An Evolutionary Conserved Pathway of Nuclear Factor-κB Activation Involving Caspase-mediated Cleavage and N-end Rule Pathway-mediated Degradation of IκBα

The Drosophila nuclear factor-κB (NF-κB)-like transcription factor Relish is activated by an endoproteolytic cleavage step mediated by the Drosophila caspase Dredd. We have examined the contribution of the caspase cascade to NF-κB activation via TRAIL, a mammalian tumor necrosis factor family ligand that is a potent activator of caspases. Our results demonstrate that TRAIL activates NF-κB in two phases as follows: an early caspase independent phase and a late caspase dependent phase. The late phase of the TRAIL-induced NF-κB is critically dependent on caspase 8 and can be blocked by pharmacological and genetic inhibitors of caspase 8 activation, such as benzyloxycarbonyl-VAD-fluoromethyl ketone, benzyloxycarbonyl-IETD-fluoromethyl ketone, and small interfering RNA targeting caspase 8 and FADD. Whereas caspase 3 is required for TRAIL-induced apoptosis, it is not involved in TRAIL-induced NF-κB activation. The late phase of TRAIL-induced NF-κB activation involves caspase mediated cleavage of IκBα between Asp31 and Ser32 residues to generate an N-terminal truncated fragment that is degraded by the proteasome via the N-end rule pathway. Our results demonstrate that caspases play an evolutionarily conserved role as regulated entry points to the N-end rule pathway and in NF-κB activation in mammalian cells.

Nuclear factor-B (NF-B) 1 is a critical transcription factor involved in the regulated expression of several genes involved in the inflammatory and immune response (1)(2)(3)(4)(5). Five known members of this family have been characterized to date and include c-Rel, NF-B1 (p50 and its precursor p105), NF-B2 (p52 and its precursor p105), p65 (RelA), and RelB (4,5). Although many dimeric forms of NF-B have been described, the classical NF-B complex is a heterodimer of the p65/RelA and p50 subunits and is found in most cells in association with a family of inhibitory proteins, called IBs, of which the most common is IB␣ (2,4). The IB proteins retain NF-B in the cytoplasm by masking its nuclear localization signal. Stimulation by a number of cytokines, such as TNF␣ and IL-1, results in the activation of a multisubunit IB kinase (IKK) complex, which leads to the inducible phosphorylation of IB proteins at two conserved serine residues located within their N-terminal region (2,4). Phospho-IB proteins are ubiquitinated and subsequently degraded by the 26 S proteasome, thereby releasing NF-B from their inhibitory influence (2). Once released, NF-B is free to migrate to the nucleus and bind to the promoter of specific genes possessing its cognate binding site.
The members of the tumor necrosis factor receptor (TNFR) superfamily and their ligands have been recognized to play a crucial role in normal development and in the regulation of immune and inflammatory response (6,7). In addition, a subgroup of these receptors, such as TNFR1, Fas/CD95, death receptor 3 (DR3/TRAMP), death receptor 4 (DR4/TRAIL-R1), death receptor 5 (DR5/TRAIL-R2), and death receptor 6 (DR6) are capable of inducing programmed cell death or apoptosis and are collectively referred to as death receptors (6,8). The apoptosis inducing ability of these receptors has been mapped to a conserved cytoplasmic domain of 60 -80 amino acids, called the death domain (DD) (6). TNFR1 is the prototypical and perhaps the best characterized death receptor (9). Ligand-induced trimerization of TNFR1 leads to the recruitment of the DD-containing adaptor protein TRADD, which helps in the formation of a plasma membrane-bound complex (complex I) via the recruitment of RIP1 and TRAF2 (10 -12). Assembly of complex I occurs in lipid rafts and leads to NF-B activation via RIP1-mediated recruitment of the IKK complex, whereas the c-Jun N-terminal kinase is activated via TRAF2-mediated activation of the mitogen-activated protein 3-kinase (10,11). Subsequently, TRADD and RIP1 dissociate from complex I and associate with a cytoplasmic complex (complex II) consisting of DD-containing protein FADD and procaspase 8, the apical caspase of the caspase cascade (10). Under the conditions favoring TNFR1-induced apoptosis, procaspase 8 is activated upon recruitment to complex II and subsequently results in the activation of downstream caspases, such as caspase 3, 6, and 7, and eventual cell death (10). However, in most cell types, activation of the caspase cascade by TNFR1 is blocked by the concurrent activation of the NF-B pathway, which is believed to protect cells against apoptosis by up-regulating the expression of anti-apoptotic molecules, such as cFLIP L and c-IAP1 (10).
