Selective Cleavage of BLM, the Bloom Syndrome Protein, during Apoptotic Cell Death*

Bloom syndrome (BS) is an autosomal recessive disorder characterized by a high incidence of cancer and genomic instability. BLM, the protein defective in BS, is a RECQ-like helicase that is presumed to function in mammalian DNA replication, recombination, or repair. We show here that BLM, but not the related RECQ-like helicase WRN, is rapidly cleaved in cells undergoing apoptosis. BLM was cleaved to 47- and 110-kDa major fragments, with kinetics similar to the apoptotic cleavage of poly(A)DP-ribose polymerase. BLM cleavage was prevented by a caspase 3 inhibitor and did not occur in caspase 3-deficient cells. Moreover, recombinant BLM was cleaved to 47- and 110-kDa fragments by caspase 3, but not caspase 6,in vitro. The caspase 3 recognition sequence412TEVD415 was verified by mutating aspartate 415 to glycine and showing that this mutation rendered BLM resistant to caspase 3 cleavage. Cleavage did not abolish the BLM helicase activity but abolished BLM nuclear foci and the association of BLM with condensed DNA and the insoluble matrix. The results suggest that BLM, but not WRN, is an early selected target during the execution of apoptosis.

Apoptosis, or programmed cell death, is critical for many biological processes, including tissue morphogenesis and homeostasis, the development of immunity and host defense mechanisms, and elimination of damaged or potentially neoplastic cells (1)(2)(3). Moreover, defective apoptosis may cause or exacerbate a variety of human diseases, including Alzheimer's and Huntington's diseases, autoimmunity, and cancer (4 -6). Cells that are dying by apoptosis share many cytological and molecular features, regardless of the initiating signal. Cytological changes include cytoplasmic and nuclear shrinkage, plasma membrane blebbing, and chromatin condensation. At the molecular level, a family of cysteine proteases, termed caspases, selectively cleaves a series of protein substrates. Initiator caspases, such as caspases 8 and 9, initiate a proteolytic cascade that eventually culminates in the cleavage and activation of downstream caspases, such as caspases 3 and 6. The downstream or execution caspases then cleave selected target proteins (7)(8)(9). Subsequently, a caspase-activated deoxyribonuclease (CAD 1 ) cleaves genomic DNA (10), at which point cell death is imminent.
Most caspases recognize a four-amino acid sequence having a C-terminal aspartate, after which they cleave. Caspases are highly selective in their target proteins and generally cleave at only one or a few specific sites within the target. Caspasemediated cleavage can lead to either functional activation or inactivation of the target protein, which presumably facilitates the orderly execution of apoptosis (7)(8)(9)(10)(11)(12)(13). The most prominent execution caspase is caspase 3, also known as CPP32, Yama, and apopain. Several proteins that sense DNA damage and/or participate in DNA repair have been identified as caspase 3 substrates. These proteins include ATM (14), RAD51 (15,16), the catalytic subunit of DNA-dependent protein kinase (DNA-PKcs) (17), and poly(A)DP-ribose polymerase (PARP) (18,19). Caspase 3 cleavage effectively inactivates these proteins and, subsequently, CAD degrades DNA (10). Thus, in advance of DNA fragmentation, caspases ensure that cells destined for apoptotic death inactivate proteins involved in maintaining genomic integrity, thereby assuring the orderly execution of apoptosis.
Recently, a family of genes related to Escherichia coli RECQ has been implicated in maintaining genomic integrity in human cells (20). RECQ encodes a DNA helicase that participates in homologous recombination and suppresses illegitimate recombination (21,22). At least five human RECQ-like genes have been identified: RECQL (23,24), BLM (25), WRN (26), RECQL4, and RECQL5 (27). Among these, BLM was the first to be linked to a hereditary disease (25). Defects in BLM cause Bloom's syndrome (BS), an autosomal recessive disorder characterized by multiple abnormalities, including immunodeficiency, pre-and post-natal growth retardation, and a high incidence of cancer (28). Cancer is the primary cause of death in BS and generally occurs before the third decade of life. Cells from individuals with BS are hypermutable. BS cells accumulate chromatid gaps and breaks and, most prominently, numerous sister chromatid exchanges (28,29).
