Pro-caspase-3 Is a Major Physiologic Target of Caspase-8*

The apoptotic signal triggered by ligation of members of the death receptor family is promoted by sequential activation of caspase zymogens. We show here that in a purified system, the initiator caspases-8 and -10 directly process the executioner pro-caspase-3 with activation rates (k cat/K m ) of 8.7 × 105 and 2.8 × 105 m −1 s−1, respectively. These rates are of sufficient magnitude to indicate direct processingin vivo. Differentially processed forms of caspase-3 that accumulate during its activation have similar rates of activation, activities, and specificities. The pattern and rate of caspase-8 induced activation of pro-caspase-3 in cytosolic extracts was the same as in a purified system. Moreover, immunodepletion of a putative intermediary in the pathway to activation, pro-caspase-9, was without consequence. Taken together these data demonstrate that the initiator caspase-8 can directly activate pro-caspase-3 without the requirement for an accelerator. The in vitro data thus help to deconvolute previous in vivo transfection studies which have debated the role of a direct versus indirect transmission of the apoptotic signal generated by ligation of death receptors.

Regulation of apoptosis is vital to the development and long term survival of metazoan animals. Apoptosis is required to maintain the balance between cell proliferation and cell death and, therefore, disruptions in the apoptotic program are associated with pathologies such as cancer, where there is too little cell death, and degenerative diseases, where there is too much cell death. Apoptosis can be initiated by at least three types of signals: (i) specific ligation of members on the tumor necrosis factor receptor (TNFR-1) 1 family, which includes Fas/Apo-1/ CD95; (ii) cellular stress, which includes genotoxic damage and anti-neoplastic drugs; and (iii) delivery of granule-associated serine proteases from cytotoxic lymphocytes into target cells. Key mediators that initiate and execute the apoptotic program are members of the caspase family of cysteine proteases whose activation is believed to be essential for virtually all forms of apoptosis (1)(2)(3). Caspases-3, -6, and -7 are involved in the execution of cells in response to many apoptotic stimuli including ligation of death receptors of the TNFR-1 receptor family, resulting in cleavage of a number of proteins whose limited proteolysis is definitive of apoptosis. However, these executioner caspases are not directly activated by receptor ligation, but rely on the proteolytic activity of upstream initiator caspases-8 and -10 (4 -6). In the case of caspase-8 the activation occurs by recruitment of the zymogen to the cytosolic face of the death receptor, such that the initial proteolytic signal originates by autoprocessing of the clustered zymogen (7,8).
At the execution phase, caspase-3 seems to be upstream of caspases-6 and -7 and, therefore, its activation represents a key point in transmission of the proteolytic signal (9). However, the exact mechanism of how the death signal is conveyed from caspase-8 to caspase-3 remains unresolved. Is the apoptotic signal transmitted by direct activation of the executioners by the initiators, thus constituting a minimal two-step cascade that serves to mediate the apoptotic signals, or is the signal further amplified by the presence of additional factors as suggested by Scaffidi et al. (10).
To address these issues we have performed a detailed kinetic study of the activation of pro-caspase-3 by caspases-8 and -10 using recombinant zymogens and active proteases in a defined system, and compared this to the activation of the zymogen by caspases-8 and -10 in cytosolic extracts. This allows us to predict the sequence of events that results in transmission of the proteolytic death signal originating from the initiator caspases to the executioners, and on the basis of in vitro observations, test the hypothesis that additional amplifiers and regulators of the apoptotic signals are required in vivo.

