Oncogene (2000) 19, 4451 ± 4460 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc Activation of the NF-kB pathway by Caspase 8 and its homologs

Preet M Chaudhary*,1, Michael T Eby1, Alan Jasmin1, Arvind Kumar1, Li Liu1 and Leroy Hood2

1Hamon Center for Therapeutic Oncology Research, UT Southwestern Medical Center, Dallas, Texas, TX 75390-8593, USA; 2Department of Molecular Biotechnology, University of Washington, Seattle, Washington, WA 98195 USA

Caspase 8 is the most proximal caspase in the caspase al., 1996). The prodomain of Caspase 8 consists of two cascade and has been known for its role in the mediation homologous DEDs and serves to keep Caspase 8 in an of cell death by various death receptors belonging to the inactive form (Chinnaiyan and Dixit, 1997). DEDs- TNFR family. We have discovered that Caspase 8 can containing prodomains are also found in two addi- activate the NF-kB pathway independent of its activity tional cellular proteins; Caspase 10 (Mch4, FLICE2), a as a pro-apoptotic protease. This property is localized to proteolytically active Caspase 8 homolog (Fernandes- its N-terminal prodomain, which contains two homo- Alnemri et al., 1996; Vincenz and Dixit, 1997), and logous death e€ector domains (DEDs). Caspase 10 and MRIT (Casper, c-FLIP, I-FLICE, FLAME, CASH, MRIT, two DEDs-containing homologs of Caspase 8, CLARP), a Caspase 8 homolog which is devoid of can similarly activate the NF-kB pathway. Dominant- protease activity (Goltsev et al., 1997; Han et al., 1997; negative mutants of the Caspase 8 prodomain can block Hu et al., 1997; Inohara et al., 1997; Irmler et al., 1997; NF-kB induced by Caspase 8, FADD and several death Shu et al., 1997; Srinivasula et al., 1997). receptors belonging to the TNFR family. Caspase 8 can In the present study present evidence that the DEDs interact with multiple proteins known to be involved in of Caspase 8 and its homologs are also involved in the the activation of the NF-kB pathway, including the activation of the NF-kB pathway. Our study suggests serine-threonine kinases RIP, NIK, IKK1 and IKK2. the existence of a functional and biochemical link Thus, DEDs-containing caspases and caspase homolog(s) between the caspase and the kinase cascades down- may have functions beyond their known role in the stream of the adaptor molecule TRADD and indicates mediation of cell death. Oncogene (2000) 19, 4451 ± that Caspase 8 and its homologs have roles beyond 4460. their activity as cell death proteases.

Keywords: NF-kB; caspase; FADD; death e€ector domain; MRIT; cFLIP Results

Caspase 8 activates NF-kB Introduction We began by testing the ability of Caspase 8 to activate Tumor Necrosis Factor Receptor 1 (TNFR1) is the an NF-kB-driven luciferase reporter construct in 293T best characterized death receptor of the TNFR family cells (Berberich et al., 1994). Expression of Caspase 8 in (Ashkenazi and Dixit, 1998; Baker and Reddy, 1996). these cells had no major toxicity and led to signi®cant On ligand induced aggregation, death domain of activation of the NF-kB/luciferase reporter construct in TNFR1 binds the homologous domain of TRADD, a a dose- and time-dependent manner (Figure 1a ± c). death domain-containing cytoplasmic adaptor protein Caspase 8 failed to activate luciferase reporter constructs (Hsu et al., 1995, 1996; Varfolomeev et al., 1996). driven by mutant NF-kB binding-sites, NF-IL6 or TRADD subsequently activates a kinase cascade, NFAT, which demonstrated the speci®city of the assay consisting of the NF-kB and JNK/SAPK pathways, (Figure 1a and data not shown). The NF-kB inducing via the recruitment of death domain-containing protein ability of Caspase 8 was not limited to the 293T cells RIP (Receptor Interacting Protein) and TRAF2, which since it activated NF-kB in the MCF7 and 293 EBNA lacks a death domain (TNF Receptor Associated cells as well (Figure 1d and data not shown). The ability Factor 1). TRADD also activates a caspase cascade of Caspase 8 to activate the NF-kB pathway was further via the recruitment of FADD/MORT1, another con®rmed by an electrophoretic mobility shift assay, adaptor protein, which possesses a C-terminal death which also con®rmed the speci®city of this response domain and an N-terminal `death e€ector domain' (Figure 1e). However, Caspase 8 was a relatively weaker (Boldin et al., 1995; Chinnaiyan et al., 1995, 1996; Hsu stimulator of the NF-kB pathway as compared to et al., 1996; Varfolomeev et al., 1996). The DED of TNFR1 (Figure 1e). FADD binds to the N-terminal prodomain of Caspase 8 (also called FLICE, MACH or Mch5), a pro- The prodomain of Caspase 8 mediates NF-kB activation apoptotic apical caspase of the caspase cascade (Boldin et al., 1996; Fernandes-Alnemri et al., 1996; Muzio et The observed ability of Caspase 8 to activate the NF- kB pathway could be secondary to its ability to activate the cell death pathway and the resultant activation of stress-response genes. To separate the *Correspondence: PM Chaudhary, Hamon Center for Therapeutic NF-kB inducing ability of Caspase 8 from its ability to Oncology Research, UT Southwestern Medical Center, 5323 Harry Hines Blvd., Dallas, Texas, TX 75390-8593, USA act as a pro-apoptotic caspase, we tested the ability of Received 22 May 2000; revised 30 May 2000; accepted 19 July various deletion and point mutants of Caspase 8 to 2000 activate the NF-kB pathway. Activation of NF-kB by caspases PM Chaudhary et al 4452

