Caspase-2 induces apoptosis by releasing proapoptotic proteins from mitochondria.

Caspase-2 is one of the earliest identified caspases, but the mechanism of caspase-2-induced apoptosis remains unknown. We show here that caspase-2 engages the mitochondria-dependent apoptotic pathway by inducing the release of cytochrome c (Cyt c) and other mitochondrial apoptogenic factors into the cell cytoplasm. In support of these observations we found that Bcl-2 and Bcl-xL can block caspase-2- and CRADD (caspase and RIP adaptor with death domain)-induced cell death. Unlike caspase-8, which can process all known caspase zymogens directly, caspase-2 is completely inactive toward other caspase zymogens. However, like caspase-8, physiological levels of purified caspase-2 can cleave cytosolic Bid protein, which in turn can trigger the release of Cyt c from isolated mitochondria. Interestingly, caspase-2 can also induce directly the release of Cyt c, AIF (apoptosis-inducing factor), and Smac (second mitochondria-derived activator of caspases protein) from isolated mitochondria independent of Bid or other cytosolic factors. The caspase-2-released Cyt c is sufficient to activate the Apaf-caspase-9 apoptosome in vitro. In combination, our data suggest that caspase-2 is a direct effector of the mitochondrial apoptotic pathway.

Apoptosis is a form of cellular suicide that is essential for development and tissue homeostasis of all metazoan organisms. Caspases, a family of cysteine-dependent aspartate-directed proteases, play critical roles in initiation and execution of apoptosis (1). Determining the cellular processes, which lead to activation of caspases during apoptosis, and identifying the relevant intracellular caspase substrates and regulators have been the subjects of intensive investigation in the last several years.
Two relatively well-characterized apoptotic pathways have been identified. The first pathway is mediated by death receptors, such as Fas or Tumor Necrosis Factor (TNF) 1 receptor (2).
Specific adaptor proteins such as FADD (also known as MORT1) (3,4) or CRADD (also known as RAIDD) (5,6) bind to the ligand-bound receptor complexes. Interaction between the adaptor molecules and the prodomain of initiator caspases 2, 8, or 10 triggers sequestration-mediated autoactivation of these caspases (7)(8)(9). The activated initiator caspases in turn cleave and activate the downstream effector caspases 3, 6, and 7, inducing a cascade of caspase activation (10,11). In the second pathway, Cyt c is released from mitochondria to the cytoplasm in cells exposed to chemotherapeutic drugs, UV irradiation, growth factor withdrawal, or ligation of Fas and TNF receptors (12)(13)(14)(15)(16). Released Cyt c binds to Apaf-1 (apoptotic protease-activating factor-1) and promotes its oligomerization (17,18). Recruitment of procaspase-9 to this active apoptosome results in its autoactivation and subsequent activation of caspase-3 by the active caspase-9⅐Apaf-1 complex (18 -20). In both pathways activation of effector caspases by initiator caspases amplifies the apoptotic signal to ensure fast and irreversible cell death. In addition, there appears to be cross-talk between the death receptor and the mitochondrial apoptotic pathways. Caspase-8, an initiator caspase in the death receptor pathway, can connect death receptors to the core apoptotic machinery by directly cleaving downstream executioner caspases (11). Alternatively, it may activate the caspase cascade indirectly by cleaving Bid, a proapoptotic Bcl-2 family member. The cleaved Bid fragment translocates from the cytosol to the outer mitochondrial membrane resulting in disruption of the outer mitochondrial membrane and release of Cyt c from the mitochondrial intermembrane space (21,22). Engagement of the mitochondria-dependent pathway is more efficient than the mitochondria-independent pathway, as it only requires a small amount of active caspase-8 (23). Recent reports suggest that induction of Cyt c release from mitochondria is not only restricted to caspase-8. Effector caspases, caspase-3, -6, and -7, can also facilitate rapid Cyt c release from the mitochondria when other cytosolic factors are present (24).
Because of the critical role of mitochondria in apoptosis, it has been the subject of intensive research recently. Proapoptotic factors such as Cyt c (13,17), procaspases 2, 3, 9 (25, 26), apoptosis-inducing factor (AIF) (27), and second mitochondriaderived activator of caspases protein Smac (also known as Diablo) (28 -31) are safely sequestered within the mitochondrial intermembrane space in non-apoptotic cells. Upon apo-ptotic challenge, however, a rapid release of these factors through the outer mitochondrial membrane into the cytoplasm signals the initiation of the apoptotic process. The mechanism of release of these mitochondrial apoptotic factors into the cytoplasm remains unclear.
