Heat Shock Induces Apoptosis Independently of Any Known Initiator Caspase-activating Complex*

Adaptive responses to mild heat shock are among the most widely conserved and studied in nature. More intense heat shock, however, induces apoptosis through mechanisms that remain largely unknown. Herein, we present evidence that heat shock activates an apical protease that stimulates mitochondrial outer membrane permeabilization and processing of the effector caspase-3 in a benzyloxycarbonyl-VAD-fluoromethyl ketone (polycaspase inhibitor)- and Bcl-2-inhibitable manner. Surprisingly, however, neither FADD·caspase-8 nor RAIDD·caspase-2 PIDDosome (p53-induced protein with a death domain) complexes were detected in dying cells, and neither of these initiator caspases nor the endoplasmic reticulum stress-activated caspases-4/12 were required for mitochondrial outer membrane permeabilization. Similarly, although cytochrome c was released from mitochondria following heat shock, functional Apaf-1·caspase-9 apoptosome complexes were not formed, and caspase-9 was not essential for the activation of caspase-3 or the induction of apoptosis. Thus, heat shock does not require any of the known initiator caspases or their activating complexes to promote apoptotic cell death but instead relies upon the activation of an apparently novel apical protease with caspase-like activity.

Adaptive responses to mild heat shock are among the most widely conserved and studied in nature. More intense heat shock, however, induces apoptosis through mechanisms that remain largely unknown. Herein, we present evidence that heat shock activates an apical protease that stimulates mitochondrial outer membrane permeabilization and processing of the effector caspase-3 in a benzyloxycarbonyl-VAD-fluoromethyl ketone (polycaspase inhibitor)-and Bcl-2-inhibitable manner. Surprisingly, however, neither FADD⅐caspase-8 nor RAIDD⅐caspase-2 PIDDosome (p53-induced protein with a death domain) complexes were detected in dying cells, and neither of these initiator caspases nor the endoplasmic reticulum stress-activated caspases-4/12 were required for mitochondrial outer membrane permeabilization. Similarly, although cytochrome c was released from mitochondria following heat shock, functional Apaf-1⅐ caspase-9 apoptosome complexes were not formed, and caspase-9 was not essential for the activation of caspase-3 or the induction of apoptosis. Thus, heat shock does not require any of the known initiator caspases or their activating complexes to promote apoptotic cell death but instead relies upon the activation of an apparently novel apical protease with caspase-like activity.
Two pathways, referred to as the death receptor (extrinsic) and mitochondrial (intrinsic) pathways, are widely regarded as being responsible for most, if not all, caspase-dependent apoptosis (1,2). In both cases, caspase cascades are initiated through formation of a protein complex that contains a specific adapter protein and its associated "initiator" caspase (i.e. caspase-2, -8, -9, -10, or -12) (3). Once activated, most initiator caspases proteolytically activate the downstream "effector" caspases-3, -6, and/or -7, which in turn cleave specific cellular substrates resulting in chromatin compaction, membrane blebbing, and cell shrinkage.
Stimulation of death receptors (such as CD95 and tumor necrosis factor (TNF) 3 receptor 1) with their cognate ligands or agonistic anti-bodies induces receptor aggregation and formation of a "death-inducing signaling complex" (4). Although important differences exist among the various death receptor complexes, they all utilize the adapter protein Fas-associated death domain (FADD) to recruit and induce a conformational change in caspase-8 (and -10), which results in its activation and autoprocessing (induced proximity model) (3). In so-called type I cells, it is thought that death receptor stimulation activates sufficient amounts of caspase-8 within the death-inducing signaling complex so as to efficiently activate caspase-3 and execute the apoptotic program. In contrast, in type II cells, such as Jurkat T cells, caspase-8 indirectly mediates caspase-3 activation by first cleaving and activating the proapoptotic Bcl-2 family member Bid (5).
Similar to Bid, other BH3-only proteins, such as Bim, Puma, etc., serve as "stress sensors" that initiate the mitochondrial pathway to apoptosis by either inhibiting anti-apoptotic Bcl-2 family members (e.g. Bcl-2 and Bcl-x L ) and/or directly activating the pro-apoptotic Bcl-2 family members Bax or Bak (2). These actions trigger mitochondrial outer membrane permeabilization (MOMP), which is arguably the defining moment in the intrinsic pathway, as it leads to the release of cytochrome c from the intermembrane space into the cytoplasm (1). Cytochrome c then binds to apoptotic protease-activating factor-1 (Apaf-1) and, in the presence of dATP or ATP, stimulates oligomerization of Apaf-1 into a large heptameric apoptosome complex (1). The apical caspase-9 is subsequently recruited to the apoptosome, where it undergoes activation and autocatalytic processing at Asp-315 to produce p35/p12 caspase-9. Importantly, once activated, the effector caspase-3 can also participate in the processing of procaspase-9 at Asp-330 through a feedback loop to generate p37/p10 caspase-9 (6).
