Cell, Vol. 85, 817–827, June 14, 1996, Copyright 1996 by Cell PressFLICE, A Novel FADD-Homologous ICE/CED-3–likeProtease, Is Recruited to the CD95 (Fas/APO-1)Death-Inducing Signaling ComplexMarta Muzio,*kArul M. Chinnaiyan,*khave been identified, the way in which they interact toexecute the apoptotic program is poorly understood.Frank C. Kischkel,†kKaren O’Rourke,*kIt is becoming apparent that cysteine proteases re-Andrej Shevchenko,‡kJian Ni,k§Carsten Scaffidi,†lated to the Caenorhabditis elegans cell death geneJames D. Bretz,* Mei Zhang,§Reiner Gentz,§ced-3 represent the effector components of the apop-Matthias Mann,‡Peter H. Krammer,†toticmachinery. Thefirstmammalian homologof CED-3Marcus E. Peter,†# and Vishva M. Dixit*#identified was interleukin-1b–converting enzyme (ICE;*University of Michigan Medical SchoolYuan et al., 1993). Overexpression of ICE or CED-3 inDepartment of PathologyRat-1 fibroblasts induced apoptosis, suggesting thatAnn Arbor, Michigan 48109ICE was functionally, as well as structurally, related to†Tumorimmunology ProgramCED-3 (Miura et al., 1993). However, such evidence isGerman Cancer Research Centeronly a correlation, as ectopic expression of a numberIm Neuenheimer Feld 280of proteases, including chymotrypsin,proteinase K,and69120 Heidelbergtrypsin, causesignificant apoptosis(Williams andHenk-Federal Republic of Germanyart, 1994).‡Protein & Peptide GroupFurther studies suggest that proteases relatedto ICE,European Molecular Biology Laboratoryrather than ICE itself, may play a more important role inMeyerhofstrasse 1the apoptotic mechanism. First, a number of cell types69012 Heidelbergstably secrete mature IL-1b without undergoing apop-Federal Republic of Germanytosis. Second, ICE deficient mice, although unable to§Human Genome Sciences, Incorporatedgenerate active IL-1b, fail to exhibit a prominent cell9410 Key West Avenuedeath–defective phenotype (Kuida et al., 1995; Li et al.,Rockville, Maryland 20850-33381995). Third, in an in vitro model of apoptosis, con-demned phase extracts prepared from chicken DU249cells fail to cleave the primary substrate of ICE, pro–IL-Summary1b (Lazebnik et al., 1994). Instead, a proteolytic activityin these extracts, termed prICE, cleaves the DNA repairTo identify CAP3 and CAP4, components of the CD95enzyme poly (ADP-ribose) polymerase (PARP) into sig-(Fas/APO-1) death-inducing signaling complex, wenature apoptotic fragments (Lazebnik et al., 1994). Puri-utilized nano-electrospray tandem mass spectrome-fied ICE fails to cleave PARP (Lazebnik et al., 1994;try, a recently developed technique to sequence fem-Tewari et al., 1995), suggesting that prICE is distincttomole quantities of polyacrylamide gel–separatedfrom ICE.proteins. Interestingly, CAP4 encodes a novel 55 kDaTodate,seven homologs ofCED-3 andICEhave beenprotein, designated FLICE, which has homology tocharacterized, including Nedd-2/ICH-1 (Kumar et al.,both FADD and the ICE/CED-3 family of cysteine pro-1994; Wang et al., 1994), Yama/CPP-32/Apopain (Fer-teases. FLICE binds to the death effector domain ofnandes-Alnemri et al.,1994;Nicholson etal., 1995;Tew-FADD and upon overexpression induces apoptosisariet al., 1995), Tx/ICH-2/ICErel-II(Faucheu et al., 1995;that is blocked by the ICE family inhibitors, CrmA andKamens et al., 1995; Munday et al., 1995), ICE rel-IIIz-VAD-fmk. CAP3 was identified as the FLICE prodo-(Munday et al., 1995), Mch-2 (Fernandes-Alnemri et al.,main which likely remains bound to the receptor after1995a), ICE-LAP3/Mch-3/CMH-1 (Duan et al., 1996a;proteolytic activation. Taken together, this is uniqueFernandes-Alnemri et al., 1995b; Lippke et al., 1996),biochemical evidence to link a death receptor physi-and ICE-LAP6 (Duan et al., 1996b). Ectopic expressioncally to the proapoptotic proteases of the ICE/CED-3of theseICE/CED-3homologs inavarietyofcellscausesfamily.apoptosis. Only Yama and ICE-LAP3 have been showntobeproteolyticallyactivatedbyapoptoticstimuli(Chin-Introductionnaiyan et al., 1996a; Duan et al., 1996a; Schlegel et al.,1996). Future studies will delineate which family mem-Apoptosis, or programmed cell death, is a geneticallybershavean importantroleintheapoptoticmechanism.regulated mechanism with a central role in both meta-Although it is clearthat CED-3–like proteases are dis-zoandevelopment andhomeostasis (Raff, 1992; Steller,tal effectors of the cell death pathway, the proximal1995).Thecelldeathmachineryisconservedthroughoutcomponents that mediate their activation remain to beevolution(Vaux et al., 1994) and is composed of severalidentified. Two cell surface cytokine receptors, CD95distinctparts, includingeffectors, inhibitors, and activa-(Fas/APO-1) and TNFR-1, have been shown to triggertors (Chinnaiyan and Dixit, 1996; Steller, 1995). Inverte-apoptosis by their natural ligands or specific agonistbratemodelsystems havebeeninvaluable inidentifyingantibodies (Baglioni, 1992; Itoh et al., 1991; Trauth etand characterizing the genes that control apoptosisal., 1989). Both death receptors are members of the(Hengartner, 1996). While numerous candidate genestumor necrosis factor (TNF)/nerve growth factor recep-tor family, which also includes TNFR-2, low affinityNGFR, CD40, and CD30, among others (Smith et al.,kThe first six authors contributed equally to this manuscript.#M. E. P. and V. M. D. share senior authorship.1990; Tewari and Dixit, 1995). While family membersCell818are defined by the presence of cysteine-rich repeats in electrospray (Fennet al.,1989), in combinationwith tan-dem mass spectrometry (Hunt et al., 1986), was refinedtheirextracellular domains,CD95andTNFR-1alsosharea region of homology, appropriately designated the to allow sequencingof femtomolequantities ofproteinsdirectly isolated from silver-stained gels (Wilm et al.,“death domain,” required to signal apoptosis (Itohand Nagata, 1993; Tartaglia et al., 1993). This shared 1996). The peptide sequence tags (Mann and Wilm,1994) generated can then be used to screen sequencedeathdomain suggeststhat bothreceptorsinteract witha related set of signal-transducing molecules that, databasesto obtain matchingsequencesleading toiso-lation of full-length clones for functional characteriza-until recently, remained unidentified. Using the two-hybrid system, three death domain–containing mole- tion. In its first reported use, an antiangiogenic factorderived from mycoplasma was characterized using thiscules, TNFR1-associated death domain (TRADD),