EXPERIMENTAL PROCEDURES
Materials-Active caspase-3, -8, and -10 were expressed in Escherichia coli and isolated as described previously (4,6,11). The expression constructs for caspase-3 contained a His-6 tag at the C terminus of the full-length protein, while caspases-8 and -10 contained N-terminal His tags. Pro-caspase-3 was obtained by reducing the induction time in the presence of 0.2 mM isopropyl-1-thio-␤-D-galactopyranoside to 30 min. The concentrations of the purified caspase-3 zymogen was determined from the absorbance at 280 nm based on the molar absorption coefficient calculated from the Edelhoch relationship (12): caspase-3 (⑀ 280 ϭ 26000 M Ϫ1 cm Ϫ1 ). The concentration of caspases-8 and -10 were determined by active site titration using carbobenzoxy-Val-Ala-Asp-fluorom-ethyl ketone from Bachem AG, Switzerland. Carbobenzoxy-Asp-Glu-Val-Asp-7-amino-4-trifluoromethyl coumarin (Z-DEVD-AFC) and carbobenzoxy-Ile-Glu-Thr-Asp-7-amino-4-trifluoromethyl coumarin (Z-IETD-AFC) were purchased from Enzyme System Products (Dublin, CA). Dithiothreitol was from Diagnostic Chemicals Limited (Oxford, CT). Sucrose was from Mallickrodt (Paris, KY). All other chemicals were from Sigma. Mouse monoclonal anti-caspase-3 was purchased from Signal Transduction Laboratories. All SDS-PAGE were performed using 8 -18% acrylamide gels in the 2-amino-2-methyl-1,3-propandiol/ glycine discontinuous buffer system as described in Ref. 13. After electrophoresis the gels were either stained using Gel-Code (Pierce) according to the manufacture's protocol or Western blotted to Immobilon-P (Millipore) according to the procedure of Matsudaira (14). Western blot analysis were performed using the ECL kit (Amersham) according to the manufacture's protocol using rabbit anti-caspase-3, anti-caspase-9, or anti-PARP (Biomol). Rabbit anti-caspase-9 was generated from animals immunized with purified recombinant caspase-9 that had been expressed in E. coli, essentially as described by (15). Rabbit anti-sera for caspase-3 was prepared as described previously (16). Granzyme B was purified as described previously (17). Dr. Guy Porier kindly provided purified bovine poly(ADP-ribose)polymerase (PARP). The baculovirus caspase inhibitor p35 was obtained as a recombinant protein purified following expression in E. coli (18).
Design of Pro-peptide Mutants of Caspase-3-The D9A and D28A single mutants were prepared by PCR using the High Fidelity DNA polymerase (Boehringer) as described previously (19). PCR reactions were performed using the full-length gene encoding caspase-3 in pET-23b (Novagen) and the primers D9A (5Ј-ACTGAAAACTCAGTGGcTag-cAAATCCATTAAAAAT-3Ј) and D28A (5Ј-CATGGAAGCGAATCAATG-GcCTCcGGAATATCCCTG-3Ј) which introduces a NheI and a BspEI restriction site (bold), respectively, combined with standard primers matching the regions flanking the polylinker of pET-23b. Mutated nucleotides are indicated in lowercase. The truncated form of caspase-3 was generated in a standard PCR reaction by introducing a start codon at Ser 29 using the primer S29M (5Ј-CATGGAAGCGAATCAATGcatat-gGGAATATCCCTG-3Ј) which introduces a NdeI site at the new start codon, combined with a standard reverse primer matching the region downstream of the polylinker in pET-23b. The resulting PCR products were all introduced directly into pET-23b (caspase-3) expression vector as a NdeI-HindIII fragment. The D9A/D28A double mutant were prepared using the D9A primer combined with a standard reverse primer matching the region downstream of the polylinker in pET-23b and the D28A single mutant as template for the PCR reaction. The resulting PCR products were all introduced directly into pET-23b (caspase-3 D9A) expression vector as a NheI-HindIII fragment, thus, utilizing the restriction site introduced during the first round of mutagenesis. The sequence of the mutated DNA was verified using the Model 373A DNA sequencing system and the Dye terminator cycle sequencing kit from Applied Biosystems Inc.