Recently, it has been suggested that, upstream of mitochondria, other forms of cellular stress may induce the formation of distinct caspaseactivating complexes. For example, following DNA damage, caspase-2 is reportedly activated within a large molecular weight complex containing the adapter protein RAIDD and the p53-inducible scaffold protein PIDD (7). Caspase-2 is the most archaic of the mammalian caspases, exhibiting features of both initiator and effector caspases; however, it appears unable to efficiently activate the effector caspase-3 (8). Thus, caspase-2 stimulates MOMP either directly or indirectly through cleavage of Bid and relies upon the Apaf-1 apoptosome to activate effector caspases and induce apoptosis (8 -10).
Endoplasmic reticulum (ER) stress likewise may induce the formation of a complex containing the ER transmembrane kinase IRE1 and TRAF2 (TNF receptor-associated factor 2), which in turn recruits and activates the apical caspase-12 (11). Alternatively, caspase-12 may also be processed by calpains, which are often activated by calcium during ER stress (12). Active caspase-12 may then activate caspase-9 independently of the Apaf-1 apoptosome, which in turn activates the effector caspases -3 and -7 (13). However, the caspase-12 gene is not expressed in the vast majority of humans because of a premature stop codon. Thus, it has been proposed that caspase-4 may substitute for caspase-12 in man. Indeed, caspase-4 is cleaved upstream of mitochondria in Bcl-2-or Bclx L -overexpressing cells, but similar to caspase-2, it must engage the intrinsic pathway to induce apoptotic cell death (14).
Responses to heat shock are among the most highly conserved and important in nature. Consequently, numerous studies have focused on how uni-and multicellular organisms cope with this common stress (15). Previous work makes it clear that up-regulation of various heat shock proteins (Hsps), including Hsp-70 and to a lesser extent Hsp-27 and Hsp-90, represents an adaptive response that can protect cells from subsequent pro-apoptotic stimuli. However, far fewer studies have rigorously examined the mechanism(s) by which heat shock induces cell death. In this study, we have demonstrated that intense heat shock induces caspase-dependent apoptosis but does so via mechanisms that do not involve any of the established initiator caspase-activating complexes and thus may represent an archaic and novel pathway to cell death.
Cell Death Treatments-Cells were challenged with the DNA-damaging agents etoposide (60 M) or cisplatin (100 M); the kinase inhibitor staurosporine (250 nM); the translation inhibitor and p38/JNK activator anisomycin (25 M); the lysomotropic agent Leu-Leu methyl ester (25 M); the ER stress agents brefeldin A (500 nM) or thapsigargin (500 nM); or the death receptor ligands anti-CD95 (CH-11, 100 ng/ml, Upstate Biotechnology), TNF-related apoptosis-inducing ligand (TRAIL) (1.0 -1.6 g/ml), or a combination of TNF (300 ng/ml) and cycloheximide (2 g/ml). Cells treated with heat shock were incubated at 45°C for 2 h followed by post-heat shock incubation at 37°C for 4 h, unless otherwise indicated. All of the treatments were performed in cell culture incubators with a humidified atmosphere of 5% CO 2 . To inhibit protease-dependent cell death, cells were preincubated for 1 h Flow Cytometric Detection of Inner Mitochondrial Transmembrane Potential (⌬ m ) and Cell Death-Phosphatidylserine (PS) exposure and propidium iodide (PI) staining was determined as previously described (16). Changes in ⌬ m were measured by staining cells with tetramethylrhodamine ethyl ester (TMRE, 120 nM; Molecular Probes) at 37°C for 10 min. After staining, the cells were immediately analyzed by flow cytometry (EPICS XL, Beckman-Coulter).
Gel Filtration Studies-Cell lysates were prepared from control or apoptotic cells and fractionated by size exclusion chromatography using a fast protein liquid chromatography protein purification system with a Superose-6 HR 10/30 column (Amersham Biosciences), as previously described (17). Fractions were subsequently mixed with 5ϫ loading buffer, resolved by SDS-PAGE, and immunoblotted with a rabbit polyclonal antibody to PIDD (Alexis Biochemicals) or mouse monoclonal antibodies to RAIDD (clone 4B12, MBL) or caspase-2 (clone 35, BD Transduction Laboratories).