BLM encodes a 1417-amino acid 3Ј 3 5Ј DNA helicase that localizes to the nuclear matrix and discrete nuclear foci known as PML or ND-10 bodies (25, 30 -34). BLM foci also contain the recombination/recombinational repair protein RAD51 and associate with sites of putative DNA repair after damage by ionizing radiation. 2 These attributes of BLM, and the cellular phenotypes of BS cells, suggest that BLM is important for maintaining genomic integrity.
Because of its importance in maintaining genomic stability, we asked whether BLM was among the proteins targeted for selective degradation during the execution phase of apoptosis. We show here that BLM, but not the related RECQ-like protein WRN, is proteolytically cleaved at a single site in cells induced to undergo apoptosis by multiple stimuli, and identify caspase 3 as the responsible protease. BLM cleavage was an early apoptotic event that did not abrogate BLM helicase activity but caused disappearance of BLM foci and detachment from condensed DNA and the insoluble matrix. Thus, BLM, but not WRN, is targeted for degradation during the early execution phase of apoptosis.

EXPERIMENTAL PROCEDURES
Reagents-Tosyl-L-lysine chloromethyl ketone (TLCK) and leupeptin were purchased from Roche Diagnostics, acetyl-Tyr-Val-Ala-Asp fluoromethyl ketone (ZVAD-FMK) was from Enzyme System Products, and acetyl-Asp-Glu-Val-Asp aldehyde (Ac-DEVD-CHO) was from BD PharMingen. Other reagents were molecular biology or cell culture grade, purchased as indicated, and prepared according to the manufacturer's instructions.
Cell Culture and Induction of Apoptosis or Necrosis-Cells were obtained from the American Type Culture Collection and cultured under standard conditions in RPMI 1640 or Dulbecco's modified Eagle's medium (Life Technologies, Inc.) and 10% fetal calf serum. To induce apoptosis, Jurkat cells (1 ϫ 10 6 cells/ml) were incubated with 100 ng/ml anti-FAS antibody (MBL International Corp.) or 2 M staurosporine (Sigma Chemical Co.), HeLa cells were incubated with 30 ng/ml tumor necrosis factor-␣ (Calbiochem) and 10 g/ml cycloheximide (Sigma) or 2 M staurosporine, and MCF-7 cells were incubated with 30 ng/ml tumor necrosis factor-␣ and 10 g/ml cycloheximide. To induce necrosis, Jurkat cells were washed and suspended in serum-free medium lacking glucose and containing 2 mM pyruvate. After adaptation to the medium (37), they were incubated with 2.5 M oligomycin (Sigma). Pretreatment with 2.5 M oligomycin was for 45 min, followed by incubation for 6 h with or without anti-FAS. Intracellular ATP levels were determined using a commercial kit (Sigma, FL-ASC) and protocol furnished by the supplier. For protease inhibitor studies, cells were preincubated with the indicated concentrations for 30 min, anti-FAS was added, and cells were harvested 6 h later, unless noted otherwise.
Nuclear Extracts, Total Cell Lysates, and Western Analyses-Nuclear extracts were prepared from 5-10 ϫ 10 6 cells by rapid salt extraction. Briefly, cells were washed twice in ice-cold phosphate buffered saline (PBS) and stored at Ϫ80°C. Cell pellets were thawed on ice, suspended in 100 l per 5 ϫ 10 6 cells in buffer used by Dignam et al. (38): 0.42 M NaCl, 10% glycerol, 20 mM HEPES, pH 7.9, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride. The extract was clarified by centrifugation at 10,000 ϫ g for 15 min. Total cell lysates were prepared by lysing 1-2 ϫ 10 6 cells in 100 l of SDS-PAGE sample buffer without dye. Protein concentrations were determined by Bio-Rad detergent-compatible protein assay. Nuclear extracts (30 g) or total cell lysates (50 g) were separated by 4 -15% gradient (Bio-Rad) SDS-PAGE and analyzed by Western blotting as described previously (39). Antibodies were detected by chemiluminescence, using the SuperSignal West Pico detection kit (Pierce).
Digitonin Extraction-Digitonin extraction was performed as described by Adam et al. (40). Briefly, 3-6 h after induction of apoptosis, cells were washed twice in ice-cold PBS, pelleted at 190 ϫ g for 5 min, and gently resuspended in PBS containing 1% digitonin (Sigma). After a 5-min incubation on ice, cells were pelleted to separate cytosolic proteins (supernatant) from insoluble proteins (pellet). Detergent was removed from supernatant and pellet proteins by precipitating in methanol-chloroform (41). The precipitate was solubilized in 2ϫ SDS-PAGE sample buffer, and the proteins were analyzed by Western blotting.