Figure 1 Caspase 8 induces NF-kB. (a) 293T cells were transfected with a Caspase 8 expression vector (750 ng) or an empty vector (750 ng) along with an NF-kB/luciferase reporter construct (75 ng) as well as a RSV/LacZ (b-gal) reporter construct (75 ng) in duplicate and reporter assay for NF-kB activation performed as described in the Materials and methods section. A control experiment was performed with a mutant NF-kB/luciferase reporter construct in parallel. The values shown are averages (mean+s.e.m.) of one representative experiment out of three in which each transfection was performed in duplicate. (b) Dose- response of Caspase 8-induced NF-kB activation. 293T cells were transfected with empty vector or indicated amounts (in ng/well) of expression vector for Flag-Caspase 8 C360S in duplicate in each well of a 24-well plate. Western blot analysis with a Flag antibody demonstrates the expression of Caspase 8 at each dose level (right panel). (c) Time course of Caspase 8-induced NF-kB activation. 293T cells were transfected with empty vector or Caspase 8 C360S expression plasmid (250 ng/well) and cell extracts prepared for the measurement of luciferase activity at the indicated time points. (d) Induction of NF-kB by Caspase 8 in the MCF7 cells. MCF7 cells (16105) were transfected with a Caspase 8 expression construct or an empty vector (750 ng each) along with an NF-kB/ luciferase reporter construct (75 ng) in duplicate. Luciferase activity was measured 20 h later. (e) Electrophoretic mobility shift assay. 293T cells were (36105) were transfected with 5 mg of an empty vector (lane 1) or an expression vector encoding the Caspase 8 (lane 2 ± 4) or TNFR1 (lane 5). After 16 h, nuclear extracts were prepared essentially as described previously (Yeh et al., 1997). Nuclear extracts (2 mg) were incubated for 30 min at room temperature with a 32P-labeled NF-kB duplex oligonucleotide (Promega, Madison, WI, USA) in a bu€er containing 10 mM HEPES (pH 7.9); 50 mM KCl, 0.2 mM EDTA, 2.5 mM DTT, 2 mg poly (dC:dI), 10% glycerol and 0.5% NP-40. Competition was carried out with 100-fold excess of cold NF-kB oligo duplex or a non-speci®c oligo duplex. Protein-DNA complexes were resolved on a 5% native polyacrylamide and run in Tris-glycine bu€er. Gel was dried and autoradiographed. The position of the induced NF-kB complex is marked by an arrow while * marks the position of a constitutive NF-kB complex. Descriptions of various lanes are as follows: lane 1, empty vector, lane 2, Caspase 8; lane 3, Caspase 8 plus cold non-speci®c competitor duplex; lane 4, Caspase 8 plus cold NF-kB duplex; lane 5, TNFR1

A construct encoding the full-length prodomain of the catalytically active site and is, therefore, proteoly- Caspase 8, containing its two Death E€ector Domains tically inactive, was as e€ective as the wild-type (DEDs) (amino acids 1 ± 180), was able to activate NF- Caspase 8 in activating the NF-kB pathway (Figure 2). kB to an even greater extent than the full-length Caspase 8, whereas deletion constructs encoding either Ability of other caspases to activate NF-kB DED1 (aa 1 ± 103) or DED2 (aa 104 ± 180) alone failed to do so (Figure 2). Similarly, constructs encoding the We also tested the ability of several other caspase full-length protease (caspase homology) domain or the family members to activate the NF-kB pathway. individual p20 or p10 sub-domains failed to activate Unlike Caspase 8, over-expression of Caspase 10 NF-kB (Figure 2 and data not shown). Caspase 8 (Mch4 isoform) led to massive apoptosis in 293T cells C360S, which contains a cysteine to serine mutation at but failed to induce NF-kB (Figure 3a and data not

Oncogene Activation of NF-kB by caspases PM Chaudhary et al 4453 the protease activity of both proximal and distal caspases. Both CrmA and p35 were ine€ective in blocking NF-kB activation by Caspases 8 or 10, thus con®rming that NF-kB activation by these caspases is independent of their ability to activate the caspase cascade (Figure 3d). Similarly, zVAD.fmk a broad- spectrum cell-permeable peptide caspase inhibitor, was incapable of blocking Caspase 8-induced NF-kB activation (Figure 3e). Surprisingly, both CrmA and p35e€ectivelyabolishedNF-kBactivationbytheMRITa1 and MRITb1 isoforms (Figure 3d and data not shown), which suggests that MRIT-induced NF-kB activation requires caspase activation. MRIT-a1-in- duced NF-kB was also blocked by zVAD-fmk (not shown).

