Caspase-7 Is Directly Activated by the ∼700-kDa Apoptosome Complex and Is Released as a Stable XIAP-Caspase-7 ∼200-kDa Complex*

MCF-7 cells lack caspase-3 but undergo mitochondrial-dependent apoptosis via caspase-7 activation. It is assumed that the Apaf-1-caspase-9 apoptosome processes caspase-7 in an analogous manner to that described for caspase-3. However, this has not been validated experimentally, and we have now characterized the caspase-7 activating apoptosome complex in MCF-7 cell lysates activated with dATP/cytochrome c. Apaf-1 oligomerizes to produce ∼1.4-MDa and ∼700-kDa apoptosome complexes, and the latter complex directly cleaves/activates procaspase-7. This ∼700-kDa apoptosome complex, which is also formed in apoptotic MCF-7 cells, is assembled by rapid oligomerization of Apaf-1 and followed by a slower process of procaspase-9 recruitment and cleavage to form the p35/34 forms. However, procaspase-9 recruitment and processing are accelerated in lysates supplemented with caspase-3. In lysates containing very low levels of Smac and Omi/HtrA2, XIAP (X-linked inhibitor of apoptosis) binds tightly to caspase-9 in the apoptosome complex, and as a result caspase-7 processing is abrogated. In contrast, in MCF-7 lysates containing Smac and Omi/HtrA2, active caspase-7 is released from the apoptosome and forms a stable ∼200-kDa XIAP-caspase-7 complex, which apparently does not contain cIAP1 or cIAP2. Thus, in comparison to caspase-3-containing cells, XIAP appears to have a more significant antiapoptotic role in MCF-7 cells because it directly inhibits caspase-7 activation by the apoptosome and also forms a stable ∼200-kDa complex with active caspase-7.

The MCF-7 cell line was derived from a patient with metastatic breast cancer and is an often-used model system for studying estrogen receptor-positive breast cancer (for review, see Ref. 1). Because many of the problems associated with breast cancer treatment involve the development of chemo-resistance to apoptosis-inducing anti-cancer agents, there is extensive interest in using MCF-7 cells as a model for investigating the mechanisms of apoptosis in breast epithelial cells. Caspase activation is a key event in triggering the morphological and biochemical changes associated with cell death (2)(3)(4). There are two primary caspase activation pathways involving either stimulation of cell surface death receptors (the extrinsic pathway) or perturbation of mitochondria (the intrinsic pathway) (5). Many anti-cancer drugs induce apoptosis by activating the intrinsic cell death pathway, which involves the release of cytochrome c and the activation of the apoptosome-catalyzed caspase cascade (6 -8). In apoptotic cells inactive procaspases are activated via this cascade mechanism in which an initiator caspase is activated and subsequently cleaves/activates an effector caspase, which then cleaves and activates the next caspase and so on.
Caspase-3 is the most active effector caspase in both the intrinsic and extrinsic pathways, where it is processed and activated by caspase-9 and caspase-8, respectively. However, MCF-7 cells do not possess caspase-3 due to a 47-base pair deletion in the caspase-3 gene (9,10). Consequently, apoptotic cell death in MCF-7 cells must be independent of caspase-3 activation, although several studies have shown that apoptotic cell death in MCF-7 cells is accompanied by caspase activation. For example, in staurosporine-treated MCF-7 cells, recognized caspase death substrates including poly(ADP-ribose) polymerase, Rb, PAK2, gelsolin, and DFF-45 are cleaved (11). In TRAIL (tumor necrosis factorrelated apoptosis-inducing ligand)-induced apoptosis, caspase-8 is activated, which cleaves Bid to release tBid, which in turn induces cytochrome c release and caspase-9 and caspase-7 processing (12). Other studies have also shown that caspases 9, 6, 2, and 7 are cleaved/processed to their active forms (13). Thus, in the absence of caspase-3, MCF-7 cells can still activate a caspase cascade irrespective of whether the apoptosis is initiated via the intrinsic or extrinsic pathway. The intrinsic pathway, which is activated by many chemicals, including anticancer drugs, involves formation of the Apaf-1 apoptosome, a large caspase-processing complex (for review, see Refs. 6 and 7) that typically activates caspase-3. Apoptosome formation can be modeled in vitro in cell-free lysates by the addition of dATP or ATP and requires at least three apoptotic protease-activating factors (Apaf-1-3) (14 -16). The CARD (caspase recruitment) domain of Apaf-1 binds to a similar domain on procaspase-9, whereas the central CED-4 domain (98 -412) is involved in Apaf-1 oligomerization. In the presence of cytochrome c and dATP, Apaf-1 undergoes conformational changes, allowing it to oligomerize to form a very large apoptosome complex. Using gel filtration chromatography, we have isolated from both dATP-activated THP.1 and B chronic lymphocytic leukemia cell lysates two apoptosome complexes with apparent molecular masses of ϳ700 kDa and ϳ1.4 MDa (17)(18)(19). Furthermore, we have shown in apoptotic cells that the ϳ700-kDa complex predominates and is the most active complex in processing exogenous procaspase-3 (18,19). More recently, we have used a proteomic approach to characterize the composition of this ϳ700-kDa apoptosome complex and have shown that it contains solely Apaf-1 and caspase-9 as its core functional proteins (20).