Immunofluorescence-Jurkat cells (1 ϫ 10 5 ) were collected using a cytospin, fixed, and stained at room temperature as described (42). Cells were incubated with primary antibodies for 2 h and secondary antibodies for 1 h. Slides were mounted in Vectashield containing DAPI (4Ј,6-diamidino-2-phenyl-indole, Vector Laboratories) and viewed by epifluorescence. The images were captured using a cooled charge-coupled device camera and merged using Canvas (Deneba).
In Vitro Cleavage of BLM-Glutathione S-transferase (GST) and GST-BLM fusion protein were expressed in Sf9 insect cells using recombinant baculoviruses and purified using a commercial kit (Life Technologies, Inc.). Nuclear lysates were prepared as described (36,43), clarified by centrifugation, and incubated for 1 h at 4°C with glutathione-Sepharose 6-CL B resin (Amersham Pharmacia Biotech). The slurry was transferred to a column, and washed with 50 column volumes of PBS plus 0.2% Nonidet P-40 and 50 volumes of PBS. Proteins were eluted with 20 mM glutathione, 100 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM dithiothreitol, 10% glycerol (elution buffer), and migrated as single bands on silver-stained SDS-PAGE. Protein concentrations were determined by Bio-Rad assay. Recombinant activated caspases were expressed in E. coli and prepared as described (44). Purified GST-BLM (20 nM) and caspase (1 M) were incubated for 1 h at 37°C in elution buffer. Reactions were terminated by adding an equal volume of 2ϫ SDS-PAGE sample buffer or directly assayed for helicase activity.
Helicase Assay-Helicase assays were performed as described by Huang et al. (45). Briefly, activity was detected by displacement of a 32 P-5Ј-labeled 20-bp oligonucleotide from a partial 20-bp/46-bp DNA duplex in which the 46-bp oligonucleotide was unlabeled. The reaction (20 l) was incubated for 10 min at 37°C and terminated by rapid cooling and addition of 5ϫ loading buffer (2% SDS, 50 mM EDTA, 3% bromphenol blue, 3% xylene cyanol, 40% glycerol). The displaced oligonucleotide was separated from the partial duplex by 12% nondenaturing PAGE. Where indicated, proteins and probe were denatured prior to assay by heating to 95°C for 5 min.
Generation of Caspase-resistant Mutant and in Vitro Translation-Replacement of aspartate with glycine at position 415 was carried out by overlapping PCR. Briefly, pSG5-Myc-BLM1, containing the fulllength BLM cDNA with an N-terminal Myc epitope tag was digested with EcoRI and BamHI. The 1.4-kb fragment containing the site to be mutagenized was subcloned into pBluescriptKS ϩ (Stratagene). Two oligonucleotides spanning the site were synthesized (BLM-5Ј, CTTCTA-ACGGAAGTAGGTTTTAATAAAAGTGATGCC and BLM-3Ј, GGCAT-CACTTTTATTAAAACCTACTTCCGTTAGAAG). Polymerase chain reaction (PCR) was used to amplify a 1.3-kb fragment using the M13 forward and BLM-3Ј primer and a 200-bp fragment using the M13 Reverse and BLM-5Ј primer. The fragments were mixed at equal molar concentrations, and a 1.5-kb fragment was amplified using the M13 Forward and Reverse primers. The PCR product was digested with EcoRI and BamHI, subcloned in pBluescriptKS, and sequenced using the T3 primer. The mutagenized 1.4-kb EcoRI-BamHI fragment was cloned into pSG5-Myc-BLM1 that was partially digested with EcoRI and BamHI to obtain full-length D415A-BLM. In vitro translation was performed using the coupled TnT reticulocyte lysate system (Promega) and [ 35 S]methionine (Redivue, Amersham Pharmacia Biotech). The translation products were untreated or incubated with 0.1 M activated recombinant caspase for 1 h at 37°C, solubilized in SDS-PAGE sample buffer, separated by 4 -15% SDS-PAGE, and visualized by autoradiography.

