Caspase Cleaved BID Targets Mitochondria and Is Required for Cytochrome c Release, while BCL-XL Prevents This Release but Not Tumor Necrosis Factor-R1/Fas Death*

“BH3 domain only” members of the BCL-2 family including the pro-apoptotic molecule BID represent candidates to connect with proximal signal transduction. Tumor necrosis factor α (TNFα) treatment induced a caspase-mediated cleavage of cytosolic, inactive p22 BID at internal Asp sites to yield a major p15 and minor p13 and p11 fragments. p15 BID translocates to mitochondria as an integral membrane protein. p15 BID within cytosol targeted normal mitochondria and released cytochrome c. Immunodepletion of p15 BID prevents cytochrome c release. In vivo, anti-Fas Ab results in the appearance of p15 BID in the cytosol of hepatocytes which translocates to mitochondria where it releases cytochrome c. Addition of activated caspase-8 to normal cytosol generates p15 BID which is also required in this system for release of cytochrome c. In the presence of BCL-XL/BCL-2, TNFα still induced BID cleavage and p15 BID became an integral mitochondrial membrane protein. However, BCL-XL/BCL-2 prevented the release of cytochromec, yet other aspects of mitochondrial dysfunction still transpired and cells died nonetheless. Thus, while BID appears to be required for the release of cytochrome c in the TNF death pathway, the release of cytochrome c may not be required for cell death.

Programmed cell death or apoptosis is critical for the successful crafting of multiple lineages, the maintenance of normal tissue homeostasis and a non-inflammatory response to toxic stimuli (1). A distinct genetic pathway apparently shared by all multicellular organisms governs apoptosis. The BCL-2 family of proteins constitutes a decisional checkpoint within the common portion of this pathway. Full members of the BCL-2 family share homology in four conserved domains designated BH1, BH2, BH3, and BH4 (2). The multidimensional NMR and x-ray crystallographic structure of a BCL-X L monomer indicated that BH1, 2, and 3 domains represent ␣ helices in close proximity which create a hydrophobic pocket presumably involved in interactions with other BCL-2 family members (3). Intriguingly, the BCL-2 family possesses pro-apoptotic (BAX, BAK, and BOK) as well as anti-apoptotic (BCL-2, BCL-X L , BCL-W, MCL-1, and A1) molecules (2). The ratio of anti-to pro-apoptotic molecules such as BCL-2/BAX determines the response to a death signal (4). A striking characteristic of many BCL-2 family members is their propensity to form homo-and heterodimers (5,6). The NMR analysis of a BCL-X L /BAK BH3 peptide complex revealed both hydrophobic and electrostatic interactions between the BCL-X L pocket and a BH3 amphipathic ␣-helical peptide from BAK (7). Deletions within BAK (8) and an extensive mutational analysis of BAX (9) argues that the BH3 domain serves as a minimal "death domain" critical for both dimerization and killing.
A divergent subset of the BCL-2 family possesses only sequence homology to the BH3 amphipathic ␣ helical domain. These "BH3 domain only" members include the mammalian BID, BAD, BIK, BIM, BLK, and HRK and EGL-1 of Caenorhabditis elegans. Of note, all of these molecules are pro-apoptotic lending credence to the thesis that BH3 represents a minimal death domain (10 -16). Where examined these BH3 domain only molecules are capable of heterodimerizing with classic BCL-2 family members. Mutagenesis of the BH3 domain of BID (10) and BAD (17) indicated that BH3 was essential for these interactions as well as the killing activity. Several of these molecules, BID and BAD lack the typical hydrophobic COOH-terminal sequence that is found in most BCL-2 family members, which for BCL-2 has been shown to function as a signal anchor segment required for its targeting mitochondria (18). Consistent with this BID and BAD have cytosolic as well as membrane based localizations (10,19). These characteristics suggested that BID and BAD may represent death ligands, sensors that receive death signals in the cytosol and translocate to membranes where they interact with membrane bound, classic BCL-2 members which serve as "receptors" (10). This was supported by the demonstration that cells in response to the survival factor interleukin-3, inactivated BAD by phosphorylation. This has the dual impact of dictating BAD's location as well as its binding partners. Phosphorylated BAD is sequestered in the cytosol bound to 14-3-3; whereas, only the active non-phosphorylated BAD heterodimerized with BCL-X L or BCL-2 at membrane sites to prevent cell death (19). Moreover, the demonstration that egl-1 regulates all the developmental deaths in C. elegans and maps upstream to ced-9 (16) argues that BH3 domain only molecules are evolutionarily conserved components of a central death pathway. This constellation suggests that such molecules are candidates to interconnect proximal signal transduction pathways with the distal death effector mechanisms based at intracellular membranes.