Caspase-2, initially described as Nedd-2/Ich-1, has been identified as a protein related to the Caenorhabditis elegans cell death protein CED-3 and mammalian interleukin-1␤-converting enzyme (caspase-1) (32). Two distinct caspase-2 mRNA species derived from alternative splicing encode two proteins, caspase-2L and caspase-2S. Overexpression of caspase-2L induces cell death, whereas overexpression of caspase-2S can antagonize cell death. Caspase-2L is the dominant isoform that is expressed in most tissues (32). The murine caspase-2 is highly expressed in mouse embryonic brain and down-regulated in adult brain (33). Mice carrying a null mutation for caspase-2 develop normally without severe phenotypic abnormalities. However, in caspase-2 knockout mice, sympathetic and motor neurons are more sensitive to death than neurons of wild type mice when deprived of nerve growth factor. In contrast, B lymphoblasts are resistant to apoptosis mediated by granzyme B and perforin. In addition, germ cells and oocytes are found to be resistant to cell death after treatment with chemotherapeutic drugs (34). Thus, for caspase-2 knockout mice, defects in apoptosis are cell-type and stimulus-dependent.
So far the mechanism by which caspase-2 can induce activation of the effector caspases is not fully understood. CRADD, an adaptor molecule for TNF receptor, recruits pro-caspase-2 to the death receptor pathway (5,6). It has been suggested that the prodomain of caspase-2 interacts with the CARD domain of CRADD (5,6), allowing the dimerization and autoactivation of procaspase-2. However, events subsequent to caspase-2 activation remain largely unknown. To understand how caspase-2 triggers the apoptotic pathway, we examined the possibility that caspase-2 employs the mitochondrial pathway to activate the caspase cascade. Our results indicate that physiological amounts of recombinant caspase-2 protein could directly provoke release of Cyt c and other proapoptotic factors from the mitochondria independent of any cytosolic protein and also through cleavage of Bid. The released proapoptotic factors trigger activation of the downstream caspases, thus setting into motion the apoptotic cascade.

EXPERIMENTAL PROCEDURES
Reagents-Anti-Cyt c antibody (7H8.2 c12, 6H2.B4) was obtained from BD PharMingen (San Diego, CA). Anti-caspase-2 antibody (directed toward amino acids 225-401) was purchased from Transduction Laboratories (Lexington, KY). Anti-AIF and anti-Bid were purchased from Santa Cruz Biotechnology Inc. Anti-caspase-3 antibody was purchased from BD PharMingen. Anti-human cytochrome oxidase subunit II (A-6404) was purchased from Molecular Probes. Anti-Smac monoclonal antibody was raised against mature Smac. VDVAD-afc was purchased from Enzyme Systems. Anti-citrate synthase polyclonal antibody was a gift from Yuri Lazebnik. Protease inhibitors were purchased from Roche Molecular Biochemicals. Protein concentrations were determined by the Bio-Rad assay kit (Hercules, CA).
Mammalian and Bacterial Expression Vectors-For apoptosis assays, we used the mammalian double-expression vector pRSC-LacZ, which allows expression of lacZ under the Rous sarcoma virus promoter, and the recombinant proteins (caspase-2 or CRADD) under the cytomegalovirus promoter. Constructs encoding CRADD, caspase-2, caspase-9-DN, Bcl-2, Bcl-xL, p35, and X-linked inhibitor of apoptosis protein have been described before (5,19,35,36). Bid wild type, cleavage site Asp-59 to Glu mutant, and double mutant (Asp-59 and Asp-75 to Glu) genes were excised from constructs generously given by Dr. J. Yuan (Harvard Medical School, Boston, MA) and then subcloned into a T7pcDNA 3 vector. For bacterial expression, cDNA was cloned in pET 21 or pET 28 (Invitrogen). Recombinant proteins with Cterminal or N-terminal His6 tags were expressed in BL-21(DE3) bacteria and purified on a resin of Ni 2ϩ affinity by standard affinity purification procedures as described previously (18).
Apoptosis Assays-MCF-7 cells were transiently cotransfected with pRSC-lacZ constructs in the presence or absence of different apoptosis inhibitors. The cells were stained with ␤-galactosidase 30 -36 h after transfection or 20 M z-VAD-fmk treatment and examined for morphological signs of apoptosis (36). Normal and apoptotic blue cells were counted by phase-contrast microscopy. The graphs depict the mean percentage of stained apoptotic cells as a fraction of the total number of blue cells after subtracting the percentage of apoptotic cell in the empty vector-transfected control cells (means Ϯ S.D.). Data provided represent the average of at least three individual experiments (n Ն 3 and S.D. Ϯ 1-4%).
