Abstract
CUL7 and the p53-associated, PARkin-like cytoplasmic protein (PARC) were previously reported to form homodimers and heterodimers, the first demonstration of cullin dimerization. Although a CUL7-based SKP1/CUL1/F-box (SCF)–like complex has been observed, little is known about the existence of a PARC-based SCF-like complex and how PARC interacts with CUL7-based complexes. To further characterize PARC-containing complexes, we examined the ability of PARC to form an SCF-like complex. PARC binds RBX1 and is covalently modified by NEDD8, defining PARC as a true cullin. However, PARC fails to bind SKP1 or F-box proteins, including the CUL7-associated FBXW8. To examine the assembly of PARC- and CUL7-containing complexes, tandem affinity purification followed by multidimensional protein identification technology were used. Multidimensional protein identification technology analysis revealed that the CUL7 interaction with FBXW8 was mutually exclusive of CUL7 binding to PARC or p53. Notably, although heterodimers of CUL7 and PARC bind p53, p53 is not required for the dimerization of CUL7 and PARC. The observed physical separation of FBXW8 and PARC is supported functionally by the generation of Parc−/−, Fbxw8−/− mice, which do not show exacerbation of the Fbxw8−/− phenotype. Finally, all of the PARC and CUL7 subcomplexes examined exhibit E3 ubiquitin ligase activity in vitro. Together, these findings indicate that the intricate assembly of PARC- and CUL7-containing complexes is highly regulated, and multiple subcomplexes may exhibit ubiquitin ligase activity. [Cancer Res 2007;67(5):2006–14]
Introduction
The targeted polyubiquitination and subsequent degradation of proteins is an important regulatory mechanism for a variety of cellular processes. The targeted proteolysis of ubiquitinated proteins is an irrevocable step, forcing processes such as the cell division cycle to move inexorably in the forward direction (1). Polyubiquitination results from the reiteration of an enzymatic cascade wherein ubiquitin is activated by the E1 ubiquitin activating enzyme, transferred to an E2 ubiquitin conjugating enzyme, and finally transferred to a lysine residue on the substrate protein, or on lysine 48 of a previously added ubiquitin moiety, by an E3 ubiquitin ligase (2). In general, E3 ligase complexes can be grouped into two main families based on the presence of a HECT domain protein or a RING domain protein (2).
The paradigm for the multi-subunit RING E3 ligase complexes is the SKP1/CUL1/F-box (SCF) complex (3, 4). The highly helical core CUL1 scaffold of SCF recruits the RING domain protein RBX1, which, in turn, recruits an E2 enzyme to the complex (3, 5). The NH2 terminus of CUL1 determines the specificity of the complex by binding the bridging protein SKP1 and a variable F-box– containing protein (6). F-box proteins such as SKP2, FBW7, and β-TRCP allow the same core CUL1 scaffold to target p27, cyclin E, and Iκβα, respectively, for ubiquitination and degradation (7). CUL1 itself is also tightly regulated by the covalent addition or proteolytic cleavage of the small ubiquitin-like protein NEDD8, which promotes SCF complex assembly and activity while increasing the rate of SCF component degradation.
By database searches, mammalian genomes contain genes for eight cullins: CUL1, CUL2, CUL3, CUL4A, CUL4B (henceforth CUL4A/B), CUL5, CUL7, and PARC, the p53-associated, PARkin-like cytoplasmic protein. With the exception of PARC, all have been shown to form complexes biochemically similar to the SCF complex. CUL2, CUL3, CUL4A/B, and CUL7 all bind to the RING protein RBX1, whereas CUL5 binds to the related RING protein RBX2. Additionally, CUL2, CUL3, CUL4A/B, and CUL5 have been shown to be covalently modified by NEDD8, leading to the definition of cullins as proteins binding RBX1 and being covalently modified by NEDD8 (8, 9). Although it displays weak homology with the cullins and binds the RBX1-related protein APC11, APC2, a subunit of the anaphase-promoting complex/cyclosome (APC/C), is not generally considered a cullin (4).
Although the complexes formed by the greater cullin family resemble the canonical SCF complex, they exhibit key differences, most notably, their substrate adapters. In terms of RBX1 binding and substrate adapter binding, the CUL7 complex is most like the SCF complex (10–15). CUL7 binds RBX1 through a COOH-terminal domain, and SKP1 and the F-box protein FBXW8 via an NH2-terminal domain (11). However, at 1,698 amino acids, CUL7 is the second largest cullin, containing significantly more primary sequence than CUL1, binding three additional proteins. p53 binds to the HERC2 homology domain, and glomulin (GLMN) binds near the COOH terminus, whereas PARC binding has not been localized to a discrete domain (11, 16–18).
PARC is a putative cullin protein, highly homologous to CUL7, containing multiple recognized homology domains (16, 19). Similar to CUL7, PARC binds p53, and this interaction is likely to occur through the HERC2 homology domain (19). Additionally, the RING-IBR-RING domain of PARC, which is not found in CUL7, may bind to the E2 UBCH7 (20). Previously, it was suggested that PARC served to sequester p53 in the cytoplasm to prevent p53-mediated transcription and apoptosis; however, this model of PARC function is not supported by studies using Parc-null mice and cells (16, 19).
Here, we describe our efforts to further understand the construction, regulation, and function of the PARC- and CUL7-containing complexes. First, through reasonable candidate searches of known cullin-associated proteins, we define PARC as a bona fide cullin based on covalent modification with NEDD8 and binding to RBX1 via the cullin homology domain. Our reasonable candidate search also determined that CUL7 is a unique cullin that does not undergo covalent modification with NEDD8. Second, we determined the composition of PARC- and CUL7-containing complexes using tandem affinity purifications of PARC and CUL7, and extended these studies using sequential tandem affinity purification for substoichiometric members of each complex. These affinity purification experiments showed that the PARC/CUL7 heterodimer is capable of binding p53, whereas the CUL7/FBXW8 complex does not contain PARC or p53. We observed that each complex containing CUL7 or PARC had E3 ubiquitin ligase activity in vitro. Finally, we examined the effect of Parc mutation in a mouse strain that was also mutated in Fbxw8. Double-null mice exhibited a phenotype similar to the Fbxw8-null mice, consistent with the model that PARC does not affect the CUL7/FBXW8 complex.