Heat Shock Induces a Unique Caspase Processing Profile That Culminates in
Apoptosis-In our initial experiments, we noted that diverse pro-apoptotic stimuli, including DNA damaging agents, Ser/Thr kinase inhibitors, p38/JNK activators, death receptor ligands, lysomotropic agents and ER stressors, stimulated time-dependent cleavage of caspase-9 at both Asp-315 (p35 caspase-9) and Asp-330 (p37 caspase-9) (Fig. 1A), whereas heat shock induced processing almost entirely at Asp-330 (Fig. 1B). Because cleavage of caspase-9 at Asp-315 and Asp-330 is generally mediated by the Apaf-1 apoptosome and the effector caspase-3, respectively (6), the data suggested that caspase-9 cleavage following heat shock was not mediated by the apoptosome but was instead a relatively late caspase-3-dependent event. In subsequent doseresponse experiments, heat shock induced a steady increase in the activation of caspase-3 between 42 and 45°C, which correlated with the formation of p37 caspase-9 and the amount of apoptosis (Fig. 1C, lanes [2][3][4][5]. However, as previously observed, there was little evidence for the formation of p35 caspase-9 at any temperature. Therefore, our results suggested that heat shock might utilize an apical caspase other than caspase-9 to initiate its caspase cascade and promote apoptosis. In an effort to determine the importance of caspases for heat shockinduced apoptosis, as well as the sequential order of caspase activation, we next examined heat-shocked cells pretreated with either the polycaspase inhibitor Z-VAD-fmk or the caspase-3/7 selective inhibitor DEVD-CHO (20). Heat shock alone initiated the activation of all apoptosis-related caspases (i.e. caspases -2, -3, -6, -7, -8, and -9), except for caspase-10, and induced ϳ50% apoptosis by 6 h, all of which was prevented by Z-VAD-fmk ( Fig. 2A, lanes 1-3; Fig. 2B, left panels). Next, by carefully titrating DEVD-CHO, we established a concentration of this inhibitor, which did not prevent the initial processing of procaspase-3 to its intermediate p20 form (i.e. did not inhibit the apical protease responsible for cleaving procaspase-3 at Asp-175) but totally inhibited the subsequent autocatalytic conversion of caspase-3 from its p20 to its p19/17 forms (Fig. 2A, lane 4). In this context, selective inhibition of caspase-3 also prevented cleavage of a number of direct caspase-3 substrates, including poly(ADP-ribose) polymerase, caspase-6, and as predicted above, caspase-9 ( Fig. 2A, lane 4). To our surprise, however, DEVD-CHO also prevented the processing of the remaining initiator caspases -2 and -8 ( Fig. 2A, lane 4). Thus, our initial studies with pharmacological caspase inhibitors suggested that heat shock induced the activation of caspase-3 via a Z-VAD-fmk-inhibitable protease. This apical protease, however, did not appear to be any of the known classical initiator caspases, as they were all processed downstream of caspase-3.
Heat Shock Activates an Apical Protease Capable of Inducing MOMP and a Loss in ⌬ m -Inhibition of the apical protease with Z-VAD-fmk not only prevented the activation of caspase-3 but also inhibited MOMP (e.g. mitochondrial release of cytochrome c or Smac into the cytoplasm) as well as cell death up to 24 h following heat shock ( Fig. 2A, lanes 2 and 3; Fig. 2B, left panels) (supplemental Fig. S1). By contrast, inhibition of caspase-3 with DEVD-CHO did not significantly inhibit cytochrome c release but did partially inhibit Smac release or more likely inhibited its degradation by the proteasome (21). DEVD-CHO also failed to inhibit the overall amount of heat shock-induced cell death but did promote a shift in the mode of cell death from apoptosis (PS ϩ ) to necrosis (PS Ϫ PI ϩ ) ( Fig. 2A, lane 4; Fig. 2B, left panels). Therefore, collectively, these results indicated 1) that the heat-activated apical protease was responsible for inducing MOMP and 2) that downstream caspase-3 activity was required for execution of the apoptotic program, but in its absence, cells nevertheless died by necrosis because of severe mitochondrial injury brought on by the apical protease. DEVD-CHO also failed to prevent a loss in ⌬ m following heat shock (Fig. 2B, right panels). This result was somewhat surprising given that caspase-3 is thought to be primarily responsible for triggering reductions in ⌬ m by cleaving and inactivating NDUFS1, a complex I component of the mitochondrial respiratory chain (22). Thus, the heat shock-activated apical protease could induce a loss in ⌬ m via caspase-3-independent mechanisms.
