Caspase-2 Permeabilizes the Outer Mitochondrial Membrane and Disrupts the Binding of Cytochrome c to Anionic Phospholipids*

Caspases are cysteine proteases that play a central role in the execution of apoptosis. Recent evidence indicates that caspase-2 is activated early in response to genotoxic stress and can function as an upstream modulator of the mitochondrial apoptotic pathway. In particular, we have shown previously that fully processed caspase-2 can permeabilize the outer mitochondrial membrane and cause cytochrome c and Smac/DIABLO release from these organelles. Using permeabilized cells, isolated mitochondria, and protein-free liposomes, we now report that this effect is direct and depends neither on the presence or cleavage of other proteins nor on a specific phospholipid composition of the liposomal membrane. Interestingly, caspase-2 was also shown to disrupt the interaction of cytochrome c with anionic phospholipids, notably cardiolipin, and thereby enhance the release of the hemoprotein caused by treatment of mitochondria with digitonin or the proapoptotic protein Bax. Combined, our data suggest that caspase-2 possesses an unparalleled ability to engage the mitochondrial apoptotic pathway by permeabilizing the outer mitochondrial membrane and/or by breaching the association of cytochrome c with the inner mitochondrial membrane.

Apoptosis is executed by a family of cysteine proteases known as caspases. In vertebrates, these enzymes are present constitutively as zymogens and become activated in response to numerous apoptotic stimuli (1). Once activated, caspases cleave proteins after aspartate residues and thereby bring about many of the biochemical and morphological features characteristic of apoptosis. Of the two primary caspase activation pathways (death receptor-mediated and mitochondrially mediated), the mitochondrial pathway is most commonly associated with apoptosis triggered by cytotoxic stress and involves the activation of caspase-9 within the Apaf-1-containing apoptosome complex (2). Until recently, the dominant viewpoint has been that caspase-9 activation represents the apex of the caspase cascade during stress-induced apoptosis. However, this hypothesis is currently being challenged by evidence suggesting that caspase-2 may be critical for the engagement of the mitochondrial pathway, and hence caspase-9 activation, during DNA damage-induced apoptosis (3)(4)(5).
Although it was the first mammalian apoptotic caspase to be identified (6,7) and is the evolutionarily best conserved member of this family (8), the role of caspase-2 in apoptotic cell death remains contentious. Like caspase-9 it contains a caspase recruitment domain (CARD), but recent evidence indicates that caspase-2 activation occurs in an Apaf-1-independent complex, which also includes the death domain (DD)-containing protein PIDD and the adaptor protein RAIDD (9,10). Additional findings have shown that caspase-2 retains a unique ability to engage directly the mitochondrial apoptotic pathway (4,11), an effect that appears to require processing of the zymogen but not the associated catalytic activity (12).
The aim of the current study was to determine more precisely the molecular requirements for caspase-2-mediated engagement of the mitochondrial apoptotic pathway. The results indicate that processed recombinant human caspase-2 triggers a rapid and concentration-dependent release of cytochrome c from isolated mitochondria as well as liposomes. This occurs independently of other proteins and does not require a specific phospholipid profile within liposomes. In addition, our results demonstrate that caspase-2 has the capacity to detach cytochrome c from the inner mitochondrial membrane by compromising its association with anionic phospholipids.
Site-directed Mutagenesis-Active caspase-2 was used as a template to introduce a D316G point mutation using the QuikChange II sitedirected mutagenesis kit (Stratagene, La Jolla, CA). The mutation was confirmed by sequence analysis.
Cell Culture-Jurkat T-lymphocytes were cultured in RPMI 1640 complete medium supplemented with 10% (v/v) heat-inactivated fetal bovine serum, 2% (w/v) glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin in a humidified air/CO 2 (19:1) atmosphere at 37°C. Cells were maintained in a logarithmic growth phase for all experiments.