RESULTS
BLM Is Cleaved during Apoptosis-To determine whether BLM is cleaved during apoptosis, we initially used a well established model system, human Jurkat leukemia T cells induced to undergo apoptosis by a FAS-monoclonal antibody (46). Nuclear extracts were prepared from cells treated with anti-FAS for varying intervals and analyzed by Western blotting, using antibodies raised against either the N-terminal 431 amino acids (33) or the C-terminal 375 amino acids of BLM (35).
Cleavage of BLM was first evident 2-4 h after addition of anti-FAS, progressing to near completion over the next 8 -10 h (Fig. 1A). The 159-kDa BLM protein was cleaved to a 110-to 115-kDa fragment detected by the C-terminal antibody (Fig.  1A), and a 45-to 50-kDa fragment, detected by the N-terminal antibody (see Fig. 2). Subsequent experiments (see Fig. 4) showed that these fragments were 112 and 47 kDa, respectively. In striking contrast to BLM, WRN, a related RECQ-like helicase, was not cleaved during anti-FAS-mediated apoptosis (Fig. 1B). Likewise, Ku70, a component of DNA-dependent protein kinase, which is critical for repairing DNA doublestrand breaks by nonhomologous end-joining, remained intact (Fig. 1C). Thus, cleavage of BLM was an early and relatively selective event during apoptosis.
The initiation of BLM cleavage coincided with the initiation of PARP cleavage (Fig. 1D), a well established early event in the execution of apoptosis (18,19). On the other hand, although PARP was completely cleaved 6 -9 h after addition of anti-FAS (Fig. 1D), BLM required about 12 h for complete cleavage (Fig.  1A). The more rapid completion of PARP cleavage may reflect differences in the enzymes or kinetics by which PARP and BLM are cleaved or differences in their accessibility to apoptotic proteases. Like the PARP apoptotic fragments, the BLM apoptotic fragments were stable for several hours after the initiation of apoptosis.
BLM cleavage was not limited to anti-FAS-induced apoptosis, or to Jurkat cells. HeLa cells undergo rapid apoptosis in response to tumor necrosis factor-␣ (TNF-␣) and cycloheximide. Under these conditions, BLM cleavage was evident within 3 h, and near complete in 6 h ( Fig. 2A). BLM was also cleaved 3-6 h after either Jurkat or HeLa cells were induced to undergo apoptosis by the protein kinase inhibitor staurosporine (Fig.  2B). In all cases, BLM was cleaved to a 47-kDa fragment detected by the N-terminal antibody and to a 112-kDa fragment detected by the C-terminal antibody (Fig. 2, A and B).
In contrast to its fate during apoptosis, BLM was not cleaved when cells were induced to undergo necrotic cell death. Jurkat cells were treated with anti-FAS to induce apoptotic death, or oligomycin, which causes death by necrosis (37). Oligomycin caused a rapid loss of intracellular ATP, characteristic of necrotic death (37), compared with the slower loss of ATP caused by anti-FAS, characteristic of apoptotic death (Fig. 2C). Western analysis of extracts prepared 6 h later, when ATP levels were comparably low in the oligomycin and anti-FAS-treated cells, showed that BLM cleavage occurred only during anti-FAS-induced apoptosis; there was little or no BLM cleavage during oligomycin-induced necrosis (Fig. 2D). Moreover, oligomycin, which to a large extent inhibits apoptosis, also to a large extent prevented BLM degradation (Fig. 2D).
Taken together, these results suggest that BLM is an early target for selective apoptosis-induced proteolysis.
Identification of the Protease That Cleaves BLM-To identify the protease responsible for apoptotic cleavage of BLM, we treated Jurkat cells with anti-FAS in the presence of protease inhibitors. BLM cleavage was not prevented by inhibitors of serine proteases, leupeptin (100 or 200 M; Fig. 3A, lanes 5, 6) or tosyl-L-lysine chloromethyl ketone (TLCK) (100 M ; Fig. 3A,  lane 3). A high concentration of TLCK (200 M) inhibited apoptotic BLM cleavage (Fig. 3A, lane 4). TLCK has been shown to also inhibit the activity and activation of caspases in vitro (47), and a high concentration of TLCK (200 M) was shown to induce necrosis in Jurkat cells without features of apoptosis (48). In contrast to leupeptin and TLCK, caspase inhibitors (49) prevented apoptotic BLM cleavage at moderate to low concentrations. This was true for the broad-range inhibitor ZVAD-FMK (10 and 20 M; Fig. 3A, lanes 7, 8) and the caspase 3/caspase 7 inhibitor Ac-DEVD-CHO (100 M ; Fig. 3A, lane 9). These results suggest that the cleavage of BLM during apo-ptosis depends on the activity of a caspase 3 or caspase 7-like enzyme. Caspases 3 and 7 have identical recognition sequences. However, caspase 3 is the more likely candidate for cleaving BLM in vivo, because it, in contrast to caspase 7, is found in the nucleus (50) where BLM resides.