Cleavage Assay-Wild type caspases and Bid and Bid mutants were in vitro translated in the presence of [ 35 S]methionine in rabbit reticulocyte lysate with a T7-RNA polymerase-coupled TnT kit (Promega), using the pRSC-LacZ, T7pcDNA 3 , or pET 28 constructs as templates according to the manufacturer's recommendations. Equal amounts of the translation reactions were diluted in 10 l of interleukin-1␤-converting enzyme buffer (25 mM HEPES, 1 mM EDTA, 5 mM dithiothreitol, and 0.1% CHAPS (pH 7.5)) and incubated with purified recombinant caspase-2 and caspase-8, respectively, at 30°C for 1 h. The reactions were fractionated by 10% SDS-PAGE and analyzed by autoradiography.
Mitochondrial Preparation and Cytosolic Extracts-HeLa cells were collected by centrifugation at 600 ϫ g for 10 min. The cell pellets were washed twice with ice-cold phosphate-buffered saline (pH 7.4) and resuspended with five volumes of buffer A (250 mM sucrose, 20 mM HEPES, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride, pH 7.5). The cells were homogenized in a glass Dounce homogenizer until ϳ60% of the cells became Trypan blue-positive (60 strokes for HeLa cells). The homogenates were centrifuged twice at 750 ϫ g for 10 min at 4°C. Supernatants were centrifuged at 10,000 ϫ g for 15 min at 4°C, and the resulting mitochondrial pellets were re-suspended in buffer A (13). In some cases mitochondria were further purified on Percol gradient (Amersham Biosciences, Inc.) as per the manufacturer's recommendations. The supernatants of the 10,000 ϫ g spin were further centrifuged at 100,000 ϫ g for 1 h at 4°C, and the resulting supernatants (designated S-100) were frozen as aliquots at Ϫ80°C for subsequent experiments.
In some experiments Percol gradient-purified mouse liver mitochondria were used (37).
In Vitro Assay of Cyt c Release-A 5-l aliquot of mitochondria (2 g/l) was incubated with 10 g of cytosolic extract (S-100) and recombinant caspase-2 or caspase-8, in a final volume of 25 l of buffer A at 37°C for 1 h. The reaction mixture was then centrifuged at 12,000 ϫ g for 10 min at 4°C to pellet the mitochondria. Supernatants and the pellets were subjected to SDS-PAGE and immunoblotting. Similar procedures were used to assay the release of Smac/Diablo and AIF.

The Mitochondrial Pathway Is Required for Caspase-2-and CRADD-induced Apoptosis and Caspase-3
Processing-To gain a better understanding of how caspase-2 and its adaptor molecule CRADD engage the death pathway, we investigated their apoptotic and procaspase-3 processing ability in the presence of various apoptosis inhibitors. Among the inhibitors used, Bcl-2 and Bcl-xL inhibit apoptosis by blocking release of Cyt c and other proapoptotic factors from the mitochondria, whereas caspase-9-DN (active site cysteine 287 to alanine) interferes with formation of a functional Apaf-1⅐caspase-9 complex by a dominant negative mechanism. As shown in Fig. 1A, CRADDand caspase-2-induced apoptosis in MCF7-Fas cells was efficiently blocked by Bcl-2, Bcl-xL, and caspase-9-DN (Fig. 1A), suggesting that they require the participation of the mitochondria and the Apaf-1⅐caspase-9 complex to trigger activation of the apoptotic pathway.
To determine whether overexpression of caspase-2 and CRADD can induce processing of procaspase-3, 293T cells were transiently transfected with CRADD and caspase-2 expression plasmids together with or without expression plasmids encoding Bcl-xL or caspase-9-DN. The cell lysates were immunoblotted with antibody against the processed caspase-3-p20 subunit. As expected, overexpression of caspase-2 and CRADD induced processing/activation of procaspase-3 (Fig. 1B, lanes 2 and 3). This processing was completely blocked when 4-fold excess of caspase-9-DN or Bcl-xL was coexpressed together with CRADD or caspase-2 ( Fig. 1B, lanes 5, 6 and lanes 8,9). These observations are consistent with the result in Fig. 1A, confirming the importance of the mitochondrial pathway in activation of the effector caspases by CRADD and caspase-2.
