Inhibition of homodimerization of poly(ADP-ribose) polymerase by its C-terminal cleavage products produced during apoptosis.

The biochemical role of the C-terminal fragment of poly(ADP-ribose) polymerase (PARP) was investigated in HeLa cells undergoing UV-mediated apoptosis. During the course of apoptosis, the C-terminal cleavage product of PARP interacted with intact PARP and down-regulated PARP activity by blocking the homodimerization of PARP. The basic leucine zipper motif in the auto-modification domain of the C-terminal fragment of PARP represented the site of association, and Leu(405) was critical to the ability of the basic leucine zipper motif to associate with intact PARP. The expression of the C-terminal fragment of PARP stimulated UV-mediated apoptosis. These results suggest that the C-terminal cleavage product of PARP produced during apoptosis blocks the homodimerization of PARP and inhibits the cellular PARP activity. The inhibition of the cellular PARP activity might prevent cellular NAD(+) depletion and stimulate apoptosis by maintaining the basal cellular energy level required for the completion of apoptosis.

The biochemical role of the C-terminal fragment of poly(ADP-ribose) polymerase (PARP) was investigated in HeLa cells undergoing UV-mediated apoptosis. During the course of apoptosis, the C-terminal cleavage product of PARP interacted with intact PARP and down-regulated PARP activity by blocking the homodimerization of PARP. The basic leucine zipper motif in the auto-modification domain of the C-terminal fragment of PARP represented the site of association, and Leu 405 was critical to the ability of the basic leucine zipper motif to associate with intact PARP. The expression of the C-terminal fragment of PARP stimulated UVmediated apoptosis. These results suggest that the Cterminal cleavage product of PARP produced during apoptosis blocks the homodimerization of PARP and inhibits the cellular PARP activity. The inhibition of the cellular PARP activity might prevent cellular NAD ؉ depletion and stimulate apoptosis by maintaining the basal cellular energy level required for the completion of apoptosis.
Activation of caspases has been generally accepted as a common cellular event that occurs in cells undergoing apoptosis (1). Numerous cellular proteins are cleaved by caspases during the course of apoptosis. However, the biochemical consequence of proteolytic cleavage of the substrate proteins by caspases during the course of apoptosis has been poorly understood. PARP 1 has been known to be cleaved by caspases during apoptosis (2). The cleavage of PARP yields a 25-kDa N-terminal fragment containing two zinc-fingers and an 89-kDa C-terminal fragment containing the auto-modification domain and the NAD ϩ binding domain (3). The two zinc-finger motifs in the N-terminal DNA-binding domain bind to DNA at the sites of single-or double-stranded breaks, resulting in the catalytic activation of PARP. The DNA-bound, activated PARP utilizes NAD ϩ to synthesize poly(ADP-ribose) on various nuclear proteins, such as DNA polymerase ␣ (4) and ␤ (5), topoisomerase I (6) and II (7), histones (8), p53 (9), DNA-dependent protein kinase (10), and PARP itself (11).
Many studies have described PARP as a positive regulator of apoptosis. The overexpression of PARP in the transfected cells was known to stimulate apoptosis (12). The pro-apoptotic role of PARP activation was further supported by the finding that a specific chemical inhibitor of PARP, 3-aminobenzamide, suppresses apoptosis (13). However, the controversial roles of PARP during apoptosis have been argued in the studies using PARP-deficient mice. Wang et al. (14) proposed a dispensable role of PARP in apoptosis by showing a normal level of apoptosis in PARP-deficient mouse cells treated with various apoptosis inducers. An indispensible role of PARP on apoptosis was presented by Simbulan-Rosenthal et al. (15), who reported an early burst of poly(ADP-ribosyl)ation of nuclear proteins during Fas-mediated apoptosis in PARP ϩ/ϩ cells, whereas no induction of apoptosis was shown in PARP Ϫ/Ϫ cells by Fas stimulation.