Recently, BAX despite possessing a hydrophobic COOH terminus has been noted in the soluble fraction of cells as well as mitochondrial membranes (20,21). Induced BAX expression (22) or the enforced dimerization of BAX (21) results in a downstream program of mitochondrial dysfunction as well as caspase activation. A physiologic death stimulus, the withdrawal of interleukin-3, results in the translocation of monomeric BAX from the cytosol to the mitochondria where it is a homodimerized, integral membrane protein (21). Perhaps all pro-apoptotic BCL-2 family members will prove to have inactive forms which undergo conformational changes as part of their activation.
The best characterized signal transduction pathways that mediate apoptosis are the cell surface receptors of the TNF 1 family, including CD95 (Fas/Apo-1) and CD120a (p55 TNF-R1) (23)(24)(25). Engagement of Fas/TNF-R1 receptor leads to formation of a protein complex known as the DISC (death-inducing signaling complex) (26 -28). This complex consists of Fas/TNF-R1, FADD (MORT1), and pro-caspase-8 (MACH/FLICE/Mch5). Once caspase-8 is recruited, it is processed and released from the complex in active form to activate the downstream "effector" caspases (26,29,30). The caspase family has been divided into three groups based upon sequence homology and substrate specificity using a positional scanning substrate combinatorial library (31). The specificity of caspases 2, 3, and 7 (DEXD) suggests they function at the effector phase of apoptosis. In contrast, the optimal sequence for caspases 6, 8, and 9 ((I/L/ V)EXD) resembles activation sites in the effector caspase proenzymes, arguing they represent "initiator" caspases.
Wang and colleagues (32) described a cell-free system of apoptosis, in which S100 extracts of untreated HeLa cells induced the activation of caspase-3 and DNA fragmentation upon addition of dATP. Further purification of the cytosol identified cytochrome c, which was released from the mitochondria during hypotonic lysis of the cells. Apaf-1, a mammalian homolog of CED-4, was a second factor isolated and required for caspase activation (33). Recently, it has been demonstrated that cytochrome c, Apaf-1, and caspase-9 form a complex that initiates a downstream caspase cascade (34). In addition, it was observed that when Xenopus egg cytosol was incubated with isolated mitochondria, cytochrome c was released, leading to the activation of caspases and nuclear apoptosis (35). The phenomena of cytochrome c redistribution from mitochondria to cytosol was also reported to occur in intact cells during apoptosis (36). However, the precise molecular mechanism responsible for the release of cytochrome c from mitochondria to cytosol during apoptosis remained unknown.

EXPERIMENTAL PROCEDURES
TNF/CHX Treatment and Western Blot Analysis-Cells were treated with recombinant mouse TNF-␣ (1 ng/ml; Sigma) and cycloheximide (1 g/ml; Sigma), and lysed at the indicated times in 50 mM Tris (pH 7.5), 150 mM NaCl, 1% Triton X-100 supplemented with a protease inhibitor mixture (Sigma, added at a 1:100 dilution). Lysates were separated by SDS-PAGE, and transferred to a polyvinylidine difluoride (Bio-Rad) membrane. The membrane was first blocked with 5% milk for 1 h, followed by incubation with primary and secondary antibodies for 1 h each, and finally developed with enhanced chemiluminescence (Amersham). A rabbit anti-mouse BID polyclonal antibody (10) was used at 1:1000 dilution, anti-cytochrome c monoclonal antibody (PharMingen) was used at 1:500 dilution, and anti-cytochrome c oxidase subunit IV antibody was used at 1:1000 dilution. The horseradish peroxidaseconjugated secondary antibodies (Caltag) were used at 1:2000 dilution.
Recombinant BID Preparation and Purification-Murine BID was cloned into pGEX-KG. Glutathione S-transferase-BID fusion protein was induced in BL21DE3 by 1 mM isopropyl-1-thio-␤-D-galactopyranoside. The bacterial pellet was resuspended in lysis buffer (1% Triton X-100, 1 mM EDTA, 1 mM dithiothreitol in phosphate-buffered saline) supplemented with a protease inhibitor mixture (Sigma, added at a 1:100 dilution), and sonicated. After centrifugation at 10,000 ϫ g for 20 min, the supernatant was applied to glutathione-agarose beads (Sigma). The beads were washed with buffer and treated with 10 units of thrombin per original liter. Cleaved BID was eluted from beads and the cleavage reaction was terminated by adding 50 g/ml N ␣ -p-tosyl-Llysine chloromethyl ketone. To remove the glutathione S-transferase protein and incompletely cleaved fusion proteins, the preparation was further purified on a Mono-Q column and the proteins were eluted with a NaCl gradient.