Heat Shock Induces Apoptosis Independently of the Apaf-1⅐Caspase-9 Apoptosome-Although processing of caspase-9 appeared to be a relatively downstream event, as already noted, cytochrome c was released during heat shock-induced apoptosis ( Fig. 2A, lane 2). Therefore, in an effort to determine whether heat shock inhibited the formation and/or activity of the Apaf-1 apoptosome, we fractionated lysates from control and heat-shocked cells (by Superose-6 gel filtration chromatography) and compared them with naïve lysates activated with cytochrome c and dATP (cytochrome c/dATP). Heat shock induced partial oligomerization of Apaf-1 into large molecular weight complexes (ϳ700 kDa), which were similar in size to those apoptosome complexes formed in cytochrome c/dATP-activated lysates (Fig. 3A, fractions 8 -12). How-FIGURE 1. Heat shock induces a unique temperature-dependent caspase-9-processing and cell death profile. E6 Jurkat T cells were exposed to various pro-apoptotic stimuli for 1-24 h (as described under "Experimental Procedures") (A) or were heatshocked for 2 h at 42-45°C, followed by a 0 -6 h incubation at 37°C (B). Whole cells were then immunoblotted for caspases -9 and/or -3. ---denotes the fusion of two separate gels at the same exposure level. Cell death was assessed by flow cytometry, as described under "Experimental Procedures." pro, proform of the caspase; Bref A, brefeldin A; Thaps, thapsigargin; HS, heat-shocked; STS, staurosporine; Aniso, anisomycin; LLME, Leu-Leu methylester. ever, caspase-9 was not recruited to the Apaf-1 complex nor was there any evidence for apoptosome-dependent processing of caspase-9 at Asp-315 (data not shown; Figs. 1 and 2A). We considered the possibility that recruitment of caspase-9 might be inhibited by Hsp-70 or Hsp-90, because both are up-regulated in response to sublethal exposures to heat shock, and both reportedly inhibit the apoptosome (23). However, following a lethal exposure to heat shock, there was no increase in the expression of either Hsp-70 ( Fig. 2A, lanes 1-2) or Hsp-90 at 6 h (data not shown). Moreover, pretreatment of cells with the global translation inhibitor cycloheximide did not potentiate (nor did it inhibit) heat shock-induced MOMP, caspase-3 activation, or apoptosis (supplemental Fig. S2).
Despite the evidence against the involvement of the Apaf-1 apoptosome in heat shock-induced apoptosis, it remained plausible that an undetectable amount of p35 caspase-9, activated within the apoptosome, might process small amounts of caspase-3 and initiate a robust caspase amplification loop. We therefore examined both apaf-1 Ϫ/Ϫ and caspase-9 Ϫ/Ϫ MEFs for their apoptotic responses to heat shock, as well as to tumor necrosis factor-␣/cycloheximide (24). Heat shock triggered MOMP (loss of cytochrome c from mitochondria) and apoptosis in both the Apaf-1 and caspase-9-deficient MEFs and importantly, caused a time-dependent increase in caspase-3-like DEVDase activity (Fig. 3, B and C). Using an antibody that recognizes only the processed forms of caspase-3, we confirmed that caspase-3 was activated within 4 h of heat shock. TNF, which stimulates the formation of initiator FADD⅐caspase-8 complexes, likewise activated caspase-3 (Fig. 3, B and C). We subsequently confirmed the phenotype of the Apaf-1-and caspase-9-deficient cells by incubating them with cytochrome c/dATP. As expected, neither Apaf-1 nor caspase-9-deficient lysates could independently support cytochrome c/dATPdependent activation of caspases -9 and -3 but readily did so when combined (50:50 mix) prior to activation (supplemental Fig. S3). Thus, functionally active apoptosome complexes were not formed in response to heat shock, but more importantly, were not essential for heat shock-induced MOMP, caspase-3 activation, or cell death.
