Oncogene (2006) 25, 4534–4548 & 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00 www.nature.com/onc ORIGINAL ARTICLE Cytoplasmic localized ligase 7 binds to and promotes cell growth by antagonizing p53 function

P Andrews1,2,YJHe1 and Y Xiong1,2,3

1Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, NC, USA; 2Department of Biochemistry and Biophysics, University of North Carolina, Chapel Hill, NC, USA and 3Program in Molecular Biology and Biotechnology, University of North Carolina, Chapel Hill, NC, USA

Cullins are a family of evolutionarily conserved monomeric manner (monoubiquitination) to cause that bind to the small RING fingerprotein,ROC1, to conformational change to the substrate or in a constitute potentially a large number of distinct E3 polyubiquitin chain which signals the substrate to be ubiquitin ligases. CUL7 mediates an essential function degraded by the 26S proteasome. Both ubiquitin formouse embryodevelopment and has been linked with activating and conjugating enzymes are well character- cell transformation by its physical association with the ized and contain highly conserved functional domains. SV40 large T antigen. We report here that, like its closely The identity and mechanism of E3 ubiquitin ligases, on related homolog PARC, CUL7 is localized predominantly the other hand, has been elusive and long postulated as in the cytoplasm and binds directly to p53. In contrast to an activity responsible for both recognizing substrates PARC, however, CUL7, even when overexpressed, did not and for catalysing polyubiquitin chain formation sequesterp53 in the cytoplasm. We have identified a (Hershko and Ciechanover, 1998). sequence in the N-terminal region of CUL7 that is highly Currently, two major families of E3 ligases have been conserved in PARC and a sequence spanning the described. The HECT family of E3s contains a domain tetramerization domain in p53 that are required for homologous to E6-associated (E6AP) carboxyl CUL7–p53 binding. CUL7 and MDM2 did not form a terminus that can form thioester linkages with ubiquitin detectable tertiary complex with p53. In vitro, CUL7 (Huibregtse et al., 1995; Scheffner et al., 1995). Multiple caused only mono- ordi-ubiquitination of p53 underthe cellular proteins with diverse structures contain the conditions MDM2 polyubiquitinated p53. Co-expression HECT domain (five in budding yeast and B30 in of CUL7 reduced the transactivating activity of p53. human), implicating multiple substrates and physiolo- Constitutive ectopic expression of CUL7 increased the gical functions for the HECT family of E3 ligases rate of cell proliferation and delayed UV-induced G2 (Pickart, 2001). The RING family of E3s contains either accumulation in U2OS cells expressing functional p53, an intrinsic RING finger domain or an associated but had no detectable effect in p53-deficient H1299 cells. RING subunit essential for their ubiquitin ligase activity Deletion of the N-terminal domain of CUL7 or a mutation (Joazeiro and Weissman, 2000). The RING finger motif disrupting p53 binding abolished the ability of CUL7 to was initially defined a decade ago as a novel cysteine- increase the rate of U2OS cell proliferation. Our results rich sequence present in several otherwise unrelated suggest that CUL7 functions to promote cell growth proteins (Freemont et al., 1991). A large number of through, in part, antagonizing the function of p53. RING finger proteins are present in different eucaryotic Oncogene (2006) 25, 4534–4548. doi:10.1038/sj.onc.1209490; organisms and are involved in various cellular processes published online 20 March 2006 (>350 in ). Ubiquitin ligase activity has been experimentally demonstrated for more than a Keywords: Cullin 7; p53; ROC1; cytoplasmic localization dozen of them, and many more are believed to function as E3 ubiquitin ligases. The cullin proteins assemble a unique and large family of ligases by binding to the small RING finger protein Introduction ROC1 (also known as RBX1, HRT1 and Sag1) (Deshaies, 1999). There are three closely related cullin Protein ubiquitination typically involves a cascade of in both budding and fission yeast and six in three distinct enzymatic activities to activate (E1), worms, fruit flies and humans (Kipreos et al., 1996). In conjugate (E2) and ligate (E3) ubiquitin. A ubiquitin addition, three proteins, PARC (Nikolaev et al., 2003) molecule can be ligated to a substrate either in a and CUL7 present in mammals (Dias et al., 2002), and APC2 expressed in all eucaryotes (Yu et al., 1998; Correspondence: Dr Y Xiong, Department of Biochemistry and Zachariae et al., 1998), contain significant sequence Biophysics, University of North Carolina, Chapel Hill, NC, USA. B E-mail: [email protected] homology to over an 180 amino-acid region Received 21 June 2005; revised 11 January 2006; accepted 11 January involved in binding with ROC1 or a homologous small 2006; published online 20 March 2006 RING protein APC11, respectively. Genetic analyses in CUL7 binds to p53 P Andrews et al 4535 different organisms have linked various physiological that are most conserved among all cullins, and are functions with different cullin genes, ranging from cell also highly conserved in CUL7, PARC and APC2 cycle control, signal transduction, transcriptional reg- (Furukawa et al., 2000). We determined whether CUL7, ulation and maintenance of genomic integrity to like other cullins, physically associates with ROC1 by a organismal development. A possible explanation for coupled transfection–co-immunoprecipitation assay. such broad functions of cullins is that individual cullin Cul7 is expressed at a very low to undetectable levels proteins may assemble various distinct ligases to in most commonly used cell lines, making it more mediate the ubiquitination of multiple substrates, and feasible to examine ectopically expressed CUL7. Epi- thus carry out various functions in vivo. tope-tagged CUL7 and ROC1 were ectopically ex- Of three distantly related cullins, PARC and CUL7 pressed singularly or in combination in cultured 293T share extensive to each other (Dias cells. A specific CUL7–ROC1 association was readily et al., 2002; Nikolaev et al., 2003). In addition to the detected (Figure 1a). Our previous mutational analysis cullin homology region which includes both the ROC1- identified six amino acid residues in CUL1(610ILL- binding site as well as sequences involved in NEDD8 QYN615) whose deletion substantially reduced CUL1- conjugation, both proteins also contain a small region ROC1 binding (Furukawa et al., 2000) and are (110 residues) with limited sequence similarity (31% conserved in CUL7 (1472LLLYLN1477). Deletion of these identity) to HERC2 (hect domain and rcc1 domain six residues (CUL7DROC1) nearly abolished CUL7–ROC1 protein 2) within their N-terminal region, as well as a binding (Figure 1a). The lack of cell lines expressing DOC (DOC1/APC10) domain. Additionally, PARC detectable CUL7 prevent us from determining the contains in its C-terminal region two RING fingers formation and regulation of the endogenous CUL7– separated by a sequence in between the RING fingers ROC1 complex at present. We transfected myc-tagged (IBR), a unique sequence referred to as RING1-IBR- CUL7 into U2OS and 293T cells and isolated multiple RING2 (RBR) that was also found in multiple proteins independent CUL7 stable clones. CUL7 immunocom- from many eukaryotic species (Marin et al., 2004). The plexes derived from these stable cells exhibited high function of both PARC and CUL7 are poorly under- levels of auto-ubiquitin ligase activity which was stood. Mice lacking Cul7 die around birth with abolished by the omission of either E1 or E2 respiratory stress and Cul7-null mouse embryos devel- (Figure 1b). Together, these results suggest that CUL7, oped dermal and hypodermal hemorrhage (Arai et al., via its association with ROC1, may function as a 2003), implicating a critical function of CUL7 function ubiquitin ligase in vivo. Our results are consistent in development. with the report that ectopically expressed CUL7 binds The N-terminal region in cullins is mostly, if not to ROC1 and forms an active ubiquitin ligase (Dias entirely, responsible for substrate recruitment. Via its N- et al., 2002). terminal domain, CUL1 binds with SKP1 and then an F-box protein to constitute SCF-ROC1 ligases (Bai et al., 1996; Feldman et al., 1997; Skowyra et al., 1997), CUL7 is cytoplasmically localized and CUL3 binds to BTB proteins to constitute distinct In human cells, ROC1 and many cullins, including BTB-CUL3-ROC1 ligases (Furukawa et al., 2003; Geyer CUL1, CUL2, CUL4A and CUL5, are distributed in et al., 2003; Pintard et al., 2003; Xu et al., 2003). A both the cytoplasmic and nuclear compartments. potential target of PARC function has been identified to Nuclear import is linked with the neddylation and be the tumor suppressor p53 (Nikolaev et al., 2003), a maturation of CUL1-dependent SCF ligases (Furukawa sequence-specific transcription factor that acts in the et al., 2000). We determined the subcellular localization nucleus to mediate various cellular stresses and is exported of CUL7 by indirect immunofluorescence and found to the cytoplasm for ubiquitin-mediated degradation that even when ectopically overexpressed, CUL7, like its during normal cell growth (Ko and Prives, 1996; Levine, closest related homologue PARC (Nikolaev et al., 2003), 1997; Zhang and Xiong, 2001a). The PARC protein is localized predominantly in the cytoplasm (Figure 1c, localized predominantly in the cytoplasm and was and data not shown for other cell types). Treatment of reported to sequester p53 in the cytoplasm, leading to cells with leptomycin B (LMB), an inhibitor of the the notion that one function of PARC may be to Crm1-specific nuclear export pathway, accumulated p53 negatively regulate p53 through cytoplasmic sequestration which is actively exported out of the nucleus and is (Nikolaev et al., 2003). Given the high degree of degraded in the cytoplasm, but had no detectable effect homology between PARC and CUL7, we carried out on the cytoplasmic localization of CUL7 (Figure 1d), biochemical and functional analyses of CUL7, including indicating that CUL7, unlike its binding partner p53 investigating the possibility that CUL7, like PARC, may (see below), is not actively shuttling between the nucleus interact with and regulate the function of p53. and the cytoplasm. Previous studies have demonstrated that disruption of Results ROC1 binding accumulated CUL1 in the cytoplasm, and conversely, overexpression of ROC1 promoted CUL7 binds to ROC1 and possesses ubiquitin ligase CUL1 nuclear localization, suggesting a role of ROC1 activity in facilitating CUL1 nuclear accumulation or retaining We previously mapped the ROC1-binding domain CUL1 in the nucleus (Furukawa et al., 2000). To to approximately 100 amino-acid residues in CUL1 determine whether cytoplasmic localization of CUL7

