The Subunit CSN6 of the COP9 Signalosome Is Cleaved during Apoptosis*

The COP9 signalosome is a large multiprotein complex that consists of eight subunits termed CSN1–CSN8. The diverse functions of the COP9 complex include regulation of several important intracellular pathways, including the ubiquitin/proteasome system, DNA repair, cell cycle, developmental changes, and some aspects of immune responses. Nod1 is also thought to be an important cytoplasmic receptor involved in innate immune responses. It detects specific motifs of bacterial peptidoglycan, and this results in activation of multiple signaling pathways and changes in cell function. In this report, we performed a yeast two-hybrid screening and discovered that Nod1 interacts with several components of the COP9 signalosome through its CARD domain. Moreover, we observed that activation of the Nod1 apoptotic pathway leads to specific cleavage of the subunit CSN6. This cleavage is concomitant with caspase processing and generates a short amino-terminal peptide of 3 kDa. A complete inhibition of this cleavage was achieved in the presence of the broad spectrum pharmacological inhibitor of apoptosis, Z-VAD. Furthermore, overexpression of CLARP, a specific caspase 8 inhibitor, completely blocked cleavage of CSN6. Taken together, these results suggest a critical role of caspase 8 in the processing of CSN6. Moreover, these findings suggest that CSN6 cleavage may result in modifications of functions of the COP9 complex that are involved in apoptosis.

The COP9 signalosome is a highly conserved multiprotein complex present in nearly all organisms (1). The core COP9 complex consists of eight subunits, designated CSN1-CSN8. There is significant homology between the COP9 complex and the regulatory lid of the 26 S proteasome (2). The functions of the COP9 complex have been primarily studied in plants and lower organisms and to a limited extent in vertebrates. This complex regulates a multiplicity of pathways in all species through its associated enzymatic activities (3)(4)(5)(6)(7). At the biochemical level, the COP9 complex mediates protein phosphorylation (2,8,9), degradation (10), deneddylation (11,12), subcellular localization of target proteins (13)(14)(15), and recruitment of deubiquitination enzymes (16,17). Interestingly, the COP9 signalosome also appears to have essential roles in R gene-mediated resistance to infection in plants (18,19). This led us to speculate that the COP9 complex may subserve related functions in vertebrates. However, there are limited studies linking the COP9 complex to immune pathways in mammals. Here we describe studies that begin to bridge this gap in our knowledge.
Nod1 is a cytoplasmic protein that belongs to the Nod/NLR/ CATERPILLER protein family. Like many members of this family, Nod1 has been implicated in innate immune responses (20,21). Nod1 activators derived from bacterial peptidoglycan induce MAP kinase activation, cytokine production, and in some cells apoptosis. The minimal structural requirement for Nod1 activators derived from peptidoglycan is the presence of mesodiaminopimelic acid (DAP) 2 linked to the ␥-position of glutamic acid (22,23). Nod1 is a tripartite protein with an amino-terminal caspase recruitment domain (CARD), a central nucleotide-binding domain (NBD), and a carboxyl-terminal leucine-rich repeat (LRR) domain. Nod1 has structural homologies to the cytosolic disease-resistant R proteins of plants, with both containing NBD and LRR domains. A large number of R genes have been identified in plants and are notable for their roles in the regulation of the defense against invading pathogens. Now there is accumulating evidence that Nod1 mediates host defense responses to a variety of intracellular pathogens, suggesting that there may be common pathways when R proteins and Nod1 are compared (24 -28). As part of our efforts to understand Nod1-dependent activation pathways, we performed a yeast two-hybrid screening and identified several proteins of the COP9 complex that interact with Nod1. This prompted us to further characterize the interactions identified in the screen and to determine whether activation of Nod1 results in biochemical changes in the COP9 complex.
