TRAF1 Is a Substrate of Caspases Activated during Tumor Necrosis Factor Receptor-α-induced Apoptosis*

TRAF family proteins are signal-transducing adapter proteins that interact with the cytosolic domains of tumor necrosis factor (TNF) family receptors. Here we show that TRAF1 (but not TRAF2–6) is cleaved by certain caspases in vitro and during TNF-α- and Fas-induced apoptosis in vivo.160LEVD163 was identified as the caspase cleavage site within TRAF1, generating two distinct fragments. Significant enhancement of TNF receptor-1 (CD120a)- and, to a lesser extent, Fas (CD95)-mediated apoptosis was observed when overexpressing the C-terminal TRAF1 fragment in HEK293T and HT1080 cells. The same fragment was capable of potently suppressing TNF receptor-1- and TRAF2-mediated nuclear factor-κB activation in reporter gene assays, providing a potential mechanism for the enhancement of TNF-mediated apoptosis. Cell death induced by other death receptor-independent stimuli such as cisplatin, staurosporine, and UV irradiation did not result in cleavage of TRAF1, and overexpression of the C-terminal TRAF1 fragment did not enhance cell death in these cases. TRAF1 cleavage was markedly reduced in cells that contain little procaspase-8 protein, suggesting that this apical protease in the TNF/Fas death receptor pathway is largely responsible. These data identify TRAF1 as a specific target of caspases activated during TNF- and Fas-induced apoptosis and illustrate differences in the repertoire of protease substrates cleaved during activation of different apoptotic pathways.

From the ‡Burnham Institute, La Jolla, California 92037 and the §Department of Hematology-Oncology, University of Heidelberg, Heidelberg 69115, Germany TRAF family proteins are signal-transducing adapter proteins that interact with the cytosolic domains of tumor necrosis factor (TNF) family receptors. Here we show that TRAF1 (but not TRAF2-6) is cleaved by certain caspases in vitro and during TNF-␣-and Fas-induced apoptosis in vivo. 160 LEVD 163 was identified as the caspase cleavage site within TRAF1, generating two distinct fragments. Significant enhancement of TNF receptor-1 (CD120a)-and, to a lesser extent, Fas (CD95)-mediated apoptosis was observed when overexpressing the C-terminal TRAF1 fragment in HEK293T and HT1080 cells. The same fragment was capable of potently suppressing TNF receptor-1-and TRAF2-mediated nuclear factor-B activation in reporter gene assays, providing a potential mechanism for the enhancement of TNF-mediated apoptosis. Cell death induced by other death receptor-independent stimuli such as cisplatin, staurosporine, and UV irradiation did not result in cleavage of TRAF1, and overexpression of the C-terminal TRAF1 fragment did not enhance cell death in these cases. TRAF1 cleavage was markedly reduced in cells that contain little procaspase-8 protein, suggesting that this apical protease in the TNF/Fas death receptor pathway is largely responsible. These data identify TRAF1 as a specific target of caspases activated during TNF-and Fasinduced apoptosis and illustrate differences in the repertoire of protease substrates cleaved during activation of different apoptotic pathways.
TRAFs represent a group of structurally similar adapter proteins sharing a conserved C-terminal ϳ180-amino acid-long region known as the TRAF domain that binds the cytosolic tail of several TNF 1 family cytokine receptors, linking them to downstream signaling pathways (reviewed in Ref. 1). Thus far, six TRAF family proteins have been described in mammals, TRAF1-6 (2-9). Several of these TRAF family proteins can homodimerize/oligomerize with themselves. Some also selectively form heterocomplexes (2,10,11), the significance of which is largely unknown.
Several of the TRAFs play roles in regulating apoptosis (41,42). The best established mechanism for apoptosis regulation by TRAFs is through NF-B, although contributions by JNKactivating kinases and Akt may also play a role in some cir-cumstances (18,43). NF-B induces expression of several antiapoptotic genes, including the Bcl-2 family member Bfl-1 (44), IAP (inhibitor of apoptosis proteins) family genes (45,46), A20 (47), and IEX-1L (48). Interestingly, some TNF family cytokine receptors such as TNFR1 simultaneously stimulate pathways for apoptosis induction via receptor-associated proteins that activate caspase family cell death proteases (particularly caspase-8) and pathways for suppression via NF-B (reviewed in Refs. 41, 49, and 50). Thus, the balance of these two signaling pathways dictates whether cells ultimately live or die when stimulated through TNFR1.
