Inhibition of caspase-3-mediated poly(ADP-ribose) polymerase (PARP) apoptotic cleavage by human PARP autoantibodies and effect on cells undergoing apoptosis.

Autoantibodies directed to nuclear antigens are serological hallmarks of autoimmune rheumatic diseases such as systemic lupus erythematosus. Although much more is known about the molecular identity and functions of targeted self-antigens, with few exceptions, evidence that autoantibodies to these targets have a particular function and contribute directly to the pathological process is lacking. Here we show that human autoantibodies reacting with the zinc fingers of poly(ADP-ribose) polymerase involved in the recognition of damaged DNA totally prevent the cleavage of poly(ADP-ribose) polymerase by caspase-3, a process that normally occurs during early apoptosis. Furthermore, these antibodies, which are frequent in certain autoimmune rheumatic and bowel diseases, affect the characteristic features of apoptosis and increase cell survival ex vivo. This new observation is important, because failure to remove autoimmune or abnormal cells can give rise to prolonged autoimmune stimulation and tumor formation.

Poly(ADP-ribose) polymerase (PARP 1 ; EC 2.4.2.30) is a 116-kDa nuclear enzyme that detects and binds DNA strand breaks produced by various genotoxic agents. DNA breaks activate PARP, which in turn catalyzes the synthesis of poly(ADPribose) from its substrate, the respiratory coenzyme NAD ϩ , at the site of breakage (1,2). The ADP-ribose polymer is primarily attached to PARP itself (automodification) and to a few other acceptor proteins (heteromodification) involved in chromatin architecture and DNA metabolism, thus affecting their activity. The function of PARP is thought to be related to a number of nuclear processes that involve nicking and resealing of DNA strands, including transcription and DNA repair. PARP is probably associated with a multifunctional complex comprising factor x-ray cross-complementing-1, DNA polymerase ␤, and DNA ligase III, which are involved in the base excision repair pathway. PARP knock-out mice display a marked genomic instability and an extreme sensitivity to genotoxic agents (3,4), and it has been shown that PARP-deficient cell lines performed very limited DNA repair during the first hours after DNA damage by alkylating agents (5).
PARP has a modular organization (1) comprising an N-terminal DNA binding domain, which acts as a molecular nick sensor and contains two zinc finger motifs called F1 and F2, which are involved in the recognition of DNA breaks during DNA repair (6) and a bipartite nuclear localization signal, a central regulating segment, which contains the automodification site, and the C-terminal catalytic domain, which binds NAD ϩ (Fig. 1). The presence of IgG antibodies reacting with the N-terminal F1 and F2 zinc fingers (7,8) and, in a very few cases, with the PARP catalytic domain (9) has been described in the serum of patients with various systemic autoimmune diseases. In particular, a high prevalence of antibodies reacting with synthetic peptides corresponding to zinc fingers F1 and F2 (residues 18 -59 and 122-165, respectively) was observed in patients with certain autoimmune rheumatic diseases (notably systemic lupus erythematosus (SLE)) and autoimmune bowel diseases such as Crohn's disease and ulcerative colitis (8,10). These antibodies are essentially of the IgG1 subclass and recognize the PARP protein in dot immunoassay. The possible pathogenic properties of PARP antibodies are unknown. With a few exceptions, this is the case of most autoantibodies characteristic of systemic autoimmune diseases. However, deposition of immune complexes containing anti-double-stranded DNA antibody subsets can induce renal failure (11,12), these antibodies may also deposit directly on the glomerular basement membrane, inducing proteinuria (13). On the other hand, deposition of antibodies to Ro/SSA ribonucleoprotein in fetal cardiac tissues is thought to be associated with congenital heart block (14,15), and autoantibody subsets to ribosomal P proteins have been shown to be able to penetrate into live cells and cause cellular dysfunction (16). Because elevated rates of spontaneous and induced chromosomal damage have been reported in certain autoimmune rheumatic diseases (17,18), deficient DNA repair was observed in chronic ulcerative colitis (19) and SLE (20,21), and defects in poly(ADP-ribose) synthesis have been described in lupus patients and their healthy biological relatives (22,23), we studied the possibility that autoantibodies to F1 and F2 zinc fingers inhibit the PARP catalytic activity and cause the observed pathogenic effects. We report here that, contrary to our first expectations, PARP antibodies from autoimmune patients do not significantly affect PARP activity but efficiently prevent caspase-3-mediated PARP cleavage during apoptosis and prolong cell survival, which can have important deleterious effects.