A number of studies link Nod1 activation to apoptotic pathways. For example, among the first studies of Nod1, it was noted that Nod1 interacted with caspase 9 and enhanced caspase 9-mediated apoptosis (29). More recently, we showed that caspase 8 controlled Nod1-induced apoptosis in two breast cancer cell lines, MCF-7 and SK-BR3 cells (30,31). These data provided us with cell-based model systems to further dissect Nod1-dependent apoptotic pathways. Thus, we addressed the question of whether activation of Nod1 in cells that are undergoing apoptosis results in changes in any of the subunits of the * This work was supported by National Institutes of Health Grants AI 15136 COP9 signalosome. Here we report unanticipated findings showing that proteolysis of one COP9 complex subunit, CSN6, is induced by activation of Nod1 as well as by a number of other proapoptotic signals.
Mammalian Expression Constructs and Site-directed Mutagenesis-Human Myc-Nod1 cDNA was obtained from Dr. G. Nuñez (University of Michigan Medical School). Nod1 ⌬CARD, Nod1 ⌬LRR, and Nod1 LRR fragments were generated by PCR using WT Nod1 as template and inserted into pCDNA3-Myc. Mouse Myc-caspase 1 and human FLAGcaspase 5 were obtained from Dr. J. Tschopp (University of Lausanne, Switzerland). Open reading frames of the all CSN subunits and caspases 3, 4, and 6 were amplified by PCR and cloned into pCMV5.1-FLAG vector (Sigma). Caspase 9 was cloned into p3XFLAG-CMV-24 (Sigma). CSN6 D23A, D53A, D101A, and D132A/D135A were constructed by site-directed mutagenesis with the overlapping extension method by PCR using CSN6 WT DNA template. The nucleotide sequences were all confirmed by DNA sequencing.
Yeast Two-hybrid Screening-Nod1 fragment (amino acids 1-465) containing the CARD and NBD domains was cloned in pGADT7 vector. Screening was performed according to the Matchmaker Two-Hybrid System Protocol (Clontech) in the presence of 5 mM 3-aminotriazole. A GAL4-AD/leukocyte cDNA fusion library in pACT2 vector was screened as recommended by the manufacturer. Positive clones were selected and sequenced for further investigation.
Immunoprecipitation and Western Blot Analysis-Cell lysates were precleared once for 20 min at 4°C with 20 l of protein A-Sepharose beads and mixed with 0.2 g of M2 monoclonal antibody for 3 h at 4°C under constant agitation. Immune complexes were allowed to bind to 20 l of protein A-Sepharose beads overnight, beads were washed three times with lysis buffer, and the washed beads were resuspended in 30 l of Laemmli buffer and boiled for 7 min. Immunoprecipitates were separated on SDS-PAGE and transferred to nitrocellulose membranes. Filters were blocked with 3% bovine serum albumin in blocking buffer (Tris-buffered saline: 50 mM Tris⅐Cl, pH 7.5, 150 mM NaCl, 0.1% Tween 20) and incubated with anti-FLAG, anti-HA, or anti-Myc antibody for 2 h and with peroxidase-conjugated secondary antibody for 1 h at ambient temperature. Specific bands were revealed using the ECL Plus system (Amersham Biosciences).
Preparation of Cell-free Extracts and Cell-free Reactions-Cell-free extracts were generated from Jurkat cells and 293 cells as described previously (32). 10 7 cells were pelleted into a 2-ml Dounce homogenizer, and an equal volume of ice-cold cell extract buffer (CEB: 20 mM Hepes, pH 7.5, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 100 M phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 2 g/ml aprotinin) was added. Cells were allowed to swell under hypotonic conditions in CEB for 30 min on ice and were disrupted by homogenization with 30 strokes. Crude extracts were centrifuged for 10 min at 14,000 rpm to remove nuclei, unbroken cells, and other debris. Cell-free reactions were typically set in a 20-l reaction volume. Apoptosis was induced by the addition of bovine heart cytochrome c to extracts at a final concentration of 50 g/ml in the presence of 1 mM dATP. Extracts were incubated at 37°C for up to 2 h. Extracts were subsequently analyzed by Western blotting for determination of caspase activation.