Materials and Methods
Plasmids. The pFLXX+ lentivirus was constructed from the FIV lentiviral transfer vector pFLX-CPLks by inserting a cassette containing a SV40 X-8 promoter driving the blasticidin deamidase (bsd) marker from pEF6 (Invitrogen, Carlsbad, CA) into the SacII site of pFLX-CPLks (21, 22).
The pFLXX+ FLAG-HA-PARC, pFLXX+ PARC-FLAG-HA, and pFLXX+ FLAG-HA-CUL7 lentiviral vectors were generated by cloning the cytomegalovirus promoter and cDNAs from pcDNA 3.0 FLAG-HA-PARC, pcDNA 3.0 PARC-FLAG-HA, and pcDNA 3.0 FLAG-HA-CUL7 into pFLXX+ using MfeI and NotI (16). The NH2-terminal pcDNA 3.0 FLAG-HA constructs were generated by cloning the PARC and CUL7 cDNAs from pcDNA 3.0 HA-PARC, and pcDNA 3.0 HA-CUL7 into pcDNA 3.0 FLAG-HA N using BamHI and NotI. To generate pcDNA PARC-FLAG-HA, a 3′ BamHI-XhoI fragment of the PARC cDNA lacking the stop codon was created by PCR, subcloned into pBS SKII, before being cloned into the pcDNA 3.0 COOH-terminal FLAG-HA vector using BamHI and NotI. The 5′ sequence of the PARC cDNA was then cloned from pcDNA 3.0 HA-PARC, using BamHI. The NH2- and COOH-terminal pcDNA 3.0 FLAG-HA constructs were generated by cloning PCR fragments from pOZ-N containing the FLAG-HA sequence into pcDNA 3.0 (Invitrogen), using HindIII and BamHI (NH2-terminal) or XhoI and XbaI (COOH-terminal; ref. 23).
pcDNA 3.0 5XMyc-NEDD8 was generated from pcDNA 3.0 5XMyc-c-Jun by insertion of the NEDD8 cDNA using BamHI and NotI, and the pcDNA 3.0 Myc-RBX1 construct was generated from pcDNA 3.0 Myc-PARC by insertion of the RBX1 cDNA using BamHI and NotI. All sequences generated by PCR were confirmed by sequencing, and primer sequences are available upon request.
Cell culture. Stable cell lines stably expressing FLAG-HA-PARC, PARC-FLAG-HA, and FLAG-HA-CUL7 were constructed by lentiviral transduction of the BJAB cell line with the pFLXX+ transfer vectors described above. Lentivirus was generated as described (21). Forty-eight hours posttransduction, BJAB target cells were selected with 10 μg/mL of blasticidin (Invitrogen). Stable cell lines expressing PARC-FLAG-HA and Myc-RBX1 or 5XMyc-NEDD8 were generated by stable transfection of PARC-FLAG-HA BJAB cells with pcDNA 3.0 Myc-Rbx1 and pcDNA 3.0 5XMyc-NEDD8 using LipofectAMINE 2000 (Invitrogen) and G418 (Cellgro) selection. HeLa, HCT116, HCT116 TP53-null, and 293T cells were maintained in DMEM supplemented with 10% fetal clone serum (Hyclone, Logan, UT) and transfected using LipofectAMINE Plus (Invitrogen). BJAB cells and BJAB-derived cell lines were maintained in RPMI 1640 supplemented with 5% fetal clone serum.
Mice. Parc−/−, Fbxw8−/− mice were generated by crossing Parc−/− mice (129 backcrossed to C57BL/6 five times) to Fbxw8−/− mice (129 backcrossed to C57BL/6 twice).4
T. Tsutsumi et al., submitted for publication.
Antibodies. Rabbit polyclonal antibodies to PARC (BL616, BL617), CUL7 (BL653), and nonspecific rabbit IgG were obtained from Bethyl Laboratories. Anti-CUL7 (h1557) was generated as previously described. Rabbit polyclonal FBXW8 antibody (MDD) was generated by Bethyl Laboratories.4 Additional mouse monoclonal antibodies used for immunoprecipitation and Western blotting include anti-Myc (9B11; Cell Signaling Technology), anti-FLAG (M2; Sigma, St. Louis, MO), anti-GST (DG122), anti-p53 (DO1, Lab Vision), anti-HA (HA-11; Covance; 12CA5), SKP1 (Lab Vision), and RBX1 (Lab Vision).
Immunoprecipitation and Western blotting. HeLa, HCT116, and 293T cells were lysed in EBC [50 mmol/L Tris (pH 8), 120 mmol/L NaCl, 0.5% NP40] containing protease inhibitor cocktail set I (EMD Bioscience, San Diego, CA). Extracts were cleared by centrifugation at 12,000 × g for 10 min, and protein concentrations were determined by Bradford Assay (Bio-Rad, Richmond, CA). Immunoprecipitations were done for ≥3 h at 4°C, and the subsequent immunoprecipitates were washed thrice with EBC before boiling in SDS-sample buffer. Whole cell lysates and immunoprecipitates were separated by SDS-PAGE and transferred to nitrocellulose membranes (Bio-Rad) for Western blotting. Membranes were blocked in 5% milk/TBST before Western blotting. The detection of proteins was accomplished using the appropriate secondary antibodies conjugated to horseradish peroxidase (Pierce, Rockford, IL) at a 1:5,000 dilution in 5% milk/TBST. Western blots were developed using enhanced chemiluminescence (SuperSignal, Pierce).
Tandem affinity purification. Tandem affinity purifications for PARC and CUL7 were done from S100 fractions. Cytoplasmic extracts were generated by hypotonic lysis from 5 to 7 mL of packed cell volume and centrifuged at 100,000 × g for 1 h prior to immunoprecipitation. Anti-FLAG immunoprecipitation was done using M2 affinity resin (Sigma). FLAG immunoprecipitates were washed seven times with 310 mmol/L of EBC and once with 120 mmol/L of EBC before elution with 200 μg/μL of 3× FLAG peptide (Sigma) in 120 mmol/L of EBC. Eluted complexes were then immunoprecipitated for either the HA tag (tandem affinity purification) or a protein/tag other than PARC/CUL7 (sequential tandem affinity purification). Following the second immunoprecipitation in sequential experiments, the supernatant was subjected to immunoprecipitation for the HA tag. The second and third immunoprecipitates were each washed seven times with 310 mmol/L of EBC and once with 120 mmol/L of EBC before elution by boiling in SDS sample buffer lacking bromophenol blue and glycerol. All immunoprecipitations were done for >4.5 h at 4°C with rotation. All antibodies were cross-linked to protein A Sepharose using dimethyl pimelimidate. Ten percent of each sample was analyzed by SDS-PAGE and silver staining (Proteosilver, Sigma) prior to multidimensional protein identification technology (MudPIT) analysis (24).