Several studies have reported that caspase-7 is cleaved and activated in dATP stimulated lysates (17,21,22), and immuno-depletion studies indicate that caspase-9 is required for the processing of both caspase-3 and -7 (21). By implication, it has been assumed that the Apaf-1caspase-9 apoptosome complex directly processes procaspase-7 in an analogous manner to that described for procaspase-3. However, there is no direct evidence for this, and to address this, we have characterized the role of the apoptosome in caspase-7 activation in MCF-7 cells. Our studies show that the ϳ700-kDa apoptosome complex is also formed in dATP-activated MCF-7 cell lysates and apoptotic MCF-7 cells and directly cleaves and activates procaspase-7. However, although the kinetics of Apaf-1 oligomerization is normal, the recruitment and processing of caspase-9 in the holoenzyme complex is much slower than in caspase-3-containing cell lysates. We also show that caspase-7, after activation by the apoptosome complex, forms a XIAP-caspase-7 complex that is not disrupted even in the presence of Smac and Omi/HtrA2. These data suggest that the absence of caspase-3 enables XIAP to have an enhanced inhibitory effect on effector caspase activation and activity and, hence, a more significant anti-apoptotic role in MCF-7 cells.

EXPERIMENTAL PROCEDURES
Cell Culture, Apoptosis Assays, and Preparation of Control and Caspase-activated Cell Lysates-MCF-7-and MCF-7-transfected cells were cultured in RPMI 1640 supplemented with 10% fetal bovine serum and 2 mM Glutamax™ as described previously (12). For apoptosis assays, cells were harvested by mild tryptic digestion, and the percentage of apoptotic cells with exposed phosphatidylserine was determined by annexin V-fluorescein isothiocyanate binding (Bender Medsystems, Vienna, Austria) and fluorescence-activated cell sorter essentially as previously described (12). Loss of mitochondrial membrane potential (⌬⌿m) was measured with the lipophilic cationic fluorescent probe tetramethylrhodamine ethyl ester and a fluorescence-activated cell sorter scan analysis (23). For assaying cytochrome c release, cells were resuspended in cold phosphate-buffered saline and permeabilized with 0.025% digitonin, the cell cytosol and membrane fractions were prepared by centrifugation, and the cytochrome c content of the various fractions was then determined by SDS-PAGE and Western blotting (24). Cell lysates (100,000 ϫ g supernatants) from cells were prepared by freeze/thawing (F/T) 2 (7,18). In some experiments, to minimize mitochondrial breakage and release of pro-apoptotic proteins, MCF-7 cell lysates (digitonin and homogenization (Dig/Hom)) were prepared in an isotonic buffer (MSH), containing 210 mM mannitol, 70 mM sucrose, 5 mM Hepes, 1 mM EGTA, 1 mg/ml Pefabloc SCD, one protease inhibitor mixture tablet/10 ml (Roche Diagnostics), pH 7.4, using a modified digitonin/homogenization technique (20). Caspase activation in lysates (10 -15 mg/ml) was induced by incubation at 37°C for various times with 2 mM dATP/MgCl 2 plus or minus 2.0 M cytochrome c.
Fluorimetric Assays of Caspase and Caspase-activating Activity-DEVDase activity (i.e. primarily caspase-3 and -7) of lysates or column fractions was measured fluorimetrically with 200 l of assay buffer (20 M Ac-DEVD.AFC, 0.1% CHAPS, 10 mM dithiothreitol, 100 mM HEPES, and 10% sucrose, pH 7.0) using a Wallac Victor 2 1420 Multilabel counter (18). In some experiments the Ac-DEVD.AFC substrate concentration was increased to 100 M. The caspase processing/activating activity of soluble apoptosome complexes was assayed using purified recombinant procaspase-3 or procaspase-7 (20). Briefly, the method assays the DEVDase activity of the sample, which is a direct measure of how much procaspase-3/-7 has been processed and activated by the apoptosome. In addition, aliquots of the DEVD assay mixtures were diluted 1:1 with 2ϫ SDS loading buffer and analyzed for caspase-3/-7 processing by SDS-PAGE and Western blotting.
Fractionation of Cell Lysates by Gel Filtration-Lysates were fractionated by size-exclusion chromatography on either Superose-6 or Sephacryl S300 columns using a fast protein liquid chromatography (HR 10/30 column) protein purification system (Amersham Biosciences). Columns were eluted at 4°C with 5% (w/v) sucrose, 0.1% (w/v) CHAPS, 20 mM HEPES/NaOH, 5 mM dithiothreitol, and 50 mM NaCl, pH 7.0, calibrated with protein standards as previously described (18). Column fractions (0.5-2 ml) were analyzed for DEVDase activity and caspase-3 or -7 processing activity as described above. Aliquots of the column fractions were also analyzed by SDS-PAGE and Western blotting for Apaf-1, caspases 9, 7, and 3 and XIAP. For the Western blot analysis, column fractions 5-17 and 18 -30 were run on separate gels, and immunoblotting procedures were carried out in parallel. To ensure uniformity in signal response, the two blots from each column run were exposed simultaneously to Kodak X-Omat film. Appropriate fractions were pooled and concentrated with Vivaspin concentrators (Vivascience AG, Hannover, Germany) before immunoprecipitation experiments with the indicated antibodies.
Affinity Purification of Caspase-7 Complexes-Immunopurification of caspase-7 complexes was carried out using either an anti-active caspase-7 (Cell Signaling, New England Biolabs UK Ltd., Hertfordshire, UK) or anti-XIAP (clone 48, BD Biosciences Pharmingen) antibody covalently bound, respectively, to protein A and G Dyna Beads (20). Immunopurification of caspase complexes was carried out from MCF-7 cell lysates (ϳ30 mg/ml) that had first been precleared by incubating (1 h at 4°C) with 400 l of Sepharose protein A-coated beads/ml of lysate. The precleared lysates were then diluted to 15 mg/ml with assay buffer and dATP-activated for 2 h at 37°C before fractionation by gel filtration. Appropriate fractions containing active caspase-7 complexes were pooled and concentrated before adding antibody-tagged beads and roller mixing at 4°C for 4 h. Affinity purified proteins were eluted from the beads with SDS-PAGE sample loading buffer and separated by onedimensional SDS-PAGE. In some experiments apoptosome complexes were affinity-purified using an anti-caspase-9 antibody covalently attached to protein G Dyna Beads (20). Apoptosome complexes were captured by adding 20 l of cross-linked beads to 90 -100 l of dATPactivated MCF-7 cell lysate and roller mixing overnight at 4°C. Proteins were eluted from the beads as described above.