Unlike TNFR1, signaling via Fas, DR4/TRAIL-R1, and DR5/ TRAIL-R2 delivers a strong and rapid pro-apoptotic signal (8,13). Ligand binding to these receptors leads to DD-mediated recruitment of FADD directly without the involvement of TRADD (13,14). FADD subsequently leads to the recruitment and activation of procaspase 8 (8,13). DR4/TRAIL-R1 and DR5/TRAIL-R2 are also known to activate the NF-B pathway, although activation of this pathway via these receptors is of lower magnitude and slower kinetics as compared with TNFR1 (15)(16)(17)(18). Nevertheless, RIP1 and IKK complex-mediated phosphorylation and degradation of IB␣ has been implicated in NF-B activation via the TRAIL receptors as well (18), suggesting that they share at least some components of NF-B signaling with TNFR1. The mammalian tumor necrosis factor receptor signaling pathway shares several similarities with the Drosophila immune deficiency (imd) pathway, which controls the induction of antimicrobial peptide genes after Gram-negative bacterial infection (see Fig. 7) (1). The central transcription factor in the imd signaling pathway is Relish, a Drosophila NF-B homolog. Relish resembles the mammalian p100 and p105 proteins in overall structure and is composed of a DNA-binding Rel homology domain and an inhibitory IB-like ankyrin-repeat domain (1). However, in contrast to its mammalian counterparts, activation of Relish does not require proteasome-mediated degradation of its IB-like region. Instead, Relish is activated by endoproteolytic cleavage in the linker region between the Rel and IB-like domains (1). The cleavage of Relish requires the activities of two arms of the imd pathway (Fig. 7). The first arm includes IMD-(Drosophila RIP1 homolog) and dTAK1 (Drosophila transforming growth factor-␤ activating kinase)-mediated activation of Drosophila IB kinase (DmIKK) complex (19 -21). The DmIKK complex consists of Drosophila IKK␤/ immune response-deficient 5 (Ird5) and Drosophila IKK␥/ Kenny and phosphorylates Relish, triggering its cleavage (21)(22)(23). Actual cleavage of Relish is believed to be mediated by the second arm of the imd pathway that involves Drosophila FADD (dFADD)-mediated activation of Dredd, a Drosophila caspase ( Fig. 7) (24 -29). Dredd resembles caspase 8 in structure and is believed to participate in Relish activation by inducing its cleavage between the Rel and IB homology domains (25).
As discussed above, the first arm of imd pathway is conserved in mammalian cells and has been shown to play a critical role in NF-B activation via the TNFR family members. However, a role for caspase activation in NF-B activation in mammalian cells has been lacking so far (1). In this report we have analyzed NF-B activation via TRAIL, a TNF family ligand that is a powerful activator of the caspase cascade and is also known to induce NF-B activation. We present evidence demonstrating a conserved role for the second, caspase-dependent, arm of the imd pathway in NF-B activation via the TRAIL receptors.

EXPERIMENTAL PROCEDURES
Cell Lines and Reagents-293, 293T, and HeLa cells were maintained in Dulbecco's modified Eagle's medium with 10% fetal calf serum. Rabbit polyclonal antibodies against IB␣ and NEMO were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies against caspase 8 and cleaved PARP were from Cell Signaling, and those against FADD and RIP were from Transduction Laboratories. Human recombinant TRAIL, TNF-␣, and IL-1␤ were obtained from R&D Systems (Minneapolis, MN) and were used at concentrations of 100, 10, and 10 ng/ml, respectively. Caspase inhibitors were purchased from Calbiochem and dissolved in Me 2 SO.