In Vitro Cleavage of Recombinant CSN6-Purified recombinant CSN6 protein (200 ng) was diluted in a buffer containing 10 mM HEPES, pH 7.4, 220 mM mannitol, 68 mM sucrose, 2 mM NaCl, 2.5 mM KH 2 PO 4 , 0.5 mM EGTA, 2 mM MgCl 2 , 5 mM pyruvate, 100 M leupeptin, 1 mM phenylmethylsulfonyl fluoride, and 10 mM dithiothreitol and was incubated with active recombinant caspases (200 units/reaction; Calbiochem) in a final reaction volume of 20 l. After incubation for 2 h at 37°C, reactions were terminated by the addition of SDS-PAGE sample buffer and analyzed by Western blotting.

Nod1
Interacts with the COP9 Signalosome-To identify proteins involved in regulation of apoptosis induced by Nod1, we performed a yeast two-hybrid screening. The amino-terminal region of Nod1 containing the CARD and the NBD domains (amino acids 1-465) was fused to the DNA-binding domain of the yeast transcription factor GAL4 and used as bait against a human leukocyte cDNA library fused to the activation domain of GAL4. From the 6 ϫ 10 6 transfectants, two interacting clones were identified. These clones interacted with full-length Nod1 as well as with the CARD-NBD domain of Nod1 in yeast. The sequence of the clones was identical to two subunits of the COP9 signalosome complex, CSN3 and CSN1. When we substituted the LRR domain of Nod1 in the yeast two-hybrid assay, neither clone interacted, suggesting a specific interaction between CSN1 and CSN3 with the amino-terminal Nod1 sequences used here (data not shown).
The COP9 signalosome is a multiprotein complex that consists of eight subunits that show similarity to components of the 19 S proteasome lid complex (2). The exact functions of the COP9 signalosome are not clear. The fact that we obtained two proteins from the COP9 signalosome complex prompted us to investigate whether other components of the complex interact with Nod1. We first amplified coding sequences for each CSN component using a cDNA library obtained from HeLa cells. The resultant clones were fused to a vector encoding a FLAG epitope. To demonstrate an interaction between Nod1 and the various COP9 subunits, 293 cells were transfected with Myctagged Nod1 and FLAG-tagged CSN subunits, and co-immunoprecipitation experiments were performed. Following transfection, cell lysates were immunoprecipitated with monoclonal anti-FLAG antibody, and purified proteins were subjected to Western blot analysis (Fig. 1, A and B). As expected, purification of CSN1 and CSN3 led to co-purification of Nod1. Interestingly, subunits CSN5, CSN6, CSN7, and CSN8 also associated with Nod1. Surprisingly, CSN5 and CSN6 showed the strongest interaction of all subunits. No interaction between Nod1 and CSN2 or CSN4 could be detected.
Nod1 contains three distinct domains, including a CARD domain that functions as a protein-protein interaction domain triggering intracellular signaling. We therefore tested which of Nod1 domain was involved in association with the COP9 signalosome. We first constructed various deletion mutants, Nod1 ⌬CARD, Nod1 ⌬LRR, and Nod1 LRR. Fig. 1A shows that Nod1 mutants lacking the N-terminal CARD domain lost the ability to recruit proteins of the COP9 complex. Both Nod1 ⌬CARD and Nod1 LRR failed to associate with COP9 proteins, demonstrating that the CARD domain is necessary for association with COP9 components. Fig. 1B shows the association between Nod1 and CSN6 through its CARD domain. We then examined if the endogenous COP9 subunits and Nod1 could be immunoprecipitated together using anti-CSN1 and anti-CSN6 antibodies in HeLa cells overexpressing Myc-Nod1. Nod1 associated with both endogenous CSN6 and CSN1 (Fig. 1C). We next examined the ability of the COP9 complex to interact with Nod2, another member of the NLR family closely related to Nod1. 