Oncogene CUL7 binds to p53 P Andrews et al 4536

Figure 1 CUL7 is a cytoplasmically localized ubiquitin ligase. (a) U2OS cells were transfected with myc3-Cul7, ROC1 or myc3- Cul7DROC1 as indicated. Cell lysates were immunoprecipitated with either anti-myc or ROC1 antibodies, resolved by SDS–PAGE, and Western blotted with the indicated antibodies. (b) myc3-CUL7 immunocomplexes contain auto-ubiquitination activity. 293T or 293T cells constitutively expressing myc3-CUL7 were lysed and immunoprecipitated with the anti-myc antibody. Myc immunocomplexes were then incubated in the presence or absence of various components as indicated. Reactions were resolved by SDS–PAGE, and visualized by autoradiography. (c) CUL7 is a cytoplasmically localized protein. U2OS cells were transfected with myc3-Cul7. Subcellular localization of myc3-CUL7 was examined using indirect immunofluorescence. (d) CUL7 subcellular localization is not affected by leptomycin B (LMB) treatment. U2OS cells were transfected with CUL7, treated with LMB, fixed, and protein localization was determined using indirect immunoflourescence. (e) CUL7 is capable of forming a cytoplasmic ubiquitin ligase. U2OS cells were transfected with myc-Cul7 and ROC1. Myc-CUL7 and ROC1 subcellular localization was examined using the antibodies indicated.

was caused by an insufficient amount of ROC1 as a significant nuclear accumulation of CUL7 was observed result of CUL7 overexpression, we determined if CUL7 in these cells (Figure 1e), suggesting that CUL7 may subcellular localization was affected by ectopically assemble a productive ligase complex with ROC1 and overexpressed ROC1. CUL7 and ROC1 co-localization function in the cytoplasm. We also noted that ROC1 in the cytoplasm was evident in nearly all cells that preferentially accumulated in the nucleus when it was were positively transfected by both plasmids and no singularly expressed and was localized mostly in the

