INTRODUCTION

Apoptosis, or programmed cell death, plays an essential role in development, homeostasis, and defense in multicellular organisms (1, 2). Several cell surface receptors, such as CD95 (APO-1/Fas), TNF¹ receptor 1, DR3 (APO-3/TRAMP/Wsl-1/LARD), and DR4 (TRAILR) (3) belonging to the TNF receptor/nerve growth factor receptor superfamily, have been shown to trigger apoptosis upon binding of their cognate ligands or specific agonistic antibodies. Stimulation of CD95 has been shown to result in aggregation of its intracellular death domains, leading to the recruitment of a set of signaling proteins (CAP1-4) and the formation of the death-inducing signaling complex (DISC) (4, 5). In the DISC, CAP1 and CAP2 were identified as the adapter molecule FADD (MORT-1) (4, 6, 7) that couples through its C-terminal death domain to the cross-linked CD95 receptor. The N-terminal death effector domain of FADD enables recruitment of CAP4, which was identified as FLICE (MACH1/MCH5/caspase-8) (8-10).

FLICE belongs to a family of cysteine proteases (caspases, related to the Caenorhabditis elegans cell death gene ced-3 that have been shown to play a key role in the induction of most forms of apoptosis (11). Caspases are synthesized as inactive proenzymes that have to be activated by proteolytic cleavage after specific aspartate residues (12). Recently, we have shown that FLICE is activated by association with the CD95 DISC, leading to the release of the active subunits p18 and p10 into the cytosol (13). There, they can activate other caspases, in turn resulting in the specific cleavage of a number of "death substrates." During CD95 triggering all cytosolic FLICE is activated at the DISC (13). After activation at the DISC a part of the FLICE prodomain remains bound to the DISC.

A large number of caspases have been identified including caspase-1 (ICE) (14, 15), caspase-2 (ICH-1/Nedd-2) (16, 17), caspase-3 (CPP32/Yama/apopain) (18-20), caspase-4 (ICH-2/TX/ICE-rel-II) (21-23), caspase-5 (ICE-rel-III/TY) (23, 24), caspase-6 (Mch2) (25), caspase-7 (Mch3/ICE-LAP3/CMH-1) (26-28), caspase-8 (FLICE/MACH/Mch5) (8-10), caspase-9 (Mch6/ICE-LAP6) (29, 30), caspase-10 (Mch4/FLICE2) (10, 31), and caspase-11 (ICH-3) (32). However, the role of these caspases in different cell death pathways in various tissues remains elusive. Caspase-3, for example, has been shown to be proteolytically activated upon CD95-induced cell death. In mice deficient of caspase-3, however, the CD95 apoptosis pathway was not affected in most tissues (33).

In addition to the large number of different caspases, various isoforms of these molecules have been described at the mRNA level. Some of these isoforms have been found to be inactive splice variants such as ICE (34), MCH2 (25), or MCH3 (26). Others function as dominant inhibitors of apoptosis such as ICE (34) or ICH-1s (17). For caspase-8, eight different isoforms (designated as caspase-8/a-h), including FLICE (CAP4/MACH1) (8), MACH2 and MACH3, MACH1-4 (9), and Mch5 (10) have been described at the mRNA level.

In this study, expression of the different FLICE isoforms on the protein level in various cell lines was determined by monoclonal antibodies covering the three functional domains of caspase-8. Only two caspase-8 isoforms were detected on the protein level of all cell lines tested. Both isoforms were recruited and activated by the CD95 DISC with identical kinetics.

EXPERIMENTAL PROCEDURES

The monocytic cell line MonoMac, the T-cell lymphoma HUT78, the B-lymphoblastid cell line SKW6.4, the Burkitt lymphoma Raji, the Burkitt-like lymphoma BJAB, the colon carcinoma cell line HT-29, the breast carcinoma cell line MCF-7, the cervix carcinoma HeLa, the myorhabdosarcoma cell line KYM-1 (kind gift from M. Grell, Stuttgart, Germany), the small lung cell carcinoma line SCLC22H (kind gift from J. Fischer, Heidelberg, Germany), and the neuroblastoma SHEP (kind gift from M. Schwab, Heidelberg, Germany) were cultured in RPMI + 10% fetal calf serum, 0.05 mg/ml gentamycin, and 0.05 mg/ml HEPES. The hepatoma cell line HepG2, the gastric cancer line HS746T (both were a kind gift from M. Müller-Schilling, Heidelberg, Germany), and the embryonic kidney line 293T were cultured in Dulbecco's modified Eagle's medium + 10% fetal calf serum, 0.05 mg/ml gentamycin, and 0.05 mg/ml HEPES. All cells were of human origin.