Heat Shock Does Not Utilize ER Stress-activated Caspases to Induce Apoptosis-Given that heat can induce protein misfolding, we speculated that heat shock might induce apoptosis through an ER stress pathway. ER stress reportedly induces apoptosis independently of the apoptosome by activating the apical caspase-12 (mouse), or possibly caspase-4 (human), both of which are localized to the outer ER membrane (11)(12)(13)(14). Nevertheless, neither caspase-4 nor caspase-1 (another pro-inflammatory caspase) were expressed in our Jurkat T cell model (Fig. 4A), and YVAD-CHO, which potently inhibits both caspases (20), failed to prevent heat shock-induced MOMP, caspase activation, or cell death (Fig. 2A, lane 5). Consistent with these results, caspase-12-deficient MEFs also displayed no defects in their ability to undergo MOMP or activate caspase-3 following heat shock (Fig. 4B). Moreover, although calpains are thought to activate ER-associated caspases, neither PD150605 nor Z-VF-CHO, two distinct mechanism-based calpain Caspase inhibitors had no effect on control cells, as shown in Fig. 7B. B, cell death (annexin V/PI) and ⌬ m (TMRE) measurements were determined by flow cytometry, as indicated under "Experimental Procedures." ‫ء‬ denotes nonspecific bands; aCasp-6 and pCasp-10 refer to the active and proforms of caspases -6 and -10, respectively.

Bcl-2 Inhibits Heat Shock-induced MOMP and Cell Death, but Does Not Prevent the Upstream Processing of Procaspase-3-Z-VAD-fmk
clearly inhibited an apical protease responsible for heat shock-induced MOMP. However, it remained unclear whether MOMP was required for the activation of caspase-3 (e.g. by stimulating the release of a factor from mitochondria other than cytochrome c, which led to the activation of caspase-3) or whether the apical protease simultaneously activated caspase-3 upstream of mitochondria. To address these questions, we utilized cells stably transfected with Bcl-2 (a potent inhibitor of MOMP) and reasoned that if the apical protease could directly process procaspase-3 upstream of mitochondria, then Bcl-2 should not entirely prevent this cleavage from occurring. In fact, as previously demonstrated, Bcl-2 prevented MOMP and cell death induced by the agonistic CD95 antibody CH-11 but did not inhibit the initial caspase-8-dependent processing of procaspase-3 to its intermediate p20 form (Fig. 5, lanes  16 and 18) (16). Similarly, Bcl-2 inhibited heat shock-induced MOMP, as well as cell death, but did not prevent (although it delayed) the initial processing of procaspase-3 to its p20 form (Fig. 5, lane 12).
In each instance, Bcl-2 likely prevented the autocatalytic conversion of caspase-3 from its p20 to its fully mature p17 form, because it inhibited the release of Smac from mitochondria (Fig. 5, lanes 12, 16 and 18) and thus preserved the inhibition of p20 caspase-3 by IAPs (16). Indeed, the addition of recombinant Smac to lysates prepared from heatshocked Bcl-2-overexpressing cells resulted in increased caspase-3 DEVDase activity relative to non-treated cells (supplemental Fig. S4) (16). Similar to DEVD-CHO, IAP inhibition of caspase-3 activity also inhibited caspase amplification, resulting in a decrease in caspase-8 and -2 processing (Fig. 5, lanes 12, 16, and 18) (16). Thus, first and foremost, the data demonstrated that the heat-activated apical protease could induce the initial processing of procaspase-3 to its p20 form, upstream of mitochondria. Furthermore, they implicated a role for Smac in propagating the caspase amplification loop and promoting cell death by antagonizing the inhibition of p20 caspase-3 by IAPs (see Fig. 9).