Peptide mixtures generated by endoproteinase LysC and trypsin digestion were analyzed as described (25). SEQUEST was used to match tandem mass spectra to 40,877 Homo sapiens protein sequences from the National Center for Biotechnology (March 3, 2006 release) complemented with 177 usual contaminants, and 102 epitope-tagged proteins (26). To estimate false discovery rates, each entry was randomized, leading to a final database size of 82,242 amino acid sequences. Spectra/peptide matches were sorted and filtered using DTASelect (27). Spectra had to match full tryptic peptides at least seven amino acids long, with a maximum Sp score of 10, a normalized difference in cross-correlation scores (DeltCn) of at least 0.08, and minimum cross-correlation scores (Xcorr) of 1.8 for singly charged, 2.5 for doubly charged, and 3.5 for triply charged spectra. Proteins had to be identified by at least two peptides or on a peptide with two independent spectra. These selection criteria led to false discovery rates of at most 0.65% (0.1% in average), for the 14 affinity-purified protein mixtures and 12 negative controls analyzed (Supplemental Table S1). Contaminants, human keratins, and proteins that were subsets of others were removed from the final list and normalized spectral abundance factors (NSAF) were calculated as described (28).
In vitro ubiquitination. PARC and CUL7 complexes were tandem affinity–purified with extensive washing and left immobilized on protein A Sepharose following the second immunoprecipitation. A reaction mix containing ubiquitination buffer [25 mmol/L HEPES (pH 7.4), 10 mmol/L NaCl, 3 mmol/L MgCl2, 0.05% Triton X-100, 0.5 mmol/L DTT], 0.03 μg E1 (Biomol), 0.05 μg E2 (Boston Biochem), 1 μg of GST-ubiquitin (Boston Biochem), and 3 mmol/L of Mg-ATP (Boston Biochem) was added to bead-bound complexes, and the reaction was incubated at 37°C for 35 min before boiling in SDS sample buffer.
Results
PARC forms an SCF-like complex. The cullin family has been defined by three biochemical features: RBX1 or RBX2 binding, covalent modification by NEDD8, and NH2-terminal substrate adapter binding (3). Although the cullin homology of PARC has been noted by several groups, it has not been shown to biochemically function as a cullin (16, 19). Within the cullin homology domain, PARC has ∼60% identity to CUL7, yet only 25% identity to CUL1, making it a distant relative to the cullins similar to APC2, the core subunit of the APC/C (4).
To determine whether PARC is a bona fide cullin, we did a reasonable candidate search of known cullin-associated proteins, including SKP1, RBX1, and NEDD8. To examine PARC binding to RBX1, HA-tagged PARC and Myc-tagged RBX1 were cotransfected into HeLa cells for analysis by immunoprecipitation and Western blotting using the HA and Myc tags. Additionally, to determine if the cullin homology domain of PARC was required for RBX1 binding to PARC, a mutant of PARC lacking amino acids 1697 to 1785 in the cullin homology domain, analogous to an RBX1-binding deficient CUL1, was cotransfected with Myc-RBX1 for analysis (29). Immunoprecipitations for the HA tag–precipitated wild-type HA-PARC or HA-PARCΔCullin, and HA-PARC, but not HA-PARCΔCullin, efficiently coimmunoprecipitated Myc-RBX1 (Fig. 1A). Similar results were obtained with immunoprecipitation for Myc-RBX1, which could coimmunoprecipitate HA-PARC but not HA-PARCΔCullin, showing that PARC binds RBX1 in vivo and that binding requires the cullin homology domain of PARC.
PARC binds RBX1. A, HeLa cells transfected with Myc-RBX1 and either HA-PARC or HA-PARCΔCullin were immunoprecipitated for the Myc (9B11) or HA (12CA5) epitope tags prior to SDS-PAGE and Western blotting with Myc (9B11) and anti-HA (HA11). B, HeLa cell lysates were immunoprecipitated using anti-CUL7 (h1557), anti-PARC (BL617), or nonspecific rabbit IgG before Western blotting for PARC (BL617), CUL7 (BL653), FBXW8 (MDD), SKP1, or RBX1. One hundred micrograms of whole cell lysate was included to shows protein expression levels.
PARC binds RBX1. A, HeLa cells transfected with Myc-RBX1 and either HA-PARC or HA-PARCΔCullin were immunoprecipitated for the Myc (9B11) or HA (12CA5) epitope tags prior to SDS-PAGE and Western blotting with Myc (9B11) and anti-HA (HA11). B, HeLa cell lysates were immunoprecipitated using anti-CUL7 (h1557), anti-PARC (BL617), or nonspecific rabbit IgG before Western blotting for PARC (BL617), CUL7 (BL653), FBXW8 (MDD), SKP1, or RBX1. One hundred micrograms of whole cell lysate was included to shows protein expression levels.
The evolutionary relationship and degree of identity between PARC and CUL7 suggested that PARC, like CUL7 and CUL1, might also bind to SKP1 and an F-box–containing protein, in particular, the CUL7-specific F-box FBXW8. To examine PARC for SKP1 and FBXW8 binding, immunoprecipitation for PARC and Western blotting for PARC, CUL7, FBXW8, SKP1, and RBX1 was done. As a positive control for SKP1, FBXW8, and RBX1 binding, immunoprecipitation for CUL7 was also done. As shown in Fig. 1B, whereas PARC efficiently coimmunoprecipitated CUL7 and RBX1, PARC failed to bind either FBXW8 or SKP1. CUL7 was able to efficiently coimmunoprecipitate PARC, RBX1, FBXW8, and SKP1.
To further examine the potential of PARC and CUL7 to bind F-box proteins other than FBXW8, a library of 34 different F-box proteins, half of the predicted human F-box family, were tested by transfection and coimmunoprecipitations with Western blotting (6). PARC failed to bind any F-box protein assayed, whereas CUL7 bound only FBXW8 (data not shown).
To further define PARC as a cullin, the ability of PARC to be covalently modified by NEDD8 was examined. Neddylation occurs at a defined motif that is conserved in all cullins but is not conserved in the cullin-related protein APC2 (9). The cullin neddylation motif is also conserved in PARC, suggesting that PARC could also function as a substrate for neddylation (Fig. 2A). To examine PARC for neddylation, a construct expressing 5XMyc-NEDD8 was generated. The addition of a 5XMyc tag to NEDD8 created a fusion protein of 16 kDa compared with the 9 kDa unmodified NEDD8, permitting the discrimination of any supershifted bands from the normal migration of PARC.