Reagents and Western Blot Analysis-Cell culture media and materials were as previously described (12,25). Most other reagents and cells, unless indicated otherwise, were obtained from published sources (20). MCF-7 cells stably transfected with caspase-3 (MCF-7/casp-3) or pcDNA3 vector only (MCF-7/vector) cells were a gift from Dr. Alan Porter, (National University of Singapore) and originally characterized elsewhere (9). Column, cell lysates, and immunopurified samples were analyzed for various proteins by SDS-PAGE and Western blotting as previously described (20). An antibody to procaspase-7 was obtained from BD Biosciences Pharmingen. Antibodies to active caspase-3 and -7 were sourced from Cell Signaling.

Caspase Activation in dATP/Cytochrome c-treated MCF-7 Cell-free
Lysates Does Not Require Caspase-3-In this study we wished to investigate apoptosome formation and function in MCF-7 cells. It was, therefore, important to characterize cytochrome c and dATP-dependent caspase activation in the cell lysates. This in vitro model has been used . Aliquots were also taken at the end of the incubations, diluted with 2ϫ SDS-PAGE loading buffer, and analyzed by SDS-PAGE/Western blotting (W.B.) for caspase-9 and -7 (20 g/lane) and caspase-3 (7.5 g/lane). In B, the time course of caspase-9, -3, and -7 processing was studied in cell lysates (10 mg/ml) prepared from MCF-7/pcDNA3 vector and MCF-7/casp-3-transfected cells. The cell lysates were incubated with dATP/ MgCl 2 and cytochrome c as described above, and at the indicated times, 10-l aliquots were removed, mixed with 90 l of SDS-PAGE loading buffer, and analyzed by SDS-PAGE/ Western blotting. Note in panel B that a longer exposure (15 min) was used to highlight the appearance of the p22 subunit of caspase-7. **, *, and † indicate protein bands non-specifically reacting with the antibody.
with a number of cell lines, and caspase-3-containing THP.1 cell lysates undergo a marked increase (ϳ30 -50-fold) in DEVDase activity (largely due to caspase-3 processing) when activated with dATP/cytochrome c (17). However, dATP/cytochrome c treatment of MCF-7 lysates produced only a very small increase in DEVDase activity ( Fig. 1A) even though SDS-PAGE/Western blot analysis showed that caspase-9 was processed to its p35/p34 forms and caspase-7 was cleaved to its p19 active form (lane 3, Fig. 1A). In contrast, MCF-7 cell lysates, activated with dATP and then incubated with recombinant procaspase-3, exhibited a marked increase in DEVDase activity that was accompanied by procaspase-3 processing to its p19 and p17 forms (lane 6, Fig. 1A). Caspase-3 markedly influenced the processing of procaspase-9, which was initially cleaved to the p37 (caspase-3 dependent) and p35 forms when the cell lysate was activated by dATP alone (lane 5, Fig. 1A) and then fully processed primarily to the p37 and p35 forms in the presence of dATP/cytochrome c (lane 6, Fig. 1A). Also, caspase-3 accelerated the processing of procaspase-7 to its fully cleaved p19 active form (lane 6, Fig. 1A). Interestingly, in the presence of recombinant procaspase-3, the p32 form of caspase-7 was detected when the cell lysate was activated with dATP alone (lane 5, Fig. 1A). This agrees with previous studies showing that the p32 subunit is formed by caspase-3 cleavage of procaspase-7, which removes the prodomain at Asp-23 (26). The modest increase in DEVDase activity observed in the dATP/cytochrome c-treated MCF-7 lysates could possibly be explained by the fact that caspase-7 has a higher K m and a lower k cat value for the DEVD synthetic peptide substrate (26 -28). The results also suggested that caspase activation in MCF-7 cell lysates was slower than in caspase-3-containing lysates. We investigated this by comparing dATP-dependent caspase activation in MCF-7 cells stably transfected with either pcDNA3 (vector) or pcDNA-casp-3. Although in MCF-7/vector cell lysates the p35 subunit was detected 5 min after dATP/cytochrome c activation, processing of the procaspase-9 was not complete until 120 min (lane 7, Fig.  1B). The p34 subunit was not detected until 60 min after dATP/cytochrome c activation, whereas in MCF-7/casp-3 cell lysates, procaspase-9 cleavage was accelerated and almost complete after 10 min (lane 12, Fig. 1B). Thus, the p35 subunit was detected at 5 min, reached a maximum at 30 min, and remained constant until 120 min before declining to barely detectable levels by 360 min. The p37 subunit was detected after the appearance of the p35 subunit at 10 min and was maximal after 30 min before disappearing by 60 -120 min.
The time course also showed that caspase-7 activation in MCF-7/ vector cell lysates was subsequent to caspase-9 processing, as the p19 subunit of caspase-7 was not detected until 30 min after treating with dATP/cytochrome c (lane 5, Fig. 1B). By this time there was already significant processing of procaspase-9 to its p35 form. The p22 subunit of caspase-7 (see 15-min exposure, Fig. 1B) was also detected, indicating that removal of the prodomain to produce the fully processed p19 form does not require caspase-3. In contrast, in caspase-3-containing cell lysates, caspase-7 processing was faster and more extensive than in the MCF-7/vector cell lysates and was paralleled by rapid and extensive processing of caspase-3 to the p20, p19, and p17 forms.