Heat Shock Induces Apoptosis Independently of FADD and Caspase-8-Based on our studies with DEVD-CHO and Bcl-2 overexpression, caspase-8 appeared to be processed rather late in the heat shockinduced caspase cascade (Fig. 2A, lane 4; Fig. 5). Recent reports, however, suggest that, in some cases, caspase-8 may exhibit proteolytic activity in its proform (25). Therefore, to definitively rule in/out an essential role for caspase-8 (or the caspase-8/10 adapter protein FADD) in heat shock-induced apoptosis, we utilized Jurkat T cells that lacked the expression of either caspase-8 or FADD (26). As expected, in wildtype parental (A3) cells, CD95 stimulation with CH-11 led to the activation of caspases, release of cytochrome c and Smac, and the induction of apoptosis, whereas these events were largely inhibited in the FIGURE 3. The Apaf-1⅐caspase-9 apoptosome is not essential for heat shock-induced MOMP, caspase-3 activation, or cell death. A, lysates (30 mg/ml) prepared from naïve control and heatshocked cells were size-fractionated by Superose-6 gel filtration chromatography and immunoblotted for Apaf-1. As a positive control, naïve lysates (30 mg/ml) were also activated with cytochrome c (10 M) and dATP (2 mM) to stimulate in vitro formation of the Apaf-1 apoptosome. --denotes the fusion of two separate gels at the same exposure level. B and C, caspase-9 Ϫ/Ϫ and apaf-1 Ϫ/Ϫ MEFs were incubated for 2 h at 37°C (Control) or were heat-shocked (HS) for 2 h at 44°C, followed by an incubation period of 4 -24 h at 37°C. As a positive control, some cells were also treated with TNF for 24 h at 37°C. Cell death (PS ϩ cells) measurements and caspase-3-like DEVDase activities were determined as described under "Experimental Procedures." Whole cell and mitochondrial pellets were Western-blotted for active caspase-3 (aCasp-3) and cytochrome c (Cyt. c-p), respectively. Rel. FU, relative fluorescence units; Cyt c-p, cytochrome c retained in the mitochondrial pellet.
caspase-8 and FADD-deficient cells (Fig. 6, lanes 10 -12). In contrast, both etoposide and heat shock induced the activation of all caspases and triggered cell death in the parental, caspase-8, and FADD-deficient cells (Fig. 6, lanes 4 -9). Importantly, the absence of caspase-8 had no effect on the release of cytochrome c or Smac from mitochondria in heatshocked cells (Fig. 6, lanes 7 and 8), ruling out its essential role as the apical protease responsible for MOMP. The absence of caspase-8 did slightly inhibit the processing of procaspase-3 (Fig. 6, lanes 7 and 8), but this was likely due to a disruption in the normal caspase amplification loop, wherein once activated downstream of caspase-6, caspase-8 can participate in the cleavage of additional procaspase-3 (27). Thus, neither caspase-8 nor FADD played an essential role in mediating heat shock-induced MOMP, the loss in ⌬ m , caspase-3 activation or apoptosis.
Heat Shock Does Not Require Caspase-2 to Induce MOMP or the Activation of Caspase-3, nor Does It Stimulate the Formation of a RAIDD⅐Caspase-2 PIDDosome Complex-Caspase-2 also appeared to be activated downstream of caspase-3 following heat shock ( Figs. 2A and 5). Nevertheless, to rule out an essential role for caspase-2 in heat shock-induced apoptosis, we examined primary wild-type and caspase-2-deficient MEFs and found that both were equally sensitive to heat shock-induced MOMP, caspase-3 activation, and apoptosis (Fig. 7A). Previous reports, however, have suggested that caspase-2 may be uniquely important for apoptotic signaling in tumor cells (10,28). Therefore, we also heat shocked E6 Jurkat T cells in the presence of Z-VDVAD-fmk, a potent inhibitor of both caspases -2 and -3 (29). Because Z-VDVAD-fmk, similar to DEVD-CHO, can inhibit the activity of processed caspase-3 (29), it prevented the autocatalytic cleavage of caspase-3 from its p20 to its p19/17 forms and inhibited the downstream cleavage of all caspases, including caspase-2 (Fig.  7B, lanes 7 and 8) (data not shown). Nevertheless, Z-VDVAD-fmk failed to  inhibit the initial processing of procaspase-3 to its p20 form and did not prevent MOMP or the loss in ⌬ m , suggesting once again that caspase-2 was not the apical protease responsible for heat shock-induced apoptosis (Fig. 7B, lane 8).
