Pro-caspase-8 Is Predominantly Localized in Mitochondria and Released into Cytoplasm upon Apoptotic Stimulation*

The recruitment and cleavage of pro-caspase-8 to produce the active form of caspase-8 is a critical biochemical event in death receptor-mediated apoptosis. However, the source of pro-caspase-8 available for activation by apoptotic triggers is unknown. In human fibroblasts and mouse clonal striatal cells, confocal microscopy revealed that pro-caspase-8 immunofluorescence was colocalized with cytochrome c in mitochondria and was also distributed diffusely in some nuclei. Biochemical analysis of subcellular fractions indicated that pro-caspase-8 was enriched in mitochondria and in nuclei. Pro-caspase-8 was found in the intermembrane space, inner membrane, and matrix of mitochondria after limited digestion of mitochondrial fractions, and this distribution was confirmed by immunogold electron microscopy. Pro-caspase-8 and cytochromec were released from isolated mitochondria that were treated with an inhibitor of the ADP/ATP carrier atractyloside, which opens the mitochondria permeability transition pore. Release was blocked by the mitochondria permeability transition pore inhibitor cyclosporin A (CsA). After clonal striatal cells were exposed for 6 h to an apoptotic inducer tumor necrosis factor-α (TNF-α), mitochondria immunoreactive for cytochrome c and pro-caspase-8 became clustered at perinuclear sites. Pro-caspase-8 and cytochrome c levels decreased in mitochondrial fractions and increased, along with pro-caspase-8 cleavage products, in the cytoplasm of the TNF-α-treated striatal cells. CsA blocked the TNF-α-induced release of pro-caspase 8 but not cytochromec. Internucleosomal DNA fragmentation started at 6 h and peaked 12 h after TNF-α treatment. These results suggest that pro-caspase-8 is predominantly localized in mitochondria and is released upon apoptotic stimulation through a CsA-sensitive mechanism.

Apoptosis (programmed cell death) originally referred to an active form of cell death with stereotypic morphological characteristics occurring during development (1). A broad range of pathological conditions can induce apoptosis (2,3). Unbalanced cell proliferation and apoptosis may play a role in pathogenesis of certain types of tumors and neurodegenerative diseases (4,5). A family of novel cysteine proteases, named caspase, plays an essential role in most, if not all, forms of apoptosis (6 -8).
Caspases are produced as pro-enzymes and become activated by proteolytic cleavage at internal aspartate residues upon apoptotic stimulation (9). Two categories of caspases important for apoptosis have been recognized: the initiators and executioners. The initiator caspases, which include caspase-8, -9, and -10, are activated in the earlier phase of apoptosis; the executioner caspases, which include caspase-3, -6, and-7, are activated by initiator caspases and are responsible for dismantling cells (10,11). Caspase-3 activation, a convergent event in apoptosis, is triggered by a variety of apoptotic stimuli. Caspase-3 cleaves many cytoskeletal proteins, such as fodrin, and proteins involved in DNA repair and fragmentation, such as poly-(ADP-ribose) polymerase and DNA fragmentation factor-45 (10,12,13). Two pathways lead to the activation of caspase-3 through release of cytochrome c and cleavage of pro-caspase-9 and through ligation of death receptors by tumor necrosis factor (TNF) 1 and Fas ligand. In the pathway stimulated by TNF and Fas ligand, pro-caspase-8 is recruited and activated by the adapter molecules FADD/MORT1 (14 -17). Caspase-8 then directly activates pro-caspase-3 and cleaves BID, a member of Bcl-2 family proteins (18 -20).
The role of mitochondria in apoptosis was first appreciated in a cell-free system when nuclear apoptotic events were induced only by cytoplasmic fractions enriched in mitochondria (21). A large body of evidence now suggests that many apoptotic stimuli affect the mitochondrial permeability transition pore (MPTP) and cause the release of pro-apoptotic molecules such as cytochrome c and Apaf-1 from mitochondria (22)(23)(24)(25)(26). Cytochrome c and Apaf-1 activate pro-caspase-9 to initiate an apoptotic cascade. Recent studies indicate that caspase-2, -3, and -9 are also released from mitochondria during apoptosis (23,(27)(28)(29). These findings suggest that mitochondria may be more broadly involved in apoptosis than previously thought. The subcellular distribution of pro-caspase-8 is not clear. Since knowledge about the localization of pro-caspase-8 could be important in understanding its role in apoptosis, we evaluated the subcellular localization of pro-caspase-8 in human fibroblasts and in a mouse clonal striatal cell line. The results suggest that pro-caspase-8 is predominantly found in the intermembrane space and inner membrane of mitochondria and can be released upon apoptotic stimulation.