PARC is neddylated. A, the consensus site for cullin neddylation is conserved in PARC and CUL7. Arrow, the lysine to which NEDD8 is covalently linked in other cullins. B, 293T cells were either mock-transfected or transfected with 5XMyc-NEDD8. Lysates were immunoprecipitated for PARC (BL920) and the Myc epitope tag (9B11) prior to SDS-PAGE and Western blotting for total PARC (top, BL617), 5XMyc-NEDD8-PARC (middle, 9B10), and 5XMyc-NEDD8 monomer (bottom, 9B10). Whole cell lysates were included to show protein expression, and a peptide-blocked PARC immunoprecipitation was included as a negative control. C, HeLa cells were either mock-transfected or transfected with 5XMyc-NEDD8. Lysates were immunoprecipitated using anti-PARC (BL617), anti-CUL7 (h1557), or nonspecific rabbit IgG prior to SDS-PAGE and Western blotting for PARC (BL617) or CUL7 (BL653).
PARC is neddylated. A, the consensus site for cullin neddylation is conserved in PARC and CUL7. Arrow, the lysine to which NEDD8 is covalently linked in other cullins. B, 293T cells were either mock-transfected or transfected with 5XMyc-NEDD8. Lysates were immunoprecipitated for PARC (BL920) and the Myc epitope tag (9B11) prior to SDS-PAGE and Western blotting for total PARC (top, BL617), 5XMyc-NEDD8-PARC (middle, 9B10), and 5XMyc-NEDD8 monomer (bottom, 9B10). Whole cell lysates were included to show protein expression, and a peptide-blocked PARC immunoprecipitation was included as a negative control. C, HeLa cells were either mock-transfected or transfected with 5XMyc-NEDD8. Lysates were immunoprecipitated using anti-PARC (BL617), anti-CUL7 (h1557), or nonspecific rabbit IgG prior to SDS-PAGE and Western blotting for PARC (BL617) or CUL7 (BL653).
293T cells were transfected with 5XMyc-NEDD8, and immunoprecipitation was done for endogenous PARC or the epitope-tagged NEDD8. Immunoprecipitation using a PARC blocking peptide served as a negative control, and Western blotting for monomeric 5XMyc-NEDD8 (Fig. 2B,, bottom) confirmed efficient transfection and expression. Western blotting for PARC detected unmodified PARC as well as a supershifted band corresponding to the predicted molecular weight for 5XMyc-neddylated PARC in the whole cell lysate, PARC immunoprecipitate, and the Myc immunoprecipitate in the samples transfected with 5XMyc-NEDD8 (Fig. 2B,, top). Neither PARC band was detected in the peptide-blocked control. Additionally, the slower migrating band, but not the unmodified PARC band, was specifically detected upon Western blotting for the 5XMyc tag, further confirming the covalent nature of the modification (Fig. 2B,, middle). The observed neddylation is consistent with the migration of HA-PARC as a doublet and HA-PARCΔCullin as a single band, likely due to the requirement of RBX1 for cullin neddylation (Fig. 1A; ref. 8).
Notably, 5XMyc-neddylated PARC was able to coimmunoprecipitate unmodified PARC, suggesting that neddylated PARC is still capable of dimerizing with modified and unmodified PARC. However, only one band, corresponding to modified PARC, was present in Myc Western blots, suggesting either that neddylated PARC cannot bind CUL7 or that CUL7 is not neddylated (data not shown). To investigate this question, HeLa cells were transfected with 5XMyc-NEDD8 prior to immunoprecipitation and Western blotting for PARC and CUL7. Neddylated PARC was detected in the whole cell lysate, PARC immunoprecipitate, and CUL7 immunoprecipitate of 5XMyc-NEDD8–transfected cells. In addition, CUL7 was reciprocally coimmunoprecipitated with PARC in all samples (Fig. 2C). However, expression of 5XMyc-NEDD8 did not result in a supershifted CUL7 band in any sample lane, indicating that CUL7 was not a substrate for neddylation. The mechanism resulting in the CUL7 doublet observed in both the mock-transfected and transfected lanes remains unknown.
Defining subsets of PARC and CUL7 complexes. Although PARC and CUL7 seem to form cullin-like complexes, these complexes seem to be atypical in terms of associated proteins and oligomeric state. For example, PARC but not CUL7 is covalently modified by NEDD8, and PARC fails to bind SKP1 or multiple F-box proteins, despite the ability of CUL7, its evolutionary descendant, to bind SKP1 and FBXW8. Of particular interest, PARC failed to coprecipitate FBXW8 despite binding to CUL7.
To further define PARC- and CUL7-containing complexes, tandem affinity purification coupled with MudPIT was used (24). To examine the suitability of this approach, NH2- and COOH-terminal, FLAG-HA–tagged PARC were transduced into the BJAB B-cell line for large-scale preparations of cytoplasmic S100 extracts (Fig. 3A). Extracts were immunoprecipitated using the FLAG and HA epitopes prior to analysis by MudPIT. In both the NH2- and COOH-terminally tagged PARC immunoprecipitations, thousands of spectra for PARC were obtained, and there was no significant difference in associated proteins for the NH2-terminal versus the COOH-terminal tagged PARC (Supplemental Fig. S1). Peptides for known PARC-associated proteins, including CUL7, p53, RBX1, and NEDD8 were present, indicating that PARC complexes were efficiently immunoprecipitated. Notably, tandem affinity purifications of PARC analyzed by MudPIT lacked peptides for SKP1 or FBXW8, confirming the results obtained with immunoprecipitation and Western blotting.
Sequential tandem affinity purification of PARC and CUL7 complexes. A, BJAB cell lines expressing the indicated constructs were constructed by lentiviral transduction (PARC and CUL7) and stable transfection (Myc-RBX1 and 5XMyc-NEDD8). One hundred micrograms of whole cell extract from each cell line was analyzed by Western blotting using anti-HA (HA-11), anti-FLAG (M2), or Myc (9B11). B, control BJAB cells (lane 1) and cells expressing PARC-FLAG-HA were immunoprecipitated with FLAG antibody followed by FLAG peptide elution and re-immunoprecipitation with a p53 antibody (lanes 1 and 2). The remaining supernatant was immunoprecipitated with HA antibody to immunoprecipitate PARC complexes that remained after p53 was immunodepleted. Ten percent of the sequential tandem affinity purification was analyzed by SDS-PAGE and silver staining.