Cleavage of BID and NH 2 -terminal Sequence Analysis-Recombinant BID (5 g) was incubated for 2 h at 37°C with the soluble fraction of FL5.12 cells pretreated with TNF/CHX for 5 h. The proteins were lysed, separated by 16% SDS-PAGE, and transferred to a polyvinylidine difluoride (Bio-Rad) membrane. The membrane was first stained with Coomassie Blue and then destained with 80% methanol. The desired protein bands were cut out and subjected to NH 2 -terminal Edman degradation (37).
Subcellular Fractionation-FL5.12 cells were washed once in phosphate-buffered saline, resuspended in isotonic HIM buffer (200 mM mannitol, 70 mM sucrose, 1 mM EGTA, 10 mM HEPES, pH 7.5) supplemented with a protease inhibitor mixture (Sigma, added at a 1:100 dilution), and homogenized using a Polytron homogenizer (Brinkmann Instruments) at setting 6.5 for 10 s. Nuclei and unbroken cells were separated at 120 ϫ g for 5 min as the low speed pellet (P1). This supernatant was centrifuged at 10,000 ϫ g for 10 min to collect the heavy membrane pellet (HM). This supernatant was centrifuged at 100,000 ϫ g for 30 min to yield the light membrane pellet (LM) and final soluble fraction (S). For subcellular fractionation of mouse hepatocytes, cells were homogenized and separated by differential centrifugation as described below for preparation of mitochondria from mouse liver.
Mitochondria from Mouse Liver-For isolation of intact mitochondria, the liver from one mouse was minced and washed in ice-cold HIM buffer (supplemented with 2 mg/ml de-lipidated bovine serum albumin). The minced liver (ϳ2 g wet weight) was gently homogenized in 6 ml of HIM buffer in a 15-ml Wheaten Dounce glass homogenizer using two complete up and down cycles of a glass "B"-type pestle. The homogenate was diluted 6-fold with HIM buffer and centrifuged at 4°C for 10 min at 600 ϫ g in a Sorvall SS34 rotor. The supernatant was recovered, centrifuged at 7,000 ϫ g for 15 min, and the pellet resuspended in twice the original homogenate volume in HIM buffer without bovine serum albumin. After centrifuging at 600 ϫ g, mitochondria were recovered from the supernatant by centrifuging at 7,000 ϫ g for 15 min. The mitochondrial pellet was suspended in 0.5 ml of MRM buffer (250 mM sucrose, 10 mM HEPES, 1 mM ATP, 5 mM sodium succinate, 0.08 mM ADP, 2 mM K 2 HPO 4 , pH 7.5) at a concentration of 1 mg of mitochondrial protein per ml, and adjusted to 1 mM dithiothreitol just before use (38).
Protein Import-For a standard import reaction, 60 l of the soluble fraction of FL5.12 cells or hepatocytes was incubated with 10 l of mitochondria in MRM buffer (1 mg of protein/ml) at 37°C for 30 min. This import reaction was centrifuged at 10,000 ϫ g for 10 min to pellet the mitochondria. Both the pellet and the supernatant were analyzed by Western blot. For alkali extraction, the mitochondrial pellet was resuspended in freshly prepared 0.1 M Na 2 CO 3 (pH 11.5), and incubated for 30 min on ice. The membranes were subsequently pelleted in an ultracentrifuge (Beckman) at 75,000 ϫ g for 10 min and both the pellet and the supernatant were analyzed by Western blot.
For the BID depletion experiments, the soluble fraction of FL5.12 cells was incubated with anti-BID Ab for 1.5 h on ice. The Ab complexes were captured with protein A beads for 1 h and removed by centrifugation, and the procedure was repeated. The resulting BID-depleted supernatant was used in the protein import reaction. For the recombinant caspase experiments, the soluble fraction of FL5.12 cells was incubated with recombinant caspase-8 or -3 (1 g/60 l; PharMingen) at 37°C for 1 h and then used in the protein import reaction.

TNF␣/Cycloheximide Treatment of FL5.12 Results in Mitochondrial Dysfunction, Caspase-mediated Cleavage of BID, and
Cell Death-As part of a survey of the effects of death stimuli on the subcellular localization and post-translational modification of BCL-2 family members, we examined the response of the early hematopoietic cell line FL5.12 to TNF␣. Most nontransformed cells are resistant to TNF unless treated with a protein synthesis inhibitor (e.g. cycloheximide) which presumably eliminates a short half-life survival molecule (39). Treatment of FL5.12 with a combination of TNF␣/cycloheximide (TNF/CHX) resulted in a rapid reduction in the mitochondrial transmembrane potential (⌬⌿ m ) as assessed by the cationic, lipophilic dye dihexyloxacarbocynine iodide (DiOC 6 (3)) (Fig.  1A). The production of ROS such as superoxide as measured by hydroethidine and cell death as determined by propidium iodide dye exclusion, followed closely (Fig. 1A). Both the mitochondrial dysfunction and cell death were blocked by pretreatment with the broad caspase inhibitor, zVAD-fmk (Fig. 1A).