The ϳ700-kDa Apoptosome Complex Directly Processes Procaspase-7-We next investigated the ability of the ϳ700-kDa and ϳ1.4-MDa apoptosome complexes to directly activate effector caspases. Column fractions were incubated with procaspase-3, which is cleaved/activated by the Apaf-1-caspase-9 apoptosome. Caspase activation was determined by measuring DEVDase activity, which was predominant in fractions 9 -14 (Fig. 3A). These fractions corresponded to the ϳ700-kDa apoptosome complex, as determined by their Apaf-1 content ( Fig.  2A). Those fractions corresponding to the ϳ1.4-MDa apoptosome complex did not activate procaspase-3, and column fractions taken from fractionated non-activated cell lysates also did not activate this procaspase (Fig. 3A). From this we concluded that the ϳ700-kDa complex was the only active apoptosome complex. The fact that it processed and activated procaspase-3 indicated that it should also contain caspase-9. However, we were unable to detect caspase-9 in the column fractions ( Fig. 2A), as possibly the concentration of apoptosome was too low. To show that this was an Apaf-1-caspase-9 holoenzyme complex, we used an antibody to caspase-9 that only immunoprecipitates Apaf-1 when caspase-9 is a constituent part of an active apoptosome holoenzyme complex (20). The anti-caspase-9 antibody immunoprecipitated the p35 form of caspase-9 and Apaf-1 from the ϳ700-kDa complex ( Fig.  3B) but did not capture Apaf-1 from the ϳ1.4-MDa complex (Fig. 3B). We then incubated the respective apoptosome complexes with procaspase-7 (Fig. 3C) or procaspase-3 (data not shown) and assayed for DEVDase activity using 100 M Ac.DEVD.AFC (optimal concentration for assaying caspase-7 activity) and immunoblotted for active caspase-7 subunits. The ϳ700-kDa apoptosome complex directly processed and activated procaspase-3 (Fig. 3A) and procaspase-7 (Fig. 3C), as shown by the stimulation of DEVDase activity and the generation of the p19 subunit (Fig. 3C). The ϳ1.4-MDa complex had little or no caspase-processing activity with either procaspase-3 or -7. Thus, the ϳ 700-kDa apoptosome complex directly processes and activates procaspase-7. Interestingly, without caspase-3 it was still possible to generate the p19 subunit of caspase-7 (lane 4, Fig. 3C). Cleavage of the pro-domain or N-peptide of procaspase-7 at Asp-23 is believed to catalyzed by caspase-3. However, in caspase-3 null MCF-7/WT cells the ϳ700-kDa Procedures." Panel A shows in MCF-7/WT cell lysates oligomerization of Apaf-1 (un-oligomerized) to form ϳ1.4-MDa and ϳ700-kDa apoptosome complexes, with the attendant caspase-9 and -7 processing. The same analysis was carried out for cell lysates prepared from MCF-7 cells, stably transfected with caspase-3. Essentially similar elution profiles to the MCF-7/lysate were obtained for Apaf-1, caspase-9, and caspase-7. However, in MCF-7/casp-3 lysates, active caspase-3 was detected and eluted later than active caspase-7, demonstrating that active caspase-3 has a lower molecular mass than active caspase-7, which appears to be complexed to one or more other proteins (active caspase-7 complex. As in Fig.  1, **, *, and † indicate protein bands non-specifically reacting with antibody. apoptosome complex clearly catalyzes the formation of the p22 subunit of caspase-7, which then autocatalytically cleaves off the prodomain to produce the p19 subunit. Detection of the ϳ700-kDa Apoptosome Complex in Apoptotic MCF-7 Cells-To determine whether the ϳ700-kDa apoptosome complex is also formed in apoptotic MCF-7 cells, we treated MCF-7 cells with staurosporine, which is a well characterized inducer of cell death. Staurosporine induced apoptosis, as assessed by a time-dependent increase in annexin V binding, cytochrome c release, poly(ADP-ribose) polymerase cleavage, and a decrease in mitochondrial membrane potential (Fig. 4A). As expected, caspase-dependent poly(ADP-ribose) polymerase cleavage and annexin V binding were inhibited by the polycaspase inhibitor Z-VAD.fmk (Fig. 4A), whereas cytochrome c release was unaffected.
Next, we treated MCF-7 cells for 6 h with staurosporine and prepared cell lysates, which were analyzed by Superose-6 gel filtration chromatography and SDS-PAGE/immunoblotting. However, we could not detect Apaf-1 in the ϳ700-kDa and ϳ1.4-MDa apoptosome complexes (results not shown). This could be because the concentration of apoptosome complexes in apoptotic cell lysates is much less than can be achieved in the in vitro dATP-activated model system (17,19). Also, in vitro studies have shown that Apaf-1 and the apoptosome are cleaved and degraded by effector caspases (29). Thus, lysates obtained from apoptotic cells will contain a mixture of newly formed and partially degraded apoptosome complexes and, consequently, at any one time the concentration of the apoptosome in the lysate is likely to be low. Therefore, we pooled and concentrated the column fractions containing the apoptosome complexes and used an anti-caspase-9 antibody to immunoprecipitate the Apaf-1-caspase-9 apoptosome complexes.