Comparison of the Protease Activities of Caspase-2 and Caspase-8 Toward Other Caspase Zymogens-Caspase-8 can process and activate most known caspase zymogens (procaspases) (11). To compare the relative activities of caspase-2 and caspase-8 toward other known caspase zymogens, recombinant caspase-2 and caspase-8, were expressed in bacteria and then affinity-purified. Expression of casapase-2 and -8 zymogens in bacteria allowed autoactivation of the two zymogens to generate the two processed subunits of the mature caspase-2 and caspase-8 (here referred to as caspase-2 and caspase-8) ( Fig. 2A). Equal amounts of caspase-2 and caspase-8 were incubated with a panel of in vitro-translated [ 35 S]methioninelabeled caspase zymogens, and the reactions were then analyzed by SDS-PAGE and autoradiography. Unlike caspase-8, caspase-2 was not able to process zymogens of effector caspases 3, 6, or 7, or initiator caspases 8, 9, or 10. However, caspase-2 was able to process its own zymogen and the baculovirus IAP inhibitor, p35 (Fig. 2, B and C). This suggests that caspase-2 could execute the apoptotic program independent of other effector or initiator caspases, or by activating another unknown caspase. Alternatively, caspase-2 could activate the caspase cascade indirectly via the mitochondrial pathway by causing release of Cyt c and other apoptotic factors into the cytoplasm.
Caspase-2 Induces Cyt c Release from the Mitochondria-To examine whether caspase-2 induces Cyt c release from the mitochondria, purified mitochondria from HeLa cells were incubated with purified caspase-2 in vitro with or without cytosolic extracts. Caspase-8 was used at the same time as a control. When mitochondria were incubated with cytosol alone, almost no Cyt c release was observed in the supernatant, indicating that the mitochondrial preparation was not releasing Cyt c nonspecifically (Fig. 3A, lane 2). However, when mitochondria were incubated with cytosol and increasing amounts of caspase-2, an increase in the amount of Cyt c in the supernatant was detected by immunoblotting (Fig. 3A, lanes  3-7). The same results were obtained with caspase-8 (Fig. 3A,  lanes 8 -12). The amount of Cyt c released into the supernatant by a fixed amount of caspase-2 or caspase-8 was also increased by increasing the amount of the S-100 cytosol (Fig. 3B). This suggests that both caspases are capable of inducing Cyt c release, presumably by proteolytically activating one or more cytosolic factors. Surprisingly, when the S-100 cytosol was omitted from the reaction mixture, caspase-2, but not caspase-8, was still able to induce Cyt c release (Fig. 3, A and  B, compare lanes 13 and 14), suggesting that one or more of the presumable cytosolic factors are dispensable for stimulation of mitochondria by caspase-2 but not by caspase-8.
Caspase-2 Can Cleave Bid-Bid is a proapoptotic Bcl-2 family member that triggers Cyt c release from mitochondria after proteolytic cleavage by caspase-8 (21,22). Careful examination of the amino acid sequence of Bid identified two potential caspase cleavage sites that match the preferred cleavage sites for caspase-2 (38). We examined the ability of caspase-2, compared with caspase-8, to cleave Bid. In vitro translated 35 S-  35 S-Labeled caspase zymogens were incubated with or without purified caspase-2 or caspase-8 at 37°C for 1 h as indicated. The reactions were stopped by addition of SDS sample buffer, and the products were analyzed by SDS-PAGE and autoradiography. C, 35 Slabeled caspase-2 zymogen and baculovirus p35 were incubated with or without purified caspase-2 or caspase-8 and then analyzed as in B.
labeled wild type Bid (WT), Bid D59E mutant (D59E), or Bid D59E/D75E double mutant (DM) were incubated with recombinant caspase-2 or caspase-8. As shown in Fig. 4A, both caspase-2 and caspase-8 were able to cleave WT Bid, but not the D59E or DM mutants, into two small fragments almost to the same extent. Based on these results, caspase-2 can efficiently cleave Bid at Asp-59, suggesting that caspase-2 induces Cyt c release by proteolytically activating Bid.
To determine whether Bid is necessary for caspase-2-or caspase-8-induced Cyt c release from the mitochondria, endogenous Bid was immunodepleted from HeLa cell S-100 extracts before incubating with caspase-2 or -8 and mitochondria. As expected, depletion of Bid from the S100 extracts totally abolished the Cyt c-releasing ability of caspase-8 (Fig. 4B, compare  lane 3 with lane 7). However, depletion of Bid decreased, but did not totally inhibit the Cyt c releasing ability of caspase-2 (Fig. 4B, compare lane 4 with lane 8). This suggests that Bid may not be critical for caspase-2-induced Cyt c release but its presence could enhance this caspase-2 activity.