Interestingly, although TRAF1 may influence apoptosis indirectly through modulation of NF-B-inducible gene expression, TRAF1 was also identified together with TRAF2, cIAP1, and cIAP2 as NF-B-inducible genes that may form a protein complex capable of directly blocking caspase-8 activation induced by TNFR1 (51). In this regard, TRAF1 binds the BIR domain region of IAPs cIAP1 and cIAP2 (52)(53)(54), which directly inhibit certain caspase family cell death proteases (55). Finally, experiments with TRAF1 transgenic mice have revealed a role for TRAF1 in inhibition of antigen-induced programmed cell death of CD8-positive T-cells (56).
In this report, we provide evidence that TRAF1 is a substrate of caspase family proteases that are specifically activated by TNF family death receptors, but not by other types of stimuli that primarily activate an alternative mitochondrion-dependent pathway for caspase activation. Interestingly, one of the TRAF1 cleavage products can function as a suppressor of NF-B activation induced by TNFR1 and TRAF2 and thus could potentially play a role in committing TNF-stimulated cells to an apoptotic demise. The findings demonstrate differences in the repertoire of substrates cleaved by caspases that become activated by alternative apoptosis pathways.
In Vitro Caspase Cleavage Assays and Caspase Activity Assays-TRAF cDNAs in pcDNA3 were in vitro transcribed/translated with L-[ 35 S]methionine (Amersham Pharmacia Biotech) using reticulocyte lysates (Promega), and then 2 l of lysate were incubated at 37°C for 1 h in 100 l of caspase buffer (50 mM HEPES (pH 7.4), 100 mM NaCl, 10% sucrose, 1 mM EDTA, 0.1% CHAPS, and 10 mM dithiothreitol) containing recombinant active caspases at 1 M. Proteins were eluted in 20 l of Laemmli buffer for SDS-PAGE/autoradiography analysis. Recombinant caspases were prepared as described previously (59), and their active concentration was determined by titration against the unmethylated benzyloxycarbonyl-Val-Ala-Asp fluoromethyl ketone derivative purchased from Bachem.
Caspase activities were assayed as previously described (59,60). Briefly, caspase activity was assayed at 37°C in 100 l of caspase buffer containing the indicated fluorogenic peptide at 100 M. Activity was measured continuously over the indicated times by the release of fluorogenic AFC from the peptide substitute acetyl-DEVD-AFC using a Molecular Devices fluorometer in the kinetic mode and a 405-510-nm filter pair. To assess caspase activity generated in cells, 10 l of packed cells were lysed by addition of 20 l of buffer A (20 mM HEPES (pH 7.5), 10 mM KCl, 1.5 mM MgCl 2 , 1 mM EDTA, and 1 mM dithiothreitol) containing 0.25% Triton X-100. Suspensions were vortexed briefly and normalized for protein concentration, and 2 l of the normalized lysates were assayed for caspase activity in 100 l of caspase buffer containing 100 M DEVD-AFC at 37°C.
Cell Culture and Transfections-MCF-7-Fas and BJAB cells were a kind gift of Vishva Dixit. HEK293T and HT1080 cells were obtained from American Type Culture Collection (Manassas, VA). Wild-type and caspase-8-deficient Jurkat cells were generously provided by J. Blenis (61). MCF-7-Fas, HEK293T, and HT1080 cells were cultured in high glucose Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum (Hyclone Laboratories), 1 mM L-glutamine, and antibiotics. Jurkat and BJAB cells were similarly cultured, except that RPMI 1640 medium was used. Cisplatin, adriamycin (doxorubicin), etoposide, and staurosporine were added to cultures as indicated.
Reporter Gene Assays-For NF-B reporter gene assays, 293N cells at 60 -80% confluency in 12-well plates were transfected using a cationic lipid reagent (Superfect) with various amounts of plasmid DNA, including 0.1 g of pUC13-4xNF-B-Luc and 0.1 g of pCMV-␤gal. After 36 h, cells were suspended in 0.1 ml of Promega lysis buffer. Lysates were measured for luciferase activity using a luminometer (EG&G Berthold) and Promega luciferase kits, normalizing relative to ␤-galactosidase activity (11).