The affinity-purified rabbit anti-peptide antibodies anti-FLICE-N and anti-FLICE-C against the FLICE peptides 183-201 and 466-479, respectively, were generated as described previously (13). For the anti-FLICE mAbs, BALB/c mice were immunized four times by injection of 300 µg of either purified GST-N-FLICE or GST-C-FLICE. Spleen cells from immunized animals were fused with the Ag8 myeloma. 2 weeks after fusion culture supernatants from wells positive for growth were tested in an enzyme-linked immunosorbent assay with HIS-FLICE as coated antigen. Hybridomas that produced anti-FLICE mAbs were cloned several times by limited dilution yielding subclones positive for the desired antibody. The anti-FLICE mAbs used in this study were C5 (IgG2a), C15 (IgG2b), and N2 (IgG1). The mouse mAb anti-APO-1 (IgG3, k) recognizes an epitope on the extracellular part of human APO-1 (CD95/Fas) (35). The horseradish peroxidase-conjugated goat anti-mouse IgG1, IgG2a, and IgG2b were purchased from Dianova (Hamburg, Germany). All chemicals used were of analytical grade and purchased from Merck (Darmstadt, Germany) or Sigma.

Using standard polymerase chain reaction and cloning techniques the following fusion proteins were generated: HIS-FLICE, GST-N-FLICE (amino acids 1-180), and GST-C-FLICE (amino acids 181-478). Fusion proteins were purified as described previously (4). For immunoprecipitation mAbs (10 µg) were coupled to anti-IgG1 Agarose beads (Sigma) (N2), to protein A Sepharose beads (Sigma) (C15), or to protein A/G-plus Agarose (Santa Cruz Biotechnology) (C5). After addition of in vitro activated [³⁵S]FLICE and incubation for more than 1 h at 4 °C, beads were washed three times with lysis buffer. The amount of DISC-associated FLICE was determined as follows: 5 × 10⁶ SKW6.4 cells were either first treated with 2 µg/ml anti-APO-1 for 5 min at 37 °C and then lysed (stimulated condition) or first lysed and then supplemented with 2 µg/ml anti-APO-1 (unstimulated condition). ³⁵S labeling, cell lysis, and immunoprecipitation of CD95 were done as described elsewhere (4).

For Western blot detection of cytosolic proteins postnuclear supernatants equivalent to 1 × 10⁶ cells or 50 µg of total protein as determined by the BCA method (Pierce) were separated by 12% SDS-PAGE. After electrophoresis all samples were transferred to Hybond nitrocellulose membrane (Amersham Corp.), blocked with 2% bovine serum albumin in PBS/Tween (PBS + 0.05% Tween 20) for at least 1 h, washed with PBS/Tween, and incubated with supernatant of anti-FLICE hybridomas diluted 1:5 in PBS/Tween for 16 h at 4 °C. Blots were washed with PBS/Tween and developed with goat anti-mouse IgG1 (N2), IgG2a (C5), or IgG2b (C15) (1:20000). After washing with PBS/Tween, the blots were developed with the chemiluminescence method (ECL) following the manufacturer's protocol (Amersham Corp.).

The CD95 DISC was immunoprecipitated from 5 × 10⁷ anti-APO-1-treated SKW6.4 cells (5 min) as described above, and immunoprecipitates were incubated with in vitro translated ³⁵S-labeled FLICE (TNT, T7 coupled reticulocyte lysate system, Promega) in FLICE cleavage buffer (50 mM HEPES, pH 7.4, 100 mM NaCl, 0.1% CHAPS, 10 mM dithiothreitol, and 10% sucrose) for 24 h at 4 °C. The cleavage reactions were stopped by addition of 1% SDS. After boiling for 3 min, samples were diluted 1:10 in lysis buffer and subjected to immunoprecipitation as described above. The immunoprecipitates were separated on 15% SDS-PAGE, and the amplified dried gels were subjected to autoradiography.