Expression Constructs-A retroviral construct encoding FLAG-DR4 was made by amplifying a DNA fragment encoding FLAG-DR4 by using Pfu polymerase and subsequently subcloning into the MSCV-neo vector (Clontech). Constructs encoding IB␣ and an NF-B/luciferase reporter have been described previously (30,31). The cDNA encoding IB␣ was subcloned into the pCMVTag-1 vector (Stratagene, La Jolla, CA) to make the construct FLAG-IB␣-Myc that carries an N-terminal FLAG tag and a C-terminal Myc tag. The D31A mutant of IB␣ was made by using site-directed mutagenesis kit (Stratagene). Adeno-associated viral vectors encoding C-terminal HA-tagged IB␣ and its mutants were made by using a commercially available kit (Stratagene).
Generation of 293NF-B-DR4 Cells-Generation of 293NF-B-Luc cells has been described previously (32). 293NF-B-Luc cells were infected with an empty retrovirus vector or one encoding FLAG-tagged DR4 and selected with G418. One of the clones that responded to TRAIL-signaling was chosen for further studies.
Luciferase Reporter Assays-The NF-B reporter assay was performed essentially as described previously (33). 293NF-B vector cells and 293NF-B-DR4 cells were seeded in 24-well plates. Thirty-six hours later cells were treated in duplicates with Me 2 SO or Z-VAD-fmk for 30 min prior to treatment with the ligands. The cells were lysed 10 h following treatment, and lysates were used for determining the luciferase activity and protein concentration (Bradford-assay; Bio-Rad). Luciferase activity was normalized relative to protein concentration. All experiments were in duplicate and repeated at least two times. The values shown are average (mean Ϯ S.E.) of a representative of at least two independent experiments.

Generation and Characterization of 293NF-B-DR4 Cell
Line-In order to facilitate a comparative analysis of NF-B activation by ligands of the TNF family, we generated subclones of 293 (human embryonic kidney) cell line with stable expression of an NF-B-driven luciferase reporter construct. One of the subclones, designated 293NFB-Luc, demonstrated low level basal luciferase activity but a robust increase (ϳ100fold) upon stimulation with TNF␣, a known activator of the NF-B pathway. However, unlike TNF␣, treatment with TRAIL led to only a minimal increase in luciferase activity in 293NF-B-Luc cells, which was probably reflective of the weak expression of TRAIL receptors in these cells. In order to make 293NF-B-Luc cells responsive to TRAIL-induced NF-B activation, we used retrovirally mediated gene transfer to generate 293NF-B-luc cells stably expressing an empty vector or an N-terminal FLAG epitope-tagged DR4 (also called TRAIL-R1), one of the TRAIL receptors capable of activating the NF-B pathway. As shown in Fig. 1, A and B, 293NF-B-DR4 cells demonstrated modest expression of DR4 as measured by cell surface staining with a FLAG antibody. Consistent with the expression of DR4, 293NF-B-DR4 cells demonstrated an increase in the NF-B reporter activity upon stimulation with TRAIL as compared with the empty vector-expressing cells ( Fig. 1, C and D). However, despite the presence of the exogenously expressed DR4, NF-B activation by TRAIL was relatively modest (ϳ25-fold increase) as compared with that induced by treatment with TNF␣ (ϳ120-fold increase) (Fig. 1, C and D). Treatment of 293NF-B-DR4 cells with TRAIL also led to apoptosis in ϳ15-25% of cells, whereas treatment with TNF␣ failed to do so (not shown). TRAIL-induced apoptosis in 293NF-B-DR4 cells could be completely blocked by a broad range caspase inhibitor, Z-VAD-fmk, thereby indicating the involvement of caspases in this process (data not shown).