The importance of PARP cleavage in apoptosis has recently been recognized. For example, it has been suggested that the activity of PARP is stimulated by DNA breaks during the early course of apoptosis, but proteolytic cleavage decreases PARP activity in the late course of apoptosis (16). Oliver et al. (17) observed a delayed apoptosis in PARP Ϫ/Ϫ cells expressing an uncleavable mutant of PARP. They suggested that PARP cleavage might be a sign that cells should undergo apoptosis because cells were unable to repair the cellular injury triggered by the apoptosis inducers. More recently, the importance of PARP cleavage during apoptosis was emphasized by the finding that the extensive poly(ADP-ribosyl)ation of p53 early during apoptosis decreases, as activated caspase-3 cleaves PARP and the expression of p53-responsive pro-apoptotic genes, bax and Fas, is elevated during the late course of apoptosis (18). However, the biological relevance of PARP cleavage during apoptosis and the cellular function of the cleavage products have not been yet clarified.
Recent studies have proposed that an adequate level of intracellular ATP is required for the completion of apoptosis (19,20). Because the intracellular ATP level is directly affected by the catalytic activity of PARP, apoptosis, an energy-requiring process, may well be influenced by PARP activity. The Nterminal fragment of PARP containing the DNA-binding domain preferentially binds to DNA breaks and prevents the activation of PARP, whose activity is stimulated by DNA binding (21). The pro-apoptotic role of the N-terminal fragment of PARP on apoptosis has been strongly supported by a recent finding that the N-terminal fragment of PARP irreversibly binds to DNA ends produced during apoptosis (22). However, the role of the C-terminal fragment of PARP on apoptosis has not been delineated. In the present study, we propose a putative role for the C-terminal cleavage product of PARP on apoptosis based on the following findings: (a) the C-terminal fragment of PARP interacts with intact PARP through an association between the auto-modification domains and (b) the C-terminal fragment suppresses its catalytic activity by block-ing homodimerization of PARP. We have found that the disruption of homodimerization of PARP by the cleaved C-terminal fragment stimulates apoptosis in UV-treated HeLa cells.
Vectors-The cDNA encoding full-length human PARP and truncated PARP fragments were cloned in-frame to pEGFP-C1 (CLON-TECH) or pEBG vector to produce GFP or GST fusion products, respectively. For the active transport of GFP-N214 and GFP-M in the nuclei of HeLa cells, the PKKKRK sequence of the nuclear localization signal (NLS) in SV40 large T-antigen (23) was inserted between GFP and the N termini of PARP fragments by polymerase chain reaction using oligomers of 5Ј-GGAATTCAAGCTTCGCCAAAGAAAAAAGCGAAAGT-CGACGCGC-3Ј and 5Ј-GCGCGTCGACTTTCGCTTTTTTCTTTGGCG-AAGCTTGAATTCC-3Ј. N214-NLS and M-NLS were then prepared by polymerase chain reaction to construct pEBG-N214 and pEBG-M, respectively. The vector constructs encoding PARP fragments were transfected into cells by employing LipofectAMINE reagent (Life Technologies, Inc.).
In Vitro Protein Interaction Assay-PARP derivatives were cloned in pGEX-5X-3, and GST-fused PARP fragments were produced in Escherichia coli BL21(DE3) cells. Purified GST fusion products (1 g) were transferred to 200 l of an assay buffer consisting of 20 mM Tris (pH 7.4), 100 mM NaCl, 2 mM EDTA, 0.1% Nonidet P-40, 2 mM dithiothreitol, 0.05% bovine serum albumin, and 5% glycerol (25). The 35 S-labeled PARP was prepared by in vitro transcription/translation procedures using TNT linked kit (Promega). In vitro translated 35 S-labeled PARP was treated with 1 unit of DNase I for 30 min prior to the incubation with GST fusion products at 4°C for 1 h. The GST fusion products were recovered in glutathione-Sepharose 4B beads (Amersham Pharmacia Biotech) and washed three times with phosphate-buffered saline (pH 7.4). The 35 S-labeled PARP bound to GST fusion products was analyzed by 10% SDS-PAGE and subsequent autoradiography. Immunoprecipitation and Immunoblot Analysis-HeLa cells (10 7 ) were transfected with 2 g of pEBG, pEBG-N214, pEBG-C215, pEBG-M, pEBG-PARP⌬M or pEBG-M⌬bZip and incubated for 36 h. Cells were lysed in a lysis buffer containing 100 mM Tris (pH 7.4), 150 mM NaCl and 1% Nonidet P-40 for 30 min on ice. After centrifugation of cell lysate at 10,000 ϫ g for 5 min, the resulting supernatant was incubated with glutathione-Sepharose 4B resin for 1 h at 4°C. The resin was washed three times with phosphate-buffered saline and the bound proteins were analyzed by Western blotting using monoclonal antibody against the C terminus of PARP following by 10% SDS-PAGE.