EXPERIMENTAL PROCEDURES
Recombinant PARP and Synthetic Peptides-Recombinant human PARP (rPARP) was produced as described previously (24). PARP F2 * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. peptide and 44 -67 peptide of ribonucleoprotein SmD1 were synthesized and purified as described (7,25). The homogeneity of each peptide was checked by analytical high-performance liquid chromatography on a nucleosil C18 5-m column (4.6 ϫ 150 mm), using a linear gradient of 0.1% trifluoroacetic acid in water and acetonitrile containing 0.08% trifluoroacetic acid. The identity of purified peptides was assessed by matrix-assisted laser desorption and ionization time-of-flight mass spectrometry using a Protein TOF apparatus (Bruker Spectrospin, Bremen, Germany).
Induction of Apoptosis-The human promyelocytic leukemia HL60 cells (1 ϫ 10 5 /ml) were grown in RPMI 1640 medium supplemented with 10% heat-inactivated fetal calf serum. To study the effect of antibodies on apoptosis, HL60 cells were first preincubated at 37°C with affinity-purified antibodies or normal IgG (5 g of IgG/5 ϫ 10 4 cells per well containing 100 l of medium) or with buffer alone. After 6 h, 10 l of RPMI medium containing 50 g/ml actinomycin D were added for 18 h at 37°C. Biochemical characterization of apoptosis was routinely assessed by DNA fragmentation analysis on a 1% agarose gel and by Western immunoblotting using the monoclonal antibody (mAb) C-2-10, which recognizes the 89-kDa apoptotic PARP fragment (27).
Flow Cytometry Analysis-Cell pellets were resuspended in Hepes buffer containing 2.5 g/ml propidium iodide (PI) and annexin V-fluorescein isothiocyanate conjugate (both from PharMingen). The fluorescence intensity was measured using a FACSCalibur apparatus and CELLQuest software (Becton Dickinson, San Jose, CA). Cell debris was excluded, and 5000 gated events were acquired. Early apoptotic cells are annexin V-positive and PI-negative, whereas late apoptotic cells are both annexin V-and PI-positive.

Patients' Autoantibodies Specific for the F2 Finger Do Not
Inhibit the Catalytic Activity of PARP-Human autoantibodies are often capable of inhibiting in vitro the functional activity of the antigen they target. This has been shown, for example, in the case of autoantibodies to protein kinase NII, DNA topoisomerase I, proliferating cell nuclear antigen, and tRNA synthetases (29). The possibility that autoantibodies to F2 inhibit in vitro the PARP catalytic activity was thus investigated using different assays previously designed to test the inhibiting properties of induced PARP mAbs (30) and chemical compounds (31). However, when compared with normal IgG, affinity-purified F2 autoantibodies from patients with SLE did not alter the activity of PARP significantly in any of these specific tests. It is interesting to note that Yamanaka et al. (9,32) showed that patients' antibodies directed to the C-terminal NAD binding domain efficiently inhibited the catalytic activity of PARP in vitro. This antibody reactivity, however, is seldom detected in patients' sera (33).