TRAIL-induced NF-B Activation Is Inhibited by Broad Range Caspase
Inhibitor Z-VAD-fmk-As the 293NF-B-DR4 cells activated NF-B in response to both TNF and TRAIL but induced caspase-dependent apoptosis only in response to TRAIL, they provided a convenient system to investigate the contribution of caspases to NF-B activation. We began by studying the effect of the broad range caspase inhibitor Z-VADfmk on TNF␣-and TRAIL-induced NF-B activation in these cells. For this purpose, cells were treated with 20 M Z-VADfmk for 30 min before and during the treatment with the ligands, and luciferase activity was measured on cellular extracts obtained after 10 h of treatment. As shown in Fig. 2A, treatment with TNF␣ led to strong activation of NF-B luciferase activity which was only minimally influenced by Z-VADfmk. In contrast, treatment with Z-VAD-fmk led to an almost 5-fold reduction in TRAIL-induced NF-B luciferase activity as compared with Me 2 SO-treated cells (Fig. 2B). To confirm that the observed effect is specific to the inhibition of caspases and not cathepsin B, which is also inhibited by Z-VAD-fmk (34), we used the cathepsin B inhibitor E64. Treatment with 20 M E64 failed to block either TNF␣-or TRAIL-induced NF-B (Fig. 2, A  and B). We also ruled out the possibility that NF-B activation observed in 293NF-B-DR4 cells is a nonspecific effect of caspase activation and resultant cell death by using several agents known to induce caspase activation via receptor-independent mechanisms. Consistent with our results published previously (35), transient transfection of plasmids encoding several pro-apoptotic molecules, such as Bax, Bid, human caspase 6, 9, and 10, led to varying levels of cell death but failed to stimulate NF-B luciferase activity (Fig. 2C). Collectively, the above results indicate that NF-B activation observed following treatment with TRAIL is not a nonspecific consequence of caspase activation or cell death.
TRAIL-induced NF-B Activation Is Inhibited by Preferential Inhibitors of Caspase 8 -Caspase 8 plays a critical role in the activation of the caspase cascade via the TRAIL receptors (14). Z-IETD-fmk, a cell-permeable caspase inhibitor that preferentially blocks caspase 8, had no significant effect on TNF␣induced NF-B activation but was as effective as Z-VAD-fmk in blocking NF-B activation by TRAIL (Fig. 3, A and B). In contrast, Z-YDVAD-fmk, a preferential inhibitor of caspase 2, was only partially effective in blocking TRAIL-induced NF-B (Fig. 3, C and D). The data using peptidyl-based caspase inhibitors, such as Z-IETD-fmk and Z-YDVAD-fmk, should be interpreted with caution as they lack absolute specificity. Therefore, in order to provide genetic proof of the involvement of caspase 8 in TRAIL-induced NF-B activation, we used the RNA interference approach to down-regulate its expression. As shown in Fig. 3E, effective suppression of caspase 8 protein in 293NF-B-DR4 cells was achieved following a single round of transfection with a caspase 8-specific siRNA duplex. Similarly, an siRNA duplex targeted against FADD, the adaptor that recruits caspase 8 to the TRAIL DISC, led to near complete loss of its expression (Fig. 3F). The caspase 8 and FADD siRNAs also completely blocked TRAIL-induced apoptosis (not shown). We next tested the ability of caspase 8 and FADD siRNAs to block NF-B activation via TNF␣, IL-1␤, and TRAIL. We used siRNAs directed against lamin A/C as a control for these experiments. As shown in Fig. 3, G and H, siRNAs against caspase 8 and FADD had no significant effect on TNF␣-and IL-1␤-induced NF-B but effectively blocked TRAIL-induced NF-B activation. Taken together with the results using Z-IETD-fmk, the above studies confirm the involvement of caspase 8 in TRAIL-induced NF-B activation.
Caspase 3 Is Required for TRAIL-induced Apoptosis but Not TRAIL-induced NF-B-Caspase 3 is an executioner caspase that is known to play a key role in death receptor-induced apoptosis. We used the siRNA approach to specifically down- regulate expression of caspase 3 and to test its involvement in TRAIL-induced NF-B and apoptosis. As shown in Fig. 4, A and B, caspase 3 siRNA effectively silenced its expression and blocked TRAIL-induced apoptosis in 293NF-B DR4 cells. Most interestingly, unlike the case with caspase 8, silencing of caspase 3 had no significant effect on either TNF␣-or TRAILinduced NF-B activation (Fig. 4, C and D). Furthermore, Z-VAD-fmk could effectively block TRAIL-induced NF-B activation even in cells in which TRAIL-induced apoptosis was blocked due to silencing of caspase 3 expression (Fig. 4, E and  F). In addition to caspase 3, caspases 6 and 7 are also known to act as executioner caspases. However, siRNA-mediated downregulation of caspases 6 and 7 failed to block TRAIL-induced NF-B (Fig. 4, G and H). Taken together, the inhibitory effect of Z-VAD-fmk on TRAIL-induced NF-B is independent of its effect on TRAIL-induced apoptosis and activation of caspases 3, 6, or 7.