Caspase-2 Can Directly Release Apoptogenic Factors from the Mitochondria-Cyt c, AIF, and Smac are normally localized in the intermembrane space of the mitochondria (27,30,31). It has been demonstrated by two-color immunofluorescence detection that AIF is released shortly before Cyt c from the mitochondria upon staurosporine stimulation (39). The released AIF translocates from the mitochondria to the nucleus where it induces caspase-independent apoptosis by causing chromatin condensation and large-scale DNA fragmentation in the nucleus. Smac is also released with Cyt c into the cytoplasm during apoptosis and functions to potentiate caspase activation by inhibiting IAPs (28 -31). To further confirm the ability of caspase-2 to directly release apoptogenic factors from the mitochondria, increasing amounts of purified caspase-2 were incubated with mitochondria in vitro, and the release of Cyt c, AIF, and Smac was determined. As shown in Fig. 5A, Cyt c, Smac, and AIF were released from isolated HeLa cell mitochon-dria into the supernatant in a dose-dependent manner by purified caspase-2. The direct release of AIF and Smac from the mitochondria into the cytosol by caspases has not been reported before. Cytochrome oxidase, an integral inner mitochondrial membrane protein, and citrate synthase, a soluble mitochondrial matrix protein, were not released by caspase-2 (Fig. 5A), indicating that the inner mitochondrial membrane remains intact after caspase-2 treatment. Combined, the above data suggest that caspase-2 specifically release soluble mitochondrial intermembrane space proteins only. Similar results were obtained with Percol gradient and highly purified mouse liver mitochondria (Fig. 5B), indicating that the effect of caspase-2 is not due to contaminating cytosolic factors or a specific effect on HeLa cell mitochondria. To further rule out the possibility that the direct effect of caspase-2 on the mitochondria is mediated by contamination of the mitochondrial preparation with cytosolic Bid and/or caspase-8, we analyzed the mitochondrial and corresponding cytosolic fractions by Western blot analysis with Bid-and caspase-8-specific antibodies. As shown in Fig. 5C, no Bid or caspase-8 bands were detected in the mitochondrial fraction. Interestingly, the proapoptotic Bcl-2 protein Bax was entirely present in the mitochondrial preparation (Fig. 5C, lane  1). The role of Bax in the direct release of the mitochondrial apoptotic factors by caspase-2 is currently unknown. Taken together, our results suggest that caspase-2 could stimulate both the Apaf-1⅐caspase-9 pathway and the caspase-independent AIF apoptotic pathway by directly releasing Cyt c, Smac, and AIF from the mitochondria.
The Released Cyt c Can Activate the Downstream Caspase Cascade-Cyt c released from the mitochondria by caspase-2 could induce activation of Apaf-1 and caspase-9, which in turn could activate the downstream caspases. To test this hypothesis we reconstituted an in vitro procaspase-3 processing sys-

FIG. 3. Caspase-2 and caspase-8 induce Cyt c release from isolated mitochondria.
A, freshly isolated HeLa cell mitochondria (10 g) were incubated with increasing amounts of caspase-2 or -8 in the presence (ϩ) or absence (Ϫ) of HeLa cytosol (5 g) at 37°C for 1 h. After incubation, the mitochondria were separated from the supernatants by centrifugation at 12,000 ϫ g for 10 min. The supernatants were subjected to SDS-PAGE (15%) and immunoblotting with anti-Cyt c antibody. The pellets were immunoblotted with anti-cytochrome oxidase (Cox) antibody. B, freshly isolated HeLa cell mitochondria (10 g) were incubated with caspase-2 or -8 (50 ng) and increasing amounts of HeLa cytosol at 37°C for 1 h. The reactions were then analyzed as in A.

FIG. 4. Caspase-2 induces Cyt c release by cleaving Bid.
A, determination of the caspase-2 cleavage site in Bid. 35 S-Labeled wild type or mutant Bid proteins were incubated with purified caspase-2 or caspase-8 at 37°C for 1 h as indicated. The reactions were stopped by addition of SDS sample buffer, and the products were analyzed by SDS-PAGE and autoradiography. B, HeLa S-100 extract (40 g) or HeLa S-100 extract immunodepleted with Bid antibody (40 g) were incubated with 5 l of freshly isolated mitochondria (10 g) in the absence (lanes 2 and 6) or presence of caspase-8 (100 ng) (lanes 3 and 7) or caspase-2 (100 ng) (lanes 4 and 8). After 1 h of incubation at 37°C, the samples were centrifuged at 12,000 ϫ g for 10 min at 4°C. The resulting supernatants and pellets were subjected to SDS-PAGE and immunoblotting with anti-Cyt c and cytochrome oxidase antibodies, respectively. Mitochondria incubated with tBid (lane 9) were used as a positive control.
tem. In this system, we incubated T7-tagged procaspase-3 C-A with purified Apaf-1 and procaspase-9 in the presence or absence of dATP and Cyt c. As shown in Fig. 5D, only after addition of increasing amounts of Cyt c, procaspase-3 was processed. Caspase-2 alone, or together with Apaf-1 and procaspase-9 could not promote processing of procaspase-3 (Fig.  5D, lanes 1-5). To determine whether Cyt c released from the mitochondria by caspase-2 could also induce procaspase-3 processing in this system, we incubated increasing amounts of caspase-2 with purified mitochondria and Apaf-1, procaspase-9, and dATP. As expected, caspase-2 induced processing of procaspase-3 in a dose-dependent manner, only in the presence of mitochondria (Fig. 5E, lanes 4 -6). These results demonstrate that caspase-2 can activate the downstream caspases via the mitochondrial amplification pathway.