Using this technique, we were able to detect the ϳ700-kDa apoptosome complex in lysates isolated from staurosporine-treated cells (lane 8, Fig.  4B). Furthermore, when we inhibited caspase activity with Z-VAD.fmk, increased amounts of Apaf-1 were immunoprecipitated from those fractions corresponding to the ϳ700-kDa complex (lane 9, Fig. 4B). Interestingly, in the presence of Z-VAD.fmk we detected both the proand p35 forms of caspase-9 bound to the ϳ700-kDa apoptosome complex. We were unable to detect the ϳ1.4-MDa apoptosome complex in apoptotic MCF-7 cell lysates (Fig. 4B).
XIAP Regulates the MCF-7 ϳ700-kDa Apoptosome Complex-Previously, in THP.1 cell lysates we have shown that the proapoptotic proteins Smac and Omi/HtrA2 regulate XIAP binding to the apoptosome (20). Thus, in Smac-and Omi/HtrA2-free lysates the apoptosome contains Apaf-1, caspase-9, caspase-3, and XIAP, whereas in the presence of Smac and Omi the apoptosome complex contains only caspase-9 and Apaf-1. MCF-7 cell lysates prepared by F/T contained substantial amounts of Smac and Omi/HtrA2 (Fig. 5A). In contrast, a combination of digitonin permeabilization and homogenization (Dig/Hom) produces cell lysates with very low or negligible levels of cytochrome c, Smac, and Omi/HtrA2 (Fig. 5A). These lysates still respond to dATP/ cytochrome c and process/activate recombinant procaspase-3 as shown by the increase in DEVDase activity (lane 6, Fig. 5B). However, F/T lysates exhibited a ϳ2-fold greater increase in procaspase-3 activating activity (lane 3, Fig. 5B), indicating that the caspase activating activity of the Dig/Hom lysates was inhibited. We, therefore, used the anticaspase-9 immunoprecipitation method to assess apoptosome composition in the two types of lysate. In the F/T lysates, apoptosome formation as determined by the ability of the anti-caspase-9 antibody to capture Apaf-1 was essentially maximal at 5 min and was accompanied FIGURE 4. Staurosporine-induced apoptosis in MCF-7/WT cells is accompanied by formation of an ϳ700-kDa apoptosome. MCF-7/WT cells were seeded at 2 ϫ 10 5 /well in a 6-well plate and cultured for 18 h before treatment. Three wells per treatment were used, and in the indicated experiments the polycaspase inhibitor Z-VAD.fmk (100 M) was added 1 h before induction of apoptosis with staurosporine (STS, 1 M). In A, the time course for the induction of apoptosis in MCF-7 cells was measured by the percentage of cells showing loss of mitochondrial membrane potential (⌬⌿m) and increased phosphatidylserine exposure as described under "Experimental Procedures." The Western blots (W.B.) show caspase-dependent cleavage of the apoptotic specific poly(ADP-ribose) polymerase protein substrate to the cleaved p85 form and release of cytochrome c (Cyto c) from the mitochondria (mitos) to the cytosol. In B, cells were seeded in large flasks (3 per treatment) at 5 ϫ 10 6 cells per flask and cultured for 18 h. The cells were then incubated for 1 h plus or minus Z-VAD.fmk (100 M) before inducing apoptosis with staurosporine (1 M). After 6 h the cells were harvested and resuspended in 150 l of cell lysis buffer, and cell-free lysates were produced by freeze/ thawing as described under "Experimental Procedures." The lysates were then separated by Superose-6 gel filtration as described in Fig. 3, and the fractions were analyzed for Apaf-1 and caspase-9 (data not shown) and caspase-7 (see also Fig. 6D) by SDS-PAGE and Western blotting. The fractions corresponding to the 1.4-MDa and ϳ700-kDa apoptosome fractions were pooled and concentrated to 110 l and then incubated overnight with anticaspase-9 antibody cross-linked to Dyna beads as described in Fig. 3. The washed beads were then eluted with SDS-PAGE loading buffer and analyzed for Apaf-1 and caspase-9 by SDS-PAGE and Western blotting. The input, supernatant, and elution volumes loaded onto the gel were equivalent to 6.0, 6.0, and 66.0%, respectively, of the pooled and concentrated fractions. I.P., immunoprecipitate.  (20 M) and an aliquot (20 g) was taken for the input sample; the remaining lysate (ϳ100 l) was incubated overnight at 4°C with 20 l of anti-caspase-9 complexed to Dyna beads. The supernatants were removed, and the beads were washed 4 times before eluting in 30 l of SDS-PAGE loading buffer, which was analyzed by immunoblotting for Apaf-1, caspase-9, and XIAP. The input, supernatant, and elution volumes loaded onto the SDS-PAGE gels were 1, 2, and 66%, respectively, of the pooled and concentrated fractions. In C, an additional immunoblot is shown for caspase-7, which was only detected in the eluate fractions from the Dig/Hom lysates and only when the blot (W.B.) was overexposed. The asterisk refers to a nonspecific band, which is mitochondrial Hsp60 (D. Twiddy, . G. M. Cohen, M. MacFarlane, and K. Cain, unpublished data) and is detected in the F/T but not the Dig/Hom lysates. In E, F/T and Dig/Hom lysates were fractionated as described in Fig. 2, and the fractions (15-22), which should contain the active caspase-7 complex, were analyzed for the p19 form of caspase-7. FIGURE 6. Active caspase-7 forms an ϳ200 -300-kDa complex that is distinct from active caspase-3. As described in Fig. 2, MCF-7/WT and MCF-7/casp-3 cell lysates (15 mg/ml) were activated with 2 mM dATP/MgCl 2 and cytochrome c (2 M) for 30 min at 37°C. The lysates were separated by Superose-6 gel-filtration chromatography, and column fractions were assayed for DEVDase activity. In A, the DEVDase activity for the MCF-7/WT and MCF-7/casp-3 fractions is shown as open and solid symbols, respectively. Unactivated (Control) and cytochrome c/dATP-activated (Cyt c/dATP) lysates are shown as E and F and as छ and ࡗ, respectively. Note the different y axis scales, respectively, for the MCF-7/WT and by processing of caspase-9 to form the p35 subunit (Fig. 5D). A similar time course for apoptosome formation was seen in the Dig/Hom lysates (Fig. 5C). Thus, the decreased caspase-processing activity of the Dig/ Hom lysates was not due to reduced apoptosome formation or caspase-9 processing. We next immunoblotted for XIAP, which interacts with the ATPF motif of the p12 subunit of caspase-9, generated by autocatalytic cleavage of procaspase-9 at Asp-315 (30). In Dig/Hom lysates, co-precipitation of XIAP with Apaf-1 and caspase-9 was detected within 1 min of initiating caspase activation with dATP/cytochrome c (Fig. 5C). Furthermore, the XIAP-Apaf-1-caspase-9 holoenzyme complex was stable for at least 30 min. In marked contrast, in F/T lysates, XIAP did not co-precipitate with Apaf-1 and caspase-9 (Fig. 5D).
