Death receptor recruitment of endogenous caspase-10 and apoptosis initiation in the absence of caspase-8.

Caspase-8 is believed to play an obligatory role in apoptosis initiation by death receptors, but the role of its structural relative, caspase-10, remains controversial. Although earlier evidence implicated caspase-10 in apoptosis signaling by CD95L and Apo2L/TRAIL, recent studies indicated that these death receptor ligands recruit caspase-8 but not caspase-10 to their death-inducing signaling complex (DISC) even in presence of abundant caspase-10. We characterized a series of caspase-10-specific antibodies and found that certain commercially available antibodies cross-react with HSP60, shedding new light on previous results. The majority of 55 lung and breast carcinoma cell lines expressed mRNA for both caspase-8 and -10; however, immunoblot analysis revealed that caspase-10 protein expression was more frequently absent than that of caspase-8, suggesting a possible selective pressure against caspase-10 production in cancer cells. In nontransfected cells expressing both caspases, CD95L and Apo2L/TRAIL recruited endogenous caspase-10 as well as caspase-8 to their DISC, where both enzymes were proteolytically processed with similar kinetics. Caspase-10 recruitment required the adaptor FADD/Mort1, and caspase-10 cleavage in vitro required DISC assembly, consistent with the processing of an apoptosis initiator. Cells expressing only one of the caspases underwent ligand-induced apoptosis, indicating that each caspase can initiate apoptosis independently of the other. Thus, apoptosis signaling by death receptors involves not only caspase-8 but also caspase-10, and both caspases may have equally important roles in apoptosis initiation.

CD95 ligand (CD95L, also called Fas or APO-1 ligand) is an important inducer of physiological apoptosis, acting primarily within the immune system (1). A more recently discovered relative of CD95L, Apo2L/TRAIL (Apo2 ligand or tumor necrosis-related apoptosis-inducing ligand) (2,3), is implicated as well in apoptosis control, although its precise biological role is less well defined (4). Like most other tumor necrosis factor (TNF) 1 gene superfamily members, CD95L and Apo2L/TRAIL are homotrimeric, type II transmembrane proteins that interact with corresponding members of the TNF receptor superfamily. CD95L signals apoptosis through the "death" receptor CD95, whereas Apo2L/TRAIL signals apoptosis through the death receptors DR4 and DR5 (5). Death receptors form a subgroup within the TNF receptor superfamily of type I transmembrane proteins that share sequence homology within a cytoplasmic region defined as the "death domain" (6,7). Signaling through death receptors can be modulated by "decoy" receptors that lack functional death domains; DcR3 binds CD95L (8,9), whereas DcR1 (10,11) and DcR2 (12,13) interact with Apo2L/TRAIL. CD95L (14) and Apo2L/TRAIL (15)(16)(17) use similar signaling mechanisms to transduce a pro-apoptotic signal into target cells (18). Both ligands trigger a series of protein-protein interactions that assemble a death-inducing signaling complex (DISC) at the cytoplasmic death domain of the receptor. Upon ligand binding, CD95, DR4, and DR5 recruit the adaptor molecule FADD/Mort1 (19,20) through homophilic death domain interactions. In turn, FADD recruits the zymogen form of the apoptosis-initiating protease caspase-8, through homophilic interaction of "death effector domains" (21,22). The proximity of caspase-8 zymogens in the DISC facilitates activation through self-processing, leading to cleavage of downstream effector caspases that execute the apoptotic death program.