Western blot analysis of whole cell lysates prepared from FL5.12 cells treated with TNF/CHX revealed that the intracellular pro-apoptotic molecule p22 BID was cleaved to yield a major p15 and minor p13 and p11 fragments (Fig. 1B). The 2B4 T cell hybridoma which is also killed by TNF and displays mitochondrial dysfunction (not shown) also demonstrated the p15 fragment (Fig. 1B). Pretreatment of cells with 50 M zVADfmk markedly inhibited BID cleavage in FL5.12 cells and to a large extent in 2B4 cells (Fig. 1B).
Identification of the Cleavage Sites in BID-To determine the cleavage sites in BID, recombinant murine BID (rBID) was incubated for 2 h at 37°C with the S100 fraction of TNF/CHXtreated FL5.12 cells, S100(TNF) (Fig. 2A). Following the reaction the mixture was size fractionated by PAGE followed by Coomassie Blue staining. The S100(TNF) caused complete cleavage of p22 rBID (lane 3) which was inhibited by the inclusion of 50 M zVAD-fmk in the reaction mixture (lane 4). p22 rBID cleavage generated a major p15 and minor p13 fragment. Incubation of rBID with either recombinant active caspase-8 or caspase-3 also generated the p15 fragment (not shown). NH 2 -terminal peptide sequence analysis of these fragments revealed that p22 rBID was cleaved between amino acids Asp 59 and Gly 60 to generate p15 and between amino acids Asp 75 and Ser 76 to generate the p13 fragment (Fig. 2B). These fragments comigrated precisely with the upper two fragments detected in TNF-treated cells ( Fig. 1B and data not shown), arguing that intracellular BID is also cleaved at these sites.
To determine whether the third cleavage site responsible for the less abundant p11 seen in FL5.12 cells (Fig. 1B) was the predicted Asp 98 residue we utilized a D98A mutant. 35 S-Labeled in vitro translated BID and BID(D98A) were incubated with S100(TNF) from FL5.12 cells and analyzed by SDS-PAGE. Cleavage of wild type BID generated the three cleavage prod-

p15 BID Targets Mitochondria and Releases Cytochrome c
ucts seen in vivo, whereas cleavage of BID(D89A) generated only the p15 and p13 fragments (not shown). Taken together, BID is cleaved after three Asp residues located at positions 59, 75, or 98 generating three fragments (p15, p13, and p11) (Fig.  2B).
TNF/CHX Treatment Leads to Accumulation of p15 BID in Mitochondria as an Integral Membrane Protein and Release of Cytochrome c-To assess the location of intracellular BID, we disrupted FL5.12 cells using isotonic lysis conditions which kept mitochondria intact with a retained outer membrane. A substantial portion of p22 BID was consistently in the soluble S100 fraction (S) representing the cytosol as well as the mitochondria-enriched HM fraction as documented by the mitochondrial markers (cytochrome c, intermembrane space; cytochrome c oxidase, inner membrane) (Fig. 3A, lanes 1-4). The low speed pellet (P1) comprised of residual whole cells, nuclei, and some mitochondria, also displays BID.
At 5 h following TNF/CHX treatment the p15 BID fragment was often still present in the cytosol but was predominantly in the mitochondrial HM fraction (Fig. 3A, lanes 5-8). By 7 h p15 BID was almost exclusively in the mitochondria. The p13 and p11 minor fragments were associated exclusively with the mitochondrial fraction (not shown). In addition, following the TNF/CHX death stimulus, most of the cytochrome c was released from the mitochondrial HM fraction, found either in the S100 fraction or presumably as part of membrane fragments in the LM fraction (Fig. 3A, middle panel).
To assess the membrane association of p22 BID and its cleavage products, the mitochondria (HM fraction) from TNF/ CHX-treated FL5.12 cells were incubated in hypotonic buffer, alkaline buffer, or in high salt. The mitochondrial pellet (P) was separated from the supernatant (S) by high speed centrifugation. p22 BID was sensitive to all three treatments (Ͼ50% found in the supernatant (Fig. 3B)); whereas, p15, p13, and p11 were markedly resistant to these treatments (Fig. 3B) indicative of an integral membrane position.