EXPERIMENTAL PROCEDURES
Cell Culture-Human fibroblasts obtained from Coriell Cell Repositories were cultured in minimum Eagle's medium (Life Technologies, Gaithersburg, MD) supplemented with 15% fetal bovine, 100 units/ml penicillin G/streptomycin, 0.5 g/ml amphotericin B, 2 mM glutamine, 2ϫ final concentration minimum Eagle's medium vitamins, nonessential and essential amino acids (all from Life Technologies) at 37°C with 5% CO 2 . Clonal mouse striatal cells were cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum, 4.5 glucose, and 50 units/ml penicillin G/streptomycin (30).
Immunofluorescence and Confocal Microscopy-Clonal striatal cells or human fibroblasts were cultured on poly-L-lysine-coated microslips for 24 h. Immunohistochemistry was performed as described previously (31)(32)(33). To study the cellular localization of pro-caspase-8, cells were incubated with the rabbit polyclonal antibodies against pro-caspase-8 (SK-441, Ref. 32) or with the mouse monoclonal antibody against cytochrome c (PharMingen, San Diego, CA) for 24 h at 4°C. Microslips were washed and incubated with fluorescence-conjugated goat anti-rabbit IgG antibodies (or donkey against mouse IgG) (Molecular Probes, Eugene, OR). To study the effects of TNF-␣ on localization of procaspase-8 and cytochrome c, clonal striatal cells were treated with TNF-␣ (20 ng/ml) for 6 h and then processed for immunostaining as described above. The nuclei of these cells were stained with propidium iodide. Immunostained cells were examined with a confocal microscope (Bio-Rad 1024) and merged in Adobe Photoshop.
Immunogold and Electron Microscopy-Clonal striatal cells were cultured in 60-mm dishes, fixed, and incubated with polyclonal antibodies against pro-caspase-8 as described above and then with goldconjugated secondary antibody for 2 h at room temperature. Cells were post-fixed in 2.5% glutaraldehyde, incubated in 1% osmium tetroxide and 1% uranyl acetate, dehydrated in increasing grades of alcohol, and embedded in an ethanol-soluble epon (LX112, LADD). Embedded cells were sectioned (Ultracut E, Reichert-Jung) and examined with a JEOL 100CX electron microscope.
Isolation of Mitochondria, Mitochondrial Fractions, and Nuclei-Purification of mitochondria was performed using a Percoll gradient procedure described by Gasnier et al. (34) with minor modifications. Clonal striatal cells were harvested and rinsed in Hanks' balanced salt solution twice. Cells were suspended in 0.5 ml of buffer A containing 250 mM sucrose, 1 mM EDTA, 50 mM Tris-HCl, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, 0.28 units/ml apotinin, 50 g/ml leupeptin, and 7 g/ml pepstatin A (pH 7.4) and homogenized with a glass Pyrex microhomogenizer (30 strokes). The homogenate was centrifuged at 1000 ϫ g at 4°C for 10 min, and the resultant supernatant was transferred to a new tube and centrifuged at 10,000 ϫ g at 4°C for 20 min to obtain the mitochondrial pellet and supernatant. The supernatant was centrifuged at 100,000 ϫ g for 1 h at 4°C to generate the cytosolic fraction. The mitochondrial pellet was washed in buffer B containing 250 mM sucrose, 1 mM EGTA, 10 mM Tris-HCl (pH 7.4) three times to remove lysosome and microsome contamination. Crude mitochondria were further purified by layering the preparation on top of a gradient consisting of 2.2 ml of 2.5 M sucrose, 6.55 ml of Percoll solution (Amersham Pharmacia Biotech), and 12.25 ml of 10 mM Tris-HCl (pH 7.4), and 1 mM EDTA. The mixture was centrifuged at 60,000 ϫ g for 45 min. A dense band recovered from approximately two-thirds down the tube, corresponding to purified mitochondria, was removed with a Pasteur pipette, diluted with 1 ml of buffer B, and washed twice by centrifugation at 10,000 ϫ g at 4°C for 15 min each to remove the Percoll solution. The purity and integrity of mitochondrial preparations were validated by electron microscopy. Protein concentration was determined with BCA kit (Pierce).
The subfractionation of mitochondria was performed using a protocol described by Pederson et al. (35) with minor modifications. Briefly, crude mitochondria were resuspended in H-medium containing 220 mM D-mannitol, 70 mM sucrose, 20 mM HEPES, and 0.5 mg/ml bovine serum albumin (pH 7.4). The mitochondria suspension was mixed with equal volumes of H-medium containing 1.2% digitonin, incubated for 15 min at 4°C with mild stirring, and centrifuged for 15 min at 10,000 ϫ g after the addition of a 2-fold volume of H-medium. The resultant supernatant was centrifuged at 150,000 ϫ g for 1 h at 4°C to obtain the intermembrane space fraction (supernatant) and crude outer membrane fraction (pellet). The pellet fractions resulting from the 10,000 ϫ g centrifugation were washed once with H-medium, suspended in H-medium, son-icated for 2 min on ice, and centrifuged at 150,000 ϫ g for 1 h at 4°C. The resultant supernatants were used as the matrix fraction, and the pellet fractions were used as inner membrane fraction. Both outer and inner membrane fractions were solubilized with a solution containing 50 mM Tris-HCl, 0.5% Nonidet P-40, and 1 mM calcium chloride (TNC, pH 8.0). Protein concentrations in all fractions were determined by the BCA kit.