Evaluation of Apoptosis-HeLa cells (10 5  apoptosis, cells were fixed with 4% para-formaldehyde for 30 min, and apoptotic cells expressing GFP with condensed nuclei were counted after staining the cells with Hoechst33342 (1 g/ml) for 10 min.

RESULTS
Self-association of PARP at the bZip Motif in the Auto-modification Domain-PARP is a catalytic dimer, and the selfassociation of PARP is known to stimulate its catalytic activity (27,28). PARP domains participating in the self-association were examined as shown in Fig. 1. GST-fused PARP fragments were incubated with [ 35 S]PARP prepared by in vitro transcription/translation procedures, and [ 35 S]PARP specifically bound to GST-fused PARP fragments was resolved by 10% SDS-PAGE after GST pull-down. To exclude the possibility of intermolecular linking of PARP mediated by the DNA duplex that contacts the DNA-binding domains of separate PARP molecules, DNA employed in in vitro transcription/translation procedures was digested with DNase I before GST pull-down. The interaction between intact PARP and the N-terminal fragments that contain the DNA-binding domain disappeared after DNase I treatment, suggesting that the association was mediated by DNA. In contrast, the C-terminal fragments of PARP maintained the association with [ 35 S]PARP after DNase I treatment (Fig. 1B). Data shown in Fig. 1C indicate that the auto-modification domain of PARP is the site of intermolecular association. A bZip motif (amino acids 394 -422) known to participate in protein-protein interaction (29, 30) has been identified in the auto-modification domain of PARP. Thus, we examined whether the bZip motif in the auto-modification domain provides a structural framework for the intermolecular association. Data in Fig. 1C strongly suggest that the bZip motif in the auto-modification domain of PARP represents the site of the association.
The intermolecular association of the cellular PARP was studied in Fig. 2. GST-fused PARP fragments were expressed in HeLa cells, and the association of GST fusion products with the cellular PARP was monitored by immunoblot analysis using anti-PARP monoclonal antibody ( Fig. 2A). Data demonstrate that the bZip motif in the auto-modification domain interacts with the cellular PARP. A putative role of the Cterminal fragment of PARP on the self-association was investigated in cells co-expressing HA-tagged PARP (HA-PARP) with GST-fused PARP fragments. The association of the endogenous cellular PARP with HA-PARP was almost completely inhibited by the PARP fragments containing the auto-modification domain (C215 and M). However, a PARP fragment that lacks the auto-modification domain (PARP⌬M) or an auto-modification domain that lacks the bZip (M⌬bZip) failed to block the self-association of PARP in the transfected cells (Fig. 2B).
The bZip motif in the auto-modification domain of PARP contains invariable leucine residues (31). Thus, the function of the conserved leucine residues was evaluated in an experiment in which the leucine residues in the bZip motif were substituted, and the association of intact PARP with the mutated auto-modification domain was examined (Fig. 3). The leucine residues in the bZip motif were important for the association, and Leu 405 was of critical importance. The auto-modification domain in which Leu 405 was substituted to Arg 405 failed to interact with 35 S-labeled PARP prepared by in vitro transcription/translation. However, the association between the automodification domain and intact PARP was not disturbed by the substitution of leucine residues to valine, implicating that hydrophobic interaction is involved in the association.