In Vitro Activation of Caspases 3 by Granzyme B, Caspase-8, and Caspase-10 -The activation of the caspases-3 and -7 was investigated using an on-line activation assay as described previously (20). All reactions were performed in caspase buffer (20 mM Pipes, 100 mM KCl, 10 mM dithiothreitol, 1 mM EDTA, 0.1% CHAPS, 10% sucrose, pH 7.2) (21). The reactions were carried out in a Molecular Devices SpectraMax 340 plate reader thermostated at 37°C, operating in the kinetic mode. Rates of caspase activation were determined as described previously (20). Briefly, various zymogen concentrations in the range (69 -690 nM) were warmed at 37°C in caspase buffer containing 0.2 mM Ac-DEVD-pNA, followed by addition of caspase-8 to a final concentration of 18 nM. Substrate hydrolysis was followed at 5-s intervals and the resulting activation curve fitted by non-linear regression to the equation, Culture and Transfection of MCF-7 Cells-MCF cells were maintained in RPMI 1640 with 10% heat-inactivated fetal calf serum supplemented with 2 mM L-glutamine, 100 units/ml penicillin, and 50 g/ml streptomycin. The caspase-3 deficient MCF-7 line was kindly provided by J. Boothman (University of Wisconsin) (23). The absence of detectable caspase-3 was verified by immunoblotting with anti-caspase-3 rabbit polyclonal antibodies. The absence of caspase-3 is due to a 47base pair deletion within exon 3 of the CASP-3 gene. This deletion results in the skipping of exon 3 during pre-mRNA splicing, thereby abrogating translation of the caspase-3 mRNA (24). MCF-7 cells were transfected with caspase-3 in a retroviral vector (pBabe-puro) (provided by Dr. T. Sladek, Chicago Medical School). After the selection with puromycin, stable expression of the caspase was verified by immunoblotting. MCF-7 control cell lines were generated by transfecting with empty pBabe-puro vector. Stable expression of caspase-3 was verified by immunoblotting after selection.
Preparation of Cytosolic Extracts-For preparation of cell free lysates we used the procedure of Ellerby et al. (25) with minor modifications, the most important of which was omission of exogenous protease inhibitors. Human 293 cells from 12 10-cm Petri dishes were harvested by gentle scraping into phosphate-buffered saline at 4°C, pelleted by centrifugation for 5 min at 200 ϫ g, and subsequently washed once in the same buffer. The cell pellet was resuspended in HEB (20 mM Pipes, 10 mM KCl, 5 mM EDTA, 2 mM MgCl 2 , 1 mM dithiothreitol, pH 7.4) at 4°C and pelleted by centrifugation at 1000 ϫ g. The cell pellet was then resuspended in an equal volume of HEB and allowed to swell on ice for 30 min. The cells were then cracked by passing through a 24-gauge needle and pelleted by centrifugation at 16,000 ϫ g for 30 min, and the supernatant (cytosolic extract) recovered. The quality of the lysates were determined by the ability to support rapid generation of caspase activity by addition of cytochrome c and dATP as well as the inability to generate caspase activity upon addition of dATP alone.
Depletion of Caspases from Cell Free Extracts-Lysates from 293 cells were depleted of caspase-3 or caspase-9 by incubating 200 l of cytosolic lysate with 15 l of monoclonal anti-caspase-3 antibody, rabbit anti-caspase-9 antiserum, or as a control anti-NFB antibody on ice for 30 min. The antibody/lysate mixtures were then added onto drained protein G beads (15 l of packed beads) previously washed with HEB buffer containing 2 mM dithiothreitol and incubated on a rotator for 3 h at 4°C. The protein G beads were removed by centrifugation to allow recovery of the specifically depleted extracts. The depleted extracts were stored in appropriate portions at Ϫ80°C.

Activation Kinetics of Pro-caspase-3 by Caspases-8, -10, and
Granzyme B-To determine the relative efficiency of activation of pro-caspase-3 by physiologic activators, we produced the zymogen in E. coli by reducing the induction times normally used for expression of the mature enzyme to 30 min. Procaspase-3 was obtained as a full-length precursor under these conditions (Fig. 1), and is fully activable by the apical caspases-8 and -10, and also by the cytotoxic lymphocytes serine protease granzyme B (Table I). Thus, we are able to deter-mine the intrinsic activation kinetics of the zymogen under defined conditions. Zymogen at varied concentrations was equilibrated at 37°C in caspase buffer, followed by addition of an activator and analysis of activation rate using the reporter substrate Ac-DEVD-pNa (see "Experimental Procedures"). Background hydrolysis of the reporter substrate by caspases-8 and -10 was subtracted to allow direct analysis of the rate of zymogen activation. The observed rates of activation of caspase-3 depended on the activator. Granzyme B is by far the most efficient (k cat /K m ϭ 4.8 ϫ 10 6 M Ϫ1 s Ϫ1 ), capable of activating the caspase zymogen more than 5.5-fold faster than caspase-8 (k cat /K m ϭ 0.87 ϫ 10 6 M Ϫ1 s Ϫ1 ), which in turn catalyzes the activation about 3-fold faster than caspase-10 (k cat /K m ϭ 0.28 ϫ 10 6 M Ϫ1 s Ϫ1 ).