Oncogene CUL7 binds to p53 P Andrews et al 4537 cytoplasm when co-expressed with CUL7 (Figure 1e), localization was determined by indirect immunofluores- raising the possibility that at high levels, CUL7 may cence. As expected, CUL7 was localized predominantly interfere with the function of other cullins by sequester- in the cytoplasm and the T antigen was detected only in ing ROC1 in the cytoplasm. the nucleus (Figure 2b) when expressed separately. When the two proteins were co-expressed, both retained their respective localization and no significant change of CUL7 associates with p53 and SV40T antigen subcellular localization was seen for either protein. To gain an initial view of the function of CUL7, we These results are consistent with the relative low examined the CUL7 complex by coupled [35S]metabolic abundance of CUL7–T antigen association observed labeling and immunoprecipitation (35S-IP). To increase by co-immunoprecipitation (Figure 2a), suggesting that the specificity of detection, we constructed a double the CUL7–T antigen interaction may be transient or epitope-tagged CUL7 fused with two tandem copies of involve only a small fraction of each protein. A similar HA and three copies of FLAG at the N- and C-terminus pattern of interaction was also seen between CUL7 and of CUL7 (HA2-CUL7-FLAG3), respectively, and ecto- p53 (see below). pically expressed it in various cell lines. A representative 35S-IP analysis of CUL7 complexes is shown in Figure 2. CUL7 binds to p53 In addition to CUL7, several additional proteins were By transfection and coimmunoprecipitation assays, we identified that appear to specifically associate with demonstrated that CUL7 binds with p53 (Figure 3a). CUL7, as they were absent when the FLAG antibody Mutation targeting the nuclear export signal (NES) in was blocked by a molar excess of antigen peptide, p53 (p53DNES) blocked p53 nuclear export and decreased suggesting that CUL7 may interact with multiple CUL7–p53 association, whereas mutation targeting the cellular proteins in vivo. nuclear localization signal (NLS) in p53 (p53DNLS) In 293T cells, at least five CUL7-interacting proteins blocked p53 nuclear import and increased CUL7–p53 were seen (pointed out by an arrow, compare lanes 5 association (Figure 3a). These findings, taken together and 6), including notably one at 53 kDa and one at with the observation that CUL7 localizes predominantly 97 kDa that appear to correspond to p53 and SV40 large in the cytoplasm and does not shuttle between the T antigen, respectively. A side-by-side comparison with nucleus and the cytoplasm (Figure 1), are consistent both p53 and T antigen immunocomplexes supported with an interpretation that p53–CUL7 interaction the interpretation; both 53 and 97 kDa bands exhibited occurs in the cytoplasm. We noticed, in this (lane 6) the same mobility as p53 and T antigen, and the 97 kDa and other experiments, that overexpression of p53 band was not detected in the other three cell lines consistently reduced the level of CUL7 protein. The lacking T antigen. Reciprocally, a protein apparently molecular basis for this reduction is not clear. corresponding to CUL7 was detected in p53 immuno- complexes derived from three cell lines expressing p53 (293, 293T and U2OS) but not in p53-deficient SaOS2 An N-terminal domain of CUL7 mediates p53 binding cells, and in T antigen immunocomplexes derived from We generated a series of deletion mutants of CUL7 and 293T cells, but not the other three cell lines lacking T mapped the site involved in p53 binding (Figure 3b). antigen. We carried out mass spectrometric analyses of Deletion of the N-terminal 422 residues completely CUL7 complexes derived from scaled up immunopreci- abolished CUL7’s binding activity to p53 (Figure 3c). pitations, and thus far have identified positively two The N-terminal 221 residues exhibited a low level of proteins, the 97 kDa T antigen and 90 kDa heat-shock p53-binding activity and the N-terminal 965 residues protein (HSP90, confirmatory data not shown). Mass exhibited a p53-binding activity that, when taking into spectrometric analysis of p53 was not conclusive, but a consideration its lower level of expression, is close to the p53–CUL7 interaction was confirmed by direct IP- wild-type CUL7. We were unable to detect the expres- Western and mutagenesis analyses (see below). The level sion of either CUL7N422 or CUL7N680 and suspected that of 35S-labelled (i.e. newly synthesized) CUL7 in complex both proteins might be unstable. Treatment of trans- with p53 was as high as in CUL7 complexes and fected cells with a proteasome inhibitor, MG132, substantially higher than that detected in the T antigen prevented their rapid degradation and allowed us to complex. The significance of this difference, whether examine their p53-binding activity. As shown in the resulting from a higher affinity of CUL7 to p53 than to right panel of Figure 3c, both CUL7N422 and CUL7N680 the T antigen, and how these three proteins interact with bind to p53 at a level close to that of wild-type CUL7. each other has yet to be determined. The CUL7–T To define the sequence in CUL7 required for p53 antigen interaction has been previously reported and our binding, we compared the N-terminal 422 residues of result confirms the finding (Tsai et al., 2000; Ali et al., CUL7 with PARC and identified several highly con- 2004). served sequences for site-directed mutagenesis analysis. Given their separate subcellular localization, one in Deletion of an internal 16 residue, 389–404 in human the nucleus and one in the cytoplasm, we asked whether CUL7 that is invariably conserved in mouse CUL7, in CUL7 and the T antigen, when co-expressed in the same human PARC and contains only one amino-acid change cell, could relocalize each other. Myc-tagged CUL7 and in mouse PARC, completely abolished the p53-binding an untagged T antigen were either singularly expressed activity of CUL7 without any detectable effect on or co-expressed in U2OS cells and their subcellular CUL7’s binding with ROC1 (Figure 3d). Taken

Oncogene CUL7 binds to p53 P Andrews et al 4538

Figure 2 CUL7 complex. (a) Four different cell lines were transiently transfected with a plasmid expressing tagged CUL7 and pulse labeled with [35S]methionine for 4 h. 35S-labeled lysates were immunoprecipitated with the indicated antibodies and immunoprecipitates were resolved by SDS–PAGE followed by autoradiography. (b) U2OS cells were transfected with plasmids expressing the indicated protein and subcellular localization of myc3-CUL7 and the large T antigen was examined using indirect immunoflourescence.

together, these results suggest that the N-terminal 422 of CUL7 is highly conserved with PARC (59% residues of CUL7 are both required and sufficient for identical). Searches of current databases or pairwise binding with p53. The N-terminal 422 amino-acid region comparison with known p53-binding proteins using this

Oncogene CUL7 binds to p53 P Andrews et al 4539

Figure 3 CUL7 binds p53. (a) U2OS cells were transfected with myc-CUL7, wild-type p53, and mutant p53 defective in nuclear localization (p53DNLS) or nuclear export (p53DNES). Cells were lysed and equal amount of protein lysates was immunoprecipitated with the anti-myc antibody. Cells were resolved by SDS–PAGE and visualized by Western blotting. (b) Mapping of the p53-binding region of CUL7. A schematic representation of the wild-type and deletion mutants of CUL7. (c) Myc3-tagged wild-type and mutant CUL7 were co-expressed with p53 in U2OS cells and binding was observed by IP–Western analysis using anti-myc (9E10) and anti-p53 (DO1) antibodies. CUL7N422 and CUL7N680 mutants are unstable when ectopically expressed and their binding with p53 was assayed in cells treated with MG132. (d) A 16-residue sequence (residues 389–404 in human CUL7, underlined) invariably conserved between human and mouse CUL7 is compared with the corresponding PARC proteins. Retrovirally infected pools of U2OS cells stably expressing either CUL7 or CUL7D389À404 were lysed and p53–CUL7 binding was determined by IP–Western analysis. sequence did not identify any other proteins exhibiting CUL7 binds to the tetramerization domain in p53 significant similarities, indicating that PARC and CUL7 The p53 protein contains several well-characterized have evolved a unique p53-binding sequence. domains: a N-terminal domain (residues 1–100) contain-