Isolation of Rat Liver Mitochondria-The liver of a male Sprague-Dawley rat was minced on ice, resuspended in 50 ml of MSH buffer (210 mM mannitol, 70 mM sucrose, 5 mM Hepes, pH 7.5) supplemented with 1 mM EDTA, and homogenized with a tight-fitting glass-Teflon motorized homogenizer. Homogenates were centrifuged at 600 ϫ g for 8 min at 4°C. The supernatant was decanted and centrifuged at 5,500 ϫ g for 15 min to form a mitochondrial pellet that was washed in MSH buffer, re-centrifuged at 5,500 ϫ g for 15 min, and resuspended in MSH buffer at a protein concentration of 80 -100 mg/ml. Caspase-2-induced Cytochrome c Release-Isolated rat liver mitochondria (1 mg/ml) were incubated in the presence or absence of recombinant caspase-2 in a final volume of 100 l. Incubations were performed either in buffer A, consisting of 140 mM mannitol, 46 mM sucrose, 50 mM KCl, 1 mM KH 2 PO 4 , 5 mM succinate, 1 mM EGTA, pH 7.4, or buffer B (where indicated), consisting of 150 mM KCl, 1 mM KH 2 PO 4 , 5 mM succinate, 1 mM EGTA, 5 mM Tris, pH 7.4. For caspase-2-induced cytochrome c release from Jurkat T-lymphocytes, cells (10 6 ) were washed in phosphate-buffered saline, resuspended in 100 l buffer A, and permeabilized with digitonin (5 g). Following a 1-min stabilization period, the indicated amount of recombinant caspase-2 was added.
Caspase-2-induced Release from Liposomes-Liposomes (100 g/ml) filled with cytochrome c or 20-kDa dextran-FITC were incubated with various amounts of caspase-2, mutant caspase-2, or caspase-3 in KCl buffer for 20 min at 22°C. After centrifugation at 100,000 ϫ g for 45 min, the amount of cytochrome c or dextran-FITC in the supernatant and pellet was evaluated. Cytochrome c release was monitored by Western blotting, whereas dextran release was monitored in duplicates of 150 l/well in a 96-well plate by recording the FITC fluorescence in a Fluoroscan II plate reader (Labsystems, Stockholm, Sweden) using an excitation wavelength of 485 nm and monitoring wavelength of 538 nm. KCl buffer was used as a blank. The extent of dextran-FITC release is reported as a percentage of the total after complete lysis of the liposome pellet with 1% CHAPS in KCl buffer.
Cytochrome c Binding to Sucrose-loaded Liposomes-Bovine heart cytochrome c (250 g/ml final concentration) was incubated with sucrose-loaded liposomes (100 g/ml) for 30 min at 22°C in KCl buffer. The liposomes were washed three times in KCl buffer to diminish nonspecific binding of cytochrome c to the phospholipids. Caspase-2 or mutant caspase-2 was added at the indicated concentrations, incubated for 20 min, and the supernatant was subsequently separated from the pellet by centrifugation at 100,000 ϫ g for 45 min. Cytochrome c binding and release were detected by Western blotting.
Estimation of Mitochondrial Respiration-Oxygen consumption by isolated rat liver mitochondria was measured using a Clark-type oxygen electrode (Yellow Spring Instrument Co.) at 25°C. Mitochondria with a respiratory control ratio (defined as the rate of respiration in the presence of ADP divided by the rate obtained following the expenditure of ADP) above 4 were used for all experiments. Fresh mitochondria were prepared for each experiment and used within 4 h.

RESULTS AND DISCUSSION
Caspase-2 Induces a Rapid and Concentration-dependent Release of Cytochrome c-We have shown recently that caspase-2 can trigger the release of cytochrome c and Smac/DIABLO from isolated liver mitochondria as well as permeabilized Jurkat T-lymphocytes (12). This effect appears to require processing of the zymogen but not the associated catalytic activity. To further examine the ability of caspase-2 to permeabilize mitochondria and stimulate cytochrome c release, we used wild-type (C-2), catalytically inactive (C303A), and uncleavable (D316G) recombinant caspase-2 proteins to treat permeabilized Jurkat cells, isolated liver mitochondria, or liposomes that had been loaded with either FITC-coupled 20-kDa dextran or bovine heart cytochrome c. Importantly, catalytically inactive caspase-2 (C303A) possessed no VDVADase activity (12), whereas uncleavable caspase-2 (D316G) retained ϳ20% activity relative to that of wild type, which is consistent with data reported previously (14).