Active Site Mutant Caspase-2 Is Unable to Induce Cyt c Release from Mitochondria-If caspase-2 protease activity is critical for its ability to induce Cyt c release from mitochondria, then we expect an active site mutant of caspase-2 to be inactive with respect to Cyt c release. To test this hypothesis, we expressed caspase-2 C-A (active site cysteine 303 to alanine) in bacteria and affinity-purified it. Unlike the WT caspase-2 zymogen, the inactive C-A mutant did not undergo processing in bacteria (Fig. 6A). When the same amount of caspase-2 and caspase-2 C-A was incubated with purified mitochondria, only the WT caspase-2 was able to induce Cyt c release in a dosedependent manner from the mitochondria into the supernatants (Fig. 6B, compare lanes 1-6 with lanes 7-12). The recombinant purified caspase-2 C-A used in the above experiment was not able to cleave the caspase-2 substrate VDVAD-afc (Fig.  6C). Taken together, these observations suggest that the protease activity of caspase-2 is necessary for caspase-2 to directly induce Cyt c release from the mitochondria.
Activated Caspase-2 Can Process Endogenous Procaspase-2 to Amplify the Apoptotic Signal-Procaspase-2 has been found in the cytoplasm, nucleus, Golgi complex, and mitochondria (40,41). It has been shown that, upon induction of apoptosis and PT (permeability transition) pore opening, caspase-2 is FIG. 5. Caspase-2 induces Cyt c release from isolated mitochondria and activates the downstream caspases. A, dose response of Cyt c, AIF, and Smac release by caspase-2 from isolated mitochondria. Freshly isolated HeLa cell mitochondria were incubated with increasing amounts of purified recombinant caspase-2 in buffer A at 37°C for 1 h. The samples were then centrifuged at 12,000 ϫ g for 10 min at 4°C. The resulting supernatants and pellets were fractionated by SDS-PAGE, and then the supernatants were Western-blotted with anti-Cyt c, AIF, Smac, citrate synthase (CS) and cytochrome oxidase antibodies. The corresponding pellets were Western-blotted with citrate synthase and cytochrome oxidase antibodies. B, Percol gradient-purified mouse liver mitochondria were incubated with increasing amounts of purified recombinant mature caspase-2 in buffer A at 37°C for 1 h and then analyzed by Western blotting with Cyt c antibody as described above. C, mitochondrial (M, lane 1) and cytosolic (C, lane 2) fractions were isolated from HeLa cells, and the two fractions were fractionated by SDS-PAGE and then Western-blotted with anti-Bax, cytochrome oxidase, Bid, caspase-8, and ␤-actin. D, purified recombinant T7-tagged procaspase-3 C-A was incubated with caspase-2 (100 ng), Apaf-1, caspase-9, and dATP with (lanes 6 -8) or without (lanes 1-5) increasing amounts of Cyt c (1, 10, 50 ng) for 1 h at 37°C as indicated. The reaction products were then analyzed by SDS-PAGE and immunoblotted with anti-T7-tag antibody. E, purified recombinant T7-tagged procaspase-3 C-A was incubated with Apaf-1, caspase-9, and dATP in the presence (ϩ) or absence (Ϫ) of freshly isolated HeLa cell mitochondria and increasing amounts of caspase-2 (1, 100, 200 ng) at 37°C for 1 h. The reaction products were then analyzed as in D. Lane 1, a control T7-tagged procaspase-3 C-A incubated with Apaf-1, caspase-9, dATP, and Cyt c. Lane 7, a control T7-tagged procaspase-3 C-A incubated with 200 ng of caspase-2-treated mitochondria. released from purified mitochondria and becomes activated (26). To determine whether active caspase-2 can process the endogenous procaspase-2 found in the mitochondria and cytoplasm, we incubated recombinant caspase-2 with purified mitochondria and cytoplasmic extract from three different cell lines, MCF-7, HEK293T, and HeLa cells. Endogenous mitochondrial and cytosolic procaspase-2 was cleaved to generate a major band of apparent molecular masses of 35 kDa (Fig. 7,  lanes 3, 4, 7, 8, 11, and 12). Some of the exogenous recombinant caspase-2 is seen as a p30 band that is smaller than the processed fragment of the endogenous proenzyme (Fig. 7, compare lanes 13 with the other lanes). Our data indicate that activated caspase-2 can directly process the endogenous mitochondrial and cytosolic procaspase-2. The ability of exogenous caspase-2 to process the endogenous mitochondrial procaspases-2, which is present in the intermembrane space (26), suggests that activated caspase-2 can translocate across the outer mitochondrial membrane. This suggests that, in addition to its ability to release mitochondrial intermembrane space proteins from the mitochondria, activated cytoplasmic caspase-2 could also activate mitochondrial procaspase-2 during apoptosis to further amplify the apoptotic cascade. DISCUSSION Two alternatively spliced caspase-2 isoforms, caspase-2L and caspase-2S, with