Oncogene CUL7 binds to p53 P Andrews et al 4540 ing the transcriptional activation activity, an MDM2- CUL7–p53 binding (see below), these results indicate binding sequence, several phosphorylation sites of that the N-terminal domain and an intact central DNA-damage-activated kinases and a nuclear export domain is not required for p53 to bind with CUL7, signal, a central domain (residues 100–300) responsible and that CUL7 likely binds to a sequence between for sequence-specific DNA binding, and a C-terminal residues 200 and 360 in p53 spanning the tetrameriza- domain encompassing a tetramerization domain (resi- tion domain. dues 320–360), an NES (residues 339–352) and an extreme C-terminus (residues 363–393) containing most, CUL7, p53 and MDM2 do not form a ternary complex if not all lysine residues for ubiquitination by MDM2 Mapping of CUL7-binding site to a sequence between ligase (Ko and Prives, 1996; Levine, 1997; Zhang and residues 200 and 360 suggests that CUL7-binding site is Xiong, 2001a) (Figure 4a). We generated a series of p53 different from that of MDM2. Various specific muta- deletion mutants, expressed in mammalian cells as GST tions in the N-terminal region of p53 have been fusion proteins, and tested their binding with wild-type previously characterized that disrupt MDM2–p53 bind- CUL7 by a standard GST pull-down assay. An ing, including Leu14Gln/Phe19Ser and Leu22Gln/ appreciable amount of CUL7 was precipitated from Trp23Ser (Lin et al., 1994), as well as DN25 and total cell extract with the GST-p53 fusion protein DN15. None of these mutants affected p53’s binding (Figure 4b). Deletion of up to 200 residues (p53DN100 with CUL7 (Figure 5a), further supporting the notion and p53DN200) from the N-terminus of p53 reduced that MDM2 and CUL7 bind to two separate sequences somewhat, but did not abolish CUL7 biding, and a in p53. These data also suggest that binding of CUL7 further deletion of another 100 residues (p53DN300) with p53 does not require MDM2 and is not dependent completely abolished p53’s binding activity with on MDM2-mediated p53 ubiquitination. CUL7. Together with the finding that deletion of This finding led us to determine whether these three the extreme C-terminal 30 residues did not disrupt proteins could form a CUL7–p53–MDM2 ternary complex bridged by p53. We examined the presence of MDM2 and p53 in CUL7 immunocomplexes in cells triply transfected with plasmids expressing all three proteins. Under the condition where a binary p53– CUL7 complex was readily demonstrated, no CUL7– MDM2 interaction was detected (Figure 5b). The interpretation of whether CUL7 and MDM2 could simultaneously bind to p53 is, however, complicated by the fact that MDM2 promotes rapid p53 degradation. As a result, an MDM2–p53–CUL7 ternary complex could have formed, but existed in vivo only transiently and escaped detection by the co-immunoprecipitation assay. To rule out this possibility, we repeated the same experiment with a mutant MDM2 containing a point mutation in the RING finger (C464A) that completely abolished the ligase activity of MDM2. In reciprocal IP– Western analyses, p53 was readily detected in both CUL7 and MDM2 immunocomplexes, but neither CUL7 nor MDM2 was detected in the other immuno- complex (Figure 5c). From these results, we conclude that CUL7–p53–MDM2 ternary complex, if assembled in vivo, present only in a very low level. There could be two possible explanations for this. Either p53 does not have sufficient surface areas to interact with two large proteins, CUL7 (185 kDa) and MDM2 (75 kDa) at the same time, or there was only a very small amount of p53 and Mdm2 proteins, both predominantly localized in the nucleus, present in the cytoplasm at the same time, making the detection of MDM2–p53–CUL7 ternary complex very difficult.

Figure 4 Mapping the p53-binding domain in CUL7. (a)A CUL7 is unable to promote p53 polyubiquitination in schematic representation of the wild-type and deletion mutants of vitro GST-p53. (b) Mapping of the CUL7-binding region of p53. U2OS To elucidate the functional consequences of the p53– cells were transfected with myc3-Cul7 and GST-p53 or GST-p53 mutants as indicated. Cells were lysed, proteins pulled down with CUL7 interaction, we first determined whether CUL7 glutathione agarose beads, resolved by SDS–PAGE, and visualized promotes p53 polyubiquitination by both in vitro and in by Western blotting using the indicated antibodies. vivo ubiquitination assays. To avoid the possibility of

Oncogene CUL7 binds to p53 P Andrews et al 4541

Figure 5 CUL7 does not form a ternary complex with p53 and MDM2. (a) Myc3-CUL7 or myc3-HDM2 (human MDM2) was co- expressed with wild-type p53 or p53-containing point mutations or deletions as indicated. Binding was determined by immunoprecipitation and Western blotting using anti-myc (9E10) and anti-p53 (FL393) antibodies. (b) U2OS cells were transfected with myc3-CUL7, p53 and HDM2. Cells were lysed, proteins immunoprecipitated with anti-myc antibody, resolved by SDS–PAGE, and visualized by Western blot. (c) U2OS cells were transfected with myc3-CUL7, HA-HDM2, HA-HDM2C464A and p53. Protein– protein interactions were examined by reciprocal IP–Western analyses. transcriptional upregulation of MDM2 by p53, and to Under this assay condition, MDM2 promoted an avoid a contaminating contribution of endogenous efficient poly- or multiple ubiquitination of p53 MDM2 ligase activity, we co-transfected CUL7 with a (Figure 6a). Under the same assay conditions, we transcriptionally inactive p53 mutant (p53R273H) into readily detected ubiquitination of p53 by CUL7. p53-deficient H1299 cells. CUL7–p53 complexes, and Omitting E1 and E2 abolished p53 ubiquitination, control MDM2–p53 complexes, were immunoprecipi- excluding the possibility that observed p53 ubiquitina- tated and assayed in vitro for p53 ubiquitination after tion resulted from contamination of the CUL7 im- incubating with E1 and E2-Ubc5. Expression of CUL7 munoprecipitates with ubiquitinated p53 generated and MDM2 was confirmed by direct immunoblotting. in vivo. In repeated assays, we reproducibly detected