We also immunoblotted for caspase-7 in both F/T and Dig/Hom lysates and were unable to detect significant amounts of caspase-7 in the apoptosome complex (results not shown). However, using heavily exposed immunoblots, we detected very small amounts of active caspase-7 associating with the apoptosome complex as isolated from Dig/Hom lysates (see the 30-min time point, Fig. 5C). Furthermore, the majority of the processed caspase-7 (p19 subunit) was detected in the supernatant fractions (Fig. 5C). Thus, although caspase-7 processing is abrogated in Dig/Hom lysates, it is clear that the cleaved form of caspase-7 is still not bound to the apoptosome. Therefore, we analyzed the Dig/Hom lysate by gel filtration (Fig. 5E) and found that small amounts of processed caspase-7 were instead associated with the ϳ200 -300 kDa complex (see Fig. 2). This is very different from caspase-3, which in the absence of Smac and Omi/HtrA2 binds to the apoptosome (20) by simultaneous binding to caspase-9 and XIAP (22). However, in the absence of Smac and Omi/HtrA2, XIAP binds to and is an integral component of the MCF-7 Apaf-1-caspase-9 apoptosome complex. Thus, our results show that caspase-7 is not stably associated with the apoptosome, irrespective of the presence of XIAP.
Processed Caspase-7 Elutes as an ϳ200 -300-kDa Complex-The predicted molecular mass of tetrameric active caspase-7 in dATP-activated and apoptotic cell lysates is ϳ60 kDa. However, the elution behavior of active caspase-7 in MCF-7 cell lysates indicated that it was complexed with other proteins (Fig. 2). In this respect previous studies have suggested that caspase-7 after activation can bind to other proteins and subcellular organelles (31,32). Our gel filtration results indicated that caspase-7 is cleaved/activated by the ϳ700-kDa apoptosome and then redistributes and binds with a protein(s) to form a ϳ200 -300-kDa complex (Fig. 2). In this respect caspase-7 appeared to be quite different from caspase-3, and to confirm this conclusion we analyzed column fractions from fractionated MCF-7/WT and MCF-7/casp-3 lysates for DEVDase activity. In dATP-activated MCF-7/WT lysates a small peak of DEVDase activity eluted in fractions 18 -21 (Fig. 6A). In contrast, the DEVDase activity of dATP-activated MCF-7/caspase-3 lysates was much higher (ϳ10 -15-fold) than the MCF-7/WT lysates and eluted in fractions 22-26 (Fig. 6A).
The ϳ200 -300-kDa complex was also detected in lysates obtained from apoptotic MCF-7/WT cells treated for 6 h with staurosporine (Fig.  6D). Control lysates did not contain this complex and, significantly, Z-VAD.fmk abolished caspase-7 processing and formation of the ϳ200 -300-kDa complex. Because the active caspase-7 complex was also formed in apoptotic cells, we wanted to carry out a proteomic analysis of this complex. Therefore, we used gel filtration chromatography to partially purify the complex, which was then immunoprecipitated with the active anti-caspase-7 antibody. The p19 form of caspase-7 was identified by liquid chromatography-tandem mass spectroscopy (data not shown) and immunoblotting (lane 6, Fig. 7A) in the complexes prepared from dATP-activated but not heat-activated lysates. However, mass spectrometry analysis of other captured proteins revealed that there were no significant differences between the heat-and dATP-activated samples (data not shown).
Identification of the ϳ200 -300-kDa Complex as a XIAP-Caspase-7 Complex-IAP proteins are potential candidates for binding to active caspase-7, and we therefore probed the ϳ200 -300-kDa caspase-7 complex for XIAP, cIAP1, and cIAP2. XIAP was not detected in the immunoprecipitated complex (lane 6, Fig. 7A), and cIAP1 and cIAP2, which were detected in the input fractions of both control and dATP-activated samples, were also not present in the eluate fractions (lane 6, Fig. 7A). However, some faint nonspecific bands (asterisks) were detected with the cIAP1 and cIAP2 antibodies in both the control and dATP-activated samples ( lane 5 and 6, Fig. 7A). Significantly, the intensity of these bands did not vary or correlate with the intensity of the p19 subunit of caspase-7. The cIAP1-nonspecific band runs very close to the expected position of cIAP1, so we performed a second pull-down using cell lysates. In this experiment (see Supplemental Fig. 1) we extended the running time on the SDS-PAGE and achieved a clear separation between the nonspecific band in the eluate and the cIAP1 band in the input and supernatant fractions.
