The Human COP9 Signalosome Protects Ubiquitin-conjugating Enzyme 3 (UBC3/Cdc34) from β-Transducin Repeat-containing Protein (βTrCP)-mediated Degradation*

The COP9 signalosome (CSN) is an essential multisubunit complex that regulates the activity of cullin-RING ubiquitin ligases by removing the ubiquitin-like peptide NEDD8 from cullins. Here, we demonstrate that the CSN can affect other components of the ubiquitination cascade. Down-regulation of human CSN4 or CSN5 induced proteasome-mediated degradation of the ubiquitin-conjugating enzyme UBC3/Cdc34. UBC3 was targeted for ubiquitination by the cullin-RING ubiquitin ligase SCFβTrCP. This interaction required the acidic C-terminal extension of UBC3, which is absent in ubiquitin-conjugating enzymes of the UBCH5 family. Conversely, the UBC3 acidic domain was sufficient to impart sensitivity to SCFβTrCP-mediated ubiquitination to UBCH5 enzymes. Our work indicates that the CSN is necessary to ensure the stability of selected ubiquitin-conjugating enzymes and uncovers a novel pathway of regulation of ubiquitination processes.

The COP9 signalosome (CSN) is an essential multisubunit complex that regulates the activity of cullin-RING ubiquitin ligases by removing the ubiquitin-like peptide NEDD8 from cullins. Here, we demonstrate that the CSN can affect other components of the ubiquitination cascade. Down-regulation of human CSN4 or CSN5 induced proteasome-mediated degradation of the ubiquitin-conjugating enzyme UBC3/Cdc34. UBC3 was targeted for ubiquitination by the cullin-RING ubiquitin ligase SCF ␤TrCP . This interaction required the acidic C-terminal extension of UBC3, which is absent in ubiquitin-conjugating enzymes of the UBCH5 family. Conversely, the UBC3 acidic domain was sufficient to impart sensitivity to SCF ␤TrCP -mediated ubiquitination to UBCH5 enzymes. Our work indicates that the CSN is necessary to ensure the stability of selected ubiquitin-conjugating enzymes and uncovers a novel pathway of regulation of ubiquitination processes.
Ubiquitin-mediated proteolysis regulates a wide range of cellular processes by controlling the stability of short-lived proteins, such as cell cycle factors, transcriptional regulators, and signal transducers. Protein ubiquitination requires coordinated activation of different enzymes: an E1 (ubiquitin-activating), an E2 (ubiquitin-conjugating), and an E3 (ubiquitin ligase) that is responsible for substrate recognition (1). E3 enzymes of the cullin-RING ubiquitin ligase (CRL) 2 family contain a cullin and a RING domain protein that helps recruit E2 to the CRL complex (2). One of the most extensively studied CRLs is SCF, which consists of Skp1, CUL1, Rbx/Hrt1, and one of several F-box domain-containing proteins. Substrate specificity is dictated by the F-box protein: SCF SKP2 binds a number of cell cycle regulators (cyclins D 1 and E and the cell cycle inhibitors p27 and p21) through the adaptor SKP2. Conversely, the adaptor ␤-transducin repeat-containing protein (␤TrCP) targets ␤-catenin and IB and is involved in the regulation of these transcriptional pathways (3). Together with the E2 UBC3/ Cdc34, SCF plays an essential role in cell proliferation (4 -6). In addition to UBC3, other E2 enzymes may be recruited by SCF: in humans, three E2 enzymes (UBCH5A, UBCH5B, and UBCH5C) homologous to Saccharomyces cerevisiae Ubc4 have been identified and shown to ubiquitinate several substrates in the context of SCF (7). The activity of CRLs is regulated by reversible conjugation of cullins with the ubiquitin-like protein NEDD8. Cullin neddylation is essential for CRL function, as demonstrated by its severe impairment upon disruption of the neddylation pathway (8 -10). NEDD8 is thought to stimulate CRL activity through multiple mechanisms, which include displacement of the inhibitor CAND1/TIP120 and cullin conformational alterations that bring E2 in closer proximity to the substrate (11,12). Cul1 neddylation also favors recruitment of E2 enzymes, possibly by providing a direct interaction surface for E2 (13,14).