Cytosolic p15 BID Targets Mouse Liver Mitochondria While p22 BID Does Not-To assess whether the p15 BID fragment can target mitochondria, the cytosol of FL5.12 cells 5 h after TNF/CHX treatment, S100(TNF) (Fig. 3C, lane 2) was incubated with purified, intact mitochondria from mouse liver (Fig.  3C, lane 1) in a standard protein import reaction. At 37°C Ͼ90% of p15 BID but Ͻ10% of p22 BID targeted mitochondria (Fig. 3C, lanes 3 and 4). Moreover, the targeted p15 but not p22 was resistant to alkali extraction (lanes 9 -12), indicating that p15 BID was now an integral membrane protein. Targeting of p15 BID did not occur at 4°C (lanes 5 and 6). Moreover, the inclusion of zVAD-fmk in the reaction did not inhibit targeting of pre-existing p15 (lanes 7 and 8) arguing that this event does not require an additional caspase cleavage at the mitochondria.
Targeting of Cytosolic p15 BID to Mitochondria Is Required for the Release of Cytochrome c-The cleavage of cytosolic BID by TNF-induced caspases and the targeting of p15 BID to the mitochondria represents an attractive correlate with the mitochondrial dysfunction/cytochrome c release. We next wished to determine if the targeting of p15 to mitochondria is in and of itself required for the release of cytochrome c. When the cytosol of TNF/CHX-treated cells, S100(TNF), was incubated with mouse liver mitochondria for 30 min at 37°C, p15 BID-targeted mitochondria and up to ϳ50% of cytochrome c was released into the supernatant (Fig. 4A, lanes 1 and 2). p15 BID was resistant to alkali extraction whereas as expected cytochrome c was not (lanes 3 and 4). Strikingly, depletion of p15 BID from the S100(TNF) by an anti-BID Ab which eliminated p15 targeting also prevented the release of cytochrome c (lanes 5-8). Depleting BAX from the activated S100 fraction by using a polyclonal anti-BAX Ab (651) did not inhibit cytochrome c release from mitochondria (data not shown).
BID Is the Required Substrate of Recombinant Caspases Responsible for the Release of Cytochrome c-We next asked whether BID is also a required substrate that must be cleaved  1 and 2) or in 0.1 M Na 2 CO 3 , pH 11.5 (Alkali, lanes 3 and 4), or in 0.5 M NaCl (NaCl, lanes  5 and 6), and centrifuged at 200,000 ϫ g for 10 min to yield mitochondrial pellet (P) and supernatant (S). The fractions were analyzed by Western blot with anti-BID Ab. C, cytosolic p15 BID targets mouse liver mitochondria but full-length p22 BID does not. The soluble fraction from FL5.12 cells pretreated with TNF␣/CHX, S100(TNF) (lane 2) was incubated with purified, intact mitochondria from mouse liver (Mito) (lane 1) in a standard protein import reaction in a volume of 60 l for 30 min at 37°C (lanes 3, 4, and 7-12) or 4°C (lanes 5 and 6). 50 M zVAD-fmk was added to the import reaction in lanes 7 and 8. At the end of the reaction, mitochondria were recovered by centrifugation and the mitochondrial pellet (P) and supernatant (S) were analyzed by Western blot with anti-BID Ab (left panel) or after extraction of the mitochondrial membrane with 0.1 M Na 2 CO 3 , pH 11.5 (Alkali, lanes 11 and 12). by caspases in order for cytochrome c to be released. In this paradigm, a soluble fraction (S100) from untreated FL5.12 cells was preincubated with recombinant caspase-8 or caspase-3 (rCas-8 and -3) and then added to mouse liver mitochondria.
When either the S100 fraction or recombinant caspases were incubated separately with mitochondria there was no release of cytochrome c (Fig. 4B, lanes 2 and 3, and data not shown). However, addition of active rCas-8 to the S100-generated p15 BID (lanes 6 and 7), while addition of rCas-3 generated both p15 and p11 BID (lanes 10 and 11). In both instances, the BID fragments targeted mitochondria as integral membrane proteins (lanes 8 and 9, and data not shown) and ϳ50% of cytochrome c was released (lower panel, lanes 6, 7, 10, and 11). Addition of zVAD-fmk after BID cleavage, prior to exposing mitochondria to S100(rCas-8) did not inhibit the release of cytochrome c (data not shown). Once again, immunodepletion of p15 BID from the S100(rCas-8) prevented the release of cytochrome c when this activated cytosol was added to mitochondria (Fig. 4C, lanes 3 and 4).