TRAF1 Is Cleaved by Caspase-3, -6, -8, and -10, but Not
Caspase-7 and -9, in Vitro at Aspartic Acid 163-In vitro translated, 35 S-labeled TRAF1 was incubated with recombinant active caspase-3 and -6 -10, and in vitro cleavage was monitored by SDS-PAGE. Caspase-3, -6, -8, and -10, but not caspase-7 and -9, cleaved TRAF1 under these conditions, generating two distinct fragments (Fig. 1A) of ϳ28 and 22 kDa. The cleavage pattern revealed identical sizes for the fragments, independent of the particular caspase employed, suggesting the presence of a single cleavage site. By taking into account the size of the generated fragments, a 160 LEVD 163 motif within TRAF1 appeared to be the most likely target for cleavage. Mutation of aspartic acid 163 to alanine confirmed this hypothesis and led to a non-cleavable form of TRAF1 (Fig. 1B). In contrast, a control mutation (L160A) did not prevent caspase-mediated cleavage of TRAF1 (data not shown).
TRAF1 Is Cleaved by Caspases at D163A in Cells-To investigate whether TRAF1 cleavage occurs in cells, the fibrosarcoma cell line HT1080 was transiently transfected with a Myctagged TRAF1 expression plasmid. After inducing caspase activation by adding anti-Fas antibody (CH-11), cells were lysed at various times, and cell lysates were prepared for immunoblot analysis. Within 3 h after Fas stimulation, a faint band representing the N-terminal fragment of TRAF1 became detectable. Between 6 and 12 h, the cleaved TRAF1 product increased in relative amount, whereas the full-length ϳ49-kDa TRAF1 protein was depleted. The cleavage of TRAF1 in Fasstimulated HT1080 cells was completely prevented by preincubation (ϳ30 min) with the broad-spectrum caspase inhibitor benzyloxycarbonyl-Val-Ala-Asp fluoromethyl ketone, indicating that cleavage of TRAF1 in mammalian cells is caspase-dependent. In contrast to wild-type TRAF1, the TRAF1(D163A) protein was not cleaved in Fas-stimulated HT1080 cells, even during late stages of apoptosis when most of the HT1080 cells had become rounded, shrunken, and detached from the culture dishes (12-18 h). To investigate whether other TRAF molecules undergo cleavage by caspases in Fas-stimulated HT1080 cells, TRAF2-6 were transfected into the cells in similar experiments. No visible cleavage or degradation of the full-length TRAF molecules was observed ( Fig. 2 and data not shown).
Similar observations were made in MCF-7 breast cancer cells, a caspase-3-and -10-deficient cell line (62-64) that had been stably transfected with Fas. Interestingly, cleavage of TRAF1 was evident in these cells even before addition of anti-Fas antibody (CH-11) (Fig. 2B), presumably reflecting constitutive caspase activation in these cells as a result of overexpression of Fas. Addition of anti-Fas antibody led to rapid cleavage of the remaining wild-type TRAF1. The non-cleavable TRAF1(D163A) mutant as well as wild-type TRAF2 were not cleaved in Fas-stimulated MCF-7 cells (Fig. 2B). Similar results were obtained when HT1080 and MCF-7 cells were stimulated with TNF plus cycloheximide (data not shown), suggesting that TRAF1 cleavage occurs in response to activation of this death receptor, in addition to Fas (CD95).
Endogenous TRAF1 Is Up-regulated in Activated B-and T-cells and Cleaved during TNF-and Fas-mediated Apoptosis-To determine whether the same effect can be seen for endogenous TRAF1, we stimulated the human T-cell line Jurkat with phorbol 12-myristate 13-acetate and the human B-cell line BJAB with CD40. These stimuli result in NF-B activation and increased expression of endogenous TRAF1, a NF-B-inducible gene (38,39,65). Inducing apoptosis with either anti-Fas antibody or TNF-␣ (ϩcycloheximide) led to the complete disappearance of endogenous full-length TRAF1 within 24 h. Cleavage of TRAF1 was evident within 6 h in these experiments (Fig. 3, A and B). In contrast, apoptosis induction by a death receptor-independent stimulus, staurosporine (10 M) (66), did not result in TRAF1 cleavage (data not shown).