CRLs can be deneddylated by the COP9 signalosome (CSN), a highly conserved complex of eight subunits (CSN1-CSN8). The CSN was first characterized in Arabidopsis thaliana as a repressor of light-dependent development. It was subsequently identified in all eukaryotes analyzed and shown to participate in the regulation of multiple cellular pathways, such as cell proliferation, DNA repair, and developmental processes (15). The pleiotropic properties of CSN may be explained in part by its ability to control ubiquitin-dependent protein degradation through several mechanisms. The CSN can affect protein ubiquitination through association with a deubiquitinating enzyme of the cysteine protease family, called Ubp12 in Schizosaccharomyces pombe (16) or USP15 in mammalian cells (17). An additional ubiquitin isopeptidase activity, consisting in cleavage of monoubiquitin from Cul4, has been ascribed to CSN5 (18). The CSN may also affect protein degradation indirectly through the recruitment of protein kinases (inositol-1,3,4triphosphate 5/6-kinase, protein kinase D, and casein II) (19,20) and possibly of some of their substrates, such as c-Jun and the p53 tumor suppressor protein (21), affecting protein phosphorylation and stability. Perhaps the best characterized biochemical property of the CSN is its deneddylase activity, which requires a zinc protease motif in CSN5, as well as the integrity of the whole complex (22,23). In vitro studies have suggested that CSN-mediated deneddylation inhibits CRL activity (18,24). However, CSN inactivation in vivo, while inducing cullin hyperneddylation, results in the accumulation of some CRL substrates, indicating that CRL function may be impaired when CSN activity is deficient (25). To explain this apparent contradiction, it has been proposed that the CSN is required to transiently inactivate CRLs (perhaps affecting E2 recruitment) to protect selected substrate adaptors, such as Skp2, from autocatalytic ubiquitination and destruction (26 -28). In this work, we investigated how loss of the CSN affects the activity of the E2 enzymes associated with SCF. We found that human UBC3, but not UBC4/5, is a target for SCF ␤TrCP -mediated ubiquitination and that the CSN is required to protect UBC3 from proteasomedependent degradation.
Cell Culture and Transient Transfection-HEK293T and HeLa cells were grown in Dulbecco's modified Eagle's medium (Invitrogen) with 10% fetal bovine serum (HyClone). Expression plasmids were transfected in HEK293T and HeLa cells with the jetPEI TM transfection reagent (Polyplus Transfection) according to the manufacturer's protocol. Analysis of transfected cells was performed 48 h after transfection.
Lentivirus Production and Transduction-2 ϫ 10 6 HEK293T cells were plated on 100-mm plates. The following day, the cells were cotransfected by calcium phosphate precipitation with 10 g of the pCMV⌬8.9 packaging vector, 5 g of the vesicular stomatitis virus G envelope vector, and 10 g of the shRNA vector pTCN (pTCN.control, pTCN.iCSN4, or pTCN.iCSN5). The next day, the medium was substituted with Dulbecco's modified Eagle's medium with 5% fetal bovine serum and 10 mM Hepes (pH 7.4). After 24 h, supernatants were collected, centrifuged at 2500 rpm for 5 min, and concentrated with Amicon Ultra filter devices (Millipore) at 3300 rpm for 10 min. 5 ϫ 10 5 HEK293T or HeLa cells seeded in a 6-well plate were incubated with the viral supernatant. To evaluate the efficiency of viral transduction, cells were analyzed by fluorescence-activated cell sorting for nerve growth factor receptor expression 4 days after the infection (murine anti-nerve growth factor receptor monoclonal antibody clone 20.4). All experiments were performed 5-10 days after transduction.
In Vivo UBC3 Ubiquitination-48 h after transfection, HEK293T cells were treated with 20 M MG132 for 5 h and lysed in 1% SDS, 50 mM Tris (pH 8.5), 5 mM EDTA, 1 mM AEBSF, and 10 mM N-ethylmaleimide. The lysates were incubated at 95°C for 5 min, diluted up to a final volume of 1 ml with nondenaturing buffer (1% Triton X-100, 50 mM Tris (pH 8.0), 300 mM NaCl, 1 mM AEBSF, and 10 mM N-ethylmaleimide), subjected to mechanical shearing, and cleared by incubation with 60 l of Pansorbin cells (Calbiochem) for 30 min at 4°C. Homogenates were spun at 14,000 rpm for 20 min, and supernatants were measured by the Bradford assay (Bio-Rad). Equal amounts of protein were incubated with 1 g of anti-HA antibody for 2 h at 4°C, followed by incubation with protein G-Sepharose TM 4 Fast Flow (GE Healthcare). After six washes with nondenaturing buffer, bound proteins were eluted with sample buffer and analyzed by SDS-PAGE and Western blotting.