Proteinase K Digestion of Isolated Mitochondria-A protocol reported by Samali et al. (36) with modifications was used. Mitochondria were incubated in H-medium containing 0.2 g/ml proteinase K or in Hmedium containing 1% Nonidet P-40 plus proteinase K (0.2 g/ml) for 3 min at room temperature. Control mitochondria were incubated with H-medium only or H-medium with 1% Nonidet P-40. Proteolysis was terminated by adding an equal volume of TNC buffer containing 15 g phenylmethylsulfonyl fluoride, 2% SDS, 20 l of loading buffer, and boiling for 5 min. Samples were analyzed immediately by immunoblotting.
Determination of Pro-caspase-8 Release from Isolated Mitochondria-We followed the protocol of Susin et al. (29) with minor modifications. A crude mitochondrial fraction, prepared as described above, was suspended in a cell-free system (CFS) containing 220 mM mannitol, 68 mM sucrose, 2 mM NaCl, 2.5 mM KH 2 PO 4 , 0.5 mM EGTA, 2 mM MgCl 2 , 5 mM pyruvate, 2 g/ml phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 1 g/ml pepstatin A, 50 g of antipain, 1 mM dithiothreitol, 10 mM HEPES-NaOH (pH 7.4). The suspension was equally divided into several test tubes determined by the needs of the experiments. Release of pro-caspase-8 and cytochrome c was induced by the ADP/ATP carrier inhibitor atractyloside (Atr, 5 or 10 mM, Sigma). Atractyloside was dissolved in CFS and incubated with mitochondria for 1 h at room temperature. Separation of mitochondria and supernatant was achieved by centrifugation (10,000 ϫ g, 15 min). The supernatant was then further cleared by centrifugation at 100,000 ϫ g for 1 h. To assess whether release of pro-caspase-8 is through MPTP, cyclosporin A (CsA, 25 M, Sigma) was added together with Atr. After drug treatment, mitochondria suspensions were centrifuged at 10,000 ϫ g to pellet mitochondria. The supernatants were centrifuged again at 100,000 ϫ g. The levels of pro-and active caspase-8 and cytochrome c were measured by Western blot analysis in mitochondrial and supernatant fractions.
Determination of Pro-caspase-8 Release from Intact Cells-Clonal striatal cells were cultured in 60-mm dishes and treated with TNF-␣ (10 or 20 ng/ml, Sigma) after cells reached 80 -90% confluence. Control cells received 10 l of phosphate-buffered saline (vehicle for TNF-␣). Cells were harvested and washed in phosphate-buffered saline 6 h after drug administration. To study the effect of CsA on TNF-␣-induced release of pro-caspase-8, clonal striatal cells were cultured in 60-mm dishes and treated with TNF-␣ (20 ng/ml plus vehicle), CsA (10 M), or TNF-␣ plus CsA. Cells were harvested 6 h after drug administration and washed in phosphate-buffered saline. Crude mitochondria and cytosol were prepared as described above. Protein concentrations in cytosol and mitochondria were determined by BCA kit. Levels of pro-and active caspase-8 and cytochrome c were determined with Western blot analysis.
Western Blot Analysis-Protein extracts were loaded onto 4 -20% SDS-polyacrylamide electrophoresis gels (Bio-Rad) and exposed to electrophoresis for 1.5 h. Proteins were then transferred to nitrocellulose membranes, which were blocked and incubated in Tris-HCl-buffered saline with 0.1% Tween 20 (pH 7.4) containing 3% nonfat dry milk. Rabbit polyclonal antibodies against pro-caspase-8 (SK-441) or active caspase-8 (SK-440) (33) or monoclonal antibodies against cytochrome c, HSP70 (StressGen Biotechnologies Corp., Victoria, BC, Canada) or cytochrome oxidase IV (Molecular Probes) were applied at 4°C overnight. Membranes were washed three times for 5 min each with Tris-HCl-buffered saline with 0.1% Tween 20, and immunoreactivity was detected by enhanced chemiluminescence (ECL kit, Amersham Pharmacia Biotech) and visualized by autoradiography. Autoradiograms were analyzed by densitometry with image analysis software (Sigma-Scan Pro 4.0) after capturing the digital images with Hewlett Packard ScanJet 4C/T according to the manufacturer's protocol. The average density and area of each band were measured. The total values of pro-caspase-8, cytochrome c, actin, or cytochrome oxidase were calcu-lated by multiplying the area and average density of each band. The values of pro-caspase-8 and cytochrome c were then normalized to a loading control (actin for cytosolic fractions and cytochrome oxidase for mitochondrial fractions) and expressed as the mean Ϯ S.E.