Overexpression of a C-terminal TRAF1 Fragment Enhances TNF-mediated Apoptosis-To investigate the possible functional significance of the cleavage of TRAF1, we coexpressed TRAF1(D163A), TRAF1 Fragment I (N-terminal), and TRAF1 Fragment II (C-terminal) together with TNFR1 (CD120a) or Fas expression plasmids. The levels of TRAF1 and TRAF1 mutants produced by transient transfection were comparable to those expressed endogenously in lymphoid cell lines (data not shown), arguing that the levels of proteins are within physiologically relevant ranges. Whereas TRAF1 and the noncleavable TRAF1(D163A) had no remarkable influence on TNFR1 or Fas cell death, TRAF1 Fragment II led to significant enhancement of TNF-mediated apoptosis (Fig. 4A). This increase in apoptosis was accompanied by elevations in caspase activity, as measured in lysates of these cells (Fig. 4B and data not shown). None of the TRAF constructs when expressed alone had any significant influence on cell viability in the absence of TNFR1 or Fas. Coexpression of TRAF1 Fragment II with Fas also resulted in enhanced apoptosis (Fig. 4C), although the effect of TRAF1 Fragment II was less potent for Fas than for TNFR1. Similar results were obtained in experiments using HT1080 cells (data not shown).

Influence of TRAF1 and Its Fragments on TNF-␣-and TRAF2-mediated NF-B Activation-Since TRAF1 has been implicated in regulation of TNF-mediated NF-B activation
and is known to bind TRAF2, the effect of the TRAF1 fragments on NF-B induction was investigated. No significant NF-B induction was observed in cells overexpressing TRAF1, TRAF1(D163A), or the TRAF1 fragments. Cotransfecting TRAF1 or TRAF1 Fragment II with TNFR1 in 293 cells resulted in dose-dependent suppression of TNFR1-mediated NF-B activation, with TRAF1 Fragment II exhibiting a more potent inhibitory effect (Fig. 5A). In contrast, the non-cleavable protein and TRAF1 Fragment I had no significant effect on TNFR1-mediated NF-B induction. Similar results were obtained when TRAF2 rather than TNFR1 was employed as the stimulus for inducing NF-B (Fig. 5B). In contrast to TRAF1 Fragment II, transfection of cells with plasmids encoding similar C-terminal fragments of TRAF3 or TRAF4 did not interfere with NF-B induction by TNF (data not shown), providing further evidence of the specificity of these results.
Cleavage of TRAF1 and Enhancement of Apoptosis Are Limited to Receptor-mediated Apoptosis-To further localize the significance of the cleavage of TRAF1 during apoptosis, HT1080 cells transfected with Myc-tagged wild-type TRAF1 were exposed to various apoptotic stimuli. Cisplatin, staurosporine (a broad-spectrum kinase inhibitor), and UV irradiation resulted in little or no cleavage of TRAF1 compared with TNF and Fas (Fig. 6). In contrast, adriamycin (doxorubicin) treatment resulted in accumulation of the cleaved TRAF1. Overexpression of TRAF1 Fragment II, however, did not enhance cell death induced by these stimuli (Fig. 6). Thus, enhancement of apoptosis by TRAF1 Fragment II appears to be specific for TNF family death receptors.
TRAF1 Cleavage Is Impaired in TNF-stimulated Cells with Reduced Levels of Caspase-8 -The preceding data demonstrated that TRAF1 cleavage occurs in cells following stimulation with activators of TNF family death receptors (TNF-␣ and anti-Fas antibody), but not after exposure to other stimuli (cisplatin, staurosporine, and UV) that activate caspases chiefly via an alternative pathway involving mitochondrial release of cytochrome c. The apical caspases in the TNF family death receptor and mitochondrion/cytochrome c pathways for apoptosis are caspase-8 and -9, respectively (reviewed in Refs. 49 and 67). We therefore compared TRAF1 cleavage in Jurkat cells and in Jurkat mutant cells containing greatly reduced amounts of procaspase-8 protein (61). Stimulation of wild-type and caspase-8-deficient Jurkat cells with TNF-␣ plus cycloheximide resulted in comparable amounts of cell death (as measured by trypan blue dye exclusion assays) and activation of downstream effector caspases (as measured by hydrolysis of the fluorogenic caspase substrate acetyl-DEVD-AFC) (Fig. 7).