BCL-X L , BCL-2 Does Not Prevent TNF␣-induced Translocation of p15 BID to Mitochondria, But Does Interfere with the Release of Cytochrome c-To assess the role of BCL-X L , BCL-2 in the TNF death pathway, we asked whether these antiapoptotic molecules would protect FL5.12 cells from TNF/CHX, whether p15 BID is generated and translocates to mitochondria, and whether cytochrome c is released. Treatment of FL5.12-BCL-X L cells with TNF/CHX resulted in rapid reduction in ⌬⌿ m , production of ROS and cell death (Fig. 5A). Both the mitochondrial dysfunction and cell death were blocked by pretreatment with the caspase inhibitor, zVAD-fmk (Fig. 5A). Similar results were obtained with FL5.12-BCL-2 cells (data not shown).
Intracellular p22 BID was cleaved in FL5.12-BCL-X L cells treated with TNF/CHX, and by 5 h following treatment the p15 FIG. 4. Targeting of cytosolic p15 BID to mouse liver mitochondria is required to release cytochrome c. A, the soluble fraction of FL5.12 cells treated with TNF␣/CHX and depleted of p15 BID fails to release cytochrome c. Purified, intact mitochondria from mouse liver were incubated for 30 min at 37°C with the soluble fraction of FL5.12 cells pretreated with TNF␣/CHX, S100(TNF), which had (lanes 5-8) or had not (lanes 1-4) been immunodepleted of p15 BID using anti-BID Ab. At the end of the reaction, mitochondria were recovered by centrifugation and the mitochondrial pellet (P) and supernatant (S) were analyzed by Western blot with anti-BID Ab (upper panel)/anti-cytochrome c mAb (lower panel) or after extraction of the mitochondrial pellets with 0.1 M Na 2 CO 3 (pH 11.5) (Alkali, lanes 3, 4, 7, and 8). B, cytosolic p15 BID generated by recombinant caspase cleavage targets mitochondria and results in release of cytochrome c. Purified, intact mitochondria from mouse liver were incubated for 30 min at 37°C with the soluble fraction (S100) from untreated FL5.12 cells (lanes 2-5) or that S100 which had been incubated with 1 g of recombinant caspase-8 for 1 h, S100 (rCas-8) (lanes 6 -9) or 1 g of recombinant caspase-3, S100(rCas3) (lanes 10 and 11). At the end of the reaction, the mitochondrial pellet (P) and supernatant (S) were analyzed as in A. C, the cytosolic fraction of FL5.12 cells incubated with recombinant caspase-8 when immunodepleted of p15 BID fails to release cytochrome c from mitochondria. Purified, intact mitochondria from mouse liver were incubated with the S100(rCas-8) as in B (lanes 1 and 2) or this S100(rCas-8) immunodepleted of p15 BID using anti-BID Ab (lanes 3  and 4). At the end of the reaction, the mitochondrial pellet (P) and supernatant (S) were analyzed as in A. p15 BID depleted by the first and second immunoprecipitation of the S100(rCas-8) is shown in lanes 5 and 6.
FIG. 5. BCL-X L ,BCL-2 does not prevent TNF␣-induced translocation of p15 BID to mitochondria, but does prevent cytochrome c release. A, TNF␣/CHX induces death and mitochondrial dysfunction in FL5.12-BCL-X L cells. The experimental procedure was identical to that in Fig. 1A. B, subcellular distribution of BID and cytochrome c in FL5.12-BCL-X L cells following TNF␣/CHX treatment. The experimental procedure was identical to that in Fig. 3A.
BID fragment while often still present in the cytosol was predominantly in the mitochondrial HM fraction (Fig. 5B, lanes  5-8). The p13 minor fragment was associated exclusively with the mitochondrial fraction (Fig. 5B). However, the translocation of p15 BID to mitochondria was not accompanied by a detectable release of cytochrome c (Fig. 5B, lower panel).
To assess the extent of membrane association, the mitochondria (HM fraction) from TNF/CHX-treated FL5.12-BCL-X L cells were incubated in hypotonic buffer, alkaline buffer, or in high salt. The mitochondrial pellet (P) was separated from the supernatant (S) by high speed centrifugation. p22 BID was sensitive to all three treatments (Ͼ50% found in the supernatant (data not shown); whereas, p15 and p13 were markedly resistant to these treatments indicative of an integral membrane position. Interestingly, the p11 BID minor fragment detected in FL5.12 parental cells (Fig. 3B) was not detected in FL5.12-BCL-X L cells (Fig. 5B and data not shown).