Caspase Activity Is Required for the Late Phase of NF-B Activation by TRAIL-To understand the mechanism by which caspases promote TRAIL-induced NF-B activation, we carried out an electrophoretic mobility shift assay. As shown in Fig. 5A, Z-VAD-fmk had no effect on the early phase (up to 6 h) of TRAIL-induced NF-B DNA binding activity. However, while TRAIL-induced NF-B DNA binding activity increased at the 8-and 10-h time points in Me 2 SO-treated cells, co-treatment with Z-VAD-fmk led to a gradual decline beginning at 8 h ( 5A). We obtained essentially similar results in HeLa cells, in which TRAIL signaling was initiated via its endogenously expressed receptors (Fig. 5B). Collectively, the above results suggest that caspase activity is required for the late phase of NF-B activation by TRAIL but is dispensable for the early phase.
Caspase Inhibitors Block the Late Phase of TRAIL-induced IB␣ Degradation-We next sought to determine whether Z-VAD-fmk blocks TRAIL-induced NF-B by influencing the level of the IB␣ protein. In the absence of Z-VAD-fmk, IB␣ almost completely disappeared following 2 h of TRAIL treatment (Fig. 5C, lane 4). The expression of IB␣ reappeared at the 4-and 8-h time points (Fig. 5C, lanes 5 and 6), which probably reflected newly synthesized protein, and disappeared again at the 10-and 12-h time points (Fig. 5C, lanes 7 and 8).
The decrease in expression of IB␣ observed at the latter time points was not a nonspecific consequence of caspase activation as we observed no significant change in the expression of several control proteins, such as actin, NEMO, and RIP1 (Fig. 5C). Most importantly, consistent with the gel-shift assay, cotreatment with Z-VAD-fmk had no effect on the early phase (2-h time point) of IB␣ decline (Fig. 5C, lane 12) but effectively blocked the delayed (10-and 12-h time points) decrease in its expression (Fig. 5C, lanes 15 and 16).
We next studied the effect of siRNA-mediated caspase 8 silencing on the late phase down-regulation of IB␣ expression following treatment with TRAIL and TNF␣. For this purpose, cells were transfected with siRNAs against caspase 8 or Lamin A/C (negative control),and at 48 h post-transfection cells were treated with TNF␣ or TRAIL for 10 h. As expected, both TRAIL and TNF␣ treatments led to a significant decline in IB␣ expression (Fig. 5D). However, while silencing of caspase 8 expression led to significant inhibition of IB␣ down-regulation following treatment with TRAIL, it had only a marginal effect on TNF␣-induced IB␣ degradation (Fig. 5D). Taken together with our results using Z-VAD-fmk, the above studies demonstrate a key role for caspase 8 in the delayed phase of IB␣ down-regulation following TRAIL treatment.
