I-FLICE, a Novel Inhibitor of Tumor Necrosis Factor Receptor-1- and CD-95-induced Apoptosis*

The pivotal discovery that the death proteases caspase 8 (FLICE) and caspase 10 (Mch4/FLICE2) are recruited to the CD-95 and tumor necrosis factor receptor-1 signaling complexes suggested a mechanism used by these cytotoxic receptors to initiate apoptosis. In this report, we describe the cloning and characterization of I-FLICE, a novel inhibitor of tumor necrosis factor receptor-1- and CD-95-induced apoptosis. The overall architecture of I-FLICE is strikingly similar to that of FLICE and Mch4/FLICE2. However, I-FLICE lacks both a catalytic active site and residues that form the substrate binding pocket, in keeping with its dominant negative inhibitory function. I-FLICE is the first example of a catalytically inert caspase that can inhibit apoptosis.

The pivotal discovery that the death proteases caspase 8 (FLICE) and caspase 10 (Mch4/FLICE2) are recruited to the CD-95 and tumor necrosis factor receptor-1 signaling complexes suggested a mechanism used by these cytotoxic receptors to initiate apoptosis. In this report, we describe the cloning and characterization of I-FLICE, a novel inhibitor of tumor necrosis factor receptor-1-and CD-95-induced apoptosis. The overall architecture of I-FLICE is strikingly similar to that of FLICE and Mch4/FLICE2. However, I-FLICE lacks both a catalytic active site and residues that form the substrate binding pocket, in keeping with its dominant negative inhibitory function. I-FLICE is the first example of a catalytically inert caspase that can inhibit apoptosis.
The cell death machinery is conserved throughout evolution and is composed of activators, inhibitors, and effectors (1). The effector arm of the cell death pathway is composed of a rapidly growing family of cysteine aspartate-specific proteases termed caspases (2). As implied by the name, these cysteine proteases cleave substrates following an aspartate residue (2,3). Caspases are normally present as single polypeptide zymogens and contain an N-terminal prodomain and large and small catalytic subunits (4 -6). The two-chain active enzyme (composed of the large and small subunits) is obtained following proteolytic processing at internal Asp residues (4 -6). As such, caspases are capable of activating each other in a manner analogous to zymogen activation that is observed in the coagulation cascade (7). The identification of FLICE and Mch4/ FLICE2 as receptor-associated caspases suggested a surprisingly direct mechanism for activation of the death pathway by the cytotoxic receptors CD-95 and TNFR-1 1 (7-10). Upon activation, both receptors use their death domains to bind the corresponding domain in the adaptor molecule FADD (Fasassociated death domain protein) (8 -10). Dominant negative versions of FADD that lack the N-terminal segment but still retain the death domain potently inhibit both CD-95-and TNFR-1-induced apoptosis (11,12). Given the importance of the N-terminal segment in engaging the death pathway, it has been termed the death effector domain (DED) (11).
Remarkably, the DED is present within the prodomain of FLICE and Mch4/FLICE2, and mutagenesis studies suggest that a homophilic interaction between the DED of FADD and the corresponding domain in FLICE or Mch4/FLICE2 is responsible for the recruitment of these proteases to the CD-95 and TNFR-1 signaling complexes (8 -11). Taken together, these data are consistent with FLICE and Mch4/FLICE2 being apical enzymes that initiate precipitous proteolytic processing of downstream caspases resulting in apoptosis (7,(13)(14)(15). A number of viral gene products antagonize CD-95-and TNFR-1mediated killing as a means to persist in the infected host (16). The poxvirus-encoded serpin CrmA and baculovirus gene product p35 are direct caspase inhibitors (3). In contrast, the molluscum contagiosum virus protein MC159 and the equine herpesvirus protein E8 encode DED-containing decoy molecules that bind to either FADD (MC159) or FLICE (E8) and disrupt assembly of the receptor signaling complex, thereby abrogating the death signal (17)(18)(19). The existence of these viral inhibitors has raised the question of whether functionally equivalent molecules are encoded in the mammalian genome.
Here, we report the cloning and characterization of a novel mammalian inhibitor designated I-FLICE (for inhibitor of ϪFLICE), a catalytically inactive structural homologue of FLICE and Mch4/FLICE2 that inhibits both TNFR-1-and CD-95-induced apoptosis. This is the first example of a naturally occurring catalytically inactive caspase that can act as a dominant negative inhibitor of apoptosis.
Cloning of I-FLICE-cDNAs corresponding to the partial open reading frame of I-FLICE were identified as sequences homologous to FLICE and Mch4/FLICE2 on searching the Human Genome Sciences data base using established expressed sequence tag methods (21,22). Full-length cDNAs were obtained by screening a random-primed human umbilical vein endothelial cell cDNA library constructed in the pcDNA1 vector (Invitrogen). The sequence of I-FLICE was confirmed by sequencing plasmid DNA template on both strands by the dideoxy chain termination method employing modified T7 DNA polymerase (Sequenase, U. S. Biochemical Corp.) Northern Blotting-Human multiple tissue and human cancer cell line poly(A) ϩ RNA blots were obtained from CLONTECH and processed according to the manufacturer's instructions.
Transfection, Coimmunoprecipitation, and Western Analysis-Transient transfections of 293T cells were performed as described previously (23). Cells were harvested 40 h following transfection, immunoprecipitated with the indicated antibodies, and analyzed by immunoblotting.
Cell Death Assay-Human embryonic kidney 293 (for TNFR-1 killing) or 293 EBNA cells (for CD-95 killing) were transiently transfected with 0.1 g of the reporter plasmid pCMV ␤-galactosidase plus 0.5 g of test plasmid in the presence or the absence of 2.0 g of inhibitory plasmids. 22-24 h following transfection, cells were fixed in 0.5% glutaraldehyde and stained with 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside. Percentage of apoptotic cells was determined by calculating the fraction of membrane blebbed blue cells as a function of total blue cells. All assays were evaluated in duplicate, and the mean and standard deviations were calculated.