Transient Transfection of Small Interfering RNA Oligonucleotides-7 ϫ 10 4 HEK293T or HeLa cells were seeded in a 12-well plate. 24 h later, the cells were transiently transfected with a small interfering RNA (siRNA) oligonucleotide (5 nM

RESULTS
Down-regulation of CSN4 or CSN5 Destabilizes UBC3-To investigate whether the CSN affects E2 recruitment to SCF Skp2 , we down-regulated the expression of CSN4 or CSN5 by lentivirus-mediated RNA interference. As shown previously (28), down-regulation of either subunit was sufficient to disrupt the CSN complex and induce cullin hyperneddylation and degradation of the F-box protein Skp2. In CSN-deficient cells (CSN kd cells), we observed a marked decrease in UBC3 protein levels, without a concomitant decrease in mRNA levels, pointing to a post-transcriptional mechanism for UBC3 loss (Fig. 1A). UBC3 decrease was observed following CSN down-regulation in transformed cell lines (293T and HeLa), as well as in primary cultures of normal human fibroblasts and T lymphocytes (Fig.  1, A and B). CSN down-regulation did not affect the levels of UBC4/5 proteins or of the small ubiquitin-like modifier E2 UBC9 (Fig. 1A). Treatment of CSN kd cells with the proteasome inhibitor MG132 recovered UBC3 protein levels without increasing UBC3 mRNA (Fig. 1C), indicating that UBC3 loss in these cells results from proteasome-dependent degradation. To confirm that the loss of UBC3 in CSN kd cells was due to increased turnover, we analyzed UBC3 levels in total lysates from control and CSN4 kd cells at different time points after the addition of the protein synthesis inhibitor CHX. UBC3 turnover was accelerated in CSN4 kd cells compared with control cells (Fig. 1D), indicating that loss of the CSN destabilizes UBC3. Similar results were obtained by down-regulation of CSN5 (data not shown).
CUL1 Is Involved in UBC3 Degradation-Ubiquitinated forms of UBC3/Cdc34 have been detected in yeast cells (4) and in S. cerevisiae Cdc34 ubiquitination is enhanced by recruitment to SCF (6). We therefore asked whether the degradation of human UBC3 observed in CSN kd cells was also dependent on CUL1. Down-regulation of CUL1 using siRNA recovered UBC3 protein levels in CSN kd cells without affecting UBC3 levels in control cells ( Fig. 2A and supplemental Fig. S1A). This effect is specific for Cul1 because inhibition of Cul4 did not influence UBC3 levels (supplemental Fig. S1B).
UBC3 Is a Substrate of SCF ␤TrCP -Mammalian UBC3 presents at its C terminus a consensus sequence for binding to the adaptor ␤TrCP (amino acids 230 -236, DSGTEES) (supplemental Fig. S2) (30). Recruitment of substrates to ␤TrCP through the classical consensus sequence (DpSGXXpS) depends on phosphorylation of the serine residues, and amino acids 231, 233, and 236 of UBC3 have been shown to be phosphorylated in vivo (31). We therefore explored the possibility that UBC3 is targeted for degradation through an SCF ␤TrCPdependent pathway. Endogenous UBC3 and ␤TrCP interacted in co-immunoprecipitation assays using either anti-UBC3 (middle panel) or anti-␤TrCP (right panel) antibody for the immunoprecipitation (Fig. 2B). HA-UBC3 co-immunoprecipitated both with full-length ␤TrCP and with a deletion construct (␤TrCP⌬Fbox) that lacks the F-box domain and does not bind to SCF (Fig. 2C), indicating that the interaction between UBC3 and ␤TrCP does not require binding to CUL1. We then investigated whether the observed UBC3/␤TrCP interaction resulted in UBC3 ubiquitination in vivo. Low levels of UBC3ubiquitin conjugates were visible in UBC3 immunoprecipitates in the presence of transfected Myc-ubiquitin. Overexpression of ␤TrCP1 and, to a lesser extent, ␤TrCP2 increased the amount of ubiquitinated UBC3, whereas overexpression of an unrelated F-box protein (SKP2) did not induce UBC3 ubiquitination (Fig. 2D). Consistent with the role of ␤TrCP in promoting UBC3 ubiquitination, down-regulation of endogenous ␤TrCP levels by RNA interference caused an accumulation of the endogenous UBC3 protein, as well as increased levels of a known ␤TrCP target, ␤-catenin (Fig. 2E), demonstrating that ␤TrCP plays a role in regulating UBC3 levels in vivo. Downregulation of SKP2 did not affect UBC3 levels, as predicted by Skp2 inability to promote UBC3 ubiquitination (Fig. 2F).