Patients' Autoantibodies Specific for the F2 Finger Do Inhibit the Specific Cleavage of PARP by Caspase-3-One of the characteristics associated with the execution phase of the apoptosis pathway is the specific PARP cleavage by caspases. This cleavage leads to its inactivation, thus preventing futile DNA repair cycles (27). It has been identified that caspase-3 (CPP32, Yama, or apopain) is the most efficient processing enzyme for PARP (34,35). It cleaves PARP after the consensus sequence 211 DEVD 214 located in the nuclear localization signal domain,

FIG. 2. Inhibition of caspase-3-mediated PARP cleavage by affinity-purified patients' antibodies to PARP F2. A, principle of the test setup to show the cleavage of PARP and its inhibition by PARP F2
antibodies (1, in the absence of antibodies; 2, in the presence of blocking antibodies). B, Experimental results. Lane 1, intact rPARP (control without caspase-3 and any antibody). Lane 2, rPARP incubated with caspase-3 only. Lane 3, rPARP successively incubated with affinitypurified patients' autoantibodies to F2 and active caspase-3. Lane 4, rPARP successively incubated with protein A-purified control human IgG and active caspase-3. FIG. 1. Schematic representation of the different functional PARP domains (1). Localization of the F1 and F2 zinc fingers (residues 20 -56 and 125-162, respectively) and of the bipartite nuclear localization signal (NLS) motif (residues 202-240) within the DNA binding domain (residues 1-375) is shown. The caspase-3 cleavage site within the nuclear localization signal is shown ( 211 DEVD2G 215 ). The schematic is not drawn to scale. thus generating two apoptotic-specific fragments of M r 89 and 24 kDa, respectively (27,34). To determine whether autoantibodies reacting with the PARP F2 finger located in residues 122-165 may influence the level of PARP cleavage by caspase-3, we designed a new assay illustrated in Fig. 2A in which human rPARP was preincubated with affinity-purified F2 antibodies and subsequently allowed to react with active caspase-3. After incubation, the mixture was diluted in sample buffer and subjected to SDS-polyacrylamide gel electrophoresis. PARP fragments were analyzed by Western immunoblotting using the anti-PARP mAb C-2-10 to detect the typical apoptotic fragment 89K (27). As expected, in the absence of antibodies (Fig. 2B, lane 2) or when normal human IgGs were incubated with rPARP before the addition of caspase-3 (Fig.  2B, lane 4), two fragments corresponding to the remaining intact PARP protein (116K) and the typical apoptotic 89K fragment were visualized. Most interestingly, however, when affinity-purified human antibodies to F2 were preincubated with rPARP, the apoptotic fragment 89K was not generated by caspase-3 (Fig. 2B, lane 3). The same result was found with distinct pools of affinity-purified patient antibodies to F2 as well as in the presence of affinity-purified anti-F2 peptide antibodies raised in rabbits (results not shown). No cleavage inhibition was observed when we tested non-related rabbit Abs or normal rabbit IgGs. As additional control experiments, we checked that purified IgGs were not cleaved, and thus not used as a possible secondary substrate by caspase-3. Furthermore, we verified that none of the patients' and rabbit antibodies reacted with caspase-3 itself. These data univocally indicate that in vitro, autoantibodies to F2 from patients with SLE specifically hamper PARP cleavage. A direct inactivation of caspase-3 is certainly not involved in this observation. Anti-F2 antibodies most probably act by direct or induced steric hindrance in blocking the access of the DEVD2G cleavage site to caspase-3.