DNA Fragmentation Assay-Clonal striatal cells were cultured in 60-mm dishes and treated with TNF-␣ (20 ng/ml) for 3-24 h. Cells were collected together with medium and centrifuged at 1000 ϫ g. Cell pellets were rinsed with ice-cold phosphate-buffered saline (pH 7.4) after removing the medium. Low molecular weight DNA fragments were extracted according to the protocol used by Chen et al. (37). DNA fragments were separated on 2% agarose gel (3:1 NuSive) and visualized with an UV transilluminator after ethidium bromide staining.

Immunohistochemical and Biochemical Evidence of Mitochondrial Localization of Pro-caspase-8 -
In the cytoplasm of fibroblasts and clonal striatal cells, pro-caspase-8 immunostaining displayed a rod-shaped or punctate pattern with very little or no diffuse staining. The labeling for pro-caspase-8 overlapped with the mitochondrial enzyme cytochrome c, as revealed by double immunofluorescence confocal microscopy ( Fig. 1). There was diffuse staining for pro-caspase-8 in some cell nuclei. The intensity of nuclear labeling was stronger in human fibroblasts than in clonal striatal cells (Fig. 1). The specificity of the pro-caspase-8 antibody was fully characterized previously (33) and was tested in our studies by omitting primary or secondary antibodies (data not shown).
To confirm the mitochondrial and nuclear localization of pro-caspase-8, protein extracts from crude mitochondria and Percoll gradient-purified mitochondria from clonal striatal cells and nuclei obtained from human fibroblasts were examined by Western blot. Results showed that pro-caspase-8 levels were strongly detected in both the crude mitochondria and Percoll gradient-purified mitochondria (Fig. 2, A and B), whereas the cytosolic fractions showed relatively low levels of procaspase-8. As a marker for mitochondria enrichment, cyto-chrome c or cytochrome oxidase was detected in the crude mitochondrial fraction and Percoll gradient-purified mitochondrial fraction, respectively. A similar enrichment of procaspase-8 was observed in purified mitochondria from COS-1 cells (data not shown). The purified nuclei also expressed readily detectable signal for pro-caspase-8 in Western blot (Fig.  2C). The purity of the nuclear preparation was verified by the enrichment of histone and the absence of tubulin, cytochrome c, and calnexin (data not shown).
Submitochondrial Localization of Pro-caspase-8 -To determine where pro-caspase-8 is localized in mitochondria, crude mitochondria were incubated with proteinase K in the absence or presence of the detergent Nonidet P-40 (1% final concentration). Results showed that proteinase K alone did not cause degradation of pro-caspase-8. In the presence of 1% Nonidet P-40, however, proteinase K degraded pro-caspase-8. In contrast, proteinase K alone caused partial degradation of heat shock protein (HSP70), a protein previously reported to be localized especially in the outer membrane and matrix (38 -40). When the detergent Nonidet P-40 was added, proteinase K completely degraded HSP70. Similar to pro-caspase-8, cytochrome c, a protein known to be localized inside mitochondria, was degraded by proteinase K in the presence of Nonidet P-40 but not by proteinase K alone (Fig. 3A).
The mitochondrial localization of pro-caspase-8 was further examined in mitochondrial fractions. Western blot analysis showed that pro-caspase-8 was detected in the intermembrane space, inner membrane, and matrix fractions (Fig. 3B). In several independent experiments, outer membrane always contained little or no pro-caspase-8, whereas inner membrane contained the highest levels of pro-caspase-8. Both intermembrane space and matrix contained intermediate levels of procaspase-8. Cytochrome oxidase and HSP70 were examined for the purpose of monitoring separation of outer and inner membranes. Consistent with the known distribution of these proteins, cytochrome oxidase was found only in the inner membrane fraction, whereas HSP70 was enriched in the outer membrane and in the matrix fraction (Fig. 3B).
The localization of pro-caspase-8 inside mitochondria was also examined with immunogold electron microscopy. Numerous gold particles were found inside mitochondria, but none were seen on the outer mitochondrial membrane (Fig. 4). Gold particles were absent when primary antibody was omitted (data not shown).
Release of Pro-caspase-8 from Isolated Mitochondria-To assess whether mitochondria release pro-caspase-8, crude mitochondria were suspended in the cell-free system in the presence or absence of Atr (5 or 10 mM), which opens the MPTP. After incubation, mitochondria and supernatant were separated by centrifugation. Pro-caspase-8 was modestly decreased in mitochondria and robustly detected in the supernatant after Atr treatment. Quantitative analysis of three independent experiments showed pro-caspase-8 levels in mitochondrial fractions decreased to 86.2 Ϯ 3% (5 mM) or 79.4 Ϯ 3.8% (10 mM) of control (nontreated mitochondria) and in the supernatants increased to 267.7 Ϯ 103 (5 mM) or 351.7 Ϯ 73 (10 mM) of control. Concurrently, cytochrome c was markedly decreased in the mitochondrial fraction and strongly detected in the supernatant after Atr treatment. In mitochondrial fractions, cytochrome c decreased to 57.3% Ϯ 15.1% (5 mM) or 34.8 Ϯ 8.9% (10 mM) of control (nontreated mitochondria), whereas in the supernatants the levels were 246.3 Ϯ 51.2% and 534 Ϯ 199%, respectively (Fig. 5, top).