In Vivo: Anti-Fas Ab Injection Results in Accumulation of p15 BID in the Cytosol of Hepatocytes and Its Subsequent Translocation to Mitochondria-To assess the involvement of BID in the TNF/Fas death pathway in vivo, mice were injected with anti-Fas Ab which results in massive hepatocyte cell death. To determine the subcellular location of BID, we disrupted hepatocytes using isotonic lysis conditions which kept their mitochondria intact with a retained outer membrane. The p22 BID in normal, untreated hepatocytes was predominantly in the cytosolic (S) fraction (Fig. 6A, lanes 1 and 2). However, by 1 h following anti-Fas Ab injection p15 BID appeared in the soluble S100 fraction (S) (Fig. 6A, lanes 3 and 4). Of note by 3 h following Ab injection p15 was associated exclusively with the mitochondrial fraction (HM) (Fig. 6A, lanes 5 and 6). In addition, the p15 but not p22 BID from the liver cytosol of mice treated for 1 h with anti-Fas Ab was capable of targeting mitochondria in vitro (Fig. 6B). Moreover, that same cytosol that possessed p15 at 1 h post-treatment released cytochrome c from mitochondria (Fig. 6C, lanes 3 and 4). However, the cytosolic fraction from hepatocytes 3 h post-treatment, which no longer had p15 BID present (Fig. 6A, lane 5), did not release cytochrome c substantially (Fig. 6C, lanes 5 and 6). DISCUSSION These data indicate that the TNF and Fas death signal pathways converge at BID, a shared pro-apoptotic effector belonging to the BH3 domain only subset of the BCL-2 family. Our studies suggest a model in which cytosolic p22 BID represents an inactive conformation of the molecule that is proteolytically cleaved to generate an active p15 BID (Fig. 7). In retrospect, this may account for the greater protection by caspase inhibitors of BID-induced death (10) compared with BAX-induced death (22). The p15 conformation rather selectively targets mitochondria where it resides as an integral membrane protein responsible for the release of cytochrome c (Fig. 7). A subpopulation of the full-length p22 BID is also strongly associated with the mitochondrial membrane (Fig.  3B), suggesting that a cleavage-independent pathway for BID activation may also exist. Caspase-8 presumably directly cleaves BID following its own activation by TNF-R/Fas engagement as caspase-8 prefers the Asp 59 site of BID. Removal of the NH 2 terminus would retain and potentially expose the predicted amphipathic ␣ helix, BH3, on the active COOH-terminal p15 fragment. This proteolytic cleavage may alter an inert, intramolecular folded BID or alternatively release BID from a tethering chaperone-like molecule. Immunodepletion of p15 BID from cytosols activated by either TNF-R engagement or caspase-8 addition indicates that p15 BID is requisite for the release of cytochrome c from mitochondria.
The rapid movement of p15 BID from cytosol to mitochon-drial membrane suggests a specific mechanism of targeting. One possibility is a ligand-receptor model (10) in which BH3 of p15 BID binds to membrane bound BCL-2 or BAX which serve as receptors. Alternatively, other BID-protein or BID-lipid pathways may regulate targeting. However, such an association would be a transient intermediate in that the vast majority of p15 BID at mitochondria is an alkaline-resistant integral membrane protein. p15 BID might co-integrate with such partner proteins or insert itself. But why would p15 BID be of such singular importance for the release of cytochrome c? BAX, BCL-2, and BCL-X L are able to form distinct ion conductive pores in artificial membranes (40 -43). p15 BID might regulate such channels, form a hybrid channel, or polymerize to form a channel itself. A cascade of disturbed ion homeostasis and altered transmembrane potential might ensue and could result  6) were suspended in isotonic buffer, homogenized, and separated by differential centrifugation into soluble fraction (S) representing the cytosol and heavy membrane fraction (HM) enriched for intact mitochondria as confirmed by cytochrome c Ab. There is relatively little cytochrome c within the soluble fraction reflecting degradation of released cytochrome c. Fractions were analyzed by Western blot with anti-BID Ab. B, p15 BID from the cytosol of anti-Fas Ab-treated hepatocytes targets mitochondria. The soluble fraction from liver 1 h after intravenous administration of anti-Fas Ab, S100(␣-Fas/1 h), was added to purified, intact mitochondria from liver of non-treated mice for 30 min at 37°C. At the end of the reaction, the mitochondrial pellet (P) and supernatant (S) were analyzed as in Fig.  4A. C, mouse liver cytosol with p15 BID (1-h post-anti-Fas Ab injection) releases cytochrome c from mitochondria. Purified, intact mitochondria from liver of non-treated mice were incubated for 30 min at 37°C with the S100 fraction of liver from untreated mice (lanes 1 and 2), or mice treated with anti-Fas Ab after 1 h (lanes 3 and 4) or 3 h (lanes 5 and 6). At the end of the reaction, the mitochondrial pellet (P) and supernatant (S) were analyzed by Western blot with anti-cytochrome c mAb.