Suppression of PARP Activity by Expressing the C-terminal Fragment of PARP in HeLa Cells Irradiated with UV Light-
The role of the C-terminal fragment of PARP on the cellular PARP activity was examined in co-transfected HeLa cells expressing HA-PARP and GST-fused PARP fragments. Cells were irradiated with 50 J/m 2 UV and incubated for 30 min in a (10 7 ) were transfected with 5 g of pEBG, pEBG-N214, pEBG-C215, pEBG-M, pEBG-PARP⌬M, or pEBG-M⌬bZip. Transfected cells were incubated for 48 h and GST-fused PARP fragments were isolated using glutathione-Sepharose 4B resin, and PARP bound to the resin was detected by immunoblotting using anti-PARP monoclonal antibody (upper panel). The expression of GSTfused PARP fragments was analyzed by 12% SDS-PAGE, followed by immunoblotting using anti-GST monoclonal antibody (lower panel). B, a tag encoding HA epitope was introduced to PARP (pSR␣-HA-PARP), and HA-tagged PARP was co-expressed with GST-fused PARP fragments. HA-PARP was immunoprecipitated using anti-HA monoclonal antibody, and the endogenous cellular PARP associated with HA-PARP was monitored by immunoblotting using anti-PARP monoclonal antibody following by 8% SDS-PAGE. culture medium containing [ 32 P]NA ϩ . HA-PARP was then isolated by immunoprecipitation, and the cellular PARP was assessed by the transfer of 32 P-labeled ADP-ribose moieties from NAD ϩ to HA-PARP. The expression of the C-terminal fragments that contain the bZip motif (C-215 and M) suppressed the cellular PARP activity. The auto-modification domain that lacks the bZip motif (M⌬bZip) failed to interfere with the cellular PARP activity. Data also showed that the cellular PARP activity was inhibited by the expression of PARP fragments containing N-terminal fragment (N-214 and PARP⌬M), a known trans-dominant inhibitor of PARP (32,33). These results implicate that the C-terminal fragment of PARP as well as N-terminal fragment is the trans-dominant inhibitor of the resident PARP in HeLa cells (Fig. 4).

FIG. 2. Effects of PARP fragments on the self-association of PARP in HeLa cells. A, HeLa cells
Stimulation of Apoptosis by PARP Cleavage Products-Apoptotic responses followed by UV irradiation (100 J/m 2 ) were evaluated by measuring the activation of zVAD-directed caspase and annexin-V binding to phosphatidyl serine at the surface of apoptotic HeLa cells. PARP cleavage products did not interfere with the caspase activation in cells undergoing apoptosis (Fig. 5A). The expression of the N-terminal or Cterminal fragment containing the bZip motif, induced elevated level of UV-mediated apoptosis, as assessed by the appearance of annexin-V binding at the cell surface (Fig. 5B). In addition, cells expressing the N-terminal or C-terminal fragment containing the bZip motif showed morphological changes characteristic to apoptosis, such as condensed nuclei (Fig. 5C).

DISCUSSION
Although PARP cleavage has been used as a prominent biochemical hallmark of apoptosis, the physiological relevance and functional consequence of PARP cleavage during apoptosis have not been clarified. A recent report by Oliver et al. (17) showed that Fas-induced apoptosis was delayed in cells expressing an uncleavable PARP mutant. Furthermore, Herceg and Wang (34) reported that the failure of PARP cleavage induced necrosis by the depletion of cellular energy in cells treated with various apoptotic inducers. These results suggested that the inactivation of PARP by the caspase-mediated proteolytic cleavage is a necessary requirement for the completion of apoptosis. In the present study, we attempted to interpret the biological meaning of the proteolytic cleavage of PARP during the course of apoptosis. We found that apoptosis requires the self-regulatory function of PARP, whose catalytic activity is regulated by its proteolytic cleavage products. The N-terminal fragment of PARP containing the DNA-binding domain preferentially binds to DNA breaks and thereby prevents the activation of PARP (32,33). It has been demonstrated that the N-terminal fragment of PARP irreversibly binds to DNA ends produced during apoptosis (22). The binding of the N-terminal fragment of PARP to DNA breaks has generally been known to contribute to apoptosis by blocking the access of repair enzymes to the DNA breaks generated during the course of apoptosis.