293 cells were co-transfected with FLAG-tagged CSN6 and Myc-tagged Nod2, and immunoprecipitation experiments were carried out. Fig. 1D shows that CSN6 was detected in Nod2 immunoprecipitates to the same extent as Nod1. Expression of Nod1 WT and Nod1 deletion mutant proteins was monitored in whole cell lysates by immunoblotting using anti-Myc antibody. C, endogenous COP9 complex interacts with Nod1. HeLa cells were transiently transfected with Myc-Nod1. Cell extracts were prepared and immunoprecipitated with an irrelevant IgG, anti-CSN1, and anti-CSN6 antibody. Co-precipitated Nod1 was revealed by immunoblotting using anti-Myc antibody (top panels). Membranes were stripped and blotted (WB) with anti-CSN1 (bottom left panel) and anti-CSN6 (bottom right panel) antibodies to confirm the presence of immunoprecipitated endogenous proteins. D, Nod2 also interacts with CSN6. 293 cells were transfected with CSN6-FLAG in the presence of empty vector, Myc-Nod1, or Myc-Nod2. Cell lysates were immunoprecipitated with anti-Myc antibody, and co-precipitated CSN6 protein was revealed by immunoblotting using anti-FLAG antibody. Immunoprecipitation of Nod1 and Nod2 was controlled after stripping of membranes using anti-Myc antibody. Expression of CSN6 was controlled in whole cell lysates by immunoblotting using anti-FLAG antibody.
Endogenous CSN6 Is Cleaved following Nod1 Activation, Leading to Apoptosis-We next asked whether the COP9 complex undergoes any modifications following Nod1 activation by ␥TriDAP. Previously, we showed that the addition of ␥TriDAP and TNF␣ to SK-BR3 cells resulted in enhanced Nod1-dependent cellular responses, including apoptosis (30). Thus, we treated SK-BR3 cells with a mixture of ␥TriDAP/TNF␣/CHX (␥TC) to induce apoptosis. Cell lysates were prepared following the addition of this mixture, and changes in the structure of COP9 complex subunits were monitored by Western blotting using specific antibodies directed against each CSN subunit. We observed proteolysis of the CSN6 subunit ( Fig. 2A). The polyclonal anti-CSN6 antibody we used detected a strong band at 36 kDa. Within 4 h of ␥TC treatment, the disappearance of CSN6 was clearly visible and concomitant with the appearance of a 33 kDa band. By 8 h, no intact CSN6 was visible in SK-BR3 cells. In contrast, we failed to observe any changes in the apparent molecular weight of any of the other COP9 complex components.
To define whether the cleavage of CSN6 resulted from caspase activation, we used three different approaches. First, cells were incubated with cell-permeable caspase inhibitor Z-VAD coincidentally with ␥TC. This pharmacologic inhibitor effectively blocked the cleavage of CSN6 (Fig. 2B). Second, we observed that SK-BR3 cells stably expressing the caspase 8 inhibitor c-FLIP/CLARP were totally resistant to ␥TC-induced cell death, and under these conditions we failed to observe cleavage of CSN6 (Fig. 2C). Third, it has been established that the addition of cytochrome c and dATP to Jurkat cell-free extracts triggers a cascade of caspase activation events, a model system that mimics APAF1-dependent caspase activation (32). We used this system to confirm the cleavage of CSN6. A specific cleavage of CSN6 upon exposure of Jurkat cell extracts to cytochrome c/dATP could be observed (Fig. 2D). Furthermore, the time course of activation of CSN6 was similar to the time course of caspases and PARP cleavage, strongly suggesting that cleavage of CSN6 is dependent on the onset of apoptosis and is likely to be mediated by caspases (data not shown). Taken together, these results show that activation of caspases is required for cleavage and degradation of CSN6.