The C-terminal Acidic Domain of UBC3 Is Necessary for Targeting to ␤TrCP and Is Sufficient to Destabilize UBCH5-To investigate the role of the DSGTEES motif in targeting UBC3 to ␤TrCP, we tested the ability of ␤TrCP1 to interact with a deletion mutant of UBC3 lacking this motif (UBC3-(1-229)) (Fig.  3A). Deletion of amino acids 230 -236 markedly decreased the interaction of UBC3 with endogenous ␤TrCP (Fig. 3B) or transfected ␤TrCP⌬Fbox (supplemental Fig. S3A), although some residual ␤TrCP was still able to co-immunoprecipitate with UBC3-(1-229) in both cases. Consistently, ␤TrCP overexpression still promoted some ubiquitination of UBC3-(1-229), although less efficiently compare with wild-type UBC3 (Fig.  3C). Interaction with endogenous ␤TrCP and ␤TrCP⌬Fbox was also impaired in the absence of phosphorylation of the UBC3 C-terminal domain, as shown by the analysis of a phosphorylation-defective construct (UBC3-5PT) (supplemental Fig. S3) (31). These findings indicate that the DSGTEES motif plays a major role in UBC3 binding to ␤TrCP1. However, in the absence of the DSGTEES motif, additional elements may come into play in UBC3 interaction with ␤TrCP. Several ␤TrCP sub-strates that contain variant or noncanonical recognition motifs have been identified recently (3). Notably, acidic residues have been reported to substitute for the phosphorylated residues of the classical consensus sequence (32). UBC3/Cdc34 is characterized by a C-terminal extension that contains an acidic domain essential for yeast Cdc34 cell cycle functions (supplemental Fig. S2) (4, 33). Deletion of the UBC3 acidic domain (UBC3-(1-200)) ( Fig. 3A) completely abolished the interaction between UBC3 and endogenous ␤TrCP (Fig. 3B) or ␤TrCP⌬Fbox (supplemental Fig. S3A), as well as ␤TrCP-induced ubiquitination (Fig. 3C), suggesting that multiple sequences in the C-terminal extension can interact with ␤TrCP. As a consequence of the above findings, we predicted UBC3-(1-200) to be more stable than wild-type UBC3. To assess the relative stability of these proteins, we coexpressed them with GFP as an internal control in cells treated or not with MG132. Western blot analysis and quantification of the UBC3/GFP ratio showed that MG132 treatment increased the levels of wild-type UBC3, but not of UBC3-(1-200), indicating that the latter is less sensitive to proteasome-dependent degradation (Fig. 3D).
Enzymes of the UBC4/5 family, which were stable in CSN kd cells (Fig. 1A), lack the acidic domain present in UBC3 (supplemental Fig. S2). We therefore asked whether this domain could confer instability to UBC4/5 proteins. We constructed chimeric UBCH5A and UBCH5B proteins containing UBC3 amino acids 200 -236 at their C termini (Fig. 4A). Turnover of the UBCH5Aϩ3tail chimera in CHX-treated cells was enhanced compared with turnover of wild-type UBCH5A (Fig.  4B), indicating that the acidic domain has a destabilizing effect on the protein. Similarly, proteasome inhibition did not affect the levels of wild-type UBCH5B but caused accumulation of UBCH5Bϩ3tail (Fig. 4C), showing that amino acids 200 -236 of UBC3 are sufficient to confer proteasomal sensitivity to UBCH5B. The acidic domain also conferred to the chimeric protein the ability to bind ␤TrCP⌬Fbox in co-immunoprecipitation assays (Fig. 4D). As a consequence, UBCH5Bϩ3tail was more efficiently ubiquitinated in vivo in a ␤TrCP-dependent manner (Fig. 4E). Taken together, these data demonstrate that UBC3 is a ubiquitination substrate of SCF ␤TrCP and that targeting to ␤TrCP requires the UBC3 acidic C-terminal extension.