Figure 2 Mutagenesis analysis of Caspase 8-induced NF-kB Deletion and point mutants of DEDs act as activation. Prodomain of Caspase 8 activates NF-kB. Experiment dominant-negative inhibitors of NF-kB activation was performed as described for Figure 1a. C360S=Caspase 8 C360S mutant. PRO=Caspase 8 prodomain (aa 1 ± 180). DED1 To delineate the region(s) of the DEDs critical for the and 2=Death E€ector Domains 1 (aa 1 ± 103) and 2 (aa 104 ± NF-kB inducing ability, we mutated three conserved 180) respectively. Protease domain=aa 217 ± 479. mTRAF2=- murine TRAF2. A representative (mean+s.e.m.) of three residues in the DED1 of Caspase 8. The resulting independent experiments performed in duplicate mutants, termed D73A, L74A and L75A, have the amino-acid residues at positions 73, 74 and 75, respectively, changed to alanine (Figure 4a). Both shown). However, its active site mutant (Caspase 10 D73A and L74A mutants failed to signi®cantly activate C358A) failed to induce apoptosis and led to NF-kB, whereas L75A retained partial ability to do so signi®cant induction of NF-kB (Figure 3a and data (Figure 4b). Moreover, all three mutants partially not shown). Similarly, over-expression of MRIT led to inhibited NF-kB induction by full-length Caspase 8 signi®cant activation of NF-kB in the 293T cells and and death receptors TNFR1, Fas/Apo1, DR3 and this activity was localized to its DEDs-containing DR4 (Figure 4c and data not shown). Similar results prodomain (Figure 3b and data not shown). Collec- were obtained with an N-terminal deletion construct of tively, these results con®rm that the NF-kB induction Capase 8 (ND-Caspase 8) which is missing the N- by Caspase 8 and its homologs is dependent on their terminal 63 amino acids (Figure 4b). Taken together, DEDs-containing prodomains and is independent of these results suggest a role of Caspase 8 prodomain in their protease ability. the death receptors-induced NF-kB activation. Next we sought to determine whether non-DEDs- containing caspases, such as Caspases 1, 2, 3, 6, 7 and Inhibition of FADD-induced NF-kB activation by 9, could activate NF-kB in the 293T cells. As shown in dominant-negative mutants of Caspase 8 Figure 3c, none of these other caspase family members was able to activate NF-kB, thereby con®rming the We next tested the ability of FADD to activate the importance of DEDs in this process. Caspases 7 and 9 NF-kB pathway and the role of DEDs-containing led to signi®cant cell death in the above assays. proteins in this process. As shown in Figure 5a, over- Therefore, this experiment was repeated with the active expression of FADD led to ecient activation of NF- site mutants of these caspases that are incapable of kB in 293T cells and this was e€ectively blocked by the inducing apoptosis. However, both Caspases 7 C186S D73A mutant of Caspase 8. Furthermore, unlike a and 9 C288S failed to activate NF-kB in the 293T cells previous report (Liu et al., 1996), FADD was highly as well (data not shown). e€ective in activating this pathway in MCF7 cells (Figure 5b). In these cells, FADD was capable of activating NF-kB at a relatively low dose which had no Effect of caspase inhibitors on NF-kB activation by signi®cant cytotoxicity (Figure 5b). Taken together, Caspase 8 and its homologs these results suggest that FADD can activate the NF- It may be postulated that the observed NF-kB kB pathway and that Caspase 8 or its homologs may inducing property of MRIT and the active site play a role in this process. mutants of Caspases 8 and 10 is due to the ability of these proteins to oligomerize with the endogenously Mechanism(s) of Caspase 8-, 10- and MRIT-induced expressed wild-type caspases and the resultant activa- NF-kB activation tion of the caspase cascade. Similarly, the prodomain of Caspase 8 could potentially activate the caspase The serine-threonine kinases NIK and the IKKs have cascade by oligomerizing with the endogenously been shown to be involved in the activation of the NF- expressed full-length Caspase 8. To rule out this kB pathway by the members of TNFR and ILIR possibility, we tested the ability of CrmA and p35 to families (DiDonato et al., 1997; Malinin et al., 1997; inhibit the NF-kB induced by Caspases 8, 10, and Mercurio et al., 1997; Regnier et al., 1997; Woronicz et MRIT. CrmA is a cow-pox virus protein which is an al., 1997; Zandi et al., 1997). To determine the role of inhibitor of the protease activity of proximal caspases these proteins in the Caspase 8-induced NF-kB while the baculovirus p35 protein is an inhibitor of activation, we took advantage of the dominant-

Oncogene Activation of NF-kB by caspases PM Chaudhary et al 4454

Figure 3 Ability of other caspases to activate NF-kB. (a) An active-site mutant of Caspase 10 activates NF-kB in the 293T cells. Experiment was performed essentially as described in Figure 1a using 750 ng of each construct. A representative (mean+s.e.m.) of two independent experiments performed in duplicate. (b) MRIT induces NF-kB. MRITa 1=full-length, MRITb1=prodomain (aa 1 ± 200). A representative (mean+s.e.m.) of three independent experiments performed in duplicate. (c) Non DEDs-containing caspases fail to activate NF-kB. A representative of three independent experiments performed in duplicate. (d) Ability of CrmA and p35 to block Caspase-induced NF-kB. Both CrmA and p35 fail to block Caspases 8- and 10-induced NF-kB but can block MRIT- induced NF-kB. A representative of three independent experiments performed in duplicate. (e) Caspase 8 induced NF-kB is not blocked by zVAD.fmk. Experiment was performed essentially as described for Figure 1a except cells in some of the wells were treated with zVAD.fmk (20 mM) approximately 4 h post-transfection

negative inhibitors of these kinases. As shown in kB activation was also e€ectively blocked by two Figure 6a, Caspase 8-induced NF-kB was e€ectively phosphorylation-resistant mutants of IkBa (IkBDN blocked by dominant-negative mutants of NIK and IkBa S32/36A) (Figure 6a). Collectively, these (NIKD2101 and NIKK429R) and IKK2 (IKK2- results suggest that Caspase 8 activates NF-kB by NIK K44A), whereas a similar mutant of IKK1 (IKK1- and IKK2 mediated phosphorylation of the IkB K44A) could do so only poorly. Caspase-induced NF- proteins.