RESULTS AND DISCUSSION
Sequence of I-FLICE-Sequence analysis of a full-length cDNA revealed a 1443-base pair open reading frame that encoded a novel protein with a predicted molecular mass of 55.3 kDa (Fig. 1A). Given that the protein had striking homology to FLICE and Mch4/FLICE2 but lacked an active site, making it a potential dominant negative inhibitor, it was designated I-FLICE (for inhibitor of FLICE).
The architecture of I-FLICE was strikingly similar to that of FLICE and Mch4/FLICE2, including two N-terminal DED-like tandem repeats and a region that resembled the caspase catalytic domain (Fig. 1, B and C). Overall, I-FLICE displayed 19 and 18% identity (32 and 28% similarity) to FLICE and Mch4, respectively. Importantly, I-FLICE did not contain the catalytic cysteine that is normally embedded in the conserved pentapeptide QACRG or QACQG motif present in all known caspases. Rather, the pentapeptide sequence was QNYVV. In addition, based on the x-ray crystal structure of caspase-1 (and caspase-3), amino acid residues His 237 (His 121 ), Gly 238 (Gly 122 ), and Cys 285 (Cys 163 ) are involved in catalysis, whereas residues Arg 179 (Arg 64 ), Gln 283 (Gln 161 ), Arg 341 (Arg 207 ), and Ser 347 (Ser 213 ) form a binding pocket for the carboxylate side chain of the P1 aspartic acid (4 -6). These seven residues are conserved in all caspases, but only three of them (Gly, Gln, and Ser as indicated in Fig. 1C) are found in I-FLICE. Given this lack of conservation of key residues involved in catalysis and substrate binding, it can be concluded that I-FLICE is not a cysteine protease and is incapable of binding Asp at the P1 position. Interestingly, the DED domain of I-FLICE was more related to the corresponding domains present in the viral DED-containing inhibitors, sharing 32, 30, and 29% identity (52, 49, and 39% similarity) to K13, MC159, and E8, but only 23 and 18% identity (44 and 30% similarity) to FLICE and Mch4, respectively (17,19).
Tissue Distribution of I-FLICE-Human tissue and cell line RNA blots were probed with a 32 P-labeled cDNA specific for I-FLICE. Two transcripts (7.5 and 6 kilobases) were observed, possibly due to differential polyadenylation (Fig. 2). The tissue distribution was similar to that of FLICE and Mch4/FLICE2 (8 -10, 15), being expressed in most tissues and cell lines examined except for the brain and the lymphoblastic leukemia line MOLT4. In particular, I-FLICE expression was evident in peripheral blood leukocytes, spleen, placenta, and heart. In mouse tissues the only predominant mRNA species is the 7.5kilobase form.
I-FLICE Associates with FLICE and FLICE2-Previous studies have shown that the DED domain is a protein interaction motif that mediates the binding of the adaptor molecule FADD to the effector proteases FLICE and Mch4/FLICE2 (8,10). Given the striking structural similarity, we asked if I-FLICE interacted with either FADD or other FLICE-like caspases. Co-immunoprecipitation analysis clearly revealed the ability of I-FLICE to bind FLICE and Mch4/FLICE2 (Fig.  3A), but not FADD (Fig. 3B). In this respect, I-FLICE resembles the viral DED-containing molecule E8 in that it binds FLICE but not FADD (17,18). Because there was no association between I-FLICE and FADD, I-FLICE was not recruited to the CD-95 or TNFR-1 signaling complex as evidenced by its inability to co-precipitate with these receptors (data not shown).
I-FLICE Inhibits TNFR-1 and CD-95-induced Apoptosis-Given the ability of catalytically inactive I-FLICE to complex with FLICE-like caspases, we reasoned that I-FLICE may be acting as a dominant negative inhibitor because the active form of all caspases is a tetramer derived from the processing of two zymogen forms to a four-chain assembly (4 -6). It follows that a catalytically inert zymogen, such as I-FLICE, would be processed to inactive subunits that would result in the generation of a nonfunctional tetrameric protease. Although it is presently conjecture, this putative mechanism does predict that I-FLICE should inhibit TNFR-1 and CD-95-induced apoptosis where FLICE-like caspases play an initiating role. Consistent with the prediction, overexpression of I-FLICE resulted in substantial inhibition of TNFR-1-induced cell death comparable with previously characterized inhibitors including CrmA, MC159, dominant negative FLICE, and dominant negative Mch4/ FLICE2 (Fig. 4A). However, under the present experimental conditions, I-FLICE appeared to be a less potent inhibitor of CD-95-induced cell death, possibly reflecting the more potent death signal that emanates from this receptor.
In summary, our studies have identified a catalytically inactive member of the caspase family that can serve as a dominant negative inhibitor of CD-95-and TNFR-1-induced cell death by binding and antagonizing the apical FLICE-like caspases. Additional studies will be necessary to work out in detail the exact nature of the inhibitory mechanism.