We determined that caspases 3 and 7 were capable of cleaving BLM by incubating purified recombinant BLM and activated caspases in vitro. Recombinant BLM was a fusion protein coupled to glutathione S-transferase (GST-BLM, ϳ188 kDa; see Fig. 4) and recombinant caspases were the activated forms of the executioner caspases 3, 6, and 7. Caspases 3 and 7, but not caspase 6, cleaved GST-BLM (Fig. 3B). In both cases, BLM was cleaved to a 76-kDa fragment, detected by the N-terminal antibody (Fig. 3B) and silver staining (Fig. 3C), and a 112-kDa fragment detected by silver staining (Fig., 3C). The size of the 76-kDa fragment recognized by the N-terminal antibody is consistent with the size of the GST moiety, which was fused to the BLM N terminus (see Fig. 4A) and the 47-kDa N-terminal fragment produced in vivo. Western analysis using anti-GST antibody confirmed that the 76-kDa fragment contained GST (not shown). The cleavage patterns of native BLM (47 and 112 kDa) and GST-BLM (76 and 112 kDa) suggest that the caspase 3/7 cleavage site lies in the N-terminal third of the protein.
Identification of the Caspase Cleavage Site-BLM contains only a single four-amino acid cluster, 412 TEVD 415 , that is similar to the consensus caspase 3/7 recognition sequence DEVD (11). The C-terminal aspartate in this cluster is located at amino acid 415 (Fig. 4A). Cleavage of native BLM at aspartate 415 would yield two fragments with calculated molecular masses of 47 and 112 kDa, whereas cleavage of GST-BLM at this site would yield 112-and 76-kDa fragments (Fig. 4A). These predicted sizes match the sizes of the BLM cleavage products generated in vivo and in vitro.
To ascertain whether the sequence 412 TEVD 415 is indeed the caspase recognition and cleavage site, we mutated aspartate 415 to glycine, generating the mutant protein BLM-D415G. Wild-type BLM and BLM-D415G proteins were translated and radiolabeled in vitro and then incubated with activated caspase 3. Wild-type BLM was cleaved by caspase 3, whereas BLM-D415G was resistant to caspase 3 cleavage (Fig. 4B). This result suggests that caspase 3, or possibly 7, cleaves BLM at aspartate 415 during apoptosis.
To determine whether caspase 3 or 7 cleaves BLM during apoptosis in vivo, we used the caspase 3-deficient breast cancer cell line MCF-7 (51). MCF-7 cells were induced to undergo apoptosis by TNF-␣ and cycloheximide. As reported (51), partial cleavage of PARP (Fig. 4C) and activation of caspase 7 (data not shown), which partially cleaves PARP in these cells, were apparent within 5 h. BLM, however, remained intact for at least 24 h (Fig. 4C). This result strongly suggests that caspase 3, not caspase 7, cleaves BLM in vivo during apoptotic cell death.
The Cleaved N Terminus of BLM Is Dispensable for Helicase Activity-The 112-kDa apoptotic fragment contains the helicase and DNA binding activities of BLM (Fig. 4A), but little is known about the function of the smaller 47-kDa N-terminal fragment, which contains no known protein motifs (52). To determine whether apoptotic cleavage, which separates the Nand C-terminal regions, alters BLM helicase activity, we incubated GST-BLM with activated caspase 3 in vitro under conditions in which cleavage was complete (Fig. 3, B and C). We then assayed intact and cleaved BLM proteins for helicase activity. The substrate was a partial 20/46-mer DNA duplex in which the 20-mer was radiolabeled at the 5Ј-end (Fig. 5A). Helicase activity dissociates the 20-mer from the duplex; the duplex and dissociated 20-mer were distinguished using native PAGE and autoradiography, as described (45).