Oncogene Activation of NF-kB by caspases PM Chaudhary et al 4455

Figure 4 Inhibitors of caspase-induced NF-kB activation (a) Multiple sequence alignment of the various DEDs-containing proteins. An asterisk indicates the site of various mutants used in this study (*). (b) Caspase 8 D73A, D74A, L75A and ND-Caspase 8, block Caspase 8-induced NF-kB. A representative (mean+s.e.m.) of three independent experiments performed in duplicate. (c) Inhibition of death receptors-induced NF-kB by point mutants of Caspase 8 Death E€ector Domains. A representative (mean+s.e.m.) of three independent experiments performed in duplicate

Finally, we investigated the role of TRAFs in the with the serine-threonine protein kinase RIP. As shown activation of NF-kB pathway by Caspase 8 and its in Figure 7a, Caspases 8, 10, MRIT a1andb1 homologs. An N-terminal deletion mutant of TRAF2 isoforms could coimmunoprecipitate RIP-HA but (Takeuchi et al., 1996) which has been previously failed to coimmunoprecipitate GFP-HA, thereby shown to block NF-kB activation by various death demonstrating the speci®city of the interaction. This receptors, was able to block NF-kB induction by all ability of Caspase 8 was localized to both the three DEDs-containing proteins (Figure 6b). This prodomain (aa 1 ± 180) and the caspase homology dominant-negative TRAF2 construct could also domain (aa 217 ± 479) (Figure 7b). Caspases 8, 10 and inhibit NF-kB induction by the prodomain as well MRIT could also coimmunoprecipitate with NIK as the active site mutant of Caspase 8 (data not whereas Caspase 3 failed to do so (Figure 7c). Like shown). However, an N-terminal deletion construct of the situation with RIP, both the prodomain and the TRAF5, which could block NF-kB activation by protease domains of Caspase 8 could coimmunopreci- HVEM (Hsu et al., 1997), failed to block NF-kB pitate NIK (Figure 7d). activation by Caspase 8 thereby demonstrating the We next tested the ability of the recently identi®ed speci®city of the response. These results suggest that IkB kinases, IKK1 and IKK2 to coimmunoprecipi- TRAF2 may play a role in Caspase 8-mediated NF- tate Caspase 8. As shown in Figure 7e, both IKK1 kB induction. and IKK2 could successfully coimmunoprecipitate Caspase 8. IKK1 and IKK2 were also able to immunoprecipitate MRIT (data not shown). Unlike Caspases 8, 10 and MRIT interact with RIP, NIK and the situation with RIP and NIK, while the IKKs prodomain of Caspase 8 could interact with IKK2, RIP was recently shown to be essential for TNFR1- the protease homology domain failed to do so mediated activation of the NF-kB pathway (Kelliher et (Figure 7f). These results indicate that the interaction al., 1998). Therefore, to further elucidate the mechan- of the DEDs-containing prodomain with the IKKs ism of Caspases 8-, 10- and MRIT-induced NF-kB may be crucial for the NF-kB induction by Caspase activation, we tested their ability to physically interact 8 and its homologs.

Oncogene Activation of NF-kB by caspases PM Chaudhary et al 4456

Figure 6 Mechanism(s) of NF-kB activation by DEDs-contain- ing proteins. (a) Dominant-negative inhibitors of NIK, IKKs, RIP and IkB block NF-kB activation by the DEDs-containing proteins. The experiment was performed as described for Figure 3e. The amounts of various activator and inhibitor plasmids were 100 ng/well and 750 ng/well respectively. The values shown are averages (mean+s.e.m.) of one representative experiment out of two in which each transfection was performed in duplicate. (b) Dominant negative TRAF2 blocks NF-kB induction by Caspase 8 where as dominant negative TRAF5 fails to do so. A representative (mean+s.e.m.) of three independent experiments performed in duplicate