To determine whether CSN6 cleavage is limited to SK-BR3 cells, we examined other cell types undergoing apoptosis in response to ␥TriDAP, TNF␣, or a combination of the two agonists. Cleavage of CSN6 was observed in MCF-7 cells treated with ␥TriDAP in the presence of CHX (Fig. 3A) as well as in HT-29 cells and primary MEFs treated with TNF␣ (Fig. 3, B and C). The pattern of CSN6 cleavage observed in all cell lines was identical. We recently reported that the combination of ␥TriDAP or MDP with IFN␥ induced extensive cell death in MCF-7 cells stably expressing Nod1 or Nod2, respectively (31). We therefore analyzed cleavage of CSN6 in MCF-7 Nod1 and MCF-7 Nod2 cells treated with ␥TriDAP/IFN␥ or MDP/IFN␥. Fig. 3D shows that CSN6, but not CSN5, is cleaved in response to ␥TriDAP and MDP in the presence of IFN␥. Cleavage of two markers of apoptosis, PARP and caspase 8, was also observed in these conditions. Taken together, these results demonstrate that CSN6 is the only subunit of the COP9 complex to undergo cleavage during Nod1induced apoptosis. Since this also occurred when cells were treated with different apoptotic stimuli, CSN6 cleavage may reflect a specific mechanism of apoptosis. To further examine this possibility, we exposed SK-BR3 cells to various chemotherapeutic drugs known to induce apoptosis. This included staurosporine, etoposide, camptothecin, doxorubicin, curcumin, and taxol. CSN6 cleavage was evident in cells treated with camptothecin and doxorubicin but not when treated with other compounds (Fig. 4A). Interestingly, some of these compounds, such as staurosporine, etoposide, and curcumin, induced cell death in SK-BR3 cells but did not induce CSN6 cleavage. We also studied cleavage of PARP and caspases induced by these agents. Interestingly, CSN6 correlates with PARP and caspase activation induced by camptothecin and doxorubicin. Finally, we studied CSN6 cleavage in primary MEFs. As observed with human cell lines, CSN6 was cleaved in response to doxorubicin (Fig. 4B). These results show that CSN6 cleavage occurs in cell lines as well as in primary cell cultures. Furthermore, this effect can be reproduced in cells of mouse origin.
CSN6 Is Ubiquitinated during Apoptosis-Ubiquitination is the most common mechanism of protein targeting for protea-FIGURE 2. CSN6 is cleaved during apoptosis. A, in vivo cleavage of CSN6. SK-BR3 cells were left untreated or treated with ␥TC (5 g/ml, 10 ng/ml, and 3 g/ml, respectively) for the indicated times. Cell extracts were prepared and subjected to immunoblotting with specific antibodies recognizing each CSN subunit. B, CSN6 cleavage is caspase-dependent. SK-BR3 cells were treated with ␥TC in the absence or presence of z-VAD (50 M) for 18 h. Cell extracts were prepared and subjected to immunoblotting using anti-CSN1, anti-CSN5, and anti-CSN6 antibodies. C, absence of CSN6 cleavage in SK-BR3 c-FLIP/CLARP cells. SK-BR3 cells stably expressing c-FLIP/CLARP were left untreated or treated with ␥TC for 18 h. Cell extracts were prepared, and CSN1, CSN5, and CSN6 subunits were analyzed by Western blotting. *, a nonspecific band detected by anti-CSN6 antibody. D, CSN6 cleavage in cell-free Jurkat extracts. Cell-free Jurkat extracts were incubated with or without cytochrome c (50 g/ml) in a buffer containing ATP to undergo caspase-dependent activation for the indicated times. Processing of all CSN subunits was assessed by Western blotting using specific antibodies. some-mediated degradation. To ask whether CSN6 was a target of ubiquitin, we performed an in vivo ubiquitination assay using CSN6 as substrate. FLAG-CSN6 was co-transfected in 293 cells together with a vector encoding HA-ubiquitin. As shown in Fig.  5, co-transfection of CSN6 and ubiquitin resulted in the appearance of a ladder indicating the presence of polyubiquitin CSN6 species. Co-transfection of CSN8 and ubiquitin did not result in CSN8 ubiquitination, since CSN8 could not be detected by anti-HA antibody and no ladder could be observed, suggesting the specificity of ubiquitination for CSN6. We next studied ubiquitination of endogenous CSN6 during apoptotic conditions. MCF-7 cells were treated with TNF␣ for different times in the presence of the proteasome inhibitor MG-132. Treatment of cells with MG-132 only allowed stabilization of ubiquitinated CSN6. Ubiquitinated CSN6 conjugates accumulated when cells were treated with TNF␣ (Fig. 5B). These data indicate that proteasome-mediated degradation plays an important role in CSN6 turnover during apoptosis.