Kinetics of Activation of Caspase-3 in Cytosolic Extracts-On the basis of the reaction kinetics calculated above, the kinetics of activation in more complex systems may be predicted. Therefore, this information served as a framework for the analysis of cytosolic extracts of 293 cells. First it was necessary to determine the concentration of the substrate, pro-caspase-3, in the extract. This was performed by quantitative Western blot analysis using dilutions of known concentrations of the recombinant zymogen in comparison with the lysate used throughout this study. The approximate concentration of caspases-3 was found to be 100 nM. We focused on the apical caspase-8 and asked whether the kinetics of pro-caspase-3 processing is the same in the cytosolic extract. To address these questions we compared the rate of processing of purified caspase-3 zymogen and endogenous zymogen in the 293 extracts. As shown in Fig.   2, the processing pattern is essentially identical, irrespective of the source of the caspase zymogen, natural or recombinant. Thus the processing of pro-caspase-3 by caspase-8 in the cytosolic extracts may be recapitulated in the purified system indicating that no accelerator or inhibitor of the activation process is required.
Properties of Differentially Processed Caspase-3 N-peptide-A notable feature in the caspase-3 processing pattern, both with the recombinant and natural zymogens, is the appearance of two differently processed forms of the large subunit (Fig. 2), determined by Edman degradation to begin at either residue 1 or 10. Although not apparent in Fig. 2, eventually there is further processing at Asp 28 . These differentially processed forms of the large subunit of caspase-3 have been observed many times in apoptotic cells and have been attributed to autoproteolytic cleavage at Asp 9 and Asp 28 (26). Removal of an N-peptide (sometimes called the pro-peptide) is a classic way of activating proteases (27) and in the case of caspase-1 the removal of the N-peptide has been demonstrated to have significant influence on the properties of the resulting enzyme (28). We used purified recombinant caspase-3 zymogen to test the relevance of the cleavages in the caspase-3 N-peptide to the activation of the zymogen as well as the activity of the differentially processed forms of the enzyme. To address these issues we expressed and isolated the following caspase-3 derivatives: wild type caspase-3 enzyme, caspase-3 (S29M) which lacks the entire N-peptide, caspase-3 (D28A) which lacks the initial 9 amino acids of the N-peptide, and caspase-3 (D9A/D28A) where the N-peptide cannot be removed. Additionally, we expressed the enzyme caspase-3 (D9A), however, this mutation does not prevent the autocatalytic removal of the N-peptide and thus, the resulting enzyme is identical to the wild-type. Caspase-3 and derivatives thereof that were essentially fully processed between the large and small subunits (Asp 297 ) were obtained by expression in E. coli using a 3-h induction, whereas the zymogens were obtained by reduction of the expression time to 20 min (Fig. 1). The authenticity of the N-terminal sequences was confirmed by Edman degradation.
To test whether removal of the caspase-3 N-peptide modulated catalysis we determined the k cat /K m values for the hydrolysis of Ac-DEVD-pNA for the wild type, caspase-3 (S29M), caspase-3 (D28A), and caspase-3 (D9A/D28A). Since the k cat /K m for all the enzymes tested are essentially identical (see Table  II), we rule out modulation of catalysis as a function for the N-peptide. Furthermore, comparison of the relative rates of hydrolysis of Ac-IETD-AFC and Ac-DEVD-AFC showed that the enzymes do not exhibit any significant differences in their substrate specificities (data not shown). Thus, the catalytic  apparatus does not appear to be sensitive to the presence/ absence of the N-peptide.