Sequential tandem affinity purification of PARC and CUL7 complexes. A, BJAB cell lines expressing the indicated constructs were constructed by lentiviral transduction (PARC and CUL7) and stable transfection (Myc-RBX1 and 5XMyc-NEDD8). One hundred micrograms of whole cell extract from each cell line was analyzed by Western blotting using anti-HA (HA-11), anti-FLAG (M2), or Myc (9B11). B, control BJAB cells (lane 1) and cells expressing PARC-FLAG-HA were immunoprecipitated with FLAG antibody followed by FLAG peptide elution and re-immunoprecipitation with a p53 antibody (lanes 1 and 2). The remaining supernatant was immunoprecipitated with HA antibody to immunoprecipitate PARC complexes that remained after p53 was immunodepleted. Ten percent of the sequential tandem affinity purification was analyzed by SDS-PAGE and silver staining.
To gain a better understanding of the relative abundance of each protein, normalized spectral abundance factors (NSAFs) were calculated to account for molecular weight bias and facilitate the comparison of multiple samples (28). The NSAFs suggested that many of the known PARC-associated proteins were substoichiometric in the tandem affinity purification, with one half and one tenth the molar amount of p53 and CUL7, respectively, compared with PARC (Table 1, columns 1 and 2). Although this difference in stoichiometry suggests that a PARC homodimer may bind p53, the difference could also result from the overexpression of PARC, multiple subsets of complexes, or a combination thereof.
Nonredundant NSAF values for known PARC- and/or CUL7-associated proteins in PARC and CUL7 MudPIT analyses
Cell line . | Purification . | PARC* . | CUL7* . | p53 . | RBX1 . | NEDD8 . | SKP1 . | FBXW8 . | GLMN . |
---|---|---|---|---|---|---|---|---|---|
FL-HA-PARC | FLAG-HA | 0.231 | 0.033 | 0.102 | X | 0.039 | X | X | X |
PARC-FL-HA | FLAG-HA | 0.223 | 0.031 | 0.1 | 0.002 | 0.053 | X | X | X |
PARC-FL-HA, Myc-RBX1 | FLAG-Myc | 0.265 | 0.056 | 0.131 | X† | 0.05 | X | X | X |
PARC-FL-HA, Myc-RBX1 | FLAG-Myc depletion-HA | 0.262 | 0.038 | 0.064 | 0.012 | 0.078 | X | X | X |
PARC-FL-HA, 5XMyc-NEDD8 | FLAG-Myc | 0.096 | 0.016 | 0.046 | 0.061 | 0.022 | X | X | X |
PARC-FL-HA, 5XMyc-NEDD8 | FLAG-Myc depletion-HA | 0.157 | 0.018 | 0.044 | 0.018 | 0.019 | X | X | X |
PARC-FL-HA | FLAG-p53 | 0.352 | 0.033 | 0.3 | X | 0.028 | X | X | X |
PARC-FL-HA | FLAG-p53 depletion-HA | 0.215 | 0.027 | 0.006 | 0.017 | 0.028 | X | X | X |
PARC-FL-HA | FLAG-CUL7 | 0.181 | 0.066 | 0.09 | X | 0.061 | X | X | X |
PARC-FL-HA | FLAG-CUL7 depletion-HA | 0.311 | 0.019 | 0.238 | 0.129 | X | X | X | X |
FL-HA-CUL7 | FLAG-PARC | 0.003 | 0.116 | 0.064 | X | X | 0.03 | X | X |
FL-HA-CUL7 | FLAG-PARC depletion-HA | 0.003 | 0.083 | 0.018 | 0.017 | X | 0.05 | 0.038 | 0.002 |
FL-HA-CUL7 | FLAG-p53 | 0.011 | 0.094 | 0.133 | 0.099 | X | 0.01 | 0.004 | X |
FL-HA-CUL7 | FLAG-p53 depletion-HA | 0.002 | 0.106 | 0.001 | 0.032 | X | 0.05 | 0.025 | X |
Cell line . | Purification . | PARC* . | CUL7* . | p53 . | RBX1 . | NEDD8 . | SKP1 . | FBXW8 . | GLMN . |
---|---|---|---|---|---|---|---|---|---|
FL-HA-PARC | FLAG-HA | 0.231 | 0.033 | 0.102 | X | 0.039 | X | X | X |
PARC-FL-HA | FLAG-HA | 0.223 | 0.031 | 0.1 | 0.002 | 0.053 | X | X | X |
PARC-FL-HA, Myc-RBX1 | FLAG-Myc | 0.265 | 0.056 | 0.131 | X† | 0.05 | X | X | X |
PARC-FL-HA, Myc-RBX1 | FLAG-Myc depletion-HA | 0.262 | 0.038 | 0.064 | 0.012 | 0.078 | X | X | X |
PARC-FL-HA, 5XMyc-NEDD8 | FLAG-Myc | 0.096 | 0.016 | 0.046 | 0.061 | 0.022 | X | X | X |
PARC-FL-HA, 5XMyc-NEDD8 | FLAG-Myc depletion-HA | 0.157 | 0.018 | 0.044 | 0.018 | 0.019 | X | X | X |
PARC-FL-HA | FLAG-p53 | 0.352 | 0.033 | 0.3 | X | 0.028 | X | X | X |
PARC-FL-HA | FLAG-p53 depletion-HA | 0.215 | 0.027 | 0.006 | 0.017 | 0.028 | X | X | X |
PARC-FL-HA | FLAG-CUL7 | 0.181 | 0.066 | 0.09 | X | 0.061 | X | X | X |
PARC-FL-HA | FLAG-CUL7 depletion-HA | 0.311 | 0.019 | 0.238 | 0.129 | X | X | X | X |
FL-HA-CUL7 | FLAG-PARC | 0.003 | 0.116 | 0.064 | X | X | 0.03 | X | X |
FL-HA-CUL7 | FLAG-PARC depletion-HA | 0.003 | 0.083 | 0.018 | 0.017 | X | 0.05 | 0.038 | 0.002 |
FL-HA-CUL7 | FLAG-p53 | 0.011 | 0.094 | 0.133 | 0.099 | X | 0.01 | 0.004 | X |
FL-HA-CUL7 | FLAG-p53 depletion-HA | 0.002 | 0.106 | 0.001 | 0.032 | X | 0.05 | 0.025 | X |
NOTE: Lysates from the listed cell lines were purified by sequential immunoprecipitation as indicated using the following antibodies (anti-FLAG M2, anti-HA 12CA5, anti-Myc 9B11, anti-p53 DO1, anti-CUL7 h1557, and anti-PARC BL617).