When recombinant wild-type caspase-2 was added to permeabilized Jurkat cells (Fig. 1A) or isolated liver mitochondria (Fig. 1B), cytochrome c was released in a rapid and concentrationdependent manner. In contrast, neither catalytically inactive nor uncleavable caspase-2 was able to induce any cytochrome c release (Fig. 1C). Furthermore, only wild-type caspase-2 caused a significant decline in the rate of uncoupled respiration (V CCCP ) when mitochondria were incubated with m-chlorophenylhydrazone (CCCP) (Fig. 1D), an effect that was most likely due to the prominence of cytochrome c release in the presence of caspase-2 (Fig. 1C). These findings, combined with our earlier results demonstrating that the proteolytic activity of caspase-2 is dispensable for membrane permeabilization and/or the release of cytochrome c, suggested that fully processed caspase-2 was targeting directly the outer mitochondrial membrane, or contact sites between the inner and outer mitochondrial membranes, to cause the release of proapoptotic proteins.
Caspase-2 Causes Membrane Permeabilization of Dextran-or Cytochrome c-containing Liposomes-To test this premise and to determine more precisely the molecular requirements for caspase-2-induced membrane permeabilization, experiments were performed using dextran-containing, protein-free liposomes. Importantly, liposome-based methodology has been used by several groups to faithfully mimic the behavior of mitochondria (15,16). Here, liposomes of varying composition (100% PC; 20% PI, 80% PC; and 20% CL, 80% PC) containing FITC-coupled 20-kDa dextran were prepared and incubated with different amounts of recombinant caspase-2. As illustrated in Fig. 2A, recombinant caspase-2 induced a similar concentration-dependent release of dextran from all three liposome preparations, whereas caspase-3 and the two caspase-2 mutant proteins stimulated the release of only trace amounts of dextran.
In light of these findings, the next step was to determine whether caspase-2 would trigger a similar release of cytochrome c from liposomes. However, because cytochrome c preferentially binds to anionic phospholipids, notably CL located in the inner mitochondrial membrane (17,18), we conjectured that the binding of cytochrome c to CL-containing liposomes might be different. Therefore, to ensure equal loading and binding to the lipid membrane among all liposome preparations, different amounts of cytochrome c were used for liposomes containing an anionic phospholipid (PI or CL) as compared with liposomes containing only PC (50 g of cytochrome c/ml and 200 g of cytochrome c/ml, respectively). Treatment of liposomes with caspase-2 resulted in the same release pattern of cytochrome c as was observed for the dextran experiments. Specifically, low amounts of caspase-2 led to only trace amounts of cytochrome c release, whereas robust cytochrome c release was observed when liposomes were incubated with either 1000 or 1500 ng of caspase-2 (Fig. 2B). Notably, neither caspase-3 nor mutant caspase-2 was able to induce a release of cytochrome c from the liposomes (Fig. 2C), which is consistent with our previous findings with isolated mitochondria and permeabilized cells (12).
Caspase-2 Causes Detachment of Cytochrome c from the Outer Surface of Liposomes-Interestingly, we determined during the aforementioned experiments (Fig. 2) that liposomes with cytochrome c attached to the outer (as well as the inner) surface of the membrane, as a result of fewer washing steps,  would release their cytochrome c when incubated in the presence of lower, non-permeabilizing concentrations of caspase-2 (data not shown). As expected, this observation was especially pronounced in liposomes containing CL and could possibly be explained by caspase-2-mediated detachment of cytochrome c from CL on the outer surface of liposomes.
To further investigate this possibility, experiments were performed with sucrose-loaded liposomes that had cytochrome c bound to the outer surface only (Fig. 3). These liposomes were incubated with cytochrome c (250 g/ml) and thereafter washed gently with KCl buffer. This had the effect of decreasing the amount of loosely, non-specifically bound cytochrome c to the outer surface while allowing some retention of cytochrome c binding to the anionic phospholipids, PI and CL (Fig.  3, lanes 1, 3, and 7). Next, concentrations of caspase-2 that were below those required for permeabilization (cf. Fig. 2, A and B) were used to treat these liposomes (Fig. 3). The results indicated that caspase-2 caused the detachment of cytochrome c from liposomes containing either PI or CL (Fig. 3, lanes 4, 5,  8, and 9). Moreover, caspase-2 was slightly more efficient at detaching cytochrome c from the outer surface of CL-containing liposomes (Fig. 3), suggesting that caspase-2 may have a slightly higher affinity for cardiolipin as has been described previously for Bid (19,20).