Patients' Autoantibodies Specific for the F2 Finger Decrease the Number of Apoptotic Cells and Increase Cell Survival-
Because the specific proteolytic breakdown of PARP by caspases is known to occur during the executive phase of apoptosis, we further examined whether patient-derived anti-F2 antibodies may directly affect the apoptotic events in mammalian cells. We evaluated the number of apoptotic cells in a preparation of HL60 cells incubated either with affinity-purified anti-F2 autoantibodies or with antibodies from lupus patients purified on a column containing peptide 44 -67 of nuclear SmD1 antigen (25). These antibodies were used as control, because as antibodies to F2, they recognize a major epitope of a nuclear autoantigen, they were affinity-purified using a similar procedure, and they were extracted from lupus sera. It has to be noted that because patients' antibodies to PARP are mostly directed to F1 and F2, and that other linear autoepitopes have not been described in PARP (8), autoantibodies that recognize other sites of the enzyme could not be included as other control. Antibody-preincubated cells were subjected to actinomycin D exposure to trigger apoptosis. Apoptosis was quantified by annexin V and PI staining and flow cytometry (36). Fig. 3 shows a significant decrease in the number of annexin V-and PI-positive HL60 cells (late apoptosis) after incubation with F2 autoantibodies (Fig. 3, B-D). In this representative experiment, the number of late apoptotic cells was reduced by 30.2% (67.4 versus 97.6%) compared with the number of late apoptotic cells counted in the cell sample incu- bated with control lupus autoantibodies directed to peptide 44 -67 of SmD1 (Fig. 3E).
We also used the MTT conversion assay (28) to determine the total number of living HL-60 cells exposed first to F2 and control antibodies and then to actinomycin D. As shown in Fig.  4, and in good agreement with the results in Fig. 3, the number of surviving cells was significantly increased when cells were incubated with F2 antibodies compared with the number of living cells in the sample incubated with control IgG from normal individuals.
Using both the flow cytometry analysis and MTT assay, we found that anti-F2 antibodies have no effect on HL-60 cells that have not been subsequently induced to apoptosis. This result supports the idea that anti-F2 antibodies play no deleterious role on intact dividing cells but apparently only affect preapoptotic cells. DISCUSSION The present study highlights the identification of a new, potentially pathogenic autoantibody population. Antibodies to PARP peptide F2 efficiently inhibit caspase-3-mediated PARP cleavage in vitro and prolong ex vivo survival of cell subsets undergoing apoptosis after exposure to actinomycin D. We have not expressly shown that human anti-F2 antibodies effectively enter living cells; however, the penetration of antibodies within the cells could be facilitated, because HL60 cells possess Fc receptors at their surface. In this regard, it is notable that purified F2 and SmD1 peptide antibodies were essentially of the IgG1 subclass, which is the IgG subclass showing the highest affinity for Fc receptors. Furthermore, whereas membrane integrity is not lost in the initial stages of apoptosis, cytoplasmic and nuclear membranes of apoptotic cells in the later stages are altered, and this may facilitate antibody penetration. Moreover, it has been shown that many autoantigens undergo a striking redistribution during apoptosis, becoming clustered and concentrated in the surface blebs of apoptotic cells (37,38). Among these autoantigens are nucleosomes, several ribonucleoproteins, certain ribosomal proteins, and PARP. It is possible that purified antibodies recognize PARP in apoptotic surface structures and that PARP itself mediates the penetration of anti-PARP autoantibodies, which then localize within preapoptotic cells.
Our findings suggest that as a result of antibody binding to the F2 zinc finger, a functional site that is relatively near the apoptotic caspase-3 cleavage site (Fig. 1), the access of the DEVD2G cleavage site by caspase-3 may become less easily available, and cells may be more resistant to cell death. It is notable that cells from PARP-deficient knock-out mice transfected with an uncleavable D 214 A PARP mutant present a significant apoptotic delay (39). Our finding, as well as the observation that in lupus patients the percentage of macrophages engulfing apoptotic cell material appears significantly reduced compared with control individuals (40), might be of central importance in the etiopathogenesis of SLE. These phenomena may lead to an increase in the circulation or in certain compartments of both persistent nonapoptotic and apoptotic cells that contain biochemically modified self-antigens (41). Experiments are presently being performed to analyze the morphological and immunogenic properties of this persistent and probably abnormal material in vivo and to determine whether the accumulation of certain cell subsets that would normally have been eliminated may contribute both to the local inflammatory process and to the B-and T-cell autoimmune response itself.