Oncogene CUL7 binds to p53 P Andrews et al 4542 predominantly p53 conjugated with one or two ubiqui- The observation that CUL7 can only promote mono- tin moieties. The significance of this difference between or di-ubiquitination of p53 in vitro led us to examine the MDM2 and CUL7-promoted p53 ubiquitination – possibility that in vivo CUL7 may function in p53 whether it reflects a factor required for CUL7-mediated ubiquitination and degradation through a collaboration p53 polyubiquitination that is missing in this setting, or with MDM2 in a sequential manner: p53 is first CUL7 and MDM2 are mechanistically different ligases – ubiquitinated by MDM2 in the nucleus and then the has not been determined. Surprisingly, deletion of the ubiquitin chain is elongated by CUL7, either alone, or in cullin homology and ROC1-binding sequence, from combination with MDM2, in the cytoplasm. The p53 residues 1472 to 1510, did not affect CUL7-mediated protein was ectopically expressed in U2OS cells with p53 ubiquitination (Figure 6a, lane 6), suggesting that various combinations and amounts of MDM2 and ROC1 binding is not essential for CUL7-mediated in CUL7 as well as HA-ubiquitin. At 24 h post transfec- vitro ubiquitination of p53. The nature of the CUL7- tion, cells were treated with the proteasome inhibitor associated p53 ligase is not clear at present. One possible MG132 for 4 h to accumulate and to prevent degrada- candidate ligase is APC11, which has been reported to tion of ubiquitinated proteins. Cells were directly lysed bind tightly and stoichiometrically to APC10 (Tang with 1% SDS buffer and immediately boiled for 10 min. et al., 2001). CUL7, as well as PARC, contain a DOC Diluted lysates were then immunoprecipitated with a domain highly related to APC10 (31% identical over p53 antibody, resolved by SDS–PAGE and immuno- 116 residues) and could, at least in theory, interact with blotted with anti-HA to visualize ubiquitinated p53. APC11 and recruit ubiquitin ligase activity. Under the assay conditions where MDM2 promoted

Figure 6 CUL7 promotes p53 mono- or di-ubiquitination but not p53 degradation. (a) The transcriptionally impaired p53R273H was co-transfected into p53-deficient H1299 cells with CUL7 or MDM2. CUL7–p53 and MDM2–p53 complexes were precipitated using either anti-myc (9E10) or anti-MDM2 (SMP14) antibodies. In vitro p53 ubiquitination assays were performed. Reactions were terminated by boiling in SDS sample buffer, resolved by SDS–PAGE and blotted with an anti-p53 antibody (DO1). (b) U2OS cells were co-transfected with plasmids expressing 6x (HA-Ub) and additional indicated proteins. Cells were pelleted, lysed with SDS buffer, and boiled for 10 min. Lysates were diluted 10-fold with 0.1% NP40 buffer and immunoprecipitated with a p53 antibody. Immunoprecipitates were resolved by SDS–PAGE and ubiquitinated p53 was visualized by Western blotting with an HA antibody (12CA5). (c) U2OS and U2OS cells constitutively expressing CUL7 were treated with cycloheximide to prevent protein synthesis. Total cell extracts were prepared at the indicated time points, and resolved by SDS–PAGE. The level of p53 and CUL7 was determined by direct Western blotting.

Oncogene CUL7 binds to p53 P Andrews et al 4543 efficient p53 ubiquitination, CUL7 alone was not able to promote ubiquitination of p53 in vivo (Figure 6b). When co-expressed with MDM2, CUL7 did not exhibit any appreciable enhancing effect in promoting MDM2- dependent p53 polyubiquitination.

CUL7 does not affect the half-life of the p53 protein We next determined the half-life of the p53 protein in response to CUL7 overexpression. Parental U2OS cells and U2OS cells stably expressing CUL7 were treated with cycloheximide to block new protein synthesis. Protein extracts were prepared at different time points after cycloheximide treatment and the steady-state level of p53 was determined by direct immunoblotting. The half-life of the p53 protein in logarithmically growing (i.e. unstressed) U2OS cells is approximately 30 min, and was not significantly affected by the ectopic over- expression of CUL7 (Figure 6c). The inability of CUL7 to accelerate p53 degradation is consistent with its inability to promote p53 polyubiquitination.

CUL7 does not sequester p53 in the cytoplasm PARC, like CUL7, localized predominantly in the cytoplasm and was reported to inhibit p53 function via a cytoplasmic sequestration (Nikolaev et al., 2003). Binding of p53 led us to test whether CUL7 similarly sequesters p53 in the cytoplasm. We examined p53 localization by indirect immunofluorescence in both U2OS cells and U2OS cells stably expressing CUL7. No obvious difference of p53 subcellular localization was detected (Figure 7). To increase the sensitivity of detection to confirm this finding, we ectopically ex- pressed additional p53 in both parental U2OS and CUL7-overexpressing cells and determined p53 localiza- tion. Again, no obvious difference was seen in p53 nuclear and cytoplasmic distribution by the overexpres- sion of CUL7 (Figure 7).

CUL7 represses the transcriptional activation activity of p53 and increased the rate of cell proliferation Using a luciferase reporter assay, we examined the CUL7 effects on basal p53 transcriptional activation utilizing a promoter reporter construct. We found that CUL7, when co-transfected with p53, decreased the transcriptional activation activity of p53 (Figure 8a). These results are consistent with an inhibitory activity of Figure 7 CUL7 does not sequester p53 in the cytoplasm. U2OS CUL7 toward p53. cells were transfected with either CUL7, p53 or both and protein localization was examined using indirect immunofluorescence. To assess the function of CUL7 in controlling cell growth, we determined the effect of ectopic expression of CUL7 on cell proliferation in p53-positive U2OS cells generated three different retroviruses that express the and the requirement of p53-binding activity on this wild-type CUL7 and two CUL7 mutants, one with a effect. We first established cell lines stably expressing deletion of the N-terminal 680 amino-acid residues CUL7 after transfection of U2OS cells and subsequent (referred to as CUL7DN680) containing all the p53-binding selection of G418-resistant clones and compared the rate activity, and one deleting residues 389–404, which are of cell proliferation with parental U2OS cells. Consti- required for p53 binding (CUL7D389À404). U2OS cells tutive expression of high levels of CUL7 from the CMV were transduced individually by each of the three promoter resulted in an increased rate of proliferation of viruses, stable pools were generated and the growth U2OS cells, indicating a role of CUL7 in promoting cell curve of each pool was determined (Figure 8c). growth (Figure 8b). To confirm this result and assess the The expression of each protein was confirmed by requirement of p53 binding for this activity of CUL7, we direct immunoblotting. As seen with high level CUL7

Oncogene CUL7 binds to p53 P Andrews et al 4544

Figure 8 CUL7 promotes cell proliferation and antagonizes p53 function. (a) CUL7 represses the transactivating function of p53. A synthetic luciferase reporter containing 13 tandem copies of a p53 response element (pG13-Luc, El-Deiry et al., 1992) was either singularly transfected or in combination with plasmids expressing p53 and CUL7 into U2OS cells. Relative luciferase activity was calculated as described in Materials and methods. Error bars indicate standard deviation. CUL7 repression was determined to be significant by a Student’s t-test, with P ¼ 2.8E-5. (b) A stable clone was isolated from U2OS cells after transfection of pcDNA3-myc3- CUL7 and subsequent selection of G418 resistance. The growth curves of parental U2OS and U2OS/CUL7 cells were determined. Relative proliferation was determined by dividing the cell number counted each day by the cell number counted on day 1. (c) U2OS cells were transduced with retroviruses expressing CUL7, CUL7DN680 or CUL7D389À404. Stable pools of each transduction were plated in 60 mm dishes and the rate of cell proliferation was determined by counting cells daily and dividing the number of cells by the number of cells counted on day 1. Inset: protein expression of each retrovirally infected cell pool expressing CUL7, CUL7DN680 or CUL7D389À404 was determined by Western blot analysis. (d) H1299 cells were transduced with retroviruses expressing CUL7or CUL7D389À404. Stable pools of each transduction were plated in 60 mm dishes and the rate of cell proliferation was determined by counting cells daily and divided by the number of cells on day 1. Inset: protein expression of each retrovirally infected cell pool was determined by Western blot analysis. (e) UV irradiated (40 J/M2) parental and CUL7-expressing U2OS cells were collected at 0, 24 and 48 h and analysed by FACS.