in mitochondrial swelling and ultimate rupture of the more taut outer mitochondrial membrane, which has been proposed as a mechanism of cytochrome c release (44). Alternatively, p15 BID with or without partner proteins might constitute a pore for the selective passage of cytochrome c which has been proposed for diphtheria toxin B fragment and the A fragment (45). Intracellular p22 BID was cleaved at three internal Asp sites: Asp 59 , Asp 75 , and Asp 98 (Fig. 2). Of note the minor fragments of p13 and p11 resulting from cleavage at Asp 75 and Asp 98 , respectively, are only detected in the mitochondria. While caspase-8 prefers the Asp 59 site, other caspases perhaps at the level of mitochondria may be responsible for the p13 and p11 fragments. The p11 fragment was not observed in mitochondria protected by BCL-X L lending support to this thesis. The LQTD 2 recognition motif for the predominant p15 fragment is an atypical site for initiator caspases (6, 8, 9, (I/L/ V)EXD). The DEMD 2 motif which was recognized by recombinant caspase-3 is a classic site for effector caspases (2, 3, 7 (DEXD)). Of note all three recognition sites are well conserved between mouse and human BID (Fig. 2). Caspase cleavage of BCL-2 (46) and BCL-X L (47) have been reported which convert them from anti-to pro-apoptotic molecules. Thus, caspase cleavage of the BCL-2 members may represent a feed forward loop to ensure cell death.
The observation that BCL-2, BCL-X L can in certain settings interfere with TNF/Fas-mediated cell death suggests the existence of regulatory steps beyond the activation of caspase-8 (48 -51). Recently, a cell-free system has indicated that caspase-8 activates a mitochondria dependent pathway that involves cytochrome c as well as a mitochondria independent pathway (52). Moreover, a displacement model holds that BID could participate in an additional augmentation loop. BID binding to BCL-X L could release Apaf-1 making it available to interact with cytochrome c, following its release by BID, which would activate caspase-9 and subsequently caspase-3.
However, the use of inhibitors that eliminate measurable caspase activity indicates that in certain deaths a downstream program of mitochondrial dysfunction characterized by altered ⌬⌿ m and production of ROS still runs (21,22,53). This may be of particular relevance for TNF-induced death which has been noted to be influenced by anti-oxidants (54) and potentially the profound hemorrhagic necrosis of the liver that follows in vivo administration of anti-Fas Ab (55).
Recently, two other groups have also noted the cleavage of BID by purifying a factor that released cytochrome c (56) or screening for substrates cleaved by caspase-8 (57). However, we differ from the conclusions drawn in those papers in that we noted that while the presence of BCL-X L /BCL-2 prevented the detectable release of cytochrome c, the cells still died with similar kinetics. As observed for FL5.12 cells, the presence of BCL-2 or BCL-X L does not prevent TNF/Fas killing in most cell types (51,58). This response is similar to what has been proposed as a type I cell (51). Hepatocytes may represent a type II cell in that a Bcl-2 transgene has been reported to protect mice from anti-Fas Ab-induced apoptosis (59). The findings in FL5.12 cells contrast with the capacity of BCL-X L , BCL-2 to p15 BID Targets Mitochondria and Releases Cytochrome c save these cells after interleukin-3 deprivation, by preventing BAX dimerization and translocation from cytosol to mitochondria (21). Conversely, TNF treatment of the same FL5.12-BCL-X L cells still results in cleavage and translocation of p15 BID to mitochondria where it resides as an integral membrane protein. Despite the established importance of p15 BID in releasing cytochrome c, the presence of BCL-X L prevented its detectable release. While we cannot exclude the release of a small percentage of cytochrome c (60), nevertheless the other parameters of mitochondrial dysfunction including 2⌬⌿ m and ROS production still transpired to the same extent and the cells died in a comparable time course. Thus, BID appears to be strategic in this pathway serving as the lynch pin that connects the initiator caspase to cytochrome c release. However, the robust release of cytochrome c does not appear to be required for Fas/TNF-R1-induced cell death.
Pro-apoptotic BCL-2 members are demonstrating distinct interconnections with death and survival signaling pathways. BID is integral for TNF-R1/Fas release of cytochrome c. BAX is singularly required for nerve growth factor deprivation death in neurons despite their expression of multiple pro-apoptotic members (61). BAD is inactivated by phosphorylation following interleukin-3 survival factor (19). This suggests that selected pro-apoptotic BCL-2 members will reside in distinct signal transduction pathways and in their activated conformation constitute death effectors operative at intracellular organelles, especially mitochondria.