TRAIL Treatment Leads to Cleavage of IB␣-IB␣ can be cleaved by caspases at a consensus caspase cleavage site located between Asp 31 and Ser 32 residues to generate an Nterminal truncated fragment (36,37). However, in our initial experiments, we failed to detect any cleavage products of either endogenous or overexpressed IB␣ upon TRAIL treatment. A close examination of IB␣ sequence revealed that its caspasemediated cleavage will leave an N-terminal serine residue (Ser 32 ), which is a primary (type 3) destabilizing residue in the ubiquitin-dependent mammalian N-end rule pathway that targets proteins for proteasome-mediated degradation via their destabilizing N-terminal residues (38,39). In the N-end rule pathway, the in vivo half-life of a protein is determined by a degradation signal, called the N-degron, which consists of an exposed N-terminal destabilizing residue and internal Lys residues that are the sites of ubiquitin attachment (38,39). Incidentally, IB␣ also possesses several internal Lys residues, including one (Lys 38 ) near Ser 32 , which could potentially represent the sites of substrate-linked poly-ubiquitin chains. We speculated that our failure to detect the cleavage product of IB␣ could be due to its rapid degradation by the proteasomemediated N-end rule degradation pathway. Therefore, we repeated the above experiment in the absence or presence of MG132, a proteasome inhibitor. Consistent with our previous results, treatment with TRAIL led to down-regulation of IB␣ expression at both the 4-and 12-h time points (Fig. 6A, top  panel, lanes 2 and 3), the latter of which was significantly blocked by pretreatment with Z-VAD-fmk (Fig. 6A, top panel,  compare lanes 3 and 6). Most interestingly, when the same experiment was repeated in the presence of MG132, treatment with TRAIL led to the appearance of a cleaved form of IB␣ that was weakly detected at 4 h and more prominently after 12 h of treatment with TRAIL (Fig. 6A, 2nd panel, lanes 8 and  9). Most importantly, appearance of the cleaved fragment of IB␣ was completely blocked by Z-VAD-fmk (Fig. 6A, lanes 11  and 12). In contrast, no cleavage of IB␣ was detected in cells treated with TNF␣ in the absence or presence of MG132 (Fig.  6A, lanes 13-18), thereby ruling out the possibility that the cleavage is due to treatment with MG132. We also observed that the amount of total IB␣ was significantly reduced in cells treated with TRAIL for 12 h in the presence of MG132 (Fig. 6A,  compare lanes 3 and 9), and there was only a minor increase in its level upon cotreatment with Z-VAD-fmk (Fig. 6A, compare  lanes 9 and 12). This probably reflected the inhibitory effect of MG132 on the re-synthesis of IB␣, since the latter is an NF-B-responsive gene whose transcription is blocked by inhibition of the NF-B pathway by MG132. Cleavage of IB␣ upon treatment with TRAIL in the presence of MG132 and its inhibition by Z-VAD-fmk was also demonstrated in HeLa cells (Fig.  6B). Taken together, the above results support the hypothesis that caspase activation leads to cleavage of IB␣ to yield an N-terminal destabilizing residue, which targets it for proteasome-mediated degradation and thus contributes to the late phase of NF-B activation by TRAIL.
We generated a point mutant of IB␣ containing an Asp to  15 and 16). TRAIL treatment has no effect on the expression of RIP, NEMO, or actin. Western analysis with an antibody against the cleaved form of PARP (cl.PARP) is used to demonstrate activation of the caspase cascade upon TRAIL treatment and its inhibition by Z-VAD-fmk. D, caspase 8 siRNA blocks the decline in IB␣ level in TRAIL-treated cells. 293NF-B-DR4 cells were transfected with control and caspase 8 siRNA and 48 h later treated with PBS, TNF-␣, or TRAIL for 10 h. Whole cell lysates containing equal amounts of proteins were subjected to Western blot analysis using an IB␣ antibody.
Ala mutation at the amino acid residue 31, which abolished the caspase cleavage site. This construct carried an N-terminal FLAG epitope tag and a C-terminal Myc tag to help in the detection of the cleaved fragment (Fig. 6D). We transiently transfected the wild-type and D31A mutant IB␣ constructs into 293NF-B-DR4 cells and subsequently treated them with TRAIL in the absence or presence of MG132. An N-terminal cleaved from of IB␣ was readily detected upon treatment with TRAIL and MG132 in cells transfected with the wild-type construct and was completely absent in those transfected with the D31A mutant construct (Fig. 6C). Collectively, the above results support the hypothesis that caspase-mediated cleavage of IB␣ between Asp 31 and Ser 32 residues targets it for proteasome-mediated degradation and contributes to the late phase of NF-B activation by TRAIL.
We next sought to determine the effect of the caspase-resistant mutant of IB␣ on TRAIL-induced NF-B activation. For this purpose, we generated adeno-associated viral vectors expressing C-terminal HA-tagged D31A (cleavage-resistant), S32A/S36A (phosphorylation-resistant), and D31A/S32A/S36A mutants of IB␣. As shown in Fig. 6, E and F, TRAIL-induced NF-B was partially blocked by D31A and S32A/S36A mutants, respectively. However, infection with the triple mutant (D31A/S32A/S36A) led to the most effective inhibition of TRAIL-induced NF-B, presumably reflecting the inhibition of both the early and the late phase of NF-B activation (Fig. 6, E and F).