To evaluate whether mitochondria release pro-caspase-8 through the MPTP, mitochondria were treated with Atr in the presence or absence of the MPTP inhibitor CsA (25 M). Quan- was also detected in the supernatant and was inhibited by CsA (Fig. 5, bottom).
TNF-␣ Effects on Localization of Mitochondria and Procaspase-8 -The effects of the apoptotic inducer TNF-␣ on mitochondria localization were determined by immunofluorescence and confocal microscopy. In untreated clonal striatal cells, cytochrome c and pro-caspase-8 immunoreactive mitochondria were distributed relatively evenly throughout the cytoplasm (Fig. 6, top panels, green label). Treatment with TNF-␣ for 6 h induced marked clustering, fusion, and peri-nuclear localization of mitochondria labeled for cytochrome c or procaspase-8. These changes appeared in cells with or without nuclear fragmentation, a feature of apoptosis. In some of the cells including those that were shrunken with fragmented nuclei, labeling for mitochondria with cytochrome c or procaspase-8 was severely reduced or absent, and diffuse staining for cytochrome c (Fig. 6, left middle and lower panels), or pro-caspase-8 was evident in the cytoplasm (Fig. 6, right middle and lower panels).
Release of Pro-caspase-8 from Mitochondria in Intact Cells-To evaluate mitochondrial release of pro-caspase-8 in intact cells under apoptotic conditions, clonal striatal cells were treated with TNF-␣ (10 or 20 ng/ml), for 6 h and then mitochondrial and cytosolic fractions were prepared. TNF-␣ caused a reduction in levels of pro-caspase-8 in mitochondria, whereas cytosolic levels of pro-caspase-8 levels were increased (Fig. 7A). Quantitative analysis of three independent experiments showed that in the mitochondrial fractions, pro-caspase-8 levels decreased to 78.6 Ϯ 5.2% (10 ng/ml) or 69.4 Ϯ 0.64% (20 ng/ml) of control (vehicle-treated cells), whereas in the cytosol, the levels were 140.4 Ϯ 23.7% and 331.9 Ϯ 34.5% of control, respectively. Cytochrome c levels in the mitochondrial fractions were decreased to 28 Ϯ 9.2% (10 ng/ml) or 18.1 Ϯ 3.1% (20 ng/ml) of control (vehicle-treated cells), whereas in the cytosol, cytochrome c increased to 603.2 Ϯ 362.5% (10 ng/ml) or 516.6 Ϯ 330.5% (20 ng/ml) of control. In the mitochondrial fraction, an active form of caspase-8 (p10) was detected in all cells but tended to rise in cells treated with TNF-␣. Active caspase-8 (p10 and p20) was almost undetectable in cytosolic fractions of control cells but readily detected in TNF-␣ treated cells.
To evaluate whether apoptosis was induced in clonal striatal cells after TNF-␣ administration, low molecular weight DNA was extracted, and internucleosomal DNA fragmentation was examined. Internucleosomal DNA fragmentation was barely   FIG. 2. Subcellular localization of pro-caspase-8. A, pro-caspase-8 in crude mitochondria (Mit). Clonal striatal cells were processed to obtain cytosol and crude mitochondria as described under "Experimental Procedures." A is a representative immunoblot from three independent experiments. Pro-caspase-8 was highly enriched in the mitochondria but low in the cytosol. Cytochrome c was high in the mitochondrial fraction but low in the cytosol. B, pro-caspase-8 in purified mitochondria. Crude mitochondria were purified with a Percoll gradient as described under "Experimental Procedures." B is a representative immunoblot from three independent experiments. Pro-caspase-8 was enriched in the mitochondria but was low in the cytosol. Cytochrome oxidase was only detected in the mitochondrial fraction but not in the cytosolic fraction. Cytochrome c was low in mitochondria after Percoll gradient purification; therefore cytochrome oxidase was used as a mitochondrial marker in these studies. C, pro-caspase-8 in purified nuclei. Human fibroblasts were processed to obtain crude homogenate, S1, P1, and N fractions as described under "Experimental Procedures." C is a representative immunoblot from three independent experiments. CH is crude homogenate; S1 is supernatant resulting from 2000 ϫ g centrifugation of crude homogenate. P1 is the pellet resulting from 2000 ϫ g centrifugation. N is the purified nuclear fraction obtained from P1 by iodixanol step gradients. Pro-caspase-8 is present in purified nuclei and in the P1 fraction from which nuclei were purified. 20 g of protein were loaded in each lane in all experiments.