Intact and cleaved GST-BLM completely unwound the helicase substrate within 10 min (Fig. 5A, lanes 3, 4). The activities of intact GST-BLM, GST-BLM cleaved by caspase 3, and GST-BLM incubated with heat-inactivated caspase 3 were indistinguishable at this time point (Fig. 5A, lanes 3-5), as well as earlier time points (see Fig. 5E). Control reactions showed that the labeled 20-mer was released from the duplex when heated (Fig. 5A, lane 1), whereas heat-inactivated GST-BLM, or GST treated with caspase 3, had no helicase activity (Fig. 5A, lanes  2, 6). Similar results were obtained when a longer partial duplex DNA, or G4-DNA, were used as helicase substrates (not shown). We conclude that BLM retains helicase activity after cleavage by caspase 3.
The C-terminal fragment retained helicase activity even after it was physically separated from the N-terminal fragment. This was shown by immobilizing GST-BLM on glutathione-Sepharose beads before incubating with caspase 3. The C-terminal fragment was released into the supernatant upon cleavage, whereas the N-terminal fragment containing the GST moiety remained bound until eluted with glutathione (Fig.  5B). Western analysis confirmed that the C-and N-terminal fragments were present in the appropriate fractions (Fig. 5B). The fragments, and intact protein, were tested separately for helicase activity. The C-terminal fragment (Fig. 5C, lane 4), but not the N-terminal fragment (lane 3), unwound the helicase assay substrate within 10 min, as did intact (lane 1) and cleaved (lane 2) proteins. Control experiments using varying amounts of GST-BLM showed that 2-to 5-fold differences in helicase activities could be detected by the helicase assay (Fig.  5D, compare, for example, 5 versus 10 ng and 5 versus 25 ng) and that, under assay conditions in which only a fraction of the substrate was unwound, uncleaved and cleaved BLM showed little or no difference in helicase activity.
These results suggest that the N-terminal 415 amino acids of BLM are dispensable for helicase activity, and the C-terminal fragment generated by apoptotic cleavage retains helicase activity.

FIG. 5. Helicase activity of intact and caspase-treated BLM.
Recombinant GST or GST-BLM (20 M) were untreated or cleaved with activated recombinant caspase (1 M), and one-fourth the reaction mixture was assayed for helicase activity using a 20-bp/46-bp partial duplex in which the 20-bp oligonucleotide was 32 P-labeled (asterisk) at the 5Ј-end (illustrated to the left of the upper band), as described under "Experimental Procedures." Helicase activity released the singlestranded, radiolabeled 20-mer (illustrated to the left of the lower band). A, proteins were untreated (lanes 2, 3), cleaved with caspase 3 (C3) (lanes 4 -6), or proteins or probe were heat denatured (H) (lanes 1, 2, 5) prior to assay, as described under "Experimental Procedures." Lane 1, heat-denatured probe; lane 2, heat-denatured GST-BLM; lane 3, untreated GST-BLM; lane 4, GST-BLM pretreated with caspase 3; lane 5, caspase 3 was heat-denatured prior to incubation with GST-BLM; lane 6, GST pretreated with caspase 3. B, GST-BLM was digested with caspase 3, and the reaction mixture was incubated with glutathione beads. The beads were collected by centrifugation, and the supernatant (Released) containing the released BLM C-terminal fragment was recovered. The protein bound to the beads was then eluted with glutathione, and the eluate was recovered (Bound). Released and bound fractions were analyzed by Western blotting using the anti-N-terminal (␣-BLM NT) or C-terminal (␣-BLM CT) BLM antibody, as described under "Experimental Procedures." The cleaved BLM proteins are indicated. C, GST-BLM (lane 1), GST-BLM digested with caspase 3 (lane 2), and the released and bound fractions described in B were assayed for helicase activity as described in A. D, GST-BLM (1-50 ng as indicated) was assayed for helicase activity at room temperature for 30 min, as described under "Experimental Procedures." E, GST-BLM (50 ng) was incubated at 37°C for 1 h with buffer (ϪCaspase) or 1 M caspase 3 (ϩCaspase), and then assayed for helicase activity for the indicated intervals (30 -120 s) as described under "Experimental Procedures." The unwound (Probe) and heat-denatured (Probe, H) substrates are shown in the first two lanes.