Figure 8, transient transfection of Caspase 8 or Figure 5 Activation of NF-kB by FADD. (a) Inhibition of TNFR1 in 293T cells led to the stimulation of the FADD-induced NF-kB by the Caspase 8 D73A mutant. Experiment was performed as described for Figure 4c using IKK kinase activity as measured by in vitro phosphor- 100 ng/well of FADD plasmid and 750 ng/well of Caspase 8 ylation of GST-IkBa. However, consistent with the D73A mutant plasmid. The values shown are averages (mean+ previous results (Figure 1e), TNFR1 was a much s.e.m.) of one representative experiment out of three in which stronger stimulator of the IKK kinase activity as each transfection was performed in duplicate. (b) FADD induces NF-kB in the MCF7 cells. 16105 cells were transfected with an compared with Caspase 8. empty vector or indicated amounts of FADD expression constructs along with an NF-kB/luciferase reporter construct (75 ng) and a RSV/LacZ reporter construct in duplicate. Thirty- Discussion six hours later, cells in one of the wells were ®xed and stained with X-gal to determine relative cytotoxicity according to previously described criteria (Chaudhary et al., 1997). The In the present study, we have discovered that DEDs- luciferase activity was measured in cells from the other well and containing proteins can physically and functionally normalized relative to the b-galactosidase activity. Percentage cell interact with the proteins of the kinase cascade and death was as follows: Vector, 10%; FADD (100 ng), 15%; and activate the NF-kB pathway. Based on the available FADD (500 ng) 22% data, we will like to favor a model in which the NF- kB-inducing property of Caspases 8, 10 and MRIT depends on their interactions with the protein kinases Caspase 8 stimulates the kinase activity of IKK complex NIK, RIP and the IKKs. In addition to the above Based on the above interactions of Caspase 8 with the kinases, we have observed that the various DEDs- IKKs, we were next interested in testing whether containing proteins can interact with multiple members expression of Caspase 8 could also stimulate the kinase of the TRAF family of adaptor proteins (Chaudhary et activity of IKK-signalsome complex. As shown in al., 1999). While both the prodomain and the protease

Oncogene Activation of NF-kB by caspases PM Chaudhary et al 4457

Figure 7 Interactions of DEDs-containing caspases with RIP, NIK, IKK1 and IKK2. (a) Interaction of RIP with Caspase 8, 10 and MRIT. 293T cells were transfected with the indicated expression vectors and cell lysates immunoprecipitated using beads containing an antibody against the FLAG-epitope or an irrelevant mouse control antibody (C). The coimmunoprecipitating HA- tagged RIP (upper panel) was detected by Western blot analysis using rabbit polyclonal antibodies against the HA tag. The middle panel is the lower portion of the same gel and shows the lack of coimmunoprecipitation of HA-GFP, thereby demonstrating the speci®city of the interaction. Expression of the various FLAG-tagged proteins is shown on the lower panel. (b) RIP interacts with Caspase 8 prodomain (aa 1 ± 180) and caspase homology domain (217 ± 479). The lack of immunoprecipitation of RIP in the vector lane and of GFP in all lanes demonstrates the speci®city of the interaction. (c) Interactions of Caspases 8, 10, 3 and MRIT with NIK. 293T cells were transfected with expression vectors encoding FLAG-tagged Caspases and MRIT along with HA-tagged NIK and GFP and the experiment performed as described for (a). The lack of immunoprecipitation of NIK with Caspase 3 and of GFP with MRIT (lower panel) demonstrate the speci®city of the interaction. (d) NIK interacts with Caspase 8 prodomain (aa 1 ± 180) and caspase homology domain (aa 217 ± 479). (e) Interactions of Caspase 8 with the IKKs. 293T cells were transfected with the indicated plasmids and Flag-tagged IKKs immunoprecipitated with Flag beads or control beads. Coimmunoprecipitated Caspase 8 was detected by Western analysis with a rabbit polyclonal antibody against the Myc tag. (f) IKK2 interacts with the Caspase prodomain (aa 1 ± 180) but not with its protease domain (aa 217 ± 479). Lower panel shows the expression of the two caspase constructs in cellular extracts. L, lysate; C, control beads; F, ¯ag beads

MCF7 or the HeLa cells (Liu et al., 1996). A possible explanation for the di€erence between two groups of studies may lie in the use of cytotoxic concentration of FADD in the study by Liu et al., 1996, which could have prevented the measurement of NF-kB activation due to premature death of transfected cells. However, Figure 8 Caspase 8 activates IKK kinase activity. 293T cells alternative, and not entirely mutually exclusive, ex- were transfected with the indicated plasmids and the cell lysate planations for the above results are equally plausible. immunoprecipitated with a control (C) or an anti-Nemo For example, under conditions favoring activation of polyclonal serum. The in vitro kinase reaction was performed using GST-IkBa as a substrate, as described in the Materials and the caspase cascade, receptor-bound Caspases 8/10 will methods section be rapidly cleaved by cross-proteolysis and released from the DISC (Death Inducing Signaling Complex), preventing the recruitment and assembly of the NF-kB- domain of Caspase 8 can interact with NIK, RIP and inducing complex (Figure 9a). By preventing the TRAFs, only the former can interact with the IKKs, activation of Caspases 8/10, proteins, such as CrmA, suggesting that the latter interaction may be crucial to the TRAFs, the cIAPs or IEX-1L (depicted by Factor the activation of the NF-kB pathway by the DEDs- X), may allow these caspases to stay bound to the containing protein. DISC, thereby providing sucient time for the An interesting result of our study was the ability of assembly of the NF-kB inducing complex on them FADD to activate the NF-kB in the 293T and MCF7 (Figure 9b). Furthermore, the protease homology cells. On the contrary, FADD was previously shown to domain of Caspase 8 may also play a crucial role in be involved only in the mediation of apoptosis and was NF-kB activation by recruiting and bringing RIP, NIK shown not to activate the NF-kB pathway in the and TRAF2 to the prodomain-bound IKKs. There-

Oncogene Activation of NF-kB by caspases PM Chaudhary et al 4458 The exact role of Caspase 8 and its homologs in the death domain receptors-induced NF-kB activation is not clear at present. These proteins may represent one of the several possible pathways for the induction of NF-kB by the death receptors. They may add to the overall NF-kB activation induced by the death receptors, resulting in a more robust activation of this pathway. Alternatively, they may play a role in TRAF2-independent delayed phase of NF-kB activa- tion by the death receptors and thereby serve to prolong this response. Finally, the ability of Caspase 8 and FADD to activate the NF-kB pathway may be responsible for the observed defects in cardiac development and lymphocyte functions seen in mice containing the knock-out of these genes (Varfolomeev et al., 1998; Yeh et al., 1998).