FIG. 4. Overexpression of Fragment II of TRAF1 enhances TNFR1-mediated apoptosis in 293T and HT1080 cells. A, 293T
cells were transiently cotransfected with 0.1 g of pEGFP together with 1.25 g of pcDNA3-Myc-tagged TRAF1, TRAF1(D163A), TRAF1 Fragment (Frg) I, or TRAF1 Fragment II, with or without 0.25 g of a TNFR1 (p55) expression plasmid, in six-well plates at ϳ40% cell confluence. Plates were photographed 1 day later using fluorescence microscopy, and the morphology of the apoptotic (rounded, shrunken) versus non-apoptotic (flat, adherent) green fluorescent protein-expressing successfully transfected cells was noted. B, caspase activity in lysates prepared from cells transfected as described for A was measured, based on cleavage of the substrate acetyl-DEVD-AFC. Data are expressed as relative caspase activities to TNF and are representative of one of two independent experiments. C, 293T and HT1080 cells were cotransfected as described for A with various TRAF expression plasmids and plasmids encoding either TNFR1 (p55) or Fas. Viability of the green fluorescent protein-positive cells was determined after 36 h by trypan blue staining (mean Ϯ S.E., n ϭ 3).
In contrast, whereas TRAF1 was cleared from the caspase-8containing cells in a time-dependent manner after treatment with TNF-␣/cycloheximide, little cleavage of TRAF1 was observed in the caspase-8-deficient cells. Other proteins such as ␥-tubulin were not cleaved during the time frame of these experiments (data not shown), confirming the specificity of the TRAF1 observations. We therefore conclude that cleavage of TRAF1 following exposure of cells to TNF-␣ is largely dependent on caspase-8, implying that this protease is specifically responsible for TRAF1 cleavage in vivo. The ability of TNF-␣ to kill these Jurkat mutant cells that contain greatly reduced amounts of procaspase-8 is consistent with the status of Jurkat T-cells as "Type II" cells, in which small amounts of caspase-8 activation can effectively trigger activation of downstream effector caspases via a mitochondrion-dependent amplification process (68). DISCUSSION In this report, we demonstrate that TRAF1 is a target of caspases that become activated during apoptosis inducted by TNF family death receptors. The observation that Fas and TNFR1 induce cleavage of TRAF1, whereas a variety of other apoptosis inducers do not, suggests that TRAF1 cleavage results from an upstream caspase that is uniquely activated by these TNF family cytokine receptors. In this regard, at least two major pathways for caspase activation and apoptosis have been delineated, including an "extrinsic" pathway triggered by TNF family death receptors and an "intrinsic" pathway activated by damage to mitochondria (reviewed in Refs. 69 and 70). The intrinsic and extrinsic cell death pathways both induce activation of similar downstream effector caspases such as caspase-3, -6, and -7, but differ in their activation of upstream initiator caspases, with TNF family death receptors typically activating procaspase-8 (and perhaps procaspase-10) and mitochondria triggering activation of procaspase-9 via release of cytochrome c.
Although caspase-3, -6, -8, and -10 are capable of cleaving TRAF1 in vitro, the specific caspase chiefly responsible for cleavage of TRAF1 in vivo appears to be caspase-8, based on experiments performed with a mutant of the Jurkat cell line in which levels of procaspase-8 are greatly reduced (61). Further evidence consistent with cleavage of TRAF1 by an upstream initiator-type caspase rather than a downstream effector-type caspase comes from inspection of the TRAF1 cleavage site sequence: LEVD. The optimal tetrapeptide substrate sequence for cleavage by caspase-8 has been determined to be LETD/ LEVD, based on screening of combinatorial peptide libraries (71). Also, caspase-10 is known to cleave substrates preferably at (I/L/V)EXD sites (72). In contrast, the optimal tetrapeptide cleavage sequence for the downstream effector protease, caspase-3, is DEVD/DEID. Thus, the presence of a hydrophobic residue at P4Ј in the TRAF1 cleavage site argues in favor of an upstream initiator-type caspase.
The observation that TRAF1 undergoes cleavage in MCF-7 breast cancer cells, which are entirely deficient in caspase-3 and -10 (62-64), indicates that these caspases are not essential for proteolysis of TRAF1, further supporting the contention that caspase-8 is primarily responsible for this phenomenon. Moreover, although caspase-6 can cleave TRAF1 in vitro, it has been shown that the apical caspase in the TNF/Fas pathway, caspase-8, fails to induce processing of procaspase-6 in cell extracts if caspase-3 is absent (73), suggesting that the TNFR/ Fas-induced cleavage of TRAF1 seen in caspase-3-deficient MCF-7 cells is also unlikely to involve caspase-6.