Caspase-10 is closely related by sequence to caspase-8 (23,24). The two enzymes are encoded on the same region of human chromosome 2q33-34 (23,25,26), suggesting that they arose by duplication of one ancestral gene. In vitro experiments demonstrate the ability of caspase-10 to process caspase-3 and caspase-7 (23), and overexpression of caspase-10 induces apoptosis (24). By analogy to caspase-8, it has been proposed that caspase-10 may be involved in apoptosis signaling by death receptors; however, studies investigating this hypothesis have yielded conflicting results. Upon overexpression in transfected cells, TNF receptor 1 and CD95 interacted with caspase-10 (24), and DR4 and DR5 favored recruitment of caspase-10 versus caspase-8 through an adaptor that appeared to be distinct from FADD (10,27). In addition, investigation of patients with the type II autoimmune lymphoproliferative syndrome, who carry at least one inactivating mutation in the caspase-10 gene (28,29), provided evidence for an essential role of caspase-10 in apoptosis induction by Apo2L/TRAIL in lymphocytes and dendritic cells and by CD95L in lymphocytes (28). On the other hand, the results from biochemical analyses of the Apo2L/TRAIL DISC in nontransfected BJAB B lymphoma and Jurkat T leukemia cells, which express both caspase-8 and -10, indicated that DR4 and/or DR5 recruit and activate caspase-8 but not caspase-10 (16,17). Furthermore, studies with caspase-8-deficient brain cancer cell lines (30,31), mutant Jurkat T cell lines (16,17,32,33), or mouse embryonic fibroblasts (34) suggested resistance toward apoptosis induction by Apo2L/TRAIL or CD95L. Transfection of caspase-8-deficient Jurkat cells with a caspase-8 expression plasmid restored sensitivity to both ligands (16,17,32,33). These studies implied that caspase-8 has an obligatory role in death receptor signaling, whereas caspase-10 is neither involved nor important for this function.
To reassess the involvement of caspase-10 in death receptor action, we characterized a panel of antibodies that recognize this protein in its three functional isoforms and used these antibodies to examine death receptor-mediated recruitment and processing of caspase-10 in nontransfected cells. Our results demonstrate that both Apo2L/TRAIL and CD95L recruit caspase-10 to their respective DISCs, where it is processed with kinetics similar to those of caspase-8. We found that the commercially available caspase-10/b antibodies that were used in previous studies to demonstrate that caspase-10 is present in certain cell lines but is not recruited by stimulated death receptors (16,17,32) can cross-react to heat shock protein 60 (Hsp60), which is similar in mass to caspase-10/b, thus providing a new perspective on earlier observations. By investigating cell lines that express either caspase-8 alone or caspase-10 alone, we demonstrate that each caspase can be recruited to the DISC and processed to initiate apoptosis independently of the other. These results suggest that caspase-8 and caspase-10 both may have important physiologic roles in apoptosis initiation downstream of death receptors.
Immunoprecipitation and DISC Analysis-These experiments were done essentially as described (15)(16)(17). If not otherwise stated, the cells (10 7 ) were either stimulated for 10 min or the indicated time with ligand (1 g/ml Apo2L/TRAIL-FLAG or CD95L-FLAG ϩ 2 g/ml M2) or left untreated (unstimulated). After one wash with phosphate-buffered saline the cells were lysed for 30 min on ice with lysis buffer (1% Triton X-100, 150 mM NaCl, 10% glycerol, 20 mM Tris-HCl, pH 7.5, 2 mM EDTA, 0.57 mM phenylmethylsulfonyl fluoride, protease inhibitor mixture (Complete TM , Roche Molecular Biochemicals) and centrifuged at 15,000 ϫ g for 15 min at 4°C. The postnuclear supernatants were collected and rotated at 4°C for 4 -16 h in the presence of 20 l of protein A/G beads (Pierce) alone or with an indicated antibody. After seven washes with lysis buffer, the immunoprecipitations (IPs) were analyzed by SDS-PAGE followed by electroblotting and detection through Western blot (WB).