Redistribution of BLM during Apoptosis-Concomitant with cleavage, apoptosis altered the subcellular localization of BLM. BLM is found entirely in the nucleus, mostly, but not exclusively, organized into discrete foci (Fig. 6A) that costain for PML (31)(32)(33)(34). BLM foci, detectable by indirect immunofluorescence, began to disappear within 2 h after Jurkat cells were induced to undergo apoptosis by anti-FAS (not shown). Shortly thereafter (3-6 h after initiation of apoptosis) there was a major redistribution of BLM. Much of the BLM was distributed outside the areas of condensed DNA, which is visible by DAPI fluorescence. Both the apoptotic BLM cleavage fragments, detected by N-terminal (Fig. 6B) and C-terminal (Fig. 6C)-specific antibodies, showed this marked redistribution.
Apoptotic cleavage of BLM very likely caused its release from the nuclear matrix. Digitonin was used to permeabilize cytoplasmic and nuclear membranes, and soluble, loosely bound proteins were separated from insoluble cell structures by centrifugation (40). Western analysis of the concentrated soluble (S) and insoluble (P) proteins showed that a significant fraction of the BLM fragments detached from the insoluble structures, whereas a small amount of full-length protein remained bound (Fig. 6D). Because BLM is entirely nuclear and bound to the nuclear matrix (31) 2 and the insoluble fraction contains nuclear and cytosolic matrix components, we infer that BLM detaches from the nuclear matrix upon caspase cleavage. Because of the fragility of cells undergoing apoptosis, it was not possible to obtain standard nuclear matrix preparations, for example, by the methods described by Wan et al. (53). Nonetheless, these results are consistent with those obtained by immunofluorescence, suggesting that BLM is cleaved and released from the nuclear matrix during apoptosis.

DISCUSSION
Caspases are important for both the initiation and execution phases of apoptosis (7)(8)(9). At present, there is a need to identify caspase substrates to understand how caspases execute apoptosis. In recent years, a number of caspase substrates have been identified (9,54). Here, we show that BLM is a substrate for the execution caspase 3, in vitro and in vivo.
BLM encodes a DNA helicase (30) that is expressed primarily in late S phase and G 2 (55) 2 and localizes to PML nuclear bodies (31)(32)(33)(34). The phenotypes associated with defects in BLM, its homology to RECQ, and its colocalization with proteins such as RAD51 and sites of repair after DNA damage 2 suggest that BLM participates in a homologous recombination repair pathway that resolves spontaneous and/or induced DNA damage. The components of this and other repair pathways likely exist in a large complex (56,57). Our finding that BLM is specifically cleaved in cells undergoing apoptosis, but not necrosis, supports the idea that one function of the execution caspases is to dismantle protein complexes that can repair the DNA fragments generated by apoptotic deoxyribonucleases, as proposed by Casciola-Rosen et al. (58).
In contrast to the cleavage of BLM, apoptosis did not result in cleavage of the related RECQ-like helicase WRN. The WRN amino acid sequence lacks consensus cleavage sites for caspases, suggesting that WRN participates in processes that do not need to be dismantled during apoptosis. Alternatively, WRN complexes may be targeted for disruption by apoptosis, but one or more WRN-interacting proteins, rather than WRN itself, may be subject to caspase cleavage.
Defects in WRN cause the Werner syndrome (WS). Although BS and WS have similarities, there are also differences. Both syndromes are characterized by a high incidence of cancer and cellular genomic instability (25,26). However, WS individuals are asymptomatic until after puberty, and survive much longer, generally until the fourth or fifth decade of life. There are also differences between the BLM and WRN proteins, despite similar helicase domains. WRN is not found in PML nuclear bodies, and WRN, but not BLM, has intrinsic 3Ј-5Ј exonuclease activity (45). BLM and WRN both exist in large complexes (56,57). However, the WRN complex contains many proteins that participate in DNA replication, whereas the BLM complex contains many proteins that participate in DNA damage sensing or repair. Components of the WRN and BLM complexes may exchange, depending on the cell cycle or presence of DNA damage, and there may be overlap in some functions of BLM and WRN. However, because WRN and BLM associate primarily with different complexes and nuclear structures, they may have different primary functions. Interestingly, ATM, a com-FIG. 6. Subcellular redistribution of cleaved BLM. A-C, Jurkat cells were untreated (A) or induced to undergo apoptosis using anti-FAS antibody (B, C), fixed 3 h later, and stained for DNA (DAPI) or BLM using either the anti-N-terminal (A, B) or anti-C-terminal (C) BLM antibody. Cells were viewed by epifluorescence microscopy, and the fluorescent images were digitally merged (MERGE). C, Jurkat cells were untreated (Control) or induced to undergo apoptosis using anti-FAS antibody (Apoptosis). Cells were extracted 4 h later with either SDS-PAGE sample buffer (I, input) or 1% digitonin. The digitonin extract was further separated into a soluble supernatant (S) and insoluble pellet (P), the proteins in each fraction were concentrated by precipitation and denatured in SDS-PAGE sample buffer, as described under "Experimental Procedures." Proteins were analyzed by Western blotting using anti-N-terminal (␣-BLM NT) or anti-C-terminal (␣-BLM CT) BLM antibodies, and anti-tubulin antibody (Tubulin) as a control. Intact and cleaved BLM proteins are indicated, as in Fig. 1A. ponent of the BLM complex, is also cleaved during apoptosis (14). By contrast, Ku70, a component of DNA-PK, was spared apoptotic cleavage, although the DNA-PKcs is cleaved (17,58,59). Ku was recently shown to interact with WRN (60). These findings support the idea that apoptosis selectively targets processes in which BLM, but not WRN, is a primary participant.