A Low Concentration of Caspase-2 Augments Cytochrome c Release Induced by Digitonin or Bax-Because caspase-2 seemed to be disrupting cytochrome c binding to CL in liposomes, it was of interest to test this possibility using whole mitochondria. To that end, isolated rat liver mitochondria were incubated with a non-permeabilizing amount (10 ng) of caspase-2 and either recombinant oligomeric Bax or digitonin (Fig. 4, A and B). In both instances, the inclusion of 10 ng of caspase-2 led to significantly more cytochrome c being released, presumably due to caspase-2-mediated dissociation of the interaction between cytochrome c and CL. Importantly, the cytochrome c release data correlated nicely with the functional data obtained indicating that, although 10 ng of caspase-2 alone had no effect on respiration, its inclusion with digitonin resulted in a more severe decline in both state 3 and uncoupled (CCCP) respiration than was observed with digitonin alone (Fig. 4C). The rate of uncoupled respiration was partially restored upon inclusion of cytochrome c in the incubation medium.
Concluding Remarks-In the current study, we show that caspase-2 can permeabilize the outer mitochondrial membrane in a direct manner. Studies with liposomes revealed that this process requires neither other proteins nor a particular phospholipid composition of the liposomal membrane. Permeabilization appears to have been due to pore formation as incubation of liposomes with caspase-2 did not result in any apparent change in their size distribution (data not shown). This, combined with our previous finding that AIF was not released by treatment of isolated mitochondria or permeabilized Jurkat cells with caspase-2, whereas cytochrome c and Smac/DIABLO were, argues against a detergent-like effect of caspase-2 being responsible for the release of proteins from mitochondria (12).
Binding of cytochrome c to the inner mitochondrial membrane is known to involve a unique mitochondrial phospholipid, cardiolipin (17). There is also ample evidence that a decreased level of cardiolipin in the mitochondrial membrane, as a result of inhibited synthesis, accelerated break down, or oxidative stress, leads to impaired cytochrome c binding and increases the susceptibility of cells to undergo mitochondrially mediated apoptosis (see (21) for review). Furthermore, cardiolipin was shown to be a mitochondrial target for the proapoptotic Bcl-2 protein, Bid, and to play a critical role in mitochondrial perme-abilization/cytochrome c release by tBid/Bax (15,19,20,22). The cardiolipin precursor, phosphatidylglycerol, and other anionic phospholipids (e.g. phosphatidylinositol) can partially substitute for cardiolipin, but their association with cytochrome c is considerably weaker as compared with that of cardiolipin (18). Our results show that caspase-2 has the surprising ability to disrupt the association between cytochrome c and anionic phospholipids, notably CL, thereby making additional cytochrome c available for release into the cytosol/supernatant. This effect probably explains why considerably more cytochrome c is released from mitochondria treated with caspase-2 as compared with digitonin or recombinant oligomeric Bax. An alternate explanation might be that caspase-2 can stimulate unfolding of mitochondrial cristae in a manner resembling that of tBid (20).
Considering the number of pathways reportedly linking cytotoxic stress with mitochondrial engagement (see Ref. 23 for review), it would seem reasonable that the extent of the involvement of caspase-2 is likely to depend on a number of factors, including the nature of the cytotoxic stimulus, the cell type that is being affected, and the intracellular abundance of caspase-2 relative to that of other proapoptotic proteins (e.g. Bax/Bak). Our current data suggest that in some instances caspase-2 may be responsible for direct permeabilization of the outer mitochondrial membrane, while in other instances, caspase-2 may act in conjunction with Bax/Bak to amplify cytochrome c release by causing its displacement from cardiolipin. However, additional studies are needed to better understand how these different scenarios play out under physiological conditions.