In Vitro Processing of Caspases by Ligand DISCs or Purified Caspases-Apo2L/TRAIL or CD95L DISCs were immunoprecipitated from 5 ϫ 10 6 BJAB cells. The IPs were washed four times with lysis buffer and three times with reaction buffer (1ϫ phosphate-buffered saline, 5 mM dithiothreitol). Then 40 l of reaction buffer and 0.5 l of [ 35 S]methionine-labeled, in vitro translated caspase-8/a, -10/a, -10/b, or -10/d were added to the beads and incubated at room temperature. The in vitro translations were done following the manufacturer's protocol (TNT ® quick coupled transcription/translation system) (Promega, Madison, WI) using an 35 S-labeled mix of cysteine and methionine (Tran 35 S-Label TM , ICN). Alternatively, the DISC IPs were substituted by adding 0.1 g of purified caspase-8/a or caspase-10/a to the reaction mixture (from PharMingen and BioVision, Palo Alto, CA, respectively). Various time points were taken and analyzed by SDS-PAGE using a precast 10% Bis Tris NuPAGE TM gel system, and the gels were then electrotransferred onto nitrocellulose membranes following the manufacturer's instructions (Novex, San Diego, CA). The blots were air dried for 2 h at 37°C and exposed to BAS-III imaging plates (Fuji Photo Film Inc., Greenwood, SC). The plates were read using the BAS-2000 IP Scanner with the Image Reader V1.2 software and analyzed with the MacBAS V.2.4 software (Fuji Photo Film Inc.).
Transfections-Jurkat (Casp8 def.) were transfected using Lipo-fectAMINE 2000 (Life Technologies, Inc.) essentially following the manufacturer's instructions with slight modifications. Briefly, 5 ϫ 10 5 cells were transfected in 500 l with 8 l of LipofectAMINE 2000 and 5.5 g of total DNA (4 g of green fluorescent protein in pRK5 ϩ 1.5 g of caspase in pCI). After each transfection cells were split and either left untreated or stimulated with 1 g/ml Apo2L/TRAIL-FLAG or CD95L-FLAG (both ligands cross-linked with 2 g/ml M2. After 14 h of incubation, apoptosis was quantified by the percentage of propidium iodide (5 g/ml)-stained cells in the green fluorescent protein-positive population on a FACScan using the CELLQuest software (Becton Dickinson, Franklin Lakes, NJ).
Apoptosis Assays-The cells (2 ϫ 10 6 ) were treated with 1 g/ml Apo2L/TRAIL-FLAG plus 2 g/ml M2 for the indicated time, split, and analyzed for phosphatidylserine exposure, mitochondrial membrane potential (⌬⌿ m ), and DNA fragmentation by FACScan using the CELLQuest software. Following the manufacturer's instructions Annexin V-fluorescein isothiocyanate staining was assayed to detect phosphatidylserine exposure on the cell surface (Trevigen, Inc., Gaithersburg, MD). To measure ⌬⌿ m , the cells were incubated with 3,3dihexyloxacarbocyanine iodide (460 ng/ml; FL-1) (Molecular Probes, Inc., Eugene, OR) for 10 min at 37°C in the dark and analyzed by FACScan. The percentage of DNA fragmentation was quantified through analysis of propidium iodide-stained nuclei as described (38). Changes in morphology upon apoptosis induction were visualized by confocal microscopy and by Hoechst 33342 staining (Molecular Probes). The Hoechst dye was incubated with the cells for 20 min at room temperature and immediately evaluated by fluorescence microscopy.
Characterization of Caspase-10-specific Antibodies and Expression of Caspase-10 Isoforms in Various Cell Lines-Given the controversy in the literature regarding caspase-10 interaction with the Apo2L/TRAIL and CD95L DISCs and the existence of multiple caspase-10 isoforms, we set out to characterize a panel of caspase-10 antibodies for further study. We selected six antibodies (herein designated C10.1-C10.6), which detect different sites within caspase-10 (inferred from the polypeptide used for immunization or determined by the ability to immunoprecipitate specific fragments of the processed, in vitro translated protein) ( Fig. 2A). Four antibodies (C10.2-C10.5) reacted against all three caspase-10 isoforms with no cross-reactivity to caspase-8, as revealed by WB against in vitro translated caspase protein (Fig. 2B). C10.6 showed selective reactivity against caspase-10/b by WB. The C10.1 antibody did not detect the full-length caspase-10 isoforms by WB or IP; however, it detected the product of the first processing event (p47 or p43) from each isoform by IP (Fig. 2, B and C). (By analogy to caspase-8, p47 for caspase-10/b and -10/d and p43 for caspase-10/a constitute the rest of the molecule after the p12 fragment is cleaved off.) The C10.2 antibody detected the full-length form and the p43/p47 fragment of each isoform by IP (Fig. 2C).