Oncogene CUL7 binds to p53 P Andrews et al 4545 expression in stable cell lines, retrovirally mediated CUL7. The major findings of the study are that CUL7, constitutive expression of CUL7 increased the rate of like its closely related protein, PARC, binds to ROC1, cell proliferation over the course of the experimentation. possesses ubiquitin ligase activity, localizes predomi- Deletion of the N-terminal p53-binding domain or nantly in the cytoplasm, and perhaps most significantly, mutation of a sequence required for p53 binding both interacts with p53. The tumor suppressor p53 acts as a abolished the ability of CUL7 to promote cellular sequence-specific transcription factor in the nucleus. proliferation. To further correlate the changes in Studies over the past several years have demonstrated proliferation rate by ectopic CUL7 expression with that nuclear export and subcellular localization are p53 function, we similarly transduced wild-type and critical regulatory events in p53 control (Zhang and D389–404 mutant CUL7 retrovirus into p53-deficient Xiong, 2001a; Chipuk and Green, 2004). Blocking p53 H1299 cells. Unlike p53-positive U2OS cells, CUL7 nuclear export inhibited MDM2-mediated p53 degrada- expression in H1299 cells had no significant effect on the tion and led to p53 stabilization (Roth et al., 1998; Tao rate of cell proliferation (Figure 8d). and Levine, 1999; O’Keefe et al., 2003), indicating that either p53 degradation occurs in the cytoplasm or CUL7 expression delayed UV-induced G2 accumulation requires a cytoplasmic factor(s). Potentially equally Following UV irradiation, p53-positive U2OS cells, but important as, and possibly linked with, the cytoplasmic not p53-deficient H1299 cells, undergo a G2 arrest degradation of p53 is the emerging mitochondrial role before resuming cell proliferation. While non-irradiated for p53 in promoting nuclear- and transcription- parental U2OS and H1299 cells exhibited similar cell independent (Baptiste and Prives, 2004; cycle phasing, 43 and 63% of the parental U2OS cells Chipuk and Green, 2004). Very little is known about accumulated in 24 and 48 h after UV the regulation of cytoplasmic p53, including both newly irradiation, respectively, and there is no significant G2 synthesized and exported, stability, modification, sub- accumulation seen in H1299 cells (Figure 8e). A G1 cytoplasmic localization, interacting partners and com- accumulation seen at the 48 h time point in H1299 cells plex formation. Our studies suggest that CUL7, along- might have been caused in part by the increase of cell side its homologue PARC, may function as a density. The reason why U2OS cells accumulate in G2, cytoplasmic regulator of p53. as opposed to G1 seen in other cell types, is not clear at Ectopic expression of CUL7 reduced p53’s transacti- present and MAPKAP kinase 2 has been suggested to be vating activity, increased the rate of cell proliferation, a component of this UV-activated G2/M checkpoint in and delayed cellular response to DNA damage. Disrup- U2OS cells (Manke et al., 2005). Non-irradiated CUL7- tion of p53 binding abolished the ability of CUL7 to expressing U2OS cells exhibited a similar promote cell proliferation and to repress p53-mediated distribution to parental U2OS cells, but CUL7-expres- transcription. Targeted disruption of the mouse Cul7 sing U2OS cells displayed a increased G1 population resulted in retarded proliferation of mouse embryo and a delayed G2 accumulation after exposure to UV fibroblasts (MEFs) and death of embryos soon after radiation; while 20 and 43% of the U2OS cells birth (Arai et al., 2003). Together, these results accumulated in G1 and G2 phase 24 h after UV demonstrate a growth-promoting function of CUL7. irradiation, there was an increase of G1 population to We further suggest that the growth promoting activity 31% and no significant G2 accumulation seen in CUL7- of CUL7 is attributed to, if not depends on, its ability to expressing U2OS cells. At 48 h after irradiation, 63% of bind with p53 and that increased p53 activity could the U2OS cells accumulated in G2 with only 17% cells account for, at least in part, for the retarded prolifera- at G1, compared with 56% of the CUL7-expressing cells tion of Cul7-null MEFs. and 22% in G1. There was no significant change in the CUL7 has also been reported to bind with the SV40 sub-G1 population (data not shown), suggesting that large T antigen and deletion of a sequence from the T Cul7 expression did not interfere with p53’s activity in antigen that disrupt its binding with CUL7 abolished promoting apoptosis following UV. The level of p21 the ability of the T antigen to transform primary cells CDK inhibitor protein was increased in CUL7-expres- and to promote anchorage-independent cell prolifera- sing U2OS (Figure 8e). Given the function of CUL7 in tion (Ali et al., 2004). Of the various cellular proteins repressing p53-dependent transcription, the increase of that interact with the T antigen, two major ones – tumor p21 is likely caused by a p53-independent mechanism. suppressors p53 and the Rb family of proteins – are There was little difference between H1299 and CUL7- functionally inactivated, underlying the transforming expressing H1299 cells in their response to UV irradia- activity of the T antigen (Ali and DeCaprio, 2001). The tion (Figure 8e). We therefore conclude that ectopic results generated thus far have not allowed us to build a expression of CUL7 interferes with the cellular response simple model that could reconcile two seemingly distinct to UV-generated DNA damage, likely by impairing the functions of CUL7: promoting cell growth through its function of p53. p53-binding ability, and being a cellular target of T antigen-mediated cell transformation. Binding of CUL7 with p53 apparently does not require the T antigen. It, Discussion however, remains possible that the T antigen and CUL7 may facilitate one another’s activity to cause functional In this report, we carried out a series of experiments inactivation of p53. A still puzzling issue is when and aimed at gaining insight into the cellular function of how CUL7, a cytoplasmic protein, binds to the T