Studies by Tinel and Tschopp (7) suggest that procaspase-2 can be recruited to a RAIDD PIDDosome complex following DNA damage, where it resides in its active but unprocessed form. Therefore, because it was unclear whether Z-VDVAD-fmk could inhibit full-length procaspase-2, we fractionated lysates from control, heat-shocked, and etoposide-treated cells and analyzed them for the formation of RAIDD⅐caspase-2 PIDDosome complexes (7). In control cells, fulllength PIDD eluted primarily in fractions 17-20 (M r ϳ 200,000), consistent with its presence as a dimer, whereas the naturally occurring C-terminal death domain fragment of PIDD (PIDD-DD) eluted in fractions 8 -12 (M r ϳ 800,000) (Fig. 7C, Control). RAIDD and caspase-2 eluted in fractions 22-25 (M r ϳ 60,000) and 19 -23 (M r ϳ 100,000), respectively. In fractionated lysates from both heat shock and etoposide-treated cells, a portion of PIDD-DD eluted in higher molecular weight complexes (M r ϳ 1,000,000 -1,500,000), but unexpectedly neither RAIDD nor caspase-2 were observed in these fractions (Fig. 7C,  fractions 5-7). The RAIDD⅐caspase-2 PIDDosome complex was originally identified using an in vitro model of caspase-2 activation, wherein naïve lysates prepared from PIDD-overexpressing cells were incubated at 37°C for 1 h (7). Therefore, to validate our assay, we examined Jurkat T cell lysates under the same conditions and found that indeed both procaspase-2 and PIDD-DD co-eluted in higher molecular weight fractions 5-14 (M r ϳ400,000 -1,500,000) (Fig. 7C, Lysate). However, RAIDD did not co-elute with this large caspase-2 complex, consistent with a previous report (30). Thus, we could find no evidence for the formation of a "native" RAIDD⅐caspase-2 PIDDosome complex (at least in apoptotic Jurkat T cells) following exposure to heat shock or the DNA-damaging agent etoposide.
Caspases -2 and -8 Do Not Play Redundant Roles in Heat Shockinduced Apoptosis-Although there was no a priori reason to suspect that caspases -2 and -8 might compensate for one another to promote heat shock-induced apoptosis, we heat-shocked caspase-8-deficient Jurkat T cells that were first depleted of caspase-2 by RNA interference or pretreated with the caspase-2 inhibitor Z-VDVAD-fmk. Transient transfection with caspase-2-specific small interfering RNAs reduced the expression levels of caspase-2 by Ͼ90% (Fig. 8A, lanes 1 and 3) in accordance with observed transfection efficiencies. However, loss of caspase-2 expression in caspase-8-deficient cells failed to prevent the activation of caspase-3, a loss in ⌬ m , or the induction of apoptosis (Fig.  8A, lanes 2 and 4). Moreover, pretreatment of caspase-8-deficient Jurkat T cells with Z-VDVAD-fmk did not inhibit the initial processing of caspase-3 to its p20 form but did inhibit autocatalytic processing to its p17 form and the downstream processing of caspase-2, as already described (Fig. 8B, lanes 2 and 4; Fig. 7B, lane 8). More importantly, however, it once again failed to significantly inhibit either the loss in ⌬ m or cell death (Fig. 8B, lanes 2 and 4). Thus, in summary, neither active pro-nor processed caspase-2 played an essential role in triggering heat shock-induced apoptosis, and caspases -2 and -8 did not play redundant roles in promoting heat shock-induced cell death.

DISCUSSION
Elegant studies by Strasser, Vaux, and colleagues (31-33), using apaf-1 Ϫ/Ϫ and caspase-9 Ϫ/Ϫ hematopoietic cells and MEFs, indicate that neither Apaf-1 nor caspase-9 is essential for apoptosis induced by certain pro-apoptotic stimuli, including cytokine withdrawal, dexamethasone, and ␥-irradiation. They propose that, similar to CED-9 in the worm Caenorhabditis elegans, Bcl-2 may regulate the activation of a heretofore unidentified apical caspase in mammals, which directly activates caspase-7 and induces apoptosis. They further argue that this apical caspase may have evolved so as to stimulate MOMP and induce the formation of the Apaf-1⅐caspase-9 apoptosome, which primarily serves to accelerate cell death by activating caspase-3 and initiating a caspase amplification loop. Herein, we have reported that heat shock activates a pathway, which appears to be remarkably similar to that described above, yet contains some significant differences. Indeed, we have found that heat shock activates an apical protease, which is located upstream of mitochondria and is responsible for stimulating MOMP (Figs. 5 and 9). However, this heat-activated protease differs from that previously described (31,32), in that it is effectively inhibited by Z-VAD-fmk, and it can activate caspase-3 independently of the Apaf-1⅐caspase-9 apoptosome (Figs. 2; 3, B and C; and 9). Indeed, even in cells that overexpressed Bcl-2, the heat-activated apical protease still managed to stimulate processing of procaspase-3 to its intermediate p20 form, although it required significantly more time to do so and the overall amount of processing was dramatically reduced (Fig. 5, lane 12). These results are consistent with the notion that Bcl-2 may directly inhibit the activation of the apical protease. However, Bcl-2 also inhibited the release of Smac from mitochondria, which likely pre- vented a caspase amplification loop by preserving the inhibition of caspase-3 by IAPs (Fig. 9) (34).