Next, we investigated the expression of caspase-10 isoforms in the BJAB and Jurkat cell lines, which were used in previous studies on the interaction of caspase-10 with death receptors. For IP, we used the C10.4 antibody, which recognizes a site within the large protease subunit of the three caspase-10 isoforms; we then visualized caspase-10 by WB with C10.2, which

Caspase-10 Involvement in Death Receptor Signaling
recognizes an epitope within the shared prodomain, or with C10.5, which recognizes a common sequence within the small protease subunit. By this IP/WB approach, C10.2 and C10.5 detected in BJAB cells two protein bands that migrated at positions consistent with the calculated molecular masses of caspase-10/b and -10/d (ϳ59 kDa) and of caspase-10/a (ϳ54 kDa) (Fig. 2D). C10.2 and C10.5 did not detect these bands well by direct WB analysis of BJAB cell lysates, whereas C10.3 did (Fig. 2D). Notably, C10.6, an antibody that was used in previous studies to detect caspase-10 (32, 41), did not identify these bands after IP with C10.4; instead, C10.6 detected a band of slightly higher molecular mass in direct WB analysis of BJAB cell lysates (Fig. 2D, asterisk).
To confirm the identity of the two bands detected by C10.2, C10.5, and C10.3, we performed comparative two-dimensional gel electrophoresis studies comparing in vitro translated, 35 Slabeled caspase-10/a, -10/b, and -10/d with C10.4 IPs from BJAB cells; we detected caspase-10 in these studies by radiography or by WB with C10.5, respectively (Fig. 2E). Each of the in vitro translated isoforms migrated to a position that was consistent with its expected relative molecular mass (M r ) and isoelectric point (pI). Caspase-10/a ran as the smallest protein with the most acidic pI (6.2); caspase-10/d migrated close to caspase-10/a with a larger M r and a higher pI (6.4); and caspase-10/b ran similar to caspase-10/d by size though with a more basic pI (7.4). IP/WB analysis of BJAB cells revealed spots that corresponded to caspase-10/a and -10d but not with -10/b (Fig. 2E), indicating that this cell line expresses caspase-10/a and -10/d but not -10/b. Next, we examined the expression of caspase-10 isoforms in the Jurkat T cell line as well as in two previously described mutant derivative cell lines that are deficient either in FADD or in caspase-8 (32,41). Both parental and FADD-deficient Jurkat cells expressed caspase-10/a and caspase-10/d at levels comparable with those found in BJAB cells; surprisingly, the caspase-8-deficient Jurkat clone expressed substantially less caspase-10/a and -10/d than its sister clones (Fig. 2, E and H). Moreover, none of the Jurkat clones expressed caspase-10/b, in contrast to previous observations suggesting on the basis of WB with C10.6 that parental and caspase-8-deficient Jurkat cell lines express equivalent amounts of caspase-10/b (32,41).
Given these discrepant data on caspase-10 expression, we reasoned that C10.6, a rabbit polyclonal antibody raised against caspase-10/b, may cross-react to another protein with a similar M r . To test this possibility, we performed two-dimensional gel electrophoresis of lysates from parental Jurkat cells; analysis by WB with C10.6 revealed a spot similar in M r to caspase-10/b (ϳ60 kDa) but with a much more acidic pI (comantibodies are specified. C, IP ability of antibodies for DISC-processed,   Fig. 2, F and E). We detected a spot with identical characteristics by IP/WB analysis with C10.6 ( Fig. 2F, middle panel). Mass spectrometric analysis identified this spot as Hsp60 (data not shown), and WB analysis of a C10.6 IP with anti-Hsp60 antibody confirmed this result (Fig. 2F). Further WB analysis demonstrated the ability of C10.6 to react with purified Hsp60 as well as with in vitro translated caspase-10/b (Fig. 2G). Moreover, C10.6 and anti-Hsp60 detected a single band in MCF-7 and HCC2157 breast carcinoma cells, two cell lines that do not express caspase-10 as shown by WB analysis with C10.3 (Fig. 2H). Together, these results indicate that Jurkat and BJAB cells do not express caspase-10/b and that the band recognized by C10.6 in these cells is Hsp60. Several additional antibodies generated against caspase-10/b cross-reacted to Hsp60, including the antibody used in a previous study to investigate Jurkat and BJAB cells (16) (see "Experimental Procedures"). We confirmed that Jurkat and BJAB cell expression of caspase-10/a and -10/d but not -10/b was by reverse transcriptase-polymerase chain reaction analysis with isoformspecific primers (data not shown).