RESULTS

Using either GST fusion proteins with recombinant FLICE prodomain (GST-N-FLICE, amino acids 1-180) or the protease domain (GST-C-FLICE, amino acids 181-479) as an immunogen, different mouse anti-FLICE monoclonal antibodies were generated. Specificity of the antibodies was first established by Western blotting on recombinant GST-N-FLICE or GST-C-FLICE. The antibody N2 reacted only with GST-N-FLICE, whereas the antibodies C15 and C5 reacted only with GST-C-FLICE (Fig. 1A). We have recently shown that in vitro translated ³⁵S-labeled FLICE can be activated in vitro by incubation with the immunoprecipitated DISC, resulting in the formation of cleavage intermediates p43 and p12, the prodomain p26, and the active subunits p18 and p10 (Fig. 1C and Ref. 13). For further characterization of the antibodies in vitro activated ³⁵S-labeled FLICE was used in immunoprecipitation experiments (Fig. 1B). In addition to full-length FLICE, the N2 antibody immunoprecipitated the p43 and the p26 cleavage products, both containing the prodomain of FLICE. Interestingly, the band below full-length FLICE was not immunoprecipitated by N2 (Fig. 1B). This band represents N-terminal truncated FLICE due to the usage of an internal start site in the in vitro translation (13). Therefore, the N2 mAb recognizes an epitope located within the first death effector domain of FLICE. The C15 mAb directed against the C terminus of FLICE immunoprecipitated all FLICE cleavage products containing the p18 domain. The antibody C5 precipitated p12 and p10, both representing the very C terminus of FLICE. Therefore, C5 is directed against the p10 subunit of FLICE.

[View Larger Version of this Image (31K GIF file)]

A number of isoforms of caspase-8 have been described at the mRNA level (9, 10). Our mAb against the three major domains of FLICE (the prodomain and the active subunits p18 and p10) enabled us to test which of the reported caspase-8 isoforms were actually expressed in vivo. To this end several cell lines representing different tissues were tested for FLICE expression by Western blotting using the N2, C15, and C5 anti-FLICE mAbs (Fig. 2). Surprisingly, all three antibodies detected only two bands of 55 and 53 kDa of equal intensity in almost all cells. Other caspase-8 isoforms were undetectable. The only reported caspase-8 isoform that was not expected to be detected with the antibodies used was caspase-8/e (Fig. 3 and Table I).

[View Larger Version of this Image (34K GIF file)]

[View Larger Version of this Image (26K GIF file)]

Table I. Recognition of caspase-8 isoforms by different antibodies Anti-FLICE-N and anti-FLICE-C are rabbit anti-peptide antibodies generated against the FLICE peptides 183-201 and 466-479, respectively. N2, C15, and C5 are mouse monoclonal antibodies recognizing the prodomain, the p18 subunit, and the p10 subunit of FLICE, respectively. + indicates possible recognition by antibodies, and  indicates the absence of a binding site for the indicated antibodies. Cases where binding of an antibody cannot be predicted because the exact epitope for the anti-FLICE mAbs C15 and C5 is not known are labeled with question marks. Caspase anti-FLICE-N anti-FLICE-C N26 C15 C5 8/a + + + + + 8/b   + + + + 8/c   +   ? + 8/d +   +     8/e           8/f +   + ?   8/g     + ?   8/h   + + + +

Expression levels of the two detected caspase-8 isoforms were very different, spanning a range from high expression in the B-cell line SKW6.4 or the myorhabdosarcoma cell line KYM-1 to low expression in the embryonic kidney cell line 293T. Interestingly, the small lung cell carcinoma line SCLC22H was negative for FLICE expression. The third band detected only by the C5 mAb in the hepatoma line HepG2 likely represents a nonspecific background band because a FLICE isoform of this size (43 kDa) should contain the p18 subunit of FLICE and should therefore also be detected by the C15 mAb. After longer exposure of the three immunoblots, additional bands were detectable (data not shown), most of which represent unspecific binding of the anti-FLICE antibodies with cellular proteins. In summary, of the eight isoforms of caspase-8 described at the mRNA level, only two are expressed as proteins in significant amounts in all 13 cell lines tested.