Oncogene CUL7 binds to p53 P Andrews et al 4546 antigen and p53, both of which are localized predomi- with 5% CO2. Cell transfections were carried out using nantly in the nucleus. As determined by indirect FuGene (Roche, Alameda, CA, USA) or Effectene (Qiagen, immunofluorescence, we have not observed a significant Valencia, CA, USA) transfection reagents according to the change in the respective cytoplasmic and nuclear manufacturer’s instructions. For each transfection, final localization of CUL7 and p53 or CUL7 and the T concentration of total plasmid DNA was normalized with pcDNA3 to 1, 4 or 7 mg for each six-well, 60 mm or 100 mm antigen when the two proteins were co-expressed in the dish, respectively. To establish Cul7 stable expressing cell lines, same cell. U2OS cells were transfected with pcDNA3-myc3-Cul7. At 24 h The exact biochemical mechanism of CUL7, in after transfection, cells were trypsinized and re-seeded in a particular how it affects p53, remains to be determined. 1:10 000 dilution in media containing 0.75 mg/ml geneticin It is surprising to see that, in contrast to PARC which (Gibco (Invitrogen), Carlsbad, CA, USA) for 7 days. Twenty- has been reported to bind with p53 and sequester it in four independent single colonies were selected and expression the cytoplasm (Nikolaev et al., 2003), we have not been of CUL7 was analysed by direct immunoblotting using the able to detect any cytoplasmic sequestration activity of anti-myc antibody. CUL7 toward p53. In nearly all the cells that we have examined, p53, whether expressed at a low level Antibodies, immunochemistry and indirect immunofluorescence Procedures for immunoprecipitation and immunoblotting endogenously or at a high level ectopically, remained have been described previously (Jenkins and Xiong, 1995). localized predominantly in the nucleus despite ectopic Indirect immunofluorescence was previously described in expression of CUL7 (Figure 7). There were clear detail (Zhang and Xiong, 2001b). Monoclonal to HA separations of p53 and CUL7 localization in cells (12CA5, Boehringer-Mannheim, Indianapolis, IN, USA), expressing both proteins. The basis underlying the affinity purified polyclonal to HA (#sc-805, Santa Cruz, Santa mechanistic difference between CUL7 and PARC in Cruz, CA, USA), to myc (9E10, NeoMarker, Fremont, CA, affecting p53 subcellular localization is not clear at USA), to p53 (FL393, Santa Cruz, DO1, NEOMarkers, present and requires further studies. We suspect that Fremont, CA, USA), to large T antigen (PAb416, generously separate localization, rather than relocalizing each provided by Dr Terry van Dyke), rhodamine Red-, and other, may explain the facts that CUL7-mediated fluorescein isothiocyanate (FITC)-conjugated donkey second- ary antibodies (Jackson ImmunoResearch Laboratories, West repression of p53’s transcription activity is incomplete, Grove, PA, USA) were purchased commercially. and that ectopic CUL7 expression caused only a delay, but not complete abrogation, of p53-mediated G2 arrest In vitro and in vivo ubiquitin ligase activity assays after UV irradiation. The procedure for in vitro assay of ubiquitin ligase activity was Under the conditions where MDM2 can cause p53 essentially the same as previously described (Ohta et al., 1999; polyubiquitination in vitro and p53 degradation in vivo, Furukawa et al., 2002). Purified rabbit E1 ubiquitin-activating we were not be able to detect a similar function for enzyme was purchased from Affiniti Research Products CUL7 (Figure 6). Instead, we have repeatedly observed (Exeter, UK). His-tagged E2 human UbcH5C was purified a mono- and di-ubiquitination of p53 in vitro by CUL7. using nickel beads (QIAGEN). Ub was prepared by subclon- In co-transfection assays involving both CUL7 and ing full-length Ub as a fusion protein with a 6x His-tag and MDM2, we have thus far not detected any enhancing protein kinase C recognition site (LRRASV) and purified with nickel beads (Tan et al., 1999). Briefly, for both in vitro auto- effect by CUL7 on the ability of MDM2 to degrade p53 ubiquitination (E2 activation) and p53 ubiquitination assays (data not shown). From these results, we cautiously by the Cul7 ligase, wild-type or mutant CUL7 immunocom- conclude that CUL7 is unlikely to act as a p53 inhibitor plexes were immunoprecipitated from transfected cells. Im- by promoting p53 degradation and does not appear to munocomplexes immobilized on protein A agarose beads were be the long-sought cytoplasmic ubiquitin ligase respon- washed and added to a Ub ligation reaction mixture (30 ml) sible for p53 degradation. Instead, CUL7 may regulate that contained 50 mM Tris-HCl, pH 7.4, 5 mM MgCl2,2mM p53 through non-proteolytic ubiquitination. NaF, 10 nM okadaic acid, 2 mM ATP, 0.6 mM DTT, 0.75 mg Ub, 60 ng E1 and 300 ng E2. Reactions were incubated at 371C for 45 min, terminated by adding of 30 ml2Â SDS loading Materials and methods buffer, boiled for 4 min, and resolved by SDS–PAGE followed by autoradiography to visualize the ubiquitination ladders. Plasmids Substrate p53 was co-immunoprecipitated with CUL7 from Full-length human CUL7 cDNA was obtained from the p53-deficient H1299 cells co-transfected with a transcriptional Kazusa Gene project (KIAA0076), and subcloned into inactive mutant p53R273H. p53 ubiquitination was determined pcDNA3 mammalian expression vector with various epitope by immunoblotting using a monoclonal p53 antibody (DO1). tags. Human p53, HDM2 (human homologue of MDM2) and For the in vivo p53 ubiquitination assay, U2OS cells were ARF expression vectors were previously described (Zhang transfected with plasmids expressing various proteins and HA- et al., 1998; Zhang and Xiong, 2001b). GST-p53 and derivative ubiquitin as indicated in each figure. Total amount of plasmid deletion mutants were subcloned into the pEBG mammalian DNA in each transfection was adjusted to total 2.4 mg with GST expression vector. Mutations were introduced by site- pcDNA3 empty vector when needed. At 20 h after transfec- directed mutagenesis using a Quick-Change site-directed tion, cells were treated with proteasome inhibitor MG132 mutagenesis kit (Stratagene, La Jolla, CA, USA) and verified (25 mM) for 4 h. Cells were then collected, pelleted by by DNA sequencing. centrifugation, and lysed in 200 ml of pre-boiled SDS-lysis buffer (50 mM Tris-HCl, pH 7.5, 0.5 mM EDTA, 1% SDS, Cell lines, cell culture and cell transfection 1mM DTT), and further boiled for an additional 10 min. Human U2OS cells used in this study were cultured in Lysates were clarified by centrifugation at 14 000 r.p.m. on a DMEM, supplemented with 10% FBS in a 371C incubator microcentrifuge for 10 min. Supernatant was diluted 10-fold