Mapping the Caspase Cleavage Site in CSN6-The anti-CSN6 antibody used in this study was raised against residues 283-297 located at the carboxyl terminus (Fig. 6A). This antibody recognized both full-length and the 33-kDa cleavage product of CSN6. Therefore, we predicted that the cleavage site may be located ϳ3 kDa from the amino terminus. Inspection of the amino acid sequence of CSN6 showed that it contains one potential caspase cleavage site at position MEVD 23 , a site conserved in human, mouse, rat, and Xenopus. To determine the potential cleavage sites in CSN6, we mutated several Asp residues to Ala by site-directed mutagenesis. The various FLAG-CSN6 constructs were transfected in 293 cells. 293 cell extracts were combined to Jurkat extracts, and in vitro cytochrome c/dATP experiments were carried out. Mutation of these putative caspase cleavage sites revealed that CSN6 cleavage occurred at MEVD 23 2 (Fig. 6B). We also altered residues Asp 53 , Asp 101 , and Asp 132/135 , because these reflect residues around an imperfect caspase cleavage consensus site. None prevented the generation of the cleavage fragment.
To determine whether CSN6 is in fact a substrate for caspases, we investigated whether CSN6 can be cleaved by caspases in vitro. Recombinant CSN6-HIS was incubated with active recombinant caspases. All caspases tested (caspases 3, 6, 7, and 8) were capable of inducing CSN6 cleavage (Fig. 6C). The cleavage pattern of CSN6 generated by these recombinant caspases was similar, suggesting that there may be overlapping specificity in an in vitro cleavage assay. The cleavage fragment products of CSN6 observed following cleavage by recombinant caspases was of similar size to those previously observed in cells. To further evaluate the role of the different caspases on CSN6 cleavage in a more relevant system, we overexpressed each caspase in 293 cells and analyzed cleavage of endogenous CSN6. Interestingly, only overexpression of caspase 8 induced CSN6 cleavage (Fig. 6D). Caspases 1, 3, and 4 did not have any effect on CSN6, although they induced cleavage of PARP, showing that these caspases were enzymatically active.
The Major CSN6 Cleavage Fragment Remains in the COP9 Complex-The COP9 signalosome is a multiprotein complex that exists as an intact complex and co-fractionates at an apparent molecular mass of 500 kDa in gel filtration columns. We next analyzed whether cleavage of CSN6 during apoptosis had any effect on the integrity of the complex. Crude extracts from unstimulated and ␥TC-stimulated SK-BR3 cells were fractionated through a Superose 6 column, and the fractions were analyzed by immunoblotting to identify the endogenous COP9 signalosome subunits. As shown previously, all COP9 complex subunits co-fractionate in a peak around 500 kDa. All subunits were found essentially in fractions larger than their   APRIL 27, 2007 • VOLUME 282 • NUMBER 17 monomeric size. However, in extracts from apoptotic cells, larger amounts of free CSN1 and CSN5 proteins could be detected. The gel filtration pattern for CSN3 and CSN8 remained unchanged. More importantly, the major cleaved CSN6 fragment remained in the vicinity of the complex during apoptosis (Fig. 7A).

CSN6 Cleavage during Apoptosis
To investigate whether Nod1 co-purifies with the COP9, lysates from SK-BR3 cells stably expressing Nod1 were fractionated and analyzed by immunoblotting. Interestingly, many fractions containing the COP9 complex also contained Nod1 protein, further suggesting the interaction between Nod1 and COP9. Interestingly, Nod1 is also found in larger complexes. Induction of apoptosis did not affect the gel filtration pattern of Nod1. The weaker signal observed in ␥TCtreated cells is due to loading of fewer proteins, reflecting increased cell death.