Materials and methods

Expression constructs The expression constructs for the death receptors, RIP, FADD, CrmA, MRIT isoforms and the various caspases have been described previously (Chaudhary et al., 1997; Han et al., 1997). Constructs encoding the p10 and p20 sub-units of Caspase 8, Caspases 1, 2, 3, 6, 7, and 9 were gifts from Michael Wright (University of Washington). Constructs encoding NIK and its mutants (Natoli et al., 1997), IKKs and their mutants (Nakano et al., 1998), IkBaDNand IkBaS32/36A (Brockman et al., 1995), Caspase 10 (Mch4 Figure 9 Regulation of cell death and survival during death isoform) and its C358A mutant (Fernandes-Alnemri et al., receptors-signaling by Caspase 8 and its homologs. Please see text 1996) and NF-kB/luciferase reporter constructs (Berberich et for details al., 1994) have been described previously and were obtained from the indicated sources. The various point mutants of Caspases 8, 7, 9, 10 and MRIT were generated using the fore, it is conceivable that cleavage of Caspase 8 into Quickchange site directed mutagenesis kit (Stratagene). The its prodomain and the protease domain will lead to various deletion and fusion constructs were made by PCR disruption of such a complex and inability to activate using custom primers. Epitope-tagged expression constructs NF-kB. The above model might also explain how for Caspases 3, 7, 9, MRIT, RIP and their deletion and signaling through the death receptors might lead to cell point mutants were tagged at the C-termini and the death or proliferation depending on the cellular context remaining constructs were tagged at the N-termini using custom primers. The sequence of all constructs was (Figure 9). Under conditions favoring caspase cleavage, con®rmed by automated ¯uorescent dye-terminator sequen- NF-kB pathway will not be activated and the cell will cing on an ABI 373 sequencing machine. Expression of all commit to apoptosis. On the other hand, under the constructs was con®rmed by Western analysis on total cell conditions which prevent caspase activation, signaling extracts. through the death receptors will activate the NF-kB pathway, which will not only promote cellular NF-kB luciferase reporter assay proliferation but also a€ord further protection against apoptosis by transcriptional activation of anti-apopto- For reporter gene assay, 16105 293T or MCF7 cells were seeded tic genes, such as the TRAFs, the IAPs and IEX-1L in each well of a 24-well tissue culture plate. Twenty-four hours (Wang et al., 1998; Wu et al., 1998). later cells were transfected with the indicated amounts of various expression constructs along with an NF-kB/luciferase reporter The anti-apoptotic property of MRIT has been construct (75 ng) and an RSV promoter driven b-galactosidase previously explained on the basis of its ability to reporter construct (pRcRSV/LacZ) (75 ng) in duplicate. The compete with Caspases 8 or 10 to bind to the DED of NF-kB reporter plasmid contains four copies of an NF-kB FADD (Goltsev et al., 1997; Hu et al., 1997; Inohara binding site from the promoter of the invariant chain of MHC et al., 1997; Irmler et al., 1997; Shu et al., 1997; class II and has been described previously (Berberich et al., Srinivasula et al., 1997). Our results demonstrate the 1994). The amount of test plasmid used was 750 ng/well for ability of this protein to activate the anti-apoptotic experiments involving a single test plasmid and 100 ng/well for NF-kB pathway and thus provide an alternative/ those involving a test (activator) plasmid and an inhibitor additional mechanism for its ability to protect against plasmid. The amount of inhibitor plasmid used was 750 ng/ cell death. Interestingly, activation of the NF-kB well and the total amount of transfected DNA was kept constant by adding empty vector. 293T cells were transfected pathway has been also shown to promote cell death using a calcium phosphate co-precipitation method, while under certain conditions (Grimm et al., 1996; Hett- MCF7 cells were transfected using Superfect (Qiagen) as mann et al., 1999; Jung et al., 1995; Lin et al., 1998), described previously (Chaudhary et al., 1997). Twenty to 24 h which may explain the pro-apoptotic ability of MRIT later cell extracts were prepared using the Luciferase Cell as well. Culture Lysis Reagent (Promega, Madison, WI, USA) and