␤TrCP Suppression Restores UBC3 Levels in CSN kd Cells-The decrease in UBC3 observed in CSN kd cells depended on the presence of Cul1 ( Fig. 2A). We therefore asked whether SCF ␤TrCP could be responsible for UBC3 loss under these conditions. ␤TrCP suppression by RNA interference restored UBC3 levels in CSN kd cells (Fig. 5A), indicating that the SCF ␤TrCP complex is actively ubiquitinating UBC3 in these cells. Similarly, the known SCF ␤TrCP substrate ␤-catenin was strongly accumulated upon ␤TrCP down-regulation in CSN kd cells (Fig. 5A). These findings indicate that SCF ␤TrCP is still active in these CSN kd cells, contrary to what has been observed for SCF Skp2 (27,28,34). Unlike SKP2, which is readily degraded in CSN kd cells (27,28), ␤TrCP was weakly affected by CSN down-regulation (Fig. 5B), suggesting that the residual protein is sufficient to preserve SCF ␤TrCP function.
USP15 Is Not Necessary to Protect UBC3 from Degradation-Theubiquitin-specificproteaseenzyme USP15 (Ubp12 in yeast) can be found associated with the CSN, and it has been shown to protect selected CRL components from autocatalytic ubiquitination and degradation (35). This raised the possibility that degradation of UBC3 following CSN down-regulation could be due to loss of the protective effect of USP15. Down-regulation of USP15 by RNA interference did not recover UBC3 levels in HeLa or 293T cells (Fig. 5C and data not shown), indicating that the deubiquitinating activity of USP15 is not required for UBC3 stability. ␤TrCP levels were similarly unaffected by USP15 loss. Conversely, USP15 down-regulation increased the levels of ␤-catenin, as reported recently (36).

DISCUSSION
In this work, we have shown that ubiquitin-dependent degradation controls the levels of the ubiquitinconjugating enzyme UBC3 and that the CSN is necessary to maintain UBC3 stability. In the absence of CSN activity, UBC3 is degraded through a proteasome-dependent pathway that involves CUL1. Furthermore, we found that ␤TrCP suppression recovers UBC3 levels, indicating that direct targeting by SCF ␤TrCP plays an important role in UBC3 loss in CSN-deficient cells.
Several ␤TrCP substrates present the consensus binding motif DpSGX 2-3 pS, which is also found at the C terminus of human UBC3. Deletion of the ␤TrCP consensus motif of UBC3 strongly reduced but did not completely abolish binding to ␤TrCP or ␤TrCP-dependent ubiquitination. These findings suggested that additional sequences in UBC3-(1-229) could still mediate binding to ␤TrCP. In fact, further deletion of adjacent sequences in the acidic tail of UBC3 completely abolished binding to ␤TrCP and stabilized the molecule against ubiquitin-dependent degradation. These results are in line with recent reports demonstrating that variant or noncanonical recognition motifs can bind to ␤TrCP (3). In particular, FIGURE 2. UBC3 is a target of SCF ␤TrCP . A, UBC3 protein levels in CSN knockdown cells are rescued by CUL1 suppression. shRNA-transduced 293T cells were transfected with a control oligonucleotide or with two different siRNAs targeting CUL1. UBC3 content in total lysates was analyzed by Western blotting (upper panel). Cullin mRNA levels were analyzed by quantitative reverse transcription-PCR (lower panel). B, endogenous UBC3 and ␤TrCP interact. UBC3 (middle panel) and ␤TrCP (right panel) were immunoprecipitated (IP) from HEK293T cell lysates with specific antibodies. The control used was the irrelevant mouse (left panel) or rabbit (right panel) IgG antibody. WB, Western blot. C, the F-box is not required for ␤TrCP interaction with UBC3. 293T cells were transfected with the indicated vectors (full-length ␤TrCP (␤TrCP-FL) and the F-box deletion mutant ␤TrCP⌬Fbox), and ␤TrCP was immunoprecipitated with anti-FLAG antibodies. UBC3 wt, wild-type UBC3. D, ␤TrCP overexpression promotes UBC3 ubiquitination. 293T cells were transfected as indicated. HA-UBC3 was immunoprecipitated with anti-HA antibody, and ubiquitinated forms of UBC3 were detected by anti-Myc Western blotting. E, down-regulation of ␤TrCP increases UBC3 protein levels. 293T cells were transfected with two different siRNA oligonucleotides that target both ␤TrCP1 and ␤TrCP2. F, down-regulation of SKP2 has no effect on UBC3 protein levels. 293T cells were transfected with two siRNA oligonucleotides targeting SKP2.