Given that both initiator-and effector-type caspases are capable of cleaving TRAF1 in vitro, it seems probable that factors other than preferred tetrapeptide cleavage sequences may play a role in dictating which specific caspases cleave TRAF1 in cells. One potential explanation is that interactions of TRAF1 with other proteins may place it into the proximity of certain activated caspases, particularly caspase-8. For example, TRAF1 reportedly interacts directly or indirectly with a variety of proteins that are recruited to TNF family death receptor complexes, including TRADD, TRAF2, RIP2, FLIP, cIAP1, and cIAP2 (2,13,35,37,52). Thus, these protein interactions could place TRAF1 into close proximity with active caspase-8 during triggering of TNFR1 or related receptors.
The observation that adriamycin (doxorubicin) stimulated TRAF1 cleavage in cultured tumor cell lines suggests differences in the caspases activated by this anticancer drug compared with cisplatin, staurosporine, and UV irradiation, which induced little or no TRAF1 cleavage. In this regard, doxorubicin has been reported to induce apoptosis through a Fas-dependent pathway involving autocrine up-regulation of Fas and Fas ligand expression in some tumor and leukemia cell lines (74 -76). The TRAF1 cleavage seen in HEK293 cells following treatment with doxorubicin may reflect activation of a death receptor-dependent pathway for apoptosis, thus explaining the differences in TRAF1 cleavage in tumor cells treated with various anticancer drugs. However, overexpression of TRAF1 Fragment II failed to enhance apoptosis induced by doxorubicin (although it did augment apoptosis induced by TNF and by Fas), suggesting that this anticancer drug may simultaneously activate both death receptor-dependent (TRAF1-sensitive) and -independent (TRAF1-insensitive) pathways (77).
As shown here, one of the cleavage products generated by caspase-mediated proteolysis of TRAF1 is capable of suppressing NF-B induction by TNFR1 and TRAF2. In many tumor cell lines and most normal cell types, TNF-␣ interactions with TNFR1 result in cytotoxicity only when protein or mRNA synthesis is blocked (reviewed in Refs. 50 and 78). Transfection experiments using IB dominant-negative mutants that are resistant to degradation and use of cells from gene knockout mice lacking NF-B components have provided evidence that TNF-induced NF-B is responsible for maintaining cell survival in the setting of TNF-␣ exposure (79 -82). Thus, the observation that TRAF1 Fragment II (amino acids 164 -416) enhances apoptosis induced by TNF-␣ is consistent with the finding that this cleavage fragment also blocks NF-B induction by TNFR1. We cannot exclude the possibility, however, that TRAF1 Fragment II also interferes with other aspects of TNFR1 signaling. For example, TRAF1 binds to TRAF2 via its TRAF domain, which is encompassed in the Fragment II cleavage product of TRAF1 (2). Although TRAF2 is capable of inducing NF-B, gene knockout studies suggest that TRAF2 may be primarily responsible for activation of MAP3Ks that stimulate JNK rather than for NF-B induction within the context of TNFR1 signal transduction (43). Thus, heterodimerization of cleaved TRAF1 with TRAF2 may interfere with the ability of TRAF2 to activate both NF-B as well as JNK. In this regard, a role for JNK in suppression of apoptosis induction by TNF-␣ has also been inferred from studies of transgenic mice expressing a dominant-negative form of TRAF2 (83).
In addition to NF-B induction, the TRAF1 Fragment II cleavage product hypothetically could interfere with TNFR1induced apoptosis through a transcription-independent mechanism. For example, it has been suggested that a tetrameric complex consisting of TRAF1, TRAF2, cIAP1, and cIAP2 can block procaspase-8 processing and activation by TNFR1 through an unresolved mechanism not requiring induction of NF-B (51). The Fragment II cleavage product of TRAF1 could therefore somehow disturb the anti-apoptotic function of this multiprotein complex. In this regard, it should be noted that overexpression of non-cleavable, full-length TRAF1(D163A) versus TRAF1 Fragment II resulted in different effects on TNFR1-and Fas-induced apoptosis, suggesting differences in the functions of these proteins. Consequently, by comparing the structures of full-length TRAF1 (residues 1-415) and TRAF Fragment II (residues 164 -415), it can be deduced that the N-terminal region of TRAF1 (upstream of the TRAF domain) plays a role in apoptosis suppression since absence of this domain converts TRAF1 into a pro-apoptotic protein, at least with respect to apoptosis induced by TNFR1 and Fas. Furthermore, since full-length TRAF1 generally does not induce NF-B or JNK when overexpressed, this hypothetical anti-apoptotic function of the TRAF1 N-terminal domain presumably is unrelated to those signaling pathways.