opposing effects on cell death have been identified previously (32). In this study, we focused on the predominant proapoptotic caspase-2L isoform (referred to here as caspase-2) that is widely expressed in most tissues. Caspase-2 is activated in many cell types in response to various apoptotic stimuli, including growth factor withdrawal, DNAdamaging agents, TNF-␣, Fas ligation, and antigen receptor ligation (42)(43)(44)(45). In most of these cases procaspase-2 is directly cleaved by the effector caspase-3, suggesting a positive feedback amplification loop or a possible downstream involvement of this caspase in the apoptotic cascade (41,46,47). However, because procaspase-2 has characteristics of an initiator caspase, including a large N-terminal prodomain with a CARD and the ability to interact through this CARD with the TNF-R1-associated adaptor protein CRADD (5, 6), we postulate that caspase-2 is an initiator caspase acting upstream of the effector caspases. Supporting this notion, we have previously shown that caspase-2 activation occurs upstream of caspase-3 activation in the B-cell receptor signaling pathway leading to apoptosis (45). Moreover, recent studies demonstrated that caspase-2 is activated within 10 -15 min after infection with Salmonella or induction of anoikis followed by activation of the downstream effector caspases 3, 6, and 7 (48,49). These studies demonstrated also that cytochrome c release occurs shortly after activation of caspase-2 (48,49), providing additional evidence that activated caspase-2 acts upstream of the mitochondria and could contribute to the release of cytochrome c by acting directly or indirectly on the mitochondria. However, it was not evident from these studies whether the observed activation of the effector caspases, which followed activation of caspase-2, was dependent, or not, on cytochrome c release. In the present study, using a cell-free system we provided evidence that caspase-2 plays a role as an initiator caspase in the proteolytic caspase cascade by triggering the release of apoptogenic factors, including Cyt c from the mitochondria directly and indirectly through cleavage of Bid. We show that the released Cyt c is sufficient to activate the Apaf-1⅐caspase-9 complex, which in turns activates the effector caspases.
Caspase-2 is similar to caspase-8 in its ability to cleave Bid. However, unlike caspase-8, caspase-2 cannot process and activate the effector caspases directly. A recent study, utilizing a series of cell lines stably transfected with a caspase-2 antisense construct to analyze the role of caspase-2 in the death receptor pathway, provides evidence that down-regulation of caspase-2 significantly prevents Bid cleavage and abolishes Cyt c release and caspase-3 activation after Fas stimulation (50). Another study demonstrated that ectopically expressed caspase-2 could trigger the translocation of Bid to mitochondria and release of FIG. 6. The inactive caspase-2 C-A mutant does not induce Cyt c release from isolated mitochondria. A, expression of caspase-2 and caspase-2 C-A in bacteria. His 6 -epitope-tagged caspase-2 and caspase-2 C-A mutants were expressed in E. coli, purified on Talon Ni 2ϩ -affinity resin, and then analyzed by SDS-PAGE and Coomassie Blue staining. Lane M, molecular mass markers (kDa); lane Casp-2, Talon affinity-purified caspase-2; lane Casp-2 C-A, Talon affinity-purified caspase-2 C-A; Pcasp-2, procaspase-2; LS, large subunit; SS, small subunit. B, aliquots of freshly isolated HeLa cell mitochondria (10 g each) were incubated with increasing amounts of recombinant purified caspase-2 (lanes 1-6) or caspase-2 C-A (lanes 7-12) at 37°C for 1 h. The reaction mixtures were then centrifuged at 12,000 ϫ g for 10 min at 4°C. The resulting supernatants and pellets were subjected to SDS-PAGE and immunoblotting with anti-Cyt c and cytochrome oxidase antibodies, respectively. C, enzymatic activities of reaction B were correspondingly analyzed with the penta-peptide substrate VDVAD-AFC at the same time. RFU, relative fluorescent units. FIG. 7. Endogenous procaspase-2 is cleaved by activated caspase-2 in vitro. Mitochondrial (M) and cytosolic (C) fractions were isolated from HeLa, MCF-7, and HEK293T cell lines as described under "Experimental Procedures" and incubated with (lanes 3, 4, 7, 8, 11, and 12) or without (lanes 1, 2, 5, 6, 9, and 10) purified caspase-2 protein at 37°C for 1 h. The reaction mixtures containing the mitochondrial fractions were centrifuged at 12,000 ϫ g for 10 min at 4°C, and the pellets were washed in buffer A twice to remove the exogenous caspase-2. All reaction mixtures were then analyzed by SDS-PAGE and immunoblotted with anti-caspase-2 antibody. Lane 13 is the total input of caspase-2 alone.
