspects of this enzyme and the metabolic pathway in which it is involved. The isolation and characterization of cDNAs encoding PARG from other sources will allow information concerning changes in the structure of the protein during evolution. Studies in progress of cDNA clones from human and mouse indicate that PARG is highly conserved in mammals. 2 The availability of PARP cDNA has allowed a number of molecular genetic approaches to study the function(s) of ADP-ribose polymer metabolism, and the availability of PARG cDNA should allow the design of additional molecular genetic approaches for studying this metabolism. Recently, a mouse strain containing a genetic inactivation of the gene encoding PARP has been reported (53). Although not extensively characterized, mice homozygous for the dis- rupted PARP gene are viable, indicating that the absence of PARP (and ADP-ribose polymers) has no major consequences for development. Disruption of the gene encoding PARG in mice containing a normal PARP gene will allow the determination of whether other cellular enzymes can replace PARG in the turnover of ADP-ribose polymers and/or whether development will occur in the absence of PARG. Alternatively, disruption of the PARG gene in mice containing a disrupted PARP gene may provide insights for the coordinated function of PARP and PARG.
Understanding the involvement of ADP-ribose polymer metabolism in determining the frequency of alternative cellular outcomes of DNA damage may also have a therapeutic benefit. For example, the efficacy of many antitumor agents appears to be closely related to the ability to increase the frequency of apoptosis (1,23). The ability of inhibitors of PARP to chemosensitize specifically cycling cells to the cytotoxic effects of DNA-damaging agents has demonstrated the potential of targeting ADP-ribose polymer metabolism for cancer therapy (16). In that regard, PARG is a potentially attractive target due to its low abundance, its structurally unique substrate, and the high turnover of ADP-ribose polymers that occurs following DNA damage. Modulation of PARG activity by chemical inhibitors or by molecular genetic approaches should allow assessment of the value of PARG as a new target for therapeutic development.