To date, caspases -2, -8, and -10 (and more controversially caspases -4 and -12) are the only apical caspases activated upstream of mitochondria, which may in turn induce MOMP. During the completion of our studies, Tu et al. (35) reported that caspase-2 is the apical protease responsible for initiating heat shock-induced cell death (35). They identified caspase-2 largely based on the use of a biotinylated analog of VAD-fmk (bVAD-fmk), which they used to "trap" procaspase-2 in situ, early following heat shock (35). Nevertheless, we have demonstrated that neither primary cells from caspase-2-deficient mice nor tumor cells depleted of caspase-2 by RNAi display any defects in MOMP, caspase-3 activation, or apoptosis. Moreover, pharmacological inhibition of caspase-2 failed to inhibit apoptosis, and we could find no evidence for the formation of a trimolecular RAIDD⅐caspase-2 PIDDosome complex in heat-shocked cells (Fig. 7). One possible explanation could be that caspases -2 and -8 might play redundant roles in promoting heat shockinduced apoptosis. However, this could also be ruled out, as cells deficient in either caspase-8 or FADD remained sensitive to heat shock, and neither depletion of caspase-2 by RNA interference nor pharmacologi- , were heat-shocked for 2 h at 45°C, followed by a 4 h incubation at 37°C. C, lysates prepared from control, heat-shocked, or etoposide-treated E6 Jurkat T cells were size-fractionated using Superose-6 gel filtration chromatography. As a positive control, naïve lysates were incubated at 37°C for 1 h, which induces the formation of a caspase-2-activating complex. All cell death (annexin V/PI) and ⌬ m (TMRE) measurements were determined by flow cytometry. Whole cell and mitochondrial pellets, cytosolic fractions, and gel-fractionated lysates were immunoblotted for either caspases, cytochrome c, RAIDD, full-length PIDD (FL) or its naturally occurring death domain fragment (DD). aCasp-3 and Cyt c-p refer to the active forms of caspase-3 and cytochrome c in the mitochondrial pellets, respectively. ‫ء‬ denotes nonspecific bands; ---denotes the fusion of two separate gels at the same exposure level. pro, procaspases. cal inhibition of caspase-2 in the caspase-8-deficient cells, had any effect on heat shock-induced cell death (Fig. 8).
Interestingly, heat shock did induce oligomerization of PIDD-DD into a large molecular weight complex. Benchimol and colleagues (36) have reported that PIDD-induced apoptosis in MEFs absolutely requires the adapter protein RAIDD. However, they also found that MEFs deficient in caspase-2, caspase-8, FADD, or Apaf-1 were largely (or entirely) sensitive to PIDD-induced cell death and that overexpression of a FADD dominant negative in caspase-2-deficient cells failed to enhance their resistance (36). Although not discussed by the authors, it is plausible that the partial resistance afforded the caspase-2-or caspase-8-deficient cells may have been due to a disruption in the normal caspase amplification loop. Given the apparent striking similarities between the PIDD-and heat shock-induced apoptotic pathways, it is intriguing to speculate that heat shock may somehow utilize RAIDD to activate the apical protease.
The identity of the heat-activated apical protease is currently unknown, but it does not appear to be any of the known mammalian initiator caspases previously associated with apoptosis. Subsequent attempts to implicate other cysteine proteases, such as calpains and cathepsins, in heat shock-induced apoptosis have thus far been unsuccessful ( Fig. 4C) (data not shown). Intriguingly, however, heat shock has recently been shown to induce apoptotic-like cell death in plants, and two serine proteases (coined saspases) have been isolated that are reversibly inhibited by Z-VAD-fmk and prefer substrates with an aspartate in the P 1 position (37). Thus, it is possible that heat shock may activate an upstream serine protease in mammals that exhibits caspaselike activity.
As already noted, the heat-activated apical protease stimulated apoptosis in both apaf-1 Ϫ/Ϫ and caspase-9 Ϫ/Ϫ MEFs, and processing of caspase-9 was found to be a late caspase-3-dependent event in Jurkat T cells (Figs. 1-3). However, cytochrome c was released following MOMP, and because it is currently unclear how heat shock inhibits the formation of active apoptosome complexes, we cannot rule out that the Apaf-1⅐caspase-9 apoptosome may amplify the caspase cascade downstream of mitochondria in some cell types, only that it is not essential for caspase-3 activation or apoptosis. Therefore, the identification of the heat-activated apical protease and the mechanism(s) by which it is activated and induces MOMP remain the primary focus of future studies.