It was possible that the anti-caspase-7 antibody that binds to a neoepitope in the active site could displace proteins (e.g. XIAP) which were binding to the active site of caspase-7. So we again purified the ϳ200 -300-kDa complex by gel filtration and then immunoaffinity-purified the complex with an antibody to XIAP. We then immunoblotted the pulldowns for caspase-7, XIAP, cIAP1, and cIAP2, and this showed that XIAP was complexed with caspase-7 but not with cIAP1 or cIAP2 (lane 6, Fig. 7B). Furthermore, Smac was not detected in the eluted fractions, and in the absence of dATP and cytochrome c, the XIAP antibody still pulled XIAP (Fig. 7B, compare lanes 5 with 6) but did not co-precipitate the p19 large subunit of caspase-7. Thus, after processing by the apoptosome, caspase-7 forms a stable complex with XIAP but not with cIAP1 or cIAP2.

DISCUSSION
In this study we have characterized the caspase-7-activating apoptosome complex in apoptotic MCF-7 cells and in dATP/cytochrome c-ac-tivated MCF-7 cell lysates. The rate of Apaf-1 oligomerization to form the ϳ700-kDa apoptosome complex is identical to that observed in caspase-3 containing THP.1 and Jurkat cell lysates (18,20,33). To form a fully active caspase-processing holoenzyme complex, caspase-9 must bind to the apoptosome. However, in MCF-7 cell lysates we had to immunoprecipitate caspase-9 from column fractions before we were able to detect caspase-9 in the apoptosome fractions ( Fig. 3B and 4B). This apparent absence of caspase-9 in the ϳ700-kDa apoptosome complex may simply be a detection problem due to relatively low levels of apoptosome complex in the MCF-7 cells. Alternatively it may imply that that the apoptosome complexes are heterogeneous, containing varying amounts of caspase-9. In this respect the structure of caspase-9 in the apoptosome complex is still poorly understood (for review, see Refs. 7,8,and 34). Procaspase-9, unlike effector caspases, has an unusually long flexible linker peptide which allows active site formation that gives the zymogen low but significant cleavage activity. This is markedly enhanced when caspase-9 is bound to the apoptosome in dATP/cytochrome c-activated cell lysates (18,35,36). Interestingly, non-cleavable caspase-9 mutants will also process procaspase-3 when incubated with dATP/cytochrome c-activated cytosol (22,35). This observation can be explained by the "induced-proximity model," which proposes that monomeric procaspase-9 dimerizes at high concentrations, forcing an allosteric rearrangement of one of the monomers in the dimer to produce a catalytically active site (37,38). It is suggested that Apaf-1 recruitment of caspase-9 facilitates caspase-9 dimerization/activation and that caspase-9 cleavage stabilizes the dimer (39). However, caspase-9 recruitment to the apoptosome is accompanied by simultaneous processing of the proform, as the p35 subunit was detected within 1-5 min of dATP/cytochrome c activation of MCF-7 cell lysates ( Fig. 1B and Fig 5,  C and D)). Significantly, caspase-7 processing was not detected until 30 min post-dATP activation and always lagged behind that of caspase-9 irrespective of the presence of caspase-3. Interestingly, in lysates isolated from apoptotic cells, Z-VAD.fmk seemed to freeze or stabilize procaspase-9 in the apoptosome. This suggests that caspase-9 processing destabilizes its binding to the apoptosome, resulting in the release of free caspase-9.
In the current study we provide new insights to the processing and structure of caspase-7 in cell lysates. Crystallographic studies on recombinant caspases 1, 3, and 7 have shown that the catalytically active enzymes are heterotetramers (40 -42) and at physiological concentrations form stable inactive homodimers (17,18,43,44). Cleavage of the linker peptide loop produces the large and small subunits of the heterotetramer, inducing conformational changes that form the functional active site. Although recombinant caspase-7 and caspase-3 elute on gel FIGURE 7. Caspase-7 in the ϳ200-kDa active caspase-7 complex binds to XIAP but not cIAP1 or cIAP2. In A, MCF-7/WT cell lysate (15 mg/ml) was activated with 2 mM dATP/MgCl 2 and cytochrome (Cyt) c (2 M) for 2 h at 37°C, and 3 mg of lysates were fractionated on a Superose 6 gel filtration column. The fractions corresponding to the ϳ200 -300-kDa active caspase-7 complex were pooled and concentrated to 100 l, and then 90 l were incubated overnight at 4°C with 20 l of anti-active caspase-7 antibody complexed to Dyna beads. The beads were washed 4 times and eluted in 50 l 2ϫ SDS-PAGE loading buffer. The inputs, supernatants (S'tants), and eluates corresponding to 6, 7, and 36% of the original pooled samples were then separated on SDS-PAGE and immunoblotted (W.B.) for caspase-7, XIAP, cIAP1, and cIAP2. I.P., immunoprecipitate. Similarly, in B, Sephacryl S300 gel filtration was used to partially purify the ϳ200-kDa active caspase-7 complex from 10 mg of activated lysates. Pooled fractions corresponding to the active caspase-7 complex were concentrated to 500 l and incubated for 4 h at 4°C with 150 l of Dyna beads complexed with an anti-XIAP antibody. The beads were washed 6 times and eluted with 75 l of SDS-PAGE loading buffer and analyzed for caspase-7, XIAP, cIAP1,cIAP2, and Smac. The input, supernatant, and elution volumes loaded onto the SDS-PAGE gels were 1, 1, and 13%, respectively, of the pooled and concentrated fractions. In A, proteins interacting non-specifically with the immunoblotting antibodies are indicated (asterisk). filtration columns with the predicted molecular mass of ϳ60 kDa (Ref. 17 and data not shown), our current study shows that a very different picture exists in cellular lysates (even if they contain Smac and Omi/ HtrA2). Furthermore, there are marked differences between caspases 3 and 7. First, the elution pattern of procaspase-3 and its processed form ( Fig. 2B and Fig. 6) is consistent with a molecular mass of ϳ60 kDa and agrees with studies on recombinant proteins. However, processing of procaspase-7 to produce the p19 form of caspase-7 is accompanied by a marked shift in the size of the cleaved protein, which elutes as an ϳ200-kDa complex ( Fig. 2A and Fig. 6). This complex was detected in both MCF-7/WT (casp-3 null) and MCF-7/casp-3 cells using two separate antibodies to caspase-7. Furthermore, immuno-affinity purification established that the ϳ200-kDa complex contained XIAP bound to the active form of caspase-7 (Fig. 7). The stability of this interaction seems to be very high as it was not disrupted by the presence of Smac or Omi/HtrA2 in the lysates.