The roles of the C-terminal fragment of PARP on apoptosis have not yet been described. The present study proposes that the intermolecular association of the cellular PARP with the C-terminal fragment containing the auto-modification domain contributes to apoptosis by blocking the homodimerization of PARP in cells irradiated with UV. The C-terminal fragments of PARP containing the auto-modification domain interacted with intact PARP, whereas the N-terminal fragment containing the DNA-binding domain interacted with DNA. The association between the N-terminal fragment and intact PARP disappeared after DNase I treatment, suggesting that the association was mediated by DNA (Fig. 1). We further demonstrated that the bZip motif in the auto-modification domain is the site of self-association of the cellular PARP ( Fig. 2A). The homodimerization of PARP was blocked by the expression of PARP cleavage products containing the bZip motif in HeLa cells (Fig.  2B). The hydrophobic amino acid Leu 405 in the bZip motif was of critical importance in the structural framework required for the intermolecular association and an efficient dimerization of PARP (Fig. 3). The association of PARP with the basic components of the auto-modification domain in the C-terminal fragment seem to compete with the homodimerization of PARP.
There are number of studies suggesting that oligomerization serves as an important biochemical mechanism for the regulation of protein function. Dimerization was known to provide mechanisms for the modulation of catalytic activity of enzymes (35,36) and receptor functions (37,38). Several studies have also provided evidence for the inhibition of protein dimerization by the peptides derived from the same protein. For example, the catalytic activity of E. coli ribonucleotide reductase was shown to be inhibited by the C-terminal peptide, which inhibits the homodimerization of the enzyme (35). The truncated G protein-coupled receptors (GPCRs), which inhibit the formation of dimeric arrays of GPCRs, were also identified as negative regulators of GPCR function (37). It was known that the catalytic function of PARP is maximal when PARP tends to be self-associated to the dimer form (27,28). Data in Fig. 4 implicate that the expression of the C-terminal fragment of PARP in UV-treated HeLa cells suppresses auto-poly(ADP-ribosyl)ation of PARP, probably by blocking the homodimerization of the cellular PARP. The failure of the homodimerization may interfere with the binding of PARP to DNA breaks, resulting in the catalytic inactivation of PARP. However, the expression of a mutated PARP fragment that contains the C-terminal fragment but lacks the auto-modification domain (PARP⌬M) also suppressed the cellular PARP activity. The trans-dominant inhibition of PARP activity by the N-terminal fragment might be the case with PARP⌬M. Indeed, the cellular PARP activity was inhibited by the expression of the N-terminal fragment of PARP. This result is likely to reflect the fact that the Nterminal fragment of PARP competes with intact PARP for binding at DNA breaks, which activate PARP (Fig. 4).
The pro-apoptotic role of the C-terminal fragment of PARP on apoptosis was examined in Fig. 5. Both annexin-V binding and caspase activation are well defined markers of apoptosis (39). Our results suggest that the pro-apoptotic role of the C-terminal fragment of PARP does not correlate with the enhanced caspase activation in cells undergoing UV-mediated apoptosis. The level of caspase activity in cells expressing the C-terminal fragment that lacks the bZip motif was virtually same as in cells expressing PARP cleavage products containing the bZip motif. However, the intracellular caspase activity was blocked by the expression of a known inhibitor of caspase-3, CrmA (40). However, the expression of the C-terminal fragment of PARP stimulated UV-mediated apoptosis as assessed by the appearance of annexin-V positive cells (Fig. 5B). In addition, the expression of PARP fragments containing the bZip motif (C215 and M) stimulated the induction of morphological changes characteristic to apoptosis in UV-treated HeLa cells (Fig. 5C). These results suggest that the intermolecular association of PARP at the bZip motif is required for the completion of apoptosis. The homodimerization of PARP might facilitate the progression of apoptotic cell death beyond the caspase activation step.