TABLE I Apparent k cat /K M values for the activation of caspase zymogens
It is formally possible that the presence of the N-peptide may to some degree restrict access to the active site, thereby imposing a discrimination toward protein substrates but not the short tetrapeptide substrates. To investigate this possibility we tested the ability of caspase-3, caspase-3 (D28A), and caspase-3 (D9A/D28A) to cleave the protein substrate PARP. Cleavage was monitored after addition of the caspase-3 constructs in concentration from 0.067 to 440 nM, a range which allowed a comparison of the amount of caspase required to cleave PARP in 30 min at 37°C. Although caspase-3 containing the fulllength N-peptide (D9A/D28A) cleaved PARP a bit more efficiently, the increase was considered marginal (Fig. 3). In addition, we tested the ability of the baculovirus caspase inactivator p35 to inhibit the different caspase derivatives. As it is evident from Table II the efficiency by which p35 inhibited the different caspase-3 derivatives was not affected to any significant degree and all showed comparable k obs values at the same p35 concentration.
Since the N-peptide did not influence the activity of the matured enzymes, we considered that it may influence zymogen activation, either in a defined zymogen activator system, or in cytosolic extracts. Table I demonstrates a 2-4-fold decrease in activation rate with the zymogen lacking a pro-peptide (S29M), however, we do not consider this a dramatic difference when comparisons of the magnitudes of the rates are made. To determine whether a cytosolic factor could influence activation via the caspase-3 N-peptide the endogenous zymogen was depleted with a mouse anti-caspase-3 monoclonal and the recombinant pro-caspase-3 mutant zymogens were added to the depleted extracts. Western blotting of the immunodepleted cytosolic extracts confirmed removal of at least 95% of the endogenous pro-caspase-3.
Cytosolic extracts reconstituted with recombinant procaspase-3 mutant zymogens were treated with cytochrome c plus dATP, to achieve caspase activation (Fig. 4A) and activation followed by monitoring cleavage of the reporter substrate Ac-DEVD-pNa which, in a whole extract, measures a combination of the caspases present. The response of the reconstituted extracts was identical, with the exception of the activation of the S29M reconstituted extract, which showed an activation rate about 1.5-fold higher than the others. When the identical amount of purified zymogens were activated by granzyme B (Fig. 4B), there were no observable differences in rate of activation in agreement with the close similarity of the activation rates (Table I). Taken together these results argue that the N-peptide of caspase-3 only minimally influences activity of the enzyme, or its activation.
Processing of Pro-caspase-9 -Although our data support a direct link between the initiator caspase-8 and the executioner caspase-3, others have suggested that the caspase-8 signal is indirectly passed on to pro-caspase-3 via a mitochondrial route (10) where caspase-9, the first caspase in the post-mitochondrial pathway (29), activates the caspase-3 zymogen. As shown in Fig. 5, caspase-9 processing occurs after the addition of caspase-8 to 293 cell extract. The zymogen of caspase-9 (48 kDa) was converted first to a 36-kDa form and then to one of 34 kDa. The first product is consistent with processing at Asp 330 (caspase-9 numbering) and the second represents processing at Asp 315 , previously demonstrated to be the cleavage utilized during processing of pro-caspase-9 (30). Cleavage at Asp 330 is reported to be mediated by activated caspase-3, but the enzyme responsible for processing at Asp 315 in vivo is unknown (30). Conceivably, either cut in caspase-9 could be due to the direct action of caspase-8.