The role of XIAP in mammalian apoptotic cell death is still controversial even though there is evidence that some apical and effector caspases are regulated by IAPs (45). In Drosophila, caspase inhibition by DIAP1 is essential for survival (46). However, the loss of XIAP in mammalian cells does not result in a significant phenotype (47), although overexpression of XIAP suppresses apoptosis induced by various stimuli (48). XIAP inhibits caspases 9, 3, and 7 via interactions between processed caspases and the BIR2 and BIR3 domains, and some IAPs can regulate effector caspases by binding to the catalytically active site. However, recent studies show that XIAP inhibits caspase-7 activity by multiple interactions between the N-terminal linker region of the BIR2 domain and the active site of caspase-7 and by binding to Ala-Asn-Pro, an IAP binding motif (IBM) that is revealed by cleavage at Asp-206 (49). In Drosophila, DIAP1 binds to an IBM located at the N terminus of the large subunit of Drosophila-related interleukin-converting enzyme (50), whereas cIAP1 can also bind to caspase-7 via an N-terminal Ala-Lys-Pro (AKP) IBM generated by cleavage of procaspase-7 at Asp-23. Thus, XIAP can potentially bind to the active site of caspase-7 and two separate IBM motifs. Removal of the prodomain at Asp-23 to reveal the AKP motif can be catalyzed by caspase-3 before Asp-198 cleavage by an apical caspase or granzyme B (51). However, we and others have found that active caspase-7 in vitro can also remove its own prodomain in dATPactivated MCF-7 lysates after cleavage at Asp-198 ( Fig. 1 (26)). Furthermore, recombinant procaspase-7 is processed by the ϳ700-kDa apoptosome complex to yield both the p19 and p22 forms of caspase-7 (Fig. 3C). Interestingly, the p22 form of caspase-7 was not found in the ϳ200 -300-kDa complex and instead eluted as a smaller-sized complex, indicating that it was probably not complexed with other proteins (fractions 22-24, Fig. 6B). This suggests that removal of the prodomain is required for forming the 200 -300-kDa complex. Thus, in the absence of caspase-3, apoptosome-dependent activation of caspase-7 involves cleavage at Asp-198 followed by autocatalytic cleavage at Asp-23. The exposed AKP (IBM) sequence could in theory bind cIAP1, but we could not detect this IAP in the immunoaffinity-purified ϳ200-kDa complex (Fig. 7A). Indeed, XIAP was the only IAP bound to caspase-7, and furthermore, an antibody that was raised to the active site of caspase-7 displaced XIAP in the immunoprecipitation experiments (Fig. 7A), demonstrating that XIAP interacts with active site of caspase-7. Although this confirms the presence of an active caspase-7/XIAP complex, it does not exclude the possibility that cIAP1/cIAP2 could also form separate complexes by binding to the active site of caspase-7.
The absence of caspase-3 in MCF-7 cells inevitably means that the response of these cells to apoptotic stimuli is markedly altered, as they now rely on caspase-7 to be the primary executioner caspase. In a recent study, 75% of breast tumors lacked caspase-3 transcript and protein expression (52). Thus, the MCF-7 cell line seems to be a good cell model for this particular disease, and significantly, reconstitution of caspase-3 in this cell line augments the apoptotic response of these cells to doxorubicin and other apoptotic stimuli (52,53). In this context we have delineated the activation and fate of caspase-7 during apoptosome-dependent cell death in caspase-3 null MCF-7 cells. In these cells, the rapid process of Apaf-1 oligomerization to form the ϳ700-kDa apoptosome complex is similar to caspase-3-containing cells. After binding to the apoptosome, procaspase-9 is cleaved to yield the processed p35 form of caspase-9. This process is slower in MCF-7/WT cells when compared with MCF-7/casp-3 cells. Similarly, in the absence of caspase-3, although the ϳ700-kDa Apaf-1-caspase-9 apoptosome complex catalyzes the processing of procaspase-7 to its active p19 form, it does so relatively slowly. Significantly, after processing, active caspase-7 does not associate with the apoptosome because, even in Smac/Omi-containing cells, it has such a high affinity for XIAP that it forms a relatively stable ϳ200-kDa XIAP-caspase-7 complex. This suggests that XIAP has a more pronounced anti-apoptotic effect in MCF-7 cells than it does in caspase-3-containing cells. The combination of slow processing/activation of caspase-7 and its strong binding to XIAP may explain in part why MCF-7 cells are relatively insensitive to apoptotic stimuli.