DISCUSSION
Genetic and biochemical data support a key role for caspase activation in Relish processing in Drosophila (1,40). In this report, we demonstrate that TRAIL, a TNF family ligand that is a powerful activator of the caspase cascade, activates NF-B in two phases as follows: an early caspase-independent phase and a late caspase-dependent phase. Our results further demonstrate that the late caspase-dependent phase of TRAIL-induced NF-B involves cleavage of the IB␣ protein and its subsequent degradation by proteasome. Thus, like the situation with Relish processing, caspase-mediated cleavage of IB is also involved in NF-B activation in mammalian cells, suggesting the existence of an evolutionary conserved signaling pathway (Fig. 7).
We demonstrate that none of the major downstream caspases involved in death receptor-induced apoptosis, such as caspases 3, 6, and 7, are required for TRAIL-induced NF-B. Thus, like the situation with Dredd-mediated cleavage of Relish in Drosophila, cleavage of IB␣ in mammalian cells might be primarily mediated by caspase 8 (Fig. 7). Another interesting and unexpected conclusion of this study is that activation of caspase 8 following TRAIL signaling may have two major con- sequences as follows: activation of pro-apoptotic effector caspases, such as caspases 3 and 6, leading to cell death, or cleavage of IB␣ resulting in NF-B activation (Fig. 7). It is conceivable that under conditions in which activation of caspase 3 and related apoptosis-inducing caspases is blocked due to the presence of their inhibitors, such as IAPs, TRAIL signaling preferentially leads to activation of caspases involved in NF-B activation. Most interestingly, expression of cIAP-1 and cIAP-2, two known inhibitors of caspase 3, is positively regulated by the NF-B pathway (Fig. 7). This positive feedback loop probably ensures continued suppression of the apoptosis-inducing effector caspases under conditions that favor TRAIL-induced NF-B (Fig. 7).
In the Drosophila imd pathway, Relish cleavage is dependent not only on the activity of Dredd but also on its C-terminal phosphorylation by DmIKK␤ (21)(22)(23)25). It is conceivable that a similar situation exists in mammalian cells and an additional signal is needed for effective IB␣ cleavage by caspases and subsequent degradation of the cleaved IB␣ by the proteasome. Consistent with the above hypothesis, we have observed no significant NF-B activation during apoptosis induced upon transient transfection of Bax, Bid, and human caspases 6, 9, and 10, which probably reflects the absence of the second signal. Finally, although our results provide strong evidence for caspase-mediated IB␣ cleavage in NF-B activation via TRAIL, we cannot entirely rule out the possibility that cleavage of additional cellular proteins might also contribute to this process. In this context, it is also important to point out that Relish more closely resembles NF-B1/p105 and NF-B2/p100 precursor proteins in overall structure, and it remains to be seen whether caspase activation also contributes to the activation of these NF-B subunits. Barkett et al. (36) reported previously that IB␣ can be cleaved by caspase 3 in vitro at a conserved Asp 31 -Ser 32 sequence, and an analogous sequence also represents the site of cleavage of IB␤ by caspase 3. A subsequent study demonstrated that IB␣ is cleaved in cells undergoing TNF␣-induced apoptosis or those dying as a result of IL-3 withdrawal to generate an N-terminal truncated form (⌬N-IB␣), which could bind the p65 subunit of NF-B and suppress TNF␣-induced NF-B activation (37). Although caspase-mediated cleavage of IB␣ was also encountered in our study, we have found that the N-terminal truncated form of IB␣ is rapidly degraded by the proteasome pathway resulting in NF-B activation rather than suppression. The exact cause for the discrepancy between the two studies is not known at the present. It is conceivable that proteasome-mediated degradation of ⌬N-IB␣ is governed by its signal-dependent secondary modification(s) (e.g. phosphorylation and ubiquitination) which is different between cells undergoing IL-3 withdrawal or those treated with TRAIL. Alternative explanations, including the fact that the two studies involved different cell lines, may apply as well. The ubiquitin ligase that recognizes primary (type 3) destabilizing residues of the N-end rule pathway (i.e. Ser, Ala, and Thr) in mammalian cells has yet to be identified (39). A difference in the expression and/or activity of this or related enzymes of the N-end rule pathway in different cell types or in response to different stimuli may also be a key determinant of the fate of ⌬N-IB␣ and cellular NF-B response following TRAIL stimulation.