To search for cell lines with deficits in the expression of caspase-8 and/or caspase-10 as potential experimental models, we analyzed a panel of human nonsmall cell lung carcinoma, small cell lung carcinoma, and breast carcinoma cell lines for caspase mRNA expression by quantitative reverse transcriptase-polymerase chain reaction and for caspase protein levels by WB. Of 35 lung carcinoma cell lines, 29 showed caspase-8 mRNA and protein, one showed caspase-8 mRNA but no protein, and five had neither detectable caspase-8 mRNA nor protein (Fig. 3); 20 of these lines had both caspase-10 mRNA and protein, whereas 15 had caspase-10 mRNA but little or no expression of caspase-10 protein. Of 20 breast carcinoma lines tested, all expressed caspase-8 mRNA and protein; four of them also had both caspase-10 mRNA and protein, whereas 16 had detectable caspase-10 mRNA but little or no protein (Fig. 3). Notably, the six lung carcinoma lines that lacked caspase-8 protein also lacked caspase-10 protein. Moreover, 28 of 55 of the cell lines in the entire panel had no detectable caspase-10 protein, despite having caspase-10 mRNA, and only one cell line was negative both for caspase-10 protein and mRNA, namely, HCC2157 breast carcinoma. These results suggest that in these lung and breast carcinoma cell lines, post-transcriptional mechanisms frequently down-regu-late caspase-10 expression. Loss of caspase-8 protein was notably less common and was associated with concomitant absence of the caspase-10 protein. Unlike caspase-8 and -10, FADD and caspase-9 were detected in all the cell lines, albeit at varying levels (Fig. 3). In subsequent experiments, we used the MCF-7 and HCC2157 cell lines to model cells that express caspase-8 but not caspase-10 and used the caspase-8-deficient Jurkat cell line as a model expressing caspase-10 but not caspase-8.
Recruitment of Caspase-10 to the Apo2L/TRAIL and CD95L DISCs-To re-examine caspase-10 recruitment to the DISC with the better characterized antibodies, we first studied the BJAB cell line, which expresses DR4, DR5, and CD95 as well as caspase-8 (isoforms a and b) and caspase-10 (isoforms a and d); both Apo2L/TRAIL and CD95L induce substantial apoptosis in this cell line (15)(16)(17). We incubated BJAB cells with FLAG epitope-tagged Apo2L/TRAIL or CD95L for various amounts of time and immunoprecipitated the associated DISC through each tagged ligand with an anti-FLAG antibody (M2). WB analysis showed recruitment of FADD and caspase-8 into the Apo2L/TRAIL and CD95L DISCs within minutes of ligand addition (Fig. 4), consistent with published observations (15-17, 21, 22). The results obtained with the C10.2 and C10.1 antibodies together suggested that caspase-10 was recruited equally well to both DISCs and subsequently underwent timedependent proteolytic processing (Fig. 4). The C10.2 antibody detected several specific bands, which were visible more clearly in the CD95L DISC: 1) the full-length form of caspase-10/d (p59); 2) the corresponding fragments containing the prodomain still attached to the large subunit (p47 and p43) (resulting from the first cleavage event); and 3) the prodomain alone (p25) (resulting from the second cleavage event). Full-length caspase-10/a (p54) recruited to the CD95L DISC was apparently masked by the band representing the M2 antibody heavy chain (Fig. 4, top right panel, compare 0 time point to other lanes). The signal for these bands was weaker in the C10.2 WB for the Apo2L/TRAIL DISC wherein p25 was most clearly detected; WB with the C10.1 antibody, which preferentially detects the products of the first cleavage event (p47 and p43), confirmed unequivocally that similar recruitment and processing of caspase-10 occurred in both DISCs (Fig. 4). The kinetics for caspase-8 and caspase-10 processing were similar, with some slight variation between experiments ( Fig. 4 and data not  shown). Thus, Apo2L/TRAIL and CD95L recruit not only endogenous FADD and caspase-8 but also caspase-10 to their respective endogenous death receptors in BJAB cells, leading to the processing of both caspases within a similar time frame.