DISCUSSION
The COP9 complex has been shown to physically interact with a variety of proteins that are not directly involved in proteolysis. It can associate either as a whole complex or as individual CSN subunits. For instance, p27 kip1 , protein kinase D, migration inhibitory factor, p53, and c-Jun have all been reported to be substrates for the COP9 complex (9,14,33,34). Here we report for the first time that Nod1 specifically binds to some components of the COP9 complex through its CARD domain. It strongly binds to CSN5 and CSN6 but also CSN1, CSN3, CSN7, and CSN8. No interaction with CSN2 and CSN4 could be detected. Interestingly, Nod2, another member of the NLR family that contains two CARD domains, was also found to interact with the COP9 signalosome. The CARD domain is present in many proteins involved in intracellular signaling and is involved in homotypic CARD/CARD protein interactions. Nod1 has been shown to interact with several caspases, such as caspases 1, 8, and 9, and may therefore be responsible for the recruitment of caspases to the COP9 complex that leads to cleavage of CSN6. Several proteins have been shown to be cleaved during apoptosis by caspases (35). Here we have demonstrated for the first time that the COP9 complex is another target for caspases. This contention is supported by studies of selective cleavage of the CSN6 subunit that we observed during apoptosis. The role of caspase cleavage was established by several different experi-  mental approaches. First, the broad spectrum caspase inhibitor Z-VAD completely blocked cleavage of CSN6 in ␥TC-treated cells. Second, we could not detect any cleavage of CSN6 in SK-BR3 cells stably expressing the caspase 8 inhibitor c-FLIP/ CLARP. We have previously shown that c-FLIP/CLARP completely abolished the apoptotic pathway induced by Nod1 in MCF-7 cells (31). The observation that CSN6 undergoes cleavage in MCF-7 cells that are deficient in caspase 3 indicates that this caspase is not essential for proteolysis of CSN6. Furthermore, no consensus cleavage site for caspase 3 was found in the CSN6 sequence. Our data strongly suggest that CSN6 is a direct target of caspase 8. First, in vitro studies showed that CSN6 was cleaved by recombinant caspase 8. However, these studies showed that all recombinant caspases tested were able to produce the 3-kDa fragment. Second, we report that only overexpression of caspase 8 induced CSN6 cleavage. Caspase 8 is known as an initiator caspase that can activate downstream caspases in a direct manner. However, caspase 8 has been shown to directly cleave other substrates, such as Bid (36), RIP (37,38), and PAK2 (39). On the other hand, none of the two inflammatory caspases tested, caspases 1 and 5, cleaved CSN6 when overexpressed in 293 cells.
Caspase-dependent cleavage can activate proteins to become proapoptotic, such as PAK2, MEKK1, and gelsolin (40 -42), or inactivate proteins involved in repair mechanisms of the cell cycle, such as PARP1, Rb, and protein kinase C␦ (43)(44)(45), or lead to degradation of structural proteins, such as lamins and actin (46,47). In data not reported here, we checked whether CSN6 cleavage could alter the apoptotic response in cells. We generated various MCF-7 cells stably expressing WT CSN6, a noncleavable mutant (CSN6 D23A), and the large cleaved CSN6 fragment (CSN6 ⌬23). However, none of these CSN6 proteins tested showed any significant change in sensitivity to apoptosis compared with parental cells. The functional consequence of CSN6 cleavage/COP9 alteration during apoptosis will require substantial study to identify the precise pathways that may be altered by this event.