Ten fully tryptic peptides shared by Cul7 and PARC (matching 594 spectra) were removed from the results. In each run, only peptides/spectra unique to CUL7 and PARC were used for NSAF calculations. In addition, 94 amino acid residues covered by the shared detected peptides were subtracted out of total protein lengths before calculating NSAFs.
In the Myc-RBX1 immunoprecipitation, Myc-RBX1 is visible in silver stains, and RBX1 peptides were detected when applying lower stringency criteria to the mass spectrometry data (data not shown).
NEDD8 and RBX1 promote the E3 activity of CUL1 and function as activation markers for cullins, suggesting that, in terms of cullin function, PARC subcomplexes containing NEDD8 and RBX1 are active. To examine these subcomplexes, as well as those containing CUL7 and p53, a sequential tandem affinity purification protocol was used in conjunction with cells expressing the COOH-terminally FLAG-HA–tagged PARC. Because the available antibodies for NEDD8 and RBX1 were inefficient for immunoprecipitation, 5XMyc-NEDD8 and Myc-RBX1 were stably transfected into BJAB cells expressing PARC-FLAG-HA. Additionally, a BJAB cell line stably expressing FLAG-HA-CUL7 was generated by transduction, and this line was used to examine CUL7 subcomplexes containing PARC and p53 (Fig. 3A).
To analyze PARC-containing subcomplexes, an anti-FLAG immunoprecipitation for FLAG-HA-PARC was followed by FLAG peptide elution. The eluates were re-immunoprecipitated for RBX1, NEDD8, p53, or CUL7, followed by a third immunoprecipitation of the remaining supernatant for PARC with an anti-HA antibody (Table 1). A representative experiment, using p53, is shown in Fig. 3B. In this sequential purification, the second immunoprecipitation yielded information about a specific complex, whereas the third immunoprecipitation, given efficient depletion of the subcomplex, yielded information about proteins excluded from this subcomplex. All immunoprecipitates were subsequently analyzed by MudPIT.
As before, no peptides for the CUL7-associated proteins SKP1, FBXW8, or GLMN were obtained from any purifications involving PARC-FLAG-HA (Table 1; Supplementary Fig. S1). In contrast, FLAG-HA-CUL7 was able to efficiently immunoprecipitate SKP1 and FBXW8, as well as GLMN, further supporting the observation that the heterodimeric PARC/CUL7 complex could be distinguished from the CUL7 complexes that contained FBXW8 and SKP1. Second, the sequential purification of PARC-p53 subcomplexes yielded many CUL7 peptides and the sequential purification of CUL7-p53 subcomplexes yielded many PARC peptides, indicating that heterodimeric complexes of PARC and CUL7 specifically bind to p53. The PARC-p53 subcomplex also fractionated with RBX1 and NEDD8 peptides, and the PARC-RBX1 and PARC-NEDD8 subcomplexes fractionated with p53, indicating that PARC bound to p53 may function as an active E3 ligase. Additionally, purifications of CUL7 subcomplexes did not yield NEDD8 peptides, likely due to the lack of CUL7 neddylation, the relative paucity of PARC in CUL7 complexes, and lack of tryptic peptides in NEDD8. It is also noteworthy that MudPIT analysis of neither CUL7 nor PARC complexes revealed an association with CSN subunits or CAND1, which would be expected to interact with cullins during cycles of neddylation and deneddylation (8). CAND1 binding was also not detected by transient transfection, coimmunoprecipitation, and Western blotting (data not shown).5
Kuwabara et al., submitted for publication.
While comparing NSAF data from the CUL7-p53 complex to the CUL7-p53–depleted complex, it was observed that the p53-depleted CUL7 complex contained large amounts of FBXW8 and SKP1 whereas the CUL7-p53 complex contained relatively little FBXW8 and SKP1 (Table 1; Fig. 4A). The paucity of FBXW8/SKP1 suggested that FBW8/SKP1 was excluded from the CUL7-p53 complex. To investigate the potential exclusion of p53 from the CUL7/FBXW8 complex, immunoprecipitations and Western blots were done for CUL7, p53, and FBXW8 (Fig. 4B). As previously published, CUL7 binds to PARC, p53, and FBXW8 (11, 12, 17, 18). However, p53 was only capable of binding PARC and CUL7, whereas FBXW8 only bound CUL7, indicating that the CUL7/FBXW8 complex excludes p53 and that a CUL7 complex lacking FBXW8 binds p53.
CUL7 and PARC subcomplexes exhibit E3 ubiquitin ligase activity. A, NSAF values from purification of the CUL7-p53 (black) and p53-depleted CUL7 (gray) complexes. B, HCT116 cell lysate was immunoprecipitated for PARC (BL617), CUL7 (h1557), FBXW8 (MDD), or p53 (DO-1). An anti-HA (12CA5) immunoprecipitate was used as a negative control. Immunoprecipitates were separated by SDS-PAGE and Western blotted for PARC (BL617), CUL7 (BL653), FBXW8 (MDD), and p53 (DO-1). C, lysates from HCT116 or HCT116 TP53−/− cells were immunoprecipitated with anti-PARC (BL617), anti-CUL7 (h1557), anti-p53 (DO-1), or nonspecific rabbit IgG prior to SDS-PAGE and Western blotting for PARC (BL617), CUL7 (BL653), and p53 (DO-1). D, PARC- and CUL7-containing complexes were tandem affinity–purified as indicated from BJAB-derived cell lines. Ubiquitination reactions were prepared (see Materials and Methods) using bead-bound complexes, and the reactions were separated by SDS-PAGE prior to Western blotting for GST-ubiquitin (DG122), PARC-FL-HA (12CA5), FL-HA-CUL7 (12CA5), PARC (BL617), CUL7 (BL653), and FBXW8 (MDD).