FADD/Mort1 Is an Obligatory Adaptor for Caspase-10 Recruitment to the DISC-Previous work suggested that in overexpression systems, caspase-10 can bind to DR4 and DR5 independently of FADD (10,27). To test the importance of FADD as an adaptor for endogenous caspase-10, we analyzed Apo2L/ TRAIL DISC assembly in the mutant Jurkat cell line that lacks FADD in comparison to parental or caspase-8-deficient Jurkat cells (Fig. 5). As expected, Apo2L/TRAIL co-immunoprecipitated DR5 in all three cell lines (DR4 is not expressed by Jurkat cells (17)). In addition, Apo2L/TRAIL recruited FADD into the DISC in the parental and caspase-8-deficient cells but not in the FADD-deficient cells. Apo2L/TRAIL also recruited caspase-8 and caspase-10 in the parental cells but not in the FADD-deficient cells, indicating that FADD is necessary for ligand-induced recruitment of both caspases. We obtained similar results for the CD95L DISC (Fig. 5). Thus, FADD appears to act as an obligatory adaptor that mediates the binding of caspase-8 and caspase-10 to DR5 and CD95.
Mutually Independent DISC Processing of Caspase-8 and Caspase-10 -To test whether caspase-8 or caspase-10 require each other for cleavage, we examined their DISC recruitment and processing in cell lines that express one caspase but not the other. Analysis of the caspase-8-deficient Jurkat cells line, which expresses low but detectable amounts of caspase-10 ( Fig.  2, E and H), and of the MCF-7 cell line, which expresses caspase-8 but not caspase-10 ( Figs. 2H and 3), demonstrated mutually independent Apo2L/TRAIL DISC recruitment and processing of each caspase (Fig. 5). Processing was evident by the products of the first cleavage step for caspase-8 (p43/p41) and caspase-10 (p47/p43). The amount of caspase-10 recruited to the DISC in caspase-8-deficient Jurkat cells was less than that in the parental Jurkat cells, consistent with the relatively lower caspase-10 expression level in the former cell line (Fig.  5). Thus, caspase-8 and caspase-10 can be recruited to the Apo2L/TRAIL DISC and processed independently of each other. The caspase-8-deficient Jurkat cell line has a reduced level of CD95 (Ref. 32 and data not shown), which precluded reliable analysis of the CD95L DISC in this clone.
Mutually Independent Apoptosis Initiation by Caspase-8 and Caspase-10 -Next, we investigated whether the independent processing of caspase-8 and caspase-10 is sufficient to trigger apoptosis. To this end, we used the caspase-8-deficient Jurkat cell line and the caspase-10-deficient breast carcinoma cell line HCC2157 (we excluded MCF-7 cells from this analysis because they lack caspase-3 and do not show many of the hallmarks of apoptosis). Previous work suggested on the basis of apparent resistance of the caspase-8-deficient Jurkat cell line to Apo2L/ TRAIL and CD95L that caspase-8 is absolutely necessary for apoptosis induction by these ligands; sensitivity to the ligands could be restored by transfection with caspase-8 (16,17,32). We first used the same experimental setting to ask whether caspase-10 could substitute for caspase-8. We transfected expression plasmids encoding each of the three caspase-10 isoforms or a caspase-8/a control into the caspase-8-deficient Jurkat cell line. Apo2L/TRAIL or CD95L induced significant apoptosis in caspase-8-transfected cells (Fig. 6A), consistent with previous data (17,32,33). Importantly, the two ligands induced apoptosis similarly well in cells transfected with individual caspase-10 isoforms (Fig. 6A). Because the caspase-8deficient Jurkat cells express caspase-10, albeit at a low level, the cells transfected with caspase-8 were in fact expressing both caspases. Notably, these cells did not show a greater apoptosis response than the cells transfected with caspase-10 isoforms (which were expressing endogenous plus exogenous caspase-10). Thus, there did not appear to be a synergistic induction of apoptosis in Jurkat cells containing both caspase-8 and caspase-10.