FIG. 3. Mitochondrial localization of pro-caspase-8.
A, proteinase K degradation of pro-caspase-8 in crude mitochondria. Crude mitochondria were isolated as described under "Experimental Procedures." Mitochondria were incubated in CFS in the presence or absence of proteinase K (Pro K) or proteinase K plus Nonidet P-40. A is a representative immunoblot from four independent experiments. Proteinase K alone partially degraded HSP70 but had no effect on pro-caspase-8 and cytochrome c. In the presence of Nonidet P-40, proteinase K degraded pro-caspase-8, HSP70, and cytochrome c. B, pro-caspase-8 distribution in mitochondrial fractions. Mitochondria fractions were prepared as described under "Experimental Procedures." B is a representative immunoblot from four independent experiments. Procaspase-8 was found in the matrix and intermembrane space (IMS) and at its highest levels in the inner membrane fraction (IM). HSP70 was enriched in both outer membrane (OM) and matrix fractions. Cytochrome oxidase was only found in the inner membrane fraction. 20 g of protein were loaded in each lane. detectable at 6 h after TNF-␣ administration and clearly appeared 12-24 h after TNF-␣ treatment (Fig. 7B). Since release of pro-caspase-8 and cytochrome c occurred by 6 h, these results suggested that mitochondrial release of pro-caspase-8 and cytochrome c preceded the execution of apoptosis.
To evaluate whether the release of pro-caspase-8 in intact cells is dependent on MPTP, the effect of CsA on TNF-␣ induced pro-caspase-8 release was examined in clonal striatal cells. These studies showed that pro-caspase-8 and cytochrome c levels robustly increased in the cytosol and decreased in mitochondria. Coadministration of CsA markedly attenuated TNF-␣-induced increases in the level of pro-caspase-8 in the cytosol (Fig. 7C). Treatment with CsA alone had no effect on either cytosolic or mitochondrial levels of pro-caspase-8 and cytochrome c (data not shown). Quantitative analysis of three independent experiments showed that CsA attenuated the TNF-␣-induced decrease in pro-caspase-8 levels in mitochondria from 63.7 Ϯ 9.8% (TNF-␣ 20 ng/ml) to 82.3 Ϯ 9.1% (TNF-␣ ϩ CsA) of control (vehicle-treated cells). CsA inhibited the increase in pro-caspase-8 levels in the cytosol from 209.5 Ϯ 47.8% (TNF-␣) to 102.6 Ϯ 29.6% (TNF-␣ ϩ CsA) of control (vehicle-treated cells). However, CsA was ineffective in blocking the effect of TNF-␣ on cytochrome c. In the mitochondrial fractions, cytochrome c levels were 57.7 Ϯ 11.2% in TNF-␣treated cells and 44.6 Ϯ 11.7% of control (vehicle-treated cells) in TNF-␣ plus CsA-treated cells. In the cytosol, the levels of cytochrome c were 784.5 Ϯ 267.2% and 644.4 Ϯ 104.1%, respectively. DISCUSSION A current model for the activation of caspase-8 in apoptosis after stimulation by Fas and TNF-␣ involves the recruitment of cytosolic FADD/MORT1 and pro-caspase-8 to form a deathinducing signaling complex (41)(42)(43)(44). Pro-caspase-8 has the ability to activate itself. FADD/MORT1 brings into proximity two or more pro-caspase-8 molecules, which accelerates the self-cleavage and enhances caspase-8 activation (44,45). The subcellular localization of pro-caspase-8 has not been well defined until now, making it unclear how pro-caspase-8 becomes available to FADD/MORT1 during apoptosis. We showed that mitochondria are a major site for pro-caspase-8 localization and regulation during apoptosis. By immunofluorescent staining, immunogold labeling, and Western blot analysis, pro-caspase-8 was located primarily inside mitochondria. Mitochondrial levels of pro-caspase-8 and cytochrome c were unaffected by pro-teinase K treatment, again consistent with localization inside mitochondria. The presence of pro-caspase-8 in mitochondria was found in three cell lines (human fibroblasts, monkey COS cells, mouse clonal striatal cells). Zhivotovsky et al. (46) report low levels of pro-caspase-8 in mitochondria of apoptotic Jurkat cells and no pro-caspase 8 in normal Jurkat cells by Western blot. Although only biochemical assays were performed in their study, the discrepancy with our findings is still hard to explain. Other caspases including pro-caspase-2, -3, -6, and-9 have also been identified in mitochondria (27)(28)(29). Similar to procaspase-8, caspase-9 is enriched in mitochondria compared with cytosol (27).