CUL7 and PARC subcomplexes exhibit E3 ubiquitin ligase activity. A, NSAF values from purification of the CUL7-p53 (black) and p53-depleted CUL7 (gray) complexes. B, HCT116 cell lysate was immunoprecipitated for PARC (BL617), CUL7 (h1557), FBXW8 (MDD), or p53 (DO-1). An anti-HA (12CA5) immunoprecipitate was used as a negative control. Immunoprecipitates were separated by SDS-PAGE and Western blotted for PARC (BL617), CUL7 (BL653), FBXW8 (MDD), and p53 (DO-1). C, lysates from HCT116 or HCT116 TP53−/− cells were immunoprecipitated with anti-PARC (BL617), anti-CUL7 (h1557), anti-p53 (DO-1), or nonspecific rabbit IgG prior to SDS-PAGE and Western blotting for PARC (BL617), CUL7 (BL653), and p53 (DO-1). D, PARC- and CUL7-containing complexes were tandem affinity–purified as indicated from BJAB-derived cell lines. Ubiquitination reactions were prepared (see Materials and Methods) using bead-bound complexes, and the reactions were separated by SDS-PAGE prior to Western blotting for GST-ubiquitin (DG122), PARC-FL-HA (12CA5), FL-HA-CUL7 (12CA5), PARC (BL617), CUL7 (BL653), and FBXW8 (MDD).
The exclusion of both PARC and p53 from CUL7/FBXW8 suggested the possibility that p53 could play a role in the heterodimerization of PARC and CUL7. To examine the p53 dependence of PARC and CUL7 dimerization, immunoprecipitation and Western blotting for PARC, CUL7, and p53 were done from a matched set of HCT116 cells containing wild-type TP53 or specifically deleted TP53. In both wild-type and TP53 null cells, PARC and CUL7 were able to reciprocally coimmunoprecipitate each other, showing that p53 binding is not required for PARC and CUL7 dimerization (Fig. 4C).
Functional aspects of the relationship between PARC and CUL7 complexes. Previously, we described the construction of a Parc-null mouse model and, surprisingly, these mice are phenotypically normal, with no developmental defects or late-emerging phenotypes (16). One potential explanation for the survival of the Parc-null mice was functional complementation by CUL7. Like Cul7-null mice, Fbxw8 null mice show severe intrauterine growth retardation and neonatal lethality; however, unlike Cul7 null mice, approximately one third of Fbxw8-null mice escape neonatal lethality and, despite persistent growth defects, have a normal lifespan, with no additional emerging phenotypes (11, 30). Although a double Parc−/−, Cul7−/− mouse cannot be created by mating currently existing mice, Fbxw8 is located on a different chromosome, allowing the examination of the functional overlap of PARC and CUL7 via the creation of Parc−/−, Fbxw8−/− mice. If PARC and FBXW8 have overlapping functions, it would be expected that the Parc,Fbxw8 double null mice would show exacerbation of the Fbxw8 null phenotype (i.e., increasing lethality), whereas if PARC and FBXW8 have no overlapping functions, the Parc,Fbxw8 double null mice would exhibit only the Fbxw8 null phenotype. As shown in Table 2, the ratio of double-null mice obtained from mating Parc+/−, Fbxw8+/− mice was similar to that expected with no exacerbation of the Fbxw8 null phenotype. In addition to displaying the partial lethality of the Fbxw8−/− mice, Parc−/−, Fbxw8−/− mice show growth deficiencies compared with their littermates (data not shown).
Genotypes of progeny from Parc+/−, Fbxw8+/− × Parc+/−, Fbxw8+/− cross at 3 wks
Genotype . | Number . | Percentage (%) . | Predicted percentage (%) . |
---|---|---|---|
Parc+/+, Fbxw8+/+ | 3 | 5.1 | 6.25 |
Parc+/+, Fbxw8−/− | 4 | 6.8 | 6.25 |
Parc−/−, Fbxw8+/+ | 4 | 6.8 | 6.25 |
Parc−/−, Fbxw8−/− | 2 | 3.4 | 6.25 |
Parc−/−, Fbxw8+/− | 7 | 11.9 | 12.5 |
Parc+/−, Fbxw8+/+ | 8 | 13.6 | 12.5 |
Parc+/+, Fbxw8+/− | 12 | 20.3 | 12.5 |
Parc+/−, Fbxw8−/− | 6 | 10.2 | 12.5 |
Parc+/−, Fbxw8+/− | 13 | 22.0 | 25 |
Total | 59 |
Genotype . | Number . | Percentage (%) . | Predicted percentage (%) . |
---|---|---|---|
Parc+/+, Fbxw8+/+ | 3 | 5.1 | 6.25 |
Parc+/+, Fbxw8−/− | 4 | 6.8 | 6.25 |
Parc−/−, Fbxw8+/+ | 4 | 6.8 | 6.25 |
Parc−/−, Fbxw8−/− | 2 | 3.4 | 6.25 |
Parc−/−, Fbxw8+/− | 7 | 11.9 | 12.5 |
Parc+/−, Fbxw8+/+ | 8 | 13.6 | 12.5 |
Parc+/+, Fbxw8+/− | 12 | 20.3 | 12.5 |
Parc+/−, Fbxw8−/− | 6 | 10.2 | 12.5 |
Parc+/−, Fbxw8+/− | 13 | 22.0 | 25 |
Total | 59 |
The ability of PARC- and CUL7-containing complexes to support in vitro E3 ubiquitin ligase activity was assayed. Of particular interest was the activity of the CUL7/PARC heterodimer and the p53-binding CUL7 complex, compared with the activity of the CUL7/FBXW8 complex that excludes p53. The indicated PARC-FLAG-HA and FLAG-HA-CUL7–expressing cell lines were immunoprecipitated for the FLAG epitope and eluted with FLAG peptide before a second immunoprecipitation using anti-CUL7, anti-FBXW8, anti-p53, anti-HA, or anti-Myc antibodies. The beads from the second immunoprecipitation were washed extensively and used in a substrate-independent ubiquitination reaction with GST-ubiquitin. Immunoprecipitations from the parental BJAB cell line were used as negative controls, as were reactions containing no E1 ubiquitin-activating enzyme. SDS-PAGE and subsequent Western blotting for GST-ubiquitin revealed that both FLAG-HA–purified PARC and FLAG-HA–purified CUL7 complexes exhibit E3 ligase activity, consistent with previous results (Fig. 5; ref. 17). Additionally, subcomplexes containing PARC/CUL7, PARC/p53, PARC-NEDD8, CUL7/FBXW8, and CUL7/p53 also contained E3 ligase activity, indicating that all PARC and CUL7 subcomplexes were active. Western blotting for PARC, CUL7, and FBXW8 revealed the purification of endogenous proteins with each complex.