To determine whether caspase-8 directly activated procaspase-9 during death receptor-dependent apoptosis, and if so, whether activated caspase-9 accelerated the processing of caspase-3, we performed two different experiments. We first asked whether caspase-9 processing required the presence of caspase-3. To address this question, processing of caspase-9 was examined in MCF-7 cells which are known to be deficient in caspase-3 and in transfected MCF-7 cells that express this family member (24), but which still undergo apoptosis without DNA fragmentation (23). Following treatment with TNF, caspase-3 negative MCF-7 cells failed to show evidence of caspase-9 processing. In the caspase-3 positive transfectants, however, conversion of caspase-9 was apparent after 2 h (Fig.  5B), although it does not appear to result in generation of the fragments typically associated with caspase activation. Consequently, caspase-8 activation initiated by TNFR-1 ligation (5, 31) is not associated with caspase-9 processing unless caspase-3 is present. This indicates that caspase-8 does not directly activate caspase-9 in whole MCF-7 cells. However, since the capacity of caspase-8 to activate caspase-9 may not be operative in all cell types, we then evaluated this hypothesis using our model system where the cytosolic extract was depleted of caspase-9. Although the addition of cytochrome c/dATP did not result in detectable caspase-3 activation (data not shown), the extract was fully responsive to exogenous caspase-8 (Fig. 6). In combination, these data provide further evidence that the cleavage of caspase-9 is mediated initially as a feedback from caspase-3 (32). However, more importantly caspase-9 is not cleaved directly by caspase-8 in vivo.

DISCUSSION
Caspases are commonly divided into apical and executioner subsets (1, 33). There exist two well characterized points at TABLE II Catalytic properties of caspase-3 mutants Caspase-3 mutants constructed to retain portions of the pro-peptide were compared for activity versus a synthetic substrate and for inhibition by the natural inhibitor p35. Inhibitory data for the S29M mutant were not obtained since this is essentially identical to the wild type: it contains no pro-peptide.  which apical caspases initiate apoptotic signals. One is at the cell surface where members of the TNFR-1 family of death receptors transmit a signal across the cell membrane following receptor clustering. The second point of initiation follows the release of mitochondrial factors (34,35), and although this post-mitochondrial pathway is well documented it is unclear how the mitochondrion perceives the apoptotic signal. Nevertheless, anti-neoplastic drugs, genotoxic damage, and inhibition of cellular signal transduction pathways all seem to con-verge on the mitochondrial route (36). At each point of initiation the first recognizable biochemical event is specific caspase activation, and each initiation point utilizes distinct caspases. In the death receptor pathway(s) the apical caspase-8 (and possibly 10) transmits a proteolytic signal following autoactivation at the cytosolic face of the receptor (4 -6, 37, 38). The signal for activation appears to be local clustering of procaspase-8 that possesses enough activity in its zymogen to achieve autolytic proteolytic maturation (7,8,39). In the mitochondrial route specific or nonspecific delivery of cytochrome c (29, 40 -42) to the protein Apaf-1 results in recruitment and activation of caspase-9 (29). Both caspase-8 and caspase-9 have The background activity due to the added caspase-8 has been subtracted from all rates. As it is evident from the graph, no substantial influence of the presence of pro-caspase-9 on the rate of activation was observed. been demonstrated to act on in vitro translated pro-caspases-3 and -7, the executioner caspases whose activation correlates with apoptosis. Thus there are two potential routes to activate the executioner caspases, and in this context both caspases-8 and -9 can be thought of as initiators whose pathways converge at the execution phase of apoptosis.
Although caspases-8 and -10 can activate pro-caspase-3 in vitro, it has proven difficult to determine whether the apical caspases perform this function directly or indirectly, because previous studies have relied on in vitro translated zymogens or cytosolic extracts (6,31,37,43). We here demonstrate that caspase-8 and -10 can rapidly activate caspase-3. More importantly, focusing on caspase-8 we observe that activation proceeds with the same rate in a 293 cytosolic extract, eliminating a requirement for an intermediary component. Additionally, depletion of pro-caspase-9 from the extract has no impact on caspase activation by caspase-8, and MCF-7 cells deficient in caspase-3 failed to support processing of pro-caspase-9. Therefore, processing of pro-caspase-9 in death receptor-mediated apoptosis requires the presence and activation of caspase-3 which makes it a downstream event unlikely to play a major role in caspase activation.