acidic residues have been shown to substitute for the negative charges of phosphorylated residues present in the classical ␤TrCP-binding site. Xenopus Cdc25 and human CDC25B have been described to interact with ␤TrCP through a non-phosphorylated DDG motif; the acidic context in which the DDG motif is found also affects the interaction, as shown by increased binding upon phosphorylation of adjacent sequences (32). In some instances (such as in the case of Wee1 or Cdc25), multiple binding sequences in the substrate contribute to the association with ␤TrCP (32, 37), possibly allowing cooperative binding of suboptimal sites, as seen for Sic1 binding to the F-box Cdc4 (38). Human CDC25A presents both a canonical DSG motif and a nonclassical acidic ␤TrCP-binding motif; both regions are involved in binding to ␤TrCP, although only binding through DSG is enhanced by phosphorylation (32). A similar scenario could be at play for UBC3, with primary binding mediated by the C-terminal DSGTEES motif and additional acidic residues within amino acids 201-229 cooperating to stabilize the association.
Ubiquitinated forms of UBC3/Cdc34 have been described in yeast cells (4). Although no ubiquitin ligase responsible for yeast Cdc34 ubiquitination in vivo has been described, Cdc34 from S. cerevisiae was shown to undergo autoubiquitination in vitro (39), a process that is enhanced by recruitment to SCF (6,40). Further detailed analysis identified two separate modalities for Cdc34 ubiquitination in S. cerevisiae: intramolecular ubiquitination occurs on the lysine residues that are present in the C-terminal domain of Cdc34, whereas ubiquitination of residues in the N-terminal portion of Cdc34 requires interaction with another E2 molecule in trans (41). In vivo ubiquitination and proteasome-dependent degradation of human UBC3 have not been described previously. However, human UBC3 has also been shown to undergo autoubiquitination in vitro, but the mechanisms involved (and, in particular, the contribution of the C-terminal domain) may differ from the yeast system (42). The shorter acidic tail of the human protein has no lysine residues, so intramolecular ubiquitination of the C terminus, as seen in yeast, does not take place. Consistently, Wu et al. (42) found that C terminus deleted mutants of human UBC3 (UBC3-  and UBC3-(1-208)) are still capable of autoubiquitination in vitro, with efficiencies comparable or superior to those of the wild-type protein. It remains to be determined whether autoubiquitination of human UBC3 on N-terminal sequences occurs in vivo. However, autoubiquitination does not seem to be determinant for the loss of UBC3 observed upon CSN knockdown, which may rather be explained by ␤TrCP-mediated ubiquitination. In fact, truncation of the C terminus (amino acids 200 -236) was sufficient to stabilize UBC3 equally in control and CSN kd cells (supplemental Fig. S4). This stabilizing deletion abolished ubiquitination by ␤TrCP (Fig. 3C) while not impairing recruitment to SCF (43) or autoubiquitination (42). Furthermore, downregulation of ␤TrCP in CSN-deficient cells efficiently recovered UBC3 levels (7-fold increase in CSN5 kd cells and 14-fold increase in CSN4 kd cells versus 1.6-fold increase in control cells) (Fig. 5A), suggesting that SCF ␤TrCP is targeting UBC3 more actively in cells deficient in CSN activity. An implication of these findings is that SCF ␤TrCP is competent for ubiquitination in cells with impaired CSN activity, in contrast with previous results demonstrating that CSN down-regulation is accompanied by an accumulation of substrates of another SCF complex, namely SCF Skp2 (25,28). The different behavior of the two SCF complexes in response to CSN loss may be due to intrinsic properties of their F-box subunits: whereas SKP2 is strongly destabilized in CSN-deficient cells (27,28,44), our study found that ␤TrCP levels were largely maintained and possibly sufficient to preserve SCF ␤TrCP function. The biochemical is required for interaction with ␤TrCP. 293T cells were transfected with the indicated vectors, and UBC3 constructs were immunoprecipitated (IP) with anti-HA antibody. Endogenous ␤TrCP was visualized by Western blotting. C, ␤TrCP-dependent ubiquitination of UBC3 requires the acidic C-terminal domain. 293T cells were transfected with the indicated expression constructs, and UBC3 was immunoprecipitated with anti-HA antibody. Total lysates (lower panel) and immunoprecipitated proteins (upper panels) were analyzed by Western blotting. D, the acidic domain affects UBC3 stability. 293T cells were transfected with the indicated HA-UBC3 constructs and with a GFP construct as a control of transfection efficiency. Cultures were divided in two after 24 h and treated with CHX alone (60 g/ml) or in combination with MG132 (20 M) for 6 h. UBC3 levels in lysates were quantified by phosphorimaging and normalized to GFP levels. The graph represents -fold increases in protein levels of the indicated constructs in the presence of MG132 relative to the values in the absence of MG132.