Our results involving DR4 signaling are consistent with a recent report (41) demonstrating a role for caspase activation in the induction of IL-8 gene expression in human astroglioma cells via DR5/TRAIL-R2, another TRAIL receptor that is capable of activating both caspase and NF-B pathways. In addition to the TRAIL receptors DR4 and DR5, it is conceivable that caspase activation also contributes to NF-B activation via other death receptors. However, we have observed only a marginal effect of caspase inhibitors on TNF␣-induced NF-B activation in the current study. There are several possible explanations for this difference in signaling via the various death ligands. First, unlike TRAIL, TNF␣ is a strong activator of the IKK complex and through this complex results in robust phosphorylation and near-complete degradation of IB␣. Second, unlike TRAIL, TNF␣ is a weak activator of the caspase cascade. Thus, the lack of a role for caspases in TNF␣-induced NF-B activation in this study may be a combined outcome of strong activation of the IKK complex-mediated IB␣ degradation and failure to activate the caspase cascade. However, TNF␣ is known to activate caspases in some cell types and under certain circumstances (e.g. in the presence of protein synthesis inhibitors), and it is conceivable that caspase activation does contribute to TNF␣-induced NF-B activation in these situations.
The N-end rule pathway is present in all organisms examined and relates the in vivo half-life of proteins to the identity of their N-terminal residues (38,39,42). In this pathway, the N-terminal degradation signal (N-degron) is produced by a proteolytic cleavage that yields a destabilizing N-terminal residue (38,39,42). Processing proteases that have been demonstrated to participate in the N-end rule pathway include me- FIG. 7. Parallels between the Drosophila Imd pathway and the mammalian TRAIL signaling pathway. Left panel, in the Drosophila imd pathway, binding of Gram-negative bacteria to the peptidoglycan recognition protein-LC (PGRP-LC) leads to the recruitment of IMD and activation of two parallel arms of the imd pathway (28). The first arm involves dTAK1-mediated activation of Drosophila IKK complex and the second arm leads to activation of caspases via dFADD; both of these are required for processing of Relish (28). Right panel, TRAIL signaling leads to NF-B activation in two phases. The early phase involves RIP-mediated activation of the IKK complex that leads to phosphorylation and ubiquitination of IB␣, triggering its degradation by the proteasome. The late phase of TRAIL-induced NF-B involves activation of caspase 8, which results in the cleavage of IB␣ between Asp 31 and Ser 32 residues. The resulting N-terminal truncated fragment of IB␣ is degraded by the N-end rule pathway, thereby releasing the NF-B subunits. The activation of NF-B pathway, in turn, leads to transcriptional activation of anti-apoptotic proteins, such as cIAP1 and cIAP2, which can potentially block caspase 3 and cell death. Similar to the imd pathway, it is conceivable that a second signal, which may be provided by the IKK complex, regulates caspase-mediated cleavage of IB␣ in mammalian cells. thionine aminopeptidases and separases (38,39,42). A recent study (43) demonstrated that caspase-mediated cleavage of Drosophila DIAP1 leads to its degradation by the ubiquitin-dependent N-end rule pathway, thereby establishing caspases as entry points to the N-end rule pathway as well. A large number of mammalian proteins are known to be cleaved by caspase to generate fragments that are putative N-end rule substrates (38,39,42). However, unlike Drosophila DIAP1, no caspase generated mammalian protein fragment has been actually shown to be degraded by the N-end rule pathway or to play a physiologically relevant role (39). Thus, to the best of our knowledge, caspase-mediated cleavage of IB␣ to yield an Nend rule substrate is the first demonstration of caspases as entry points to the N-end rule pathway in mammalian cells and extends the role of this pathway to cellular signaling in mammalian cells. The identification of the ubiquitin ligase involved in the type 3 branch of the N-end rule pathway will be helpful in clarifying the physiological significance of caspases as an entry point to the N-end rule pathway.