Expression of TRAF1 is highly restricted in vivo, with mRNA found at readily detectable levels only in spleen, lung, and testis (2), unlike other TRAFs, which are generally more ubiquitously expressed. Moreover, levels of TRAF1 mRNA and protein are dynamically regulated, becoming induced by activation of various cytokine receptors, anti-CD3 antibody, anti-IgM antibody, and stimuli that induce the transcription factor NF-B (38,39,65). In this regard, analysis of the promoter region of the TRAF1 gene has demonstrated the presence of multiple NF-B-binding sites and has revealed that TRAF1 is a NF-Binducible gene (65). As discussed above, full-length TRAF1 has been associated with protection from apoptosis induced by TNFR1, whereas the cleaved TRAF1 Fragment II is implicated in sensitization of cells to TNFR1-induced apoptosis. What then dictates whether up-regulation of TRAF1 is cytoprotective versus cytotoxic when cells are exposed to TNF-␣? The answer to this question seems likely to be found in the relative amounts of full-length TRAF1 versus caspase-cleaved TRAF1 that accumulate in cells. Factors that alter the ratio of these two versions of the TRAF1 protein presumably would drive the signaling process either toward or away from an apoptotic result. This ambiguity in the outcome of cellular responses may account for why overexpression of TRAF1 has been associated with protection from apoptosis and enhanced production of active NF-B in some circumstances versus induction of apoptosis and reduced NF-B induction in others (38 -40, 56, 84).
FIG. 6. Cleavage of TRAF1 occurs preferably during death receptor-mediated apoptosis. A, HT1080 cells were UV-irradiated (240 mJ) or cultured with cisplatin (100 M), doxorubicin (Dox; 2 M), or staurosporine (Stauro; 100 ng/ml) for 6 -36 h (time was adjusted for individual stimuli so that ϳ50% cell death was achieved at the time of assay). Floating and adherent cells were recovered from cultures, and lysates were prepared, normalized for total protein content (25 g), and analyzed by SDS-PAGE/immunoblotting using anti-TRAF1 antibody. B, 293T cells were transfected with control vector or TRAF1 Fragment II (TRAF1-FrgII) in combination with TNF, Fas, or control vector and, in the latter case, exposed to various drugs (300 M cisplatin, 4 M doxorubicin, or 100 nM staurosporine) or UV-irradiated (240 mJ). Cell death was determined at 24 h by trypan blue staining. (mean Ϯ S.E., n ϭ 2). Similar data were obtained using HT1080 cells (data not shown).  (61) were cultured with phorbol 12-myristate 13-acetate (0.1 M) for 2 days to up-regulate TRAF1. Subsequently, cells were stimulated with TNF-␣ (100 ng/ml) in combination with cycloheximide (CHX; 10 g/ml) for the various times as indicated. A, clearance of TRAF1 from TNF-treated cells was monitored by immunoblotting using cell lysates normalized for protein content (25 g/lane) and anti-TRAF1 antibody. Reprobing the same blot with an antibody recognizing ␥-tubulin confirmed loading of equivalent amounts of protein in all lanes (data not shown). The cleaved fragments of TRAF1 did not accumulate to detectable levels in these Jurkat cells. B, cell death was determined by failure of cells to exclude trypan blue dye at various times after TNF exposure. Data represent percentage trypan bluepositive cells (mean Ϯ S.D., n ϭ 3). C, the activity of effector caspases was measured by hydrolysis of acetyl-DEVD-AFC in cell lysates (normalized for total protein content) derived from TNF-treated Jurkat cells as described above. Data are expressed as a percentage relative to the caspase activity measured in wild-type Jurkat cells after a 24-h treatment with TNF/cycloheximide (mean Ϯ S.E., n ϭ 2).
Generation of TRAF1 knockout mice as well as knock-in mice expressing a non-cleavable TRAF1 mutant protein (D163A) will be required to reveal the in vivo functions of TRAF1 and caspase cleavage of TRAF1.