In the previous studies that gleaned an essential role for caspase-8 in apoptosis signaling by Apo2L/TRAIL or CD95L, the caspase-8-deficient Jurkat clone was assessed for apoptosis up to 8, 12, or 16 h after ligand addition (16,17,32,33). Given the ability of transfected caspase-10 to mediate ligand-induced apoptosis in the absence of caspase-8 (Fig. 6A) and the low level of caspase-10 in the untransfected caspase-8-deficient Jurkat cells (Fig. 2, E and H), we tested whether apoptosis induction in these cells might be delayed rather than abolished. Apo2L/ TRAIL induced little apoptosis in the caspase-8-deficient Jurkat cells up to 16 h; however, by 24 -48 h, these cells showed significant evidence of apoptosis, namely, increases in cell surface phosphatidylserine, loss of mitochondrial potential, and DNA fragmentation (Fig. 6B), as well as characteristic morphological changes including nuclear fragmentation (Fig. 6C). Together with the results of the transfection experiments (Fig.  6A), these data suggest that caspase-10 can mediate apoptosis independently of caspase-8. The delayed apoptosis induction in the caspase-8-deficient Jurkat cells may be explained by the lower levels of caspase-10 in this cell line as compared with the parental Jurkat cells.
Next, we analyzed the HCC2157 cell line, which expresses caspase-8 but shows no detectable caspase-10 (Figs. 2H and 6D). When incubated with Apo2L/TRAIL for 48 h, this cell line exhibited typical morphologic signs of apoptosis such as cell detachment, membrane blebbing, and nuclear fragmentation (Fig. 6C). The tight clustering of HCC2157 cells in culture prohibited flow cytometric analysis of apoptosis; however, these cells showed clear evidence for cleavage of caspase-8, caspase-3, and PARP, indicating Apo2L/TRAIL-induced apoptosis. These data suggest that caspase-8 and caspase-10 each can trigger apoptosis downstream of Apo2L/TRAIL in a mutually independent fashion.

DISCUSSION
Although Caspase-8 is thought to be essential for apoptosis induction by death receptors, the role of caspase-10 is controversial. A number of studies indicate that caspase-10 is important for apoptosis signaling by CD95L and Apo2L/TRAIL (10,27,28), whereas more recent reports suggest that caspase-10 is not important or even not involved in signaling by these ligands (16,17,32). On the basis of well characterized antibodies, our experiments with nontransfected cells demonstrate that Apo2L/TRAIL and CD95L recruit both caspase-8 and caspase-10 to their respective DISCs, and that FADD is an obligatory adaptor for both caspases. We observed indistinguishable kinetics for DISC recruitment and processing of caspase-8 and caspase-10, supporting the idea that both molecules act as initiator caspases. (There was also little kinetic difference in caspase recruitment and processing by Apo2L/ TRAIL as compared with CD95L, reinforcing the previous notion that these ligands operate similarly.) In a cell-free system, purified caspase-8 or caspase-10 enzymes readily processed in vitro translated caspase-3 as a substrate, but not caspsase-8 or caspase-10; in contrast, the isolated DISCs assembled by Apo2L/TRAIL or CD95L in BJAB cells efficiently processed caspase-8 as well as caspase-10. These data suggest that DISC assembly is important for processing of caspase-10, further supporting the conclusion that caspase-10 can act as an apoptosis initiator immediately downstream of death receptors. By using cells that express either caspase-8 or caspase-10 alone, we demonstrated DISC association and processing of each enzyme independently of the other and apoptosis initiation by each enzyme in absence of the other in response to death receptor stimulation. Further work will be needed to examine whether both caspase-8 and -10 can co-associate physically with the same DISC and whether each can activate the other in this kind of heterotypic complex.