The major execution caspases are caspases 3, 6, and 7 (61,62). GST-BLM was efficiently cleaved by caspases 3 and 7, but not caspase 6, in vitro, and degradation in vitro was indistinguishable from that observed in cells. It is not surprising that both caspases 3 and 7 cleaved recombinant BLM, because these enzymes share the same consensus cleavage recognition site (DEVD). However, caspase 7 is localized predominantly to the endoplasmic reticulum and mitochondria, whereas caspase 3 is found primarily in the nucleus and cytoplasm (50). Our finding that BLM is not cleaved during apoptosis in the caspase 3-deficient cell line MCF-7 strongly implicates caspase 3, or one of its isoforms, rather than caspase 7, in the apoptotic proteolysis of BLM in vivo.
Caspases cleave target proteins at specific sites, rather than randomly, and cleavage may either activate or inactivate the substrate (7)(8)(9)(10)(11)(12)(13). Proteolytic cleavage can provide information about the domain structure of a protein, because protease-sensitive sites are often interdomain regions that lack a defined secondary structure (63). We mapped the region in BLM targeted by caspase 3 to a single site ϳ47 kDa from the N terminus. This site, 412 TEVD 415 , resembled the consensus caspase 3/7 recognition sequence DEVD (13). The C-terminal BLM fragment generated by caspase 3 contains the helicase, ATPase, DNA binding, and nuclear localization domains. The N-terminal fragment is devoid of known protein motifs but was recently implicated in assembling BLM into a hexamer (52,64). DNA helicases are typically dimers or hexamers (65). Oligomerization is thought to be essential for processive translocation, and, in a few cases, oligomers are the active form (65)(66)(67)(68). Loss of the N-terminal fragment might prevent BLM oligomerization but clearly did not abolish helicase activity, at least in vitro. It is possible that domains outside the N-terminal region can facilitate the assembly of BLM into hexamers or that the C-terminal fragment is active as a monomer or other oligomeric form. E. coli DNA helicase II is active as a monomer, although it is capable of dimerization (69), and the WRN helicase appears to form trimers (36). Finally, although loss of the N terminus did not abolish helicase activity in vitro, it might prevent hexamerization, and hence activity, in vivo.
Whatever the biochemical outcome of caspase cleavage, immunostaining and biochemical fractionation showed that cleaved BLM lost its characteristic punctate nuclear localization, detached from an insoluble substructure, and dissociated from condensed DNA. Cleavage and loss of localization very likely obliterates the in vivo function of BLM. BLM colocalizes with a number of proteins that are essential for the repair of DNA by homologous recombination, including RAD51 (56). 2 RAD51 is also cleaved by caspase 3 during apoptosis, and cleavage abolishes the RAD51 recombinase activity (15,16). Moreover, the kinetics of BLM cleavage is similar to that of DNA-PKcs (17,58,59), ATM (14), and PARP (18,19), which are cleaved before CAD-induced DNA fragmentation occurs (10). The potential function of BLM in DNA repair suggests that its cleavage and redistribution may aid nuclear disassembly and prevent the complex in which it resides from participating in the repair of fragmented DNA molecules generated by CAD.