Discussion
PARC and CUL7 were previously reported to form homodimers and heterodimers, the first demonstration of cullin dimerization (16). Although a CUL7-based SCF-like complex has been reported, little was known about the existence of a PARC-based SCF-like complex and how this complex related to CUL7-based complexes. Furthermore, it had been suggested that neither APC2 nor PARC were true cullins (4, 11). An investigation of the potential of PARC to form an SCF-like complex revealed that PARC is indeed a bona fide cullin, as evidenced by RBX1 binding and covalent modification by NEDD8. Although PARC possesses the biochemical characteristics of a cullin, it fails to bind the CUL7-associated proteins SKP1 and FBXW8, suggesting that the PARC/CUL7 complex is distinct from the SCF-like CUL7/FBXW8 complex.
Surprisingly, CUL7, widely accepted as a true cullin, is not subject to neddylation despite binding RBX1, SKP1, and FBXW8, suggesting the definition of a true cullin is more amorphous than previously thought. The only unifying feature that separates all cullins from APC2 is RBX protein binding. However, whereas most cullins use RBX1, CUL5 uses RBX2. APC11, the RING subunit for APC2, is related to both RBX1 and RBX2, suggesting the demarcation between cullin and cullin-like is arbitrary (4, 31).
The lack of CUL7 neddylation and the absence of CAND1 binding to either CUL7 or PARC imply that CUL7 and PARC are subject to different regulatory mechanisms than other cullins. One potential regulatory mechanism is suggested by the absence of FBXW8 from CUL7/PARC complexes. Perhaps the heterodimeric complex is functionally equivalent to CAND1 binding and only homodimeric complexes are correctly assembled and active. In this model, CUL7 would be sequestered by heterodimerization with PARC and released to allow enzymatic function. This process would not be controlled by neddylation, as CUL7 binds both neddylated and non-neddylated PARC. However, substrate-independent ubiquitination assays suggest that the heterodimeric complex is active, so although the heterodimeric complex may be inactive towards FBXW8 substrates, it may remain active towards others. The biochemical data showing that PARC/CUL7 is separable from CUL7/FBXW8 is further supported by the phenotype of the Parc−/−, Fbxw8−/− mouse, which does not show an enhanced phenotype compared with the Fbxw8−/− mouse. As suggested by the lower level of CUL7 expression in the Parc knockout, the double knockout shows the survival of the Parc knockout is not due to complementation by CUL7 function (16). Although survival of the Parc−/−, Fbxw8−/− mouse alone does not prove that PARC and FBXW8 function in different pathways, the biochemical data showing the physical separation of PARC and FBXW8 suggests that PARC and FBXW8 serve different functions and make other alternative explanations of double-null survival unlikely.
PARC was initially proposed to act as a cytoplasmic anchor for p53, preventing the activation of p53-dependent target genes, but subsequent studies have found no role for either PARC or CUL7 in cytoplasmic sequestration of p53 (16, 17, 19). Additionally, p53 levels are not altered in either Parc-null or Cul7-null mouse embryonic fibroblasts, showing that PARC and CUL7 are not required for ubiquitin-mediated degradation of p53 and presenting the option that p53 serves to modify PARC and/or CUL7 function (17).6
J. R. Skaar, T. Arai, and J. A. DeCaprio, unpublished observations.
The mutually exclusive relationship of PARC and p53 with CUL7/FBXW8 raises questions regarding the role of each complex. CUL7/FBXW8 is presumed to function as an active E3 ubiquitin ligase based on its resemblance to SCF, with the exception of neddylation, although no definitive substrates have been identified. One study has reported monoubiquitination of p53 by CUL7 (17). However, the role of FBXW8 in this process was not addressed, and cullin domain deletion mutants of CUL7 retained activity (17). In conjunction with the exclusion of FBXW8 from p53-containing CUL7 complexes and the E3 ubiquitin ligase activity of CUL7/p53 and PARC/p53 complexes, this data suggests that the observed monoubiquitination is PARC/CUL7-mediated. Indeed, the PARC/CUL7 complex also seems to be active in terms of RBX1 binding and neddylation.
The functional effect of the reported p53 ubiquitination remains unclear. p53 could be a true substrate for PARC/CUL7, but alternatively, p53 could function as a novel substrate adapter or regulatory subunit for PARC/CUL7 complexes, with the reported ubiquitination of p53 analogous to F-box turnover through CUL1 (32). However, it is clear that p53 is not required for the dimerization of PARC and CUL7 (Fig. 4C).
Strictly in terms of the CUL7/FBXW8 complex, the mutually exclusive binding of FBXW8 and p53 indicates that p53 may inhibit or retarget the activity of the CUL7 complex by F-box displacement. In one possible model, increasing amounts of cytoplasmic p53 (i.e., in response to genotoxic stress) result in the inactivation of the growth-promoting activity of CUL7/FBXW8 and activation of growth-suppressive activities of PARC/CUL7, aiding p53 in inducing a cell cycle arrest. This model is consistent with the growth retardation of Cul7-null mice. Alternatively, induced p53 could affect complex assembly, driving the incorporation of more CUL7 into complexes with PARC than FBXW8. However, this option is unlikely due to the ability of PARC and CUL7 to dimerize in the absence of p53.
Taken together, our data indicate the presence of several different PARC- and CUL7-containing cullin RING ligase complexes and that each of these subcomplexes could support E3 ligase activity. The regulation of these complexes seems complex, with accessory proteins, including p53, that are not traditionally found in cullin-based complexes, and further investigation into the roles of p53 and dimerization in PARC and CUL7 function are required using in vivo analysis of PARC and CUL7 substrates.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
Current address for J.R. Skaar: NYU School of Medicine, 550 First Avenue-MSB599, New York, NY 10016.
Current address for T. Tsutsumi: Department of Infectious Diseases, Internal Medicine, Lab 113, The University of Tokyo Hospital, 81-3-5800-8803 Tokyo, Japan.
Acknowledgments
Grant support: NSF Graduate Research Fellowship (J.R. Skaar) and in part by a National Cancer Institute grant (RO1CA93804).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank William Kaelin and Susanne Schlisio for pcDNA3.0-5XMyc-c-Jun; Bert Vogelstein for the matched set of HCT116 cells; Dana Gabuzda and Andrew Mehle for help with reverse transcriptase assays; Eric McIntush at Bethyl Laboratories for help with antibody design and production; Lulu Ang, Jianping Jin, and J. Wade Harper for F-box constructs; and Michele Pagano for F-box constructs.