Activation of caspase-8 requires interactions with death domains in proteins that are present in the cytosol (47,48). Thus sequestering pro-caspase-8 in mitochondria may be a safeguard mechanism. How pro-caspase-8 enters mitochondria remains unclear. There is a mitochondrial matrix-targeting signal at the N terminus of pro-caspase-9, but other mitochondrial FIG. 4. Immunogold electron micrographs of pro-caspase-8 localization. Clonal striatal cells were immunostained using polyclonal antibody against pro-caspase-8 and gold-conjugated secondary antibody. Two representative mitochondria are shown. Gold particles (arrows) appear inside mitochondria. No gold particles appear in association with mitochondrial outer membranes (arrowheads). Scale bar ϭ 1 m.
caspases, including pro-caspase-8, do not contain a typical mitochondria localization signal. The significance of nuclear procaspase-8 remains to be determined.
TNF-␣ is thought to activate caspase-8 by a mechanism involving recruitment of pro-caspase-8 by FADD/MORT1 (14 -17). TNF-␣ and Fas have also been shown to induce release of cytochrome c and activation of caspase-9 and -3 (49 -51). In some cell types, these effects can be inhibited by Bcl-2 family proteins (52)(53). Consistent with a previous report in L929 cells (54), we found that treatment of clonal striatal cells with TNF-␣ altered mitochondrial localization, which can affect mitochondrial function. Concurrently, TNF-␣ treatment reduced pro-caspase-8 in mitochondria and increased pro-caspase-8 and its active cleavage products in the cytoplasm. Levels of cytochrome c were reduced in the mitochondria and increased in the cytoplasm in the same TNF-␣-treated cells. These events preceded internucleosomal DNA fragmentation, indicating that they occur during the initiation of apoptosis. With both Atr and TNF-␣ stimulation, the decline in signal on Western blot for pro-caspase-8 in mitochondria fractions was smaller (15-30% decrease) than for cytochrome c (50 -80% decrease). However, the actual total protein change for pro-caspase-8 may be significant since mitochondria fractions have much higher concentrations of pro-caspase-8 than cytosol. The role of MPTP in the transport of mitochondrial proteins to the cytoplasm in apoptosis remains unclear. Many stimuli, including withdrawal of nerve growth factor, UV irradiation, calcium concentration surges, oxidative stress, and certain anti-tumor drugs can cause collapse of mitochondrial transmembrane potential and induce MPTP (55)(56)(57). When MPTP is induced, proteins and other molecules participating in apoptosis can move out of mitochondria. Despite an apparent size limitation (1500 kDa) for molecules to pass through the MPTP (22), a molecule larger than pro-caspase-8, namely apoptosis-inducing factor, can be released through a CsA-sensitive MPTP (58). In addition to pro-caspase-8, mitochondria release pro-caspases-9, -6, and -3 A is a representative immunoblot from three independent experiments. In mitochondria (Mit), TNF-␣ decreased pro-caspase-8 (to 78.6 Ϯ 5.2% (10 ng/ml) or 69.4 Ϯ 0.64% (20 ng/ml) of control) and cytochrome c (to 28 Ϯ 9.2% (10 ng) or 18.1 Ϯ 3.1% (20 ng) of control). In the cytosol (Cyto), TNF-␣ increased pro-caspase-8 (increased to 140.4 Ϯ 23.7% (10 ng/ml) or 331.9 Ϯ 34.5% (20 ng/ml) of control) and cytochrome c (increased to 603.2 Ϯ362.5% (10 ng) or 516.6 Ϯ 330.5% (20 ng) of control). Active forms of caspase-8 (mainly p20) were detectable in the cytosol after TNF-␣ treatment. Protein loading per lane was 10 g for mitochondrial and 20 g for the cytosol fractions. An additional band with a molecular mass of about 60 kDa was detected in the cytosol with the pro-caspase antibody, but the identity of this band was unknown. B, clonal striatal cells were treated with TNF-␣ (20 ng/ml), and low molecular mass DNA was extracted 3-24 h after drug administration. DNA fragments were separated on 2% agarose gel. B is a representative DNA ladder from three independent experiments. The density of cells at the beginning of treatment was the same. A typical apoptotic DNA ladder appears 12-24 h after TNF-␣ treatment. M, DNA marker; C, control. C, effects of cyclosporin A on TNF-␣-induced release of pro-caspase-8 in striatal clonal cells. Clonal striatal cells were treated with TNF-␣ (20 ng/ml) with or without CsA. Cells were harvested for isolation of mitochondria and cytosol 6 h after drug administration. C is a representative immunoblot from three independent experiments. In mitochondria (Mit), CsA inhibited TNF-␣-induced decrease in pro-caspase-8 (increased to 82.3 Ϯ 9.1% (TNF-␣ plus CsA) from 63.7 Ϯ 9.8% (TNF-␣) of control) but failed to inhibit a decrease in cytochrome c (57.7 Ϯ 11.2% (TNF-␣) and 44.6 Ϯ 11.7% (TNF-␣ plus CsA) of control). In the cytosol (Cyto), CsA inhibited a TNF-␣-induced increase in pro-caspase-8 (decreased to 102.6 Ϯ 29.6% (TNF-␣ plus CsA) from 209.5 Ϯ 47.8% (TNF-␣) of control). Again, CsA was ineffective in blocking a TNF-␣-induced increase in cytochrome c (784.5 Ϯ267.2% (TNF-␣) and 644.4 Ϯ 104.1% (TNF-␣ plus CsA) of control). (27,29); the mechanisms of release are not clear.