Our experiments reveal that caspase-8-deficient Jurkat cells, which were previously believed to be resistant to Apo2L/ TRAIL and CD95L (16,17,32), in fact undergo a significant, albeit delayed apoptotic response. The low levels of caspase-10 in these cells as compared with the parental Jurkat cells, which probably were not recognized in the earlier work because of caspase-10/b antibody cross-reactivity to Hsp60, could explain the delay in apoptosis in response to death receptor ligands. It FIG. 6. Mutually independent apoptosis induction by caspase-10 and caspase-8. A, Jurkat (Casp8 def.) cells were transfected with expression plasmids encoding green fluorescent protein and the indicated caspase isoform. The cells were incubated for 14 h in the absence or presence of the indicated ligand (1 g/ml) plus M2 antibody (2 g/ml) before determination of cell death by propidium iodide staining of the green fluorescent protein-positive population. The means Ϯ S.D. of three independent experiments are shown ("specific" indicates background-subtracted). B, Jurkat (Casp-8 def.) cells were incubated for various amounts of time in the absence or presence of Apo2L/TRAIL-FLAG (1 g/ml) plus M2 antibody (2 g/ml), and apoptotic features were measured by flow cytometry. The data are presented as the means Ϯ S.D. of three independent experiments. C, indicated cell lines were left untreated (0 h) or treated with Apo2L/TRAIL as in A for 48 h and analyzed by light (confocal) or fluorescence (Hoechst staining) confocal microscopy after staining with Hoechst 33342. D, lysates of HCC2157 cells untreated or treated with Apo2L/TRAIL as in B were analyzed for processing of caspase-10, -8, and -3 and PARP. For caspase-3 the lower molecular weight cleavage products (p20, p17, and p11) are visible. For PARP the full-length (upper panel) and the p20 fragment (lower panel) are seen. Time of treatment in hours is indicated above the lanes.
is also possible that the specific absence of caspase-8 attenuates apoptosis in these cells. This is less likely, however, because the transfection experiments indicated that caspase-8 and caspase-10 are equally potent at mediating ligand-induced apoptosis. These observations, together with the finding that both caspase-8 and caspase-10 can function independently of each other, raise the question of whether the roles of these two related enzymes are redundant or whether they each have distinct and perhaps complementary functions. Embryonic fibroblasts from caspase-8 knockout mice are resistant to apoptosis induction through TNF receptor 1, DR3, or CD95, demonstrating an essential role for caspase-8 (34). However, a murine counterpart of human caspase-10 has yet to be identified, which leaves the question of potential redundancy in human cells open. It is notable in this regard that lack of caspase-10 protein expression despite mRNA presence in the 55 lung and breast cancer cell lines we analyzed was markedly more frequent than loss of caspase-8. Furthermore, all cell lines deficient in caspase-8 protein lacked caspase-10 protein as well. Given the emerging roles of death receptors in antitumor immune surveillance (4,42), these observations may reflect a greater selective pressure for apoptosis evasion in tumor cells through down-regulation of caspase-10 as compared with caspase-8, perhaps coupled with a greater propensity for down-regulation of caspase-10. Deletion or silencing of the caspase-8 gene has been observed in some childhood neuroblastomas, although the status of caspase-10 in these cells is unclear because it was analyzed with the Hsp60-cross-reactive C10.6 antibody (31). There is additional evidence for loss of caspase-8 expression in certain primitive neuroectodermal brain tumors, which appears to correlate with resistance to Apo2L/TRAIL-induced cell death (30); however, caspase-10 expression in these tumors has yet to be investigated.
In conclusion, we have established in nontransfected cells that caspase-10 is an integral part of the Apo2L/TRAIL and CD95L DISCs and that it can trigger apoptosis independently of caspase-8. Thus, both caspases may have an important physiologic role in apoptosis initiation downstream of death receptors.