We have recently shown that FLICE is recruited to the CD95 receptor in a stimulation-dependent manner forming the DISC (4, 8). To test whether the second FLICE isoform is also recruited to the CD95 receptor, we analyzed the DISC by one-dimensional as well as by two-dimensional Western blotting, using the C15 anti-FLICE mAb. As shown in Fig. 4A, both FLICE isoforms were recruited to the CD95 receptor in a stimulation-dependent manner. The comparison between the two-dimensional Western blot and the DISC precipitation from ³⁵S-labeled cells (Fig. 4B) confirmed that the upper FLICE isoform was identical with CAP4, whereas the lower isoform was hidden underneath a background spot in the ³⁵S-DISC precipitation that was also detected under unstimulated conditions.

[View Larger Version of this Image (48K GIF file)]

As both FLICE isoforms were recruited to the DISC we tested next whether both are cleaved upon triggering of CD95. Therefore, we analyzed the lysate of either untreated cells or cells stimulated with the anti-APO-1 antibody for 1 h in a Western blot experiment using the three different anti-FLICE mAbs. Prolonged stimulation of CD95 resulted in almost complete cleavage of both FLICE isoforms (Fig. 5A, lanes 2, 4, and 6). Cleavage of both full-length FLICE bands during stimulation resulted in the formation of only one p18 and p10 cleavage product as detected by the C terminus-specific mAbs C15 and C5, respectively indicating that both isoforms did not differ in their C terminus ICE-like domains (Fig. 5A, lanes 1-4). To confirm this we made use of the rabbit antibody anti-FLICE-C, which was directed against the very C terminus. This antibody was able to precipitate both FLICE isoforms, confirming that they did not differ in their C terminus (Fig. 5A, lanes 9 and 10). However, the N terminus-specific anti-FLICE mAb N2 detected two different FLICE cleavage products, p26 and p24, after CD95 stimulation (Fig. 5A, lanes 5 and 6). In addition, two bands, p43 and p41, were also detected, representing intermediates after cleavage between the p18 and the p10 subunit of FLICE, because they were also weakly detected by the C15 antibody (Fig. 5A, lanes 1, 2, 5, and 6). Therefore, the two isoforms of FLICE differed in the size of their prodomains. Given the molecular weight and pI of 53 kDa and 4.91, respectively, and the difference in the prodomain, the described isoform of caspase-8 that most likely represents the second band is caspase-8/b (MACH2) (Fig. 3). This isoform differs from FLICE in a box of 15 amino acids that is not present in caspase-8/b (9). To further test this assumption we made use of the rabbit antibody anti-FLICE-N, which was directed against these 15 amino acids. Using SDS-boiled lysates from unstimulated cells the anti-FLICE-N antibody only precipitated the upper FLICE band (Fig 5A, lanes 7 and 8). In SDS-boiled lysates from stimulated cells, the antibody only recognized the p43 intermediate and the p26 prodomain. We therefore conclude that the second FLICE isoform that is expressed on the protein level is caspase-8/b that lacks the 15 amino acids recognized by the FLICE-N antibody.

[View Larger Version of this Image (48K GIF file)]

We have recently described two novel DISC components CAP5 and CAP6 representing the prodomain of FLICE after proteolytic activation (13). To test whether these new DISC components represent the two prodomains of the expressed FLICE isoforms we analyzed the DISC in a two-dimensional Western blot using the N2 anti-FLICE mAb. The p26 and p24 cleavage products comigrated with CAP5 and CAP6, respectively (Fig. 5B and data not shown). Only CAP5 was precipitated by the anti-FLICE-N antibody, confirming that CAP6 represents the prodomain of caspase-8/b that does not contain the epitope recognized by anti-FLICE-N, whereas CAP5 represents the prodomain of caspase-8/a.