Isolated mitochondria released pro-caspase-8 when stimulated by Atr, an inhibitor of the ADP/ATP carrier and a stimulus for opening of the MPTP. The release of both pro-caspase-8 and cytochrome c was inhibited by CsA, a widely used MPTP blocker (59,60), suggesting that release of pro-caspase-8 in isolated mitochondria is mediated by a CsA-sensitive MPTP. In intact clonal striatal cells, CsA inhibited TNF-␣-induced release of pro-caspase-8 but not cytochrome c. The discrepancy in effects of CsA on cytochrome c release in isolated mitochondria and intact cells may relate to differences in preparations. The effects of Atr in isolated mitochondria are directly mediated by binding to ADP/ATP carrier, whereas in intact cells, TNF-␣ could have direct and indirect effects on mitochondria that cause release of pro-caspase-8 and cytochrome c through separate mechanisms. TNF-␣ can release cytochrome c in several cell types, but few studies have shown that this effect can be blocked by CsA treatment alone (61). Apoptotic death of mouse fibroblasts by TNF-␣ was inhibited by CsA in combination with the phospholipase A 2 inhibitor aristolochic acid, indicating involvement of multiple mechanisms (62). In our study, CsA failed to inhibit TNF-␣-induced apoptosis. One potential problem with using CsA is its toxicity in some types of cells (63). We found that clonal striatal cells developed intense internucleosomal DNA fragmentation 12 h after CsA administration alone (data not shown).
The classic model of pro-caspase-8 activation by TNF-␣ and death receptors involves recruitment of pro-caspase-8 from an unknown compartment by FADD/MORT1 through protein-protein interactions (Fig. 8). We speculate that TNF-␣-induced release of pro-caspase-8 from mitochondria could be mediated by different signaling pathways (Fig. 8). One pathway could involve TNF receptors through an as yet unknown mechanism. Another pathway could involve BID. Active caspase-8 cleaves cytoplasmic BID and causes BID to translocate to mitochondria. BID then induces release of pro-caspase-8. These possibilities will require further study. It is unclear whether the pro-caspase-8 released from mitochondria by TNF-␣ is activated directly or serves to replenish the small resident cytoplasmic pool of pro-caspase-8 depleted during apoptosis. A recent study showed that caspase-9 is enriched in mitochondria and released by Bax and ischemic insult (27). These findings together with our results on pro-caspase-8 suggest that mitochondria may be centrally involved in the activation of both initiator caspases (caspase-8 and -9) and provide a major intracellular site for integration and regulation of signaling events mediating cell death and survival.
Mitochondrial dysfunction has been implicated in certain neurological diseases, especially Huntington's disease (64). Expanded polyglutamine containing proteins including mutant huntingtin can recruit and activate caspase-8 (65). Preliminary studies in our laboratory show that the localization of mitochondria expressing pro-caspase-8 immunoreactivity is altered by increased expression of mutant huntingtin in the cytoplasm. Alterations in mitochondrial localization can affect the MPTP (54), thereby increasing the possibility for the mitochondrial release of pro-caspase-8 and subsequent activation of procaspase-8 by mutant huntingtin. Thus, the capacity for mitochondria to localize and release pro-caspase-8 may have important implications in Huntington's disease pathogenesis. Caspase-8 activates its target caspase, caspase-3. In this model, the source of pro-caspase-8 available to FADD/MORT1 is not known (solid arrows). Present results suggest that pro-caspase-8 is predominantly localized in mitochondria and is released upon TNF-␣ stimulation. We speculate that an additional pathway exists (dashed arrows) in which mitochondria are involved in death receptor-induced caspase-8 activation by releasing pro-caspase-8. It remains unclear, however, if the pro-caspase-8 released from mitochondria directly contributes to caspase-8 activation or replenishes a small pool of cytoplasmic pro-caspase-8. Signals that cause mitochondria to release procaspase-8 in response to TNF-␣ remain to be determined. One mechanism could involve BID, which can be cleaved by caspase-8 and translocate to mitochondria. Like BAX, BID induces release of cytochrome c from mitochondria upon its mitochondrial translocation. The other possible mechanism may involve signals generated directly from activation of death receptors. TNF-␣-induced release of cytochrome c can activate caspase-9, thus amplifying activation of caspase-3.