The COP9 complex is involved in multiple cellular functions, such as development, signaling processes, and ubiquitindependent protein degradation. In all of these situations, COP9 is involved in degradation of proteins conjugated to ubiquitin. Thus, targeting the COP9 complex during apoptosis may serve to shut down diverse cellular functions that rely on the activity of the complex. Recently, Azevedo et al. (18) reported that plant CSN4 and CSN5 associated with the Sgt1-Rar1 complex. Sgt1 was also described as the fifth subunit of the SCF complex. These studies suggest a link between the COP9 signalosome and SCF E3 ubiquitin ligase to promote ubiquitination of target proteins in R gene-mediated signal transduction. Our data support the contention that human Nod1, which is closely related to plant R genes, may influence the activity of the COP9 complex. One possible role of Nod1 could be to target inactivation/ degradation of COP9 complex during the apoptotic process. The fragments released from CSN6 could potentially decrease or increase cell survival but also alter or disrupt enzymatic activity of the COP9 signalosome complex, leading to its activation/inactivation. Numerous reports support a link between the ubiquitin/proteasome pathway and apoptosis. Many molecules involved in FIGURE 7. Gel filtration analysis of the COP9 complex and Nod1. SK-BR3 Nod1 cells were left untreated or treated with ␥TC for 8 h. Cell lysates were prepared by several freezing/thawing cycles, and cell extracts were fractionated on a Superose 6 column and analyzed by immunoblotting using anti-CSN1, anti-CSN3, anti-CSN5, anti-CSN6, and anti-CSN8 antibodies (A) and anti-Myc antibody (B). The column was calibrated with protein standards. Fractions corresponding to the peak position of the COP9 complex (500 kDa) and to the free subunits are labeled at the top. *, a nonspecific band detected by anti-CSN6 antibody.
cell cycle control and apoptosis, such as p53, IAP, and Bcl2 members, undergo ubiquitin-mediated degradation (48 -50). The generation of cleavage products of CSN6 is reminiscent of the findings that subunits of the 19 S regulatory complex of the proteasome, such as S1, S6Ј, and S5a, are also cleaved during apoptosis (51,52). This caspase-mediated cleavage has been shown to inhibit proteasomal degradation and induce further accumulation of proapototic molecules.
From the mutagenesis studies, the cleavage site of CSN6 was mapped at MEVD 23 2. MEVD is a preferred cleavage sequence for caspases 6 and 7. Interestingly, the cleavage of CSN6 removes a hydrophobic fragment of ϳ3 kDa containing a stretch of 10 alanines at the amino terminus. This stretch is present in human and rat proteins but not in mouse or Xenopus. Interestingly, polyalanine peptides have been shown to induce apoptosis through a caspase 8-dependent mechanism (53). Thus, cleavage of CSN6 may generate a fragment that contributes to cell death. Hydrophobicity of the peptide renders its chemical synthesis very difficult. Therefore, it is very difficult to test the hypothesis that this fragment may contribute to cell death.
Only a few apoptotic stimuli led to CSN6 cleavage in cells, including ␥TriDAP, MDP, TNF␣, camptothecin, and doxorubicin. Other stimuli tested (staurosporine, etoposide, taxol, and curcumin) did not produce the 3-kDa CSN6 fragment, suggesting a specificity of apoptotic stimuli. The observation that both doxorubicin and camptothecin induced CSN6 cleavage in breast cancer cells MCF-7 and SK-BR3 suggests differences in the caspases activated by this anticancer drug compared with staurosporine, etoposide, or curcumin, which did not induce CSN6 cleavage. In this regard, doxorubicin has been reported to induce apoptosis through a Fas-dependent apoptotic pathway (54). Furthermore, CSN6 cleavage can be observed in various cell lines, such as breast cancer cells and epithelial intestinal cells as well as primary MEFs, suggesting that CSN6 cleavage is a general event induced during apoptosis in humans and mice.
In conclusion, we have shown that apoptotic pathways induced by Nod1 and TNF␣ can alter the COP9 complex integrity. Both Nod1 ligand and TNF␣ induced degradation of CSN6. Cleavage of CSN6 by caspases during apoptosis might influence the apoptotic cascade (e.g. by altering COP9 activity and thereby stabilizing pro-or antiapoptotic factors). Further studies are under way to examine this possibility.