Because both FLICE isoforms were recruited to and activated by the CD95 DISC, we tested next whether they showed any differences in cleavage kinetics. Therefore, we analyzed the cleavage of caspase-8/a and 8/b at the DISC level and in the cytosol at various time points after CD95 stimulation. Consistent with the fast kinetics of CD95-mediated apoptosis both isoforms were recruited to the DISC within 10 s (Fig. 6A). Also after 10 s the two cleavage intermediates p43 and p41 as well as the prodomain cleavage products p26 and p24 were detectable in the DISC. In the cytosol all the FLICE cleavage products p26, p24, p18, and p10 were detectable as early as 10 s after activation (Fig. 6B). The cleavage products p26 and p24 as a readout for the activation of caspase-8/a and 8/b, respectively, increased in the cytosol during stimulation with identical kinetics, demonstrating that both isoforms were activated simultaneously. Interestingly, at the DISC level there was only a slight increase in the p26 and p24 cleavage products starting after 10 min when the amount of full-length FLICE began to decline (Fig. 6A). This suggests that there is only a limited capacity of the DISC to bind death effector domain-containing proteins. Notably, FLICE cleavage at the DISC level preceded cleavage in the cytosol, confirming that FLICE turnover takes place at the DISC level where all cytosolic FLICE is processed (Fig. 6B and Ref. 13). Taken together, our data show that only two different isoforms of caspase-8 are expressed as proteins and that both are activated simultaneously upon CD95 triggering.

[View Larger Version of this Image (32K GIF file)]

DISCUSSION

Signal transduction through the CD95 receptor has been shown to involve activation of caspases. Recently, a new member of this family of cysteine proteases, FLICE (MACH/MCH5/caspase-8) was cloned (9, 10) that was identified to be recruited to the CD95 receptor in a stimulation-dependent manner (8, 4). Therefore, FLICE is the most upstream caspase in the CD95-induced apoptotic pathway. Several different isoforms of caspase-8 have been described as cDNA clones that were identified either by yeast two-hybrid screening (9) or by data base search for homologous expressed sequence tag sequences (10). Originally, only FLICE (caspase-8/a) was shown to be expressed on the protein level, because it was cloned by purification of a protein, CAP4, that specifically associated with the CD95 DISC (8, 4). To test if any of the other caspase-8 isoforms were expressed on the protein level, we developed specific anti-FLICE mAbs covering all different functional domains of FLICE. Only one of the described isoforms (caspase-8/e) was not expected to be detectable by these mAbs (Fig. 3 and Table I). However, such a truncated version of caspase-8 may not have a physiological function. Using the anti-FLICE mAbs we now demonstrate that only two caspase-8 isoforms are expressed at detectable levels in a number of different cell lines. Interestingly, the Burkitt lymphoma line Raji also expressed only these two isoforms as protein, although five different caspase-8 mRNA species were cloned from this cell line (9).

Both expressed caspase-8 isoforms were recruited to the CD95 receptor in an activation-dependent manner. By comparison of the two-dimensional Western blot with immunoprecipitated DISC from ³⁵S-labeled cells, we could confirm that the expressed caspase-8 isoform of 55 kDa corresponds to FLICE (caspase-8/a, MACH1), originally described as CAP4 (4). We identified the second expressed caspase-8 isoform as caspase8/b (MACH2), because this protein could not be immunoprecipitated by the rabbit anti-peptide antibody anti-FLICE-N directed against a box of 15 amino acids not present in caspase-8/b (MACH2).

Different caspase-8 isoforms have been suggested to function as modulators of the activation of caspase-8 in CD95- or TNF-induced apoptosis (9). Caspase-8/c (MACH3) has been demonstrated to protect against CD95- and TNF-induced apoptosis, whereas caspase-8/d (MACH1) was suggested to enhance the cytotoxic activity of the active caspase-8 isoforms (caspase-8/a and 8/b) (9). However, none of these isoforms were detected by the anti-FLICE mAbs in significant amounts in any of the cell lines tested. Whether both expressed active isoforms caspase-8/a and 8/b have different functions remains to be determined. The fact that both were expressed in a one to one ratio in all cell lines tested as well as the identical CD95-induced activation kinetics suggests the possibility that both isoforms are necessary in equal amounts for the signal transduction of the CD95 receptor.

Recently, the number of apoptosis-inducing receptors has increased. The fact that FLICE was expressed in almost every cell line, some of which do not have the CD95 receptor or do not respond to CD95 triggering, raises the possibility that FLICE is also utilized by the signaling pathways of the other "death receptors" such as TNF receptor 1, DR3 (APO-3/TRAMP/Wsl-1/LARD), or DR4 (TRAILR) (3). Future studies should clarify the involvement of the different caspase-8 isoforms in other cell death signaling pathways.