Oncogene Activation of NF-kB by caspases PM Chaudhary et al 4459 luciferase assays performed using 20 ml of cell extract. The cell glycophosphate, 10 mM NaF, 1 mM DTT and protease lysates were diluted 1 to 20 with Phosphate Bu€ered Saline inhibitor cocktail (Roche). After 30-min incubation at 48C, (pH=7.4) and used for the measurement of the b-galactosidase supernatants were collected from centrifuge at 14 000 r.p.m. activity. Luciferase activity was normalized relative to the b- for 10 min. Protein concentration was determined using galactosidase activity to control for the di€erence in the Bradford reagent at 600 nm. Immunoprecipitation was set transfection eciency. up with 500 mg of each sample with either a control antibody or an anti-nemo / IKKg (Santa Cruz, FL-419) and carried out at 48C for 2 h. Immunoprecipitated proteins were Coimmunoprecipitation assays collected by a brief spin. Beads were then washed three For studying in vivo interaction, 26106 293T cells were times with immunoprecipitation bu€er containing 40 mM plated in a 100 mm plate and co-transfected 18 ± 24 h later Tris-HCl pH 8.0, 0.5 M NaCl, 0.1% NP-40, 6 mM EDTA, with 5 mg/plate of each epitope-tagged constructs along 1 mg 10 mM b-glycophosphate, 10 mM NaF, 1 mM DTT, and of a GFP encoding plasmid (pEGFP or HA-GFP) by calcium protease inhibitor cocktail. Kinase reaction was performed in phosphate co-precipitation. Eighteen to 36 h post-transfec- a bu€er containing 20 mM HEPES pH 7.6, 0.1 M KCl, tion, cells were lysed in 1 ml of lysis bu€er containing 0.1% 0.2 mM EDTA, 10% glycerol, 0.3 M MgCl2,30mM NaF, Triton X-100, 20 mM sodium phosphate (pH 7.4), 150 mM 30 mM b-glycophosphate, 0.15 M DTT, 1 mg BSA, 10 mM NaCl and 1 EDTA free mini-protease inhibitor tablet per ATP, 10 mg GST-IkBa (1 ± 54), and 20 mCi [g-32P]ATP (ICN, 10 ml (Boehringer Mannheim). For immunoprecipitation, 35020) per reaction at 378C for 20 min. The reaction was cells lysate (500 ml) was incubated for 1 h at 48C with 10 ml stopped by the addition of 46SDS ± PAGE sample bu€er of FLAG or control mouse Ig beads precoated with 2% and samples resolved on a 12% SDS ± PAGE gel. After BSA. Beads were washed twice with lysis bu€er, twice with a running, the gel was dried and autoradiographed. wash bu€er containing 0.1% Triton X-100, 20 mm sodium phosphate (pH 7.4), 500 mM NaCl and again with lysis bu€er. Bound proteins were eluted by boiling, separated by SDS ± PAGE, transferred to a nitrocellulose membrane and Acknowledgments analysed by Western blot. We would like to thank Drs Emad Alnemri, Vishva Dixit, David Han and Michael Wright for the CrmA, FADD and various caspase expression plasmids, Dr Edward Clark for In vitro kinase assay the NF-kB reporter plasmid, Dr Hiroyasu Nakano for Transiently transfected 293T cells were harvested 24 h post IKKs expression plasmids and Dr Gioacchino Natoli for transfection. Cells were washed with cold PBS (pH 7.4) twice NIK expression plasmids. PMC was supported by a and lysed in 500 ml lysis bu€er containing 20 mM Tris postdoctoral fellowship from the Cancer Research Fund pH 8.0, 0.5 M NaCl, 0.25% Triton X-100, 10 mM b- of the Damon Runyon-Walter Winchell Foundation.