basis for the different response of distinct F-box proteins to human CSN suppression remains to be determined. Of note, we found that SKP2 co-immunoprecipitated only deneddylated CUL1, whereas both neddylated and deneddylated CUL1 proteins were present in ␤TrCP immunoprecipitates (supplemental Fig. S5). These data suggest that, although SKP2 is highly sensitive to autocatalytic ubiquitination in the presence of hyperneddylated CUL1, ␤TrCP retains its stability in association with active CUL1. Consistent with the above findings, CSN depletion did not prevent signal-dependent degradation of the SCF ␤TrCP target IB␣ (supplemental Fig. S6), in line with published reports (45,46). In addition, we did not detect accumulation of another SCF ␤TrCP substrate, ␤-catenin, after CSN suppression (Fig. 5A). The reports in the literature regarding ␤-catenin response to CSN suppression have not been uniform, with some describing an increase in ␤-catenin levels after CSN downregulation and others indicating no changes (36,(47)(48)(49). At present, the cause of these discrepancies remains unclear, but it could reflect the different cellular contexts analyzed or the different timing of the analysis after CSN suppression.
Taken together, our data suggest that, in response to CSN down-regulation, UBC3 is degraded through increased processing by the SCF ␤TrCP complex. Further work is necessary to identify the stimuli that induce ␤TrCP-dependent ubiquitination of UBC3. No variations in the levels of UBC3 protein during cell cycle progression have been described (4,50), although serum deprivation can alter UBC3 levels in human cells (data not shown) (50). Given the importance of UBC3 for cell cycle progression, we are currently testing the hypothesis that UBC3 degradation may represent a "checkpoint" in response to DNA damage or other stress stimuli (growth factor deprivation, oxidative signals). The levels of ␤TrCP mRNA and protein are increased in response to cellular stress (42), and several known SCF ␤TrCP substrates are degraded under conditions of cellular stress (51)(52)(53)(54). It is possible that, in our experimental system, inhibition of CSN activity and the consequent disruption of cell cycle progression and intracellular signaling pathways may constitute such a stress signal. It should also be noted that, under our culture conditions and in the cell lines examined, UBC3 seems, to a certain extent, to be targeted by ␤TrCP in the absence of additional stimulation because down-regulation of ␤TrCP in HeLa and 293T cells caused an increase in UBC3 levels (Fig. 2E). In conclusion, this work uncovers a further layer of regulation of ubiquitination processes by demonstrating that UBC3 is a target itself of ␤TrCP-mediated ubiquitination and that the CSN is required to protect selected ubiquitin-conjugating enzymes from proteasome-dependent degradation. The acidic UBC3 C-terminal extension is a destabilizing element. A, shown is a schematic representation of the UBCH5 constructs. UBC3 WT, wild-type UBC3. B, the addition of the UBC3 C-terminal extension reduces UBCH5A half-life. 293T cells were transfected with HA-wild-type UBCH5A or HA-UBCH5Aϩ3tail and treated with CHX for the indicated times. Lysates were analyzed by Western blotting. UBCH5A/actin ratios are represented relative to time 0. C, the UBC3 C-terminal extension destabilizes UBCH5B. 293T cells were transfected with the indicated UBCH5 constructs and a GFP vector. Cells were treated and analyzed as described in the legend to Fig. 3D. D, the UBC3 C-terminal extension promotes interaction with ␤TrCP. 293T cells were transfected with the indicated constructs. ␤TrCP was immunoprecipitated (IP) with anti-FLAG antibodies. Total lysates (lower panel) and immunoprecipitated samples (upper panels) were analyzed by Western blotting. Lower molecular weight bands in the fourth lane represent proteolytic products of UBCH5Bϩ3tail. E, the UBC3 C-terminal extension directs ␤TrCP-dependent ubiquitination of UBCH5B. 293T cells were transfected with the indicated constructs. UBCH5B was immunoprecipitated with anti-HA antibody.