It has been proposed that mitochondria are required to transmit the apoptotic signal generated by treatment of cells by agonistic Fas antibodies, but only in a small selection of cell lines (10). This would imply that caspase-8, the apical caspase of the Fas pathway, initiates a mitochondrial signal. In support of this, the generation of caspase activity initiated by addition of human caspase-8 to cytosolic extracts of Xenopus eggs is accelerated in the presence of mitochondria (44). However, the relevance of this to homologous systems is unclear, since it is not known whether Xenopus has a caspase-8, and the kinetics of activation of a putative Xenopus caspase-3 ortholog, or whether the caspase activity in Xenopus extracts is due to such an ortholog, have not been determined. The data presented above, in contrast, do not suggest any requirement for an intermediary between caspase-8 and caspase-3. Therefore, the question is whether the mitochondrial acceleration occurs in vivo, and whether caspase-8 must transmit its signal via mitochondria to the executioners in vivo. Currently the most valuable evidence for a role of mitochondria in apoptosis triggered by death receptor ligation comes from several studies investigating expression of ectopic or transgenic Bcl-2, which is hypothesized to operate by blocking mitochondrial-dependent apoptosis (reviewed in Ref. 36). Most investigators agree that Bcl-2 prevents apoptosis triggered by genotoxic damage, glucocorticoids, and chemotherapeutic drugs, but there are inconsistencies in the data describing the protective effect of Bcl-2 against apoptosis induced by death receptors (10,33,45,46). In a survey of several cell lines, Scaffidi et al. (10) noted that most are not protected by Bcl-2 from apoptosis triggered by agonist Fas antibodies. However, the use of agonist antibodies and immortalized cell lines may not be the best way to determine a role for Bcl-2. More significantly, T-cell apoptosis in vivo, which is dependent on physiologic Fas ligation, is unaffected in Bcl-2 transgenic mice (46). In contrast, death following injection of agonist Fas antibodies in whole mice was significantly retarded in Bcl-2 transgenic mice (47). These somewhat contradictory studies can be reconciled if some cells support direct transmission of caspase-8 to caspase-3, while others require a mitochondrial accelerator (10).
Since pro-caspase-9 is not processed by caspase-8, the mitochondrial accelerator must act upstream of the caspase-9 activator complex and, therefore, presumably upstream of the mitochondrion. Possibly the pro-apoptotic mitochondrial signal is activated by the action of caspase-8 on mitochondria, or a latent protein that activates mitochondria for apoptosis. The importance of the mitochondrial route in Fas death is not clear, since evidence suggests that only a minority of cell lines require mitochondria to transmit the caspase-8 signal. If direct transmission of the signal from caspase-8 to the executioner caspase-3 occurs in most cell lines, what is the advantage of a mitochondrial intermediate? Usually, levels of complexity are added to allow additional regulation points, as clearly evidenced by the evolution of the vertebrate blood coagulation cascade. It is not immediately clear what advantage cells would achieve by adding a level of regulation to the Fas-triggered death signal, but future studies will doubtlessly focus on this issue. Regardless, it would appear that those cells in the body that are primed to undergo apoptosis during education of the immune system have evolved death receptors, caspase-8, and the caspase-8 activator complex to allow rapid direct transmission of the death signal, bypassing any mitochondrial requirement, and that pro-caspase-3 is a major physiologic substrate of caspase-8 (Fig. 7). FIG. 7. Pro-caspase-3 activation. Currently there exist two recognized points at which apical caspases are activated to initiate apoptosis. Following TNFR-1 or Fas ligation, the initiator caspase-8 is activated by adapter-mediated recruitment to the receptor's cytosolic face (7,8). Alternatively, the initiator caspase-9 is activated following release of mitochondrial components to form the Apaf complex (34,35). Both activated initiators converge on the proteolytic activation of caspase-3. In death receptor-triggered apoptosis the main pathway (bold arrows) is direct activation of pro-caspase-3 by caspase-8. In some cell types an additional pathway (light arrow) may operate by caspase-mediated delivery of a signal (question mark) to mitochondria. The importance of the mitochondrial pathway in death receptor-triggered apoptosis is unknown, but apparently subordinate to the dominant, direct pathway in most cell types. This model predicts that deficiency of caspase-9 would not affect death receptor apoptosis that has been triggered by caspase-8 activation.