Cyt c (51). Together with our data, the above evidence suggests that one way by which caspase-2 could activate the effector caspases after death receptor ligation or other apoptotic stimuli is through cleavage of Bid. Because translocation of Bid to the mitochondria is triggered by its cleavage and subsequent myristoylation at glycine 60 (52), the cleaved Bid could in turn translocate from the cytoplasm to the mitochondria causing disruption of the outer mitochondrial membrane and release of a number of mitochondrial apoptogenic factors (53,54). Among these factors, Cyt c activates the Apaf-1⅐caspase-9 apoptosome, whereas Smac inhibits IAPs and promotes further caspase activation. AIF, on the other hand, could translocate to the nucleus and induce large-scale DNA cleavage and caspaseindependent apoptosis (55).
Caspase-2 can also release Cyt c and other apoptotic factors from the mitochondria independent of Bid or other cytosolic factors. Three models have been proposed for the release of Cyt c from the mitochondria during apoptosis (56). One mechanism involves mitochondria swelling and physical rupturing of the outer mitochondrial membrane. However, this mechanism was challenged by electron microscopy studies, which revealed that apoptotic cells do not always show swelling of the mitochondria (57)(58)(59). Furthermore, the proapoptotic proteins Bid (21) and Bax (60 -62) can release Cyt c from isolated mitochondria in the absence of detectable mitochondrial swelling. The second mechanism involves formation of large ion channel in the mitochondrial membrane. This mechanism explains the activity of some Bcl-2 family members such as Bcl-2, Bcl-xL, Bid, and Bax, which can form ion channels in synthetic lipid membrane in vitro and regulate efflux of Cyt c (63)(64)(65). More recently, a third mechanism based on the mitochondria permeability transition (PT) pore was proposed to explain release of Cyt c from the mitochondria. PT pore is formed at the site of contact between mitochondrial inner and outer membranes and consists of the mitochondrial voltage-dependent anion channel (VDAC, also called porin), adenine nucleotide translocator, cyclophilin D, and other proteins (66). Through direct interaction with VDAC, Bax and Bak have been shown to induce Cyt c release by accelerating the opening of VDAC. Bcl-xL, on the other hand, prevents opening of the VDAC channel (67). However, tBid (C-terminal truncated Bid) or Bik can cause Cyt c release and cell death independent of PT pore opening (68). Thus, whether any of these mechanisms could explain the direct effect of caspase-2 on the mitochondria remains to be determined.
Nevertheless, the absence of Cyt c release from isolated mitochondria by inactive mutant of caspase-2 suggests that the caspase activity of caspase-2 is required for its direct Cyt creleasing activity. Exogenous mature caspase-2 can cleave the endogenous mitochondria procaspase-2 indicating that mature caspase-2 could freely translocate through the outer mitochondrial membrane. This raised the possibility that caspase-2 could regulate PT pore opening by cleaving VDAC. However, we find that caspase-2 does not bind to VDAC or cleave it into small fragments (Data not shown). We postulate that caspase-2 might cleave an outer mitochondrial membrane protein, perhaps a member of the Bcl-2 family, causing the release of Cyt c and other mitochondrial apoptotic factors. Supporting this hypothesis, our data clearly show that Bcl-2 or Bcl-xL can protect cells from apoptosis induced by overexpression of caspase-2, indicating that caspase-2 regulates the integrity of the outer mitochondrial membrane. Other preliminary data (not shown) also showed complete inhibition of Cyt c release when caspase-2 was incubated with mitochondria along with Bcl-xL protein.
In conclusion, our data demonstrate that caspase-2 can in-duce the release of several mitochondrial apoptotic proteins, indicating that caspase-2 induces apoptosis via the mitochondrial apoptotic pathway. However, the mechanism by which caspase-2 exerts its direct effect on the mitochondria remains to be elucidated.