References

Ashkenazi A and Dixit VM. (1998). Science, 281, 1305 ± Han DK, Chaudhary PM, Wright ME, Friedman C, Trask 1308. BJ, Riedel RT, Baskin DG, Schwartz SM and Hood L. Baker SJ and Reddy EP. (1996). Oncogene, 12, 1±9. (1997). Proc. Natl. Acad. Sci. USA, 94, 11333 ± 11338. Berberich I, Shu GL and Clark EA. (1994). J. Immunol., 153, Hettmann T, DiDonato J, Karin M and Leiden JM. (1999). 4357 ± 4366. J. Exp. Med., 189, 145 ± 158. Boldin MP, Goncharov TM, Goltsev YV and Wallach D. Hsu H, Shu HB, Pan MG and Goeddel DV. (1996). Cell, 84, (1996). Cell, 85, 803 ± 815. 299 ± 308. Boldin MP, Varfolomeev EE, Pancer Z, Mett IL, Camonis Hsu H, Solovyev I, Colombero A, Elliott R, Kelley M and JH and Wallach D. (1995). J. Biol. Chem., 270, 7795 ± Boyle WJ. (1997). J. Biol. Chem., 272, 13471 ± 13474. 7798. Hsu H, Xiong J and Goeddel DV. (1995). Cell, 81, 495 ± 504. BrockmanJA,SchererDC,McKinseyTA,HallSM,QiX, Hu S, Vincenz C, Ni J, Gentz R and Dixit VM. (1997). J. Lee WY and Ballard DW. (1995). Mol. Cell. Biol., 15, Biol. Chem., 272, 17255 ± 17257. 2809 ± 2818. Inohara N, Koseki T, Hu Y, Chen S and Nunez G. (1997). Chaudhary PM, Eby M, Jasmin A, Bookwalter A, Murray J Proc. Natl. Acad. Sci. USA, 94, 10717 ± 10722. and Hood L. (1997). Immunity., 7, 821 ± 830. Irmler M, Thome M, Hahne M, Schneider P, Hofmann K, Chaudhary PM, Eby MT, Jasmin A and Hood L. (1999). J. Steiner V, Bodmer JL, Schroter M, Burns K, Mattmann C, Biol. Chem., 274, 19211 ± 19219. RimoldiD,FrenchLEandTschoppJ.(1997).Nature, Chinnaiyan AM and Dixit VM. (1997). Semin. Immunol., 9, 388, 190 ± 195. 69 ± 76. Jung M, Zhang Y, Lee S and Dritschilo A. (1995). Science, Chinnaiyan AM, O'Rourke K, Tewari M and Dixit VM. 268, 1619 ± 1621. (1995). Cell, 81, 505 ± 512. Kelliher MA, Grimm S, Ishida Y, Kuo F, Stanger BZ and Chinnaiyan AM, Tepper CG, Seldin MF, O'Rourke K, Leder P. (1998). Immunity, 8, 297 ± 303. Kischkel FC, Hellbardt S, Krammer PH, Peter ME and Lin KI, DiDonato JA, Ho€mann A, Hardwick JM and Dixit VM. (1996). J. Biol. Chem., 271, 4961 ± 4965. Ratan RR. (1998). J. Cell. Biol., 141, 1479 ± 1487. DiDonato JA, Hayakawa M, Rothwarf DM, Zandi E and Liu ZG, Hsu H, Goeddel DV and Karin M. (1996). Cell, 87, Karin M. (1997). Nature, 388, 548 ± 554. 565 ± 576. Fernandes-Alnernri T, Armstrong RC, Krebs J, Srinivasula Malinin NL, Boldin MP, Kovalenko AV and Wallach D. SM, Wang L, Bullrich F, Fritz LC, Trapani JA, Tomaselli (1997). Nature, 385, 540 ± 544. KJ, Litwack G and Alnemri ES. (1996). Proc. Natl. Acad. Mercurio F, Zhu H, Murray BW, Shevchenko A, Bennett Sci. USA, 93, 7464 ± 7469. BL,LiJ,YoungDB,BarbosaM,MannM,ManningA Goltsev YV, Kovalenko AV, Arnold E, Varfolomeev EE, and Rao A. (1997). Science, 278, 860 ± 866. Brodianskii VM and Wallach D. (1997). J. Biol. Chem., Muzio M, Chinnaiyan AM, Kischkel FC, O'Rourke K, 272, 19641 ± 19644. Shevchenko A, Ni J, Scadi C, Bretz JD, Zhang M, Gentz Grimm S, Bauer MK, Baeuerle PA and Schulze-Ostho€ K. R, Mann M, Krammer PH, Peter ME and Dixit VM. (1996). J. Cell. Biol., 134, 13 ± 23. (1996). Cell, 85, 817 ± 827.

Oncogene Activation of NF-kB by caspases PM Chaudhary et al 4460 Nakano H, Shindo M, Sakon S, Nishinaka S, Mihara M, Vincenz C and Dixit VM. (1997). J. Biol. Chem., 272, 6578 ± Yagita H and Okumura K. (1998). Proc. Natl. Acad. Sci. 6583. USA, 95, 3537 ± 3542. Wang CY, Mayo MW, Korneluk RG, Goeddel DV and Natoli G, Costanzo A, Moretti F, Fulco M, Balsano C and Baldwin Jr AS. (1998). Science, 281, 1680 ± 1683. Levrero M. (1997). J. Biol. Chem., 272, 26079 ± 26082. Woronicz JD, Gao X, Cao Z, Rothe M and Goeddel DV. Regnier CH, Song HY, Gao X, Goeddel DV, Cao Z and (1997). Science, 278, 866 ± 869. Rothe M. (1997). Cell, 90, 373 ± 383. Wu MX, Ao Z, Prasad KV, Wu R and Schlossman SF. Shu HB, Halpin DR and Goeddel DV. (1997). Immunity, 6, (1998). Science, 281, 998 ± 1001. 751 ± 763. Yeh WC, Pompa JL, McCurrach ME, Shu HB, Elia AJ, Srinivasula SM, Ahmad M, Ottilie S, Bullrich F, Banks S, Shahinian A, Ng M, Wakeham A, Khoo W, Mitchell K, WangY,Fernandes-AlnemriT,CroceCM,LitwackG, El-Deiry WS, Lowe SW, Goeddel DV and Mak TW. Tomaselli KJ, Armstrong RC and Alnemri ES. (1997). J. (1998). Science, 279, 1954 ± 1958. Biol. Chem., 272, 18542 ± 18545. Yeh WC, Shahinian A, Speiser D, Kraunus J, Billia F, Takeuchi M, Rothe M and Goeddel DV. (1996). J. Biol. Wakeham A, de la Pompa JL, Ferrick D, Hum B, Iscove Chem., 271, 19935 ± 19942. N, Ohashi P, Rothe M, Goeddel DV and Mak TW. (1997). Varfolomeev EE, Boldin MP, Goncharov TM and Wallach Immunity, 7, 715 ± 725. D. (1996). J. Exp. Med., 183, 1271 ± 1275. Zandi E, Rothwarf DM, Delhase M, Hayakawa M and Varfolomeev EE, Schuchmann M, Luria V, Chiannilkulchai Karin M. (1997). Cell, 91, 243 ± 252. N, Beckmann JS, Mett IL, Rebrikov D, Brodianski VM, KemperOC,KolletO,LapidotT,So€erD,SobeT, Avraham KB, Goncharov T, Holtmann H, Lonai P and Wallach D. (1998). Immunity, 9, 267 ± 276.

Oncogene