Oncogene CUL7 binds to p53 P Andrews et al 4547 with 0.5% NP-40 buffer and immunoprecipitated with anti- Luciferase reporter assays p53 antibody (3.6 mg). Immunoprecipitates were washed three Cells were transfected in triplicate with various combinations times and resolved on a 7.5–15% gradient SDS–PAGE, of plasmids expressing different proteins as indicated and total followed by immunoblotting with anti-HA antibody (1 mg/ transfected DNA was normalized to 1 mg total cDNA per each ml) to detect HA-tagged ubiquitin. 10 mm plate. Luciferase reporter assays were carried out following the Dual Luciferase reporter assay protocol (Pro- Retroviral production and transduction mega, Madison, WI, USA). Briefly, cells were lysed with 500 ml Plates (10 cm) of 293T cells were transfected using the calcium of PLB (Passive Lysis Buffer) for 15 min at room temperature. phosphate method with 7 mg total DNA (3 mg pHIT-myc3- Lysates were collected and frozen at À801C to ensure complete CUL7, pHIT-myc3-CUL7DN680, pHIT-myc3-CUL7D389À404 or lysis. Lysates were thawed, cleared by centrifugation, and 20 ml pHIT-GFP for transfection and infection efficiency determina- were added to 100 ml of luciferase assay reagent. Firefly tion, plus 2 mg pCI-VSV and 2 mg pCI-PGZ). Cells were luciferase activity was assayed and recorded on a luminometer. incubated at 371C in DMEM supplemented with 10% FBS, In total, 100 ml of Stop and Glo reagent was added to each 1% Penn/Strep and 25 mM chloroquine for 18 h. The medium sample to determine Renilla luciferase activity. Individual was replaced with DMEM supplemented with 2% FBS and samples were normalized and luciferase activity was deter- incubated for 24 h at 321C. Viral supernatant was then mined. collected and filtered using a 0.45 mM syringe filter, supple- mented up to 10% with fresh FBS, and 10 mg/ml polybrene. UV irradiation and FACS analysis Six-well plates of U2OS or H1299 cells were incubated with U2OS and H1299 cells and their derivatives stably expressing 2 ml per well of the viral media for 24 h at 371C. Cell infection CUL7 were treated with 40 J/m2 of UV radiation using a efficiency was determined by GFP, and if needed, cells were Stratagene UV Stratalinker 2400. Irradiated cells were fixed subsequently infected to increase infection efficiency. with 100% ethanol to a final concentration of 75%. Fixed cells were washed with PBS and permeabilized with 0.1% Triton X- Growth curves 100, 0.1% mg/ml RNase in PBS for 30 min at 371C. Cells were U2OS and H1299 cells and their derivative lines expressing stained using propidium iodide (1 mg/ml) for 18 h at 41C and stably either the wild-type or mutant CUL7 were plated onto sorted using a Becton Dickinson FACScan and data analysed twenty 60 mm plates at a density of either 4 Â 104 (Figure 8b) using Summit software. or 2 Â 105 (Figure 8c) per plate. Each day cells from two plates were trypsinized and counted to determine the amount of cells per plate. From each plate two separate samples were taken, Acknowledgements and from each one of these samples two separate counts were made. Each plate and cell type was averaged from the eight We thank Gabrielle White Wolf for providing insightful individual counts to determine the cell number per cell line for discussion throughout this study, Toru Nishikawa for helping that day. Relative proliferation for each day was determined perform p53 ubiquitination assays and other members of by dividing the average daily cell number by the original Xiong lab for discussions. This study was supported by NIH amount of cells per plate as determined on day 1. grant CA65572 (YX).

References

Ali SH, DeCaprio JA. (2001). Cellular transformation by SV40 Feldman RMR, Correll CC, Kaplan KB, Deshaies RJ. (1997). large T antigen: interaction with host proteins. Semin Cancer A complex of Cdc4p, Skp1p, and Cdc53p/Cullin catalyzes Biol 11: 15–23. ubiquitination of the phosphorylated CDK inhibitor Sic1p. Ali SH, Kasper JS, Arai T, DeCaprio JA. (2004). Cul7/p185/ Cell 91: 221–230. p193 binding to simian virus 40 large T antigen has a role in Freemont PS, Hanson IM, Trowsdale J. (1991). A novel cellular transformation. J Virol 78: 2749–2757. cysteine-rich sequence motif. Cell 64: 483–484. Arai T, Kasper JS, Skaar JR, Ali SH, Takahashi C, DeCaprio Furukawa M, He YJ, Borchers C, Xiong Y. (2003). Targeting JA. (2003). Targeted disruption of p185/Cul7 gene results in of protein ubiquitination by BTB-Cullin 3-Roc1 ubiquitin abnormal vascular morphogenesis. Proc Natl Acad Sci USA ligases. Nat Cell Biol 5: 1001–11007. 100: 9855–9860. Furukawa M, Ohta T, Xiong Y. (2002). Activation of UBC5 Bai C, Sen P, Hofmann K, Ma L, Goebl M, Harper JW et al. ubiquitin-conjugating enzyme by the RING finger of ROC1 (1996). SKP1 connects cell cycle regulators to the ubiquitin and assembly of active ubiquitin ligases by all cullins. J Biol proteolysis machinery through a novel motif, the F-box. Cell Chem 277: 15758–15765. 86: 263–274. Furukawa M, Yanping Z, McCarville J, Ohta T, Xiong Y. Baptiste N, Prives C. (2004). p53 in the cytoplasm: a question (2000). The C-terminal sequence and ROC1 are required for of overkill? Cell 116: 487–489. efficient nuclear accumulation, NEDD8 modification, and Chipuk JE, Green DR. (2004). Cytoplasmic p53: Bax and ubiquitin ligase activity of CUL1. Mol Cell Biol 20: 8185– Forward. Cell Cycle 3: 429–431. 8197. Deshaies RJ. (1999). SCF and cullin/RING H2-based ubiqui- Geyer R, Wee S, Anderson S, Yates J, Wolf DA. (2003). BTB/ tin ligases. Annu Rev Cell Dev Biol 15: 435–467. POZ domain proteins are putative substrate adaptors for Dias DC, Dolios G, Wang R, Pan ZQ. (2002). CUL7: a DOC cullin 3 ubiquitin ligases. Mol Cell 12: 783–790. domain-containing cullin selectively binds Skp1.Fbx29 to Hershko A, Ciechanover A. (1998). The ubiquitin system. form an SCF-like complex. Proc Natl Acad Sci USA 99: Annu Rev Biochem 67: 425–479. 16601–16606. Huibregtse JM, Scheffner M, Beaudenon S, Howley PM. El-Deiry WS, Kern SE, Pietenpol JA, Kinzler KW, Vogelstein (1995). A family of proteins structurally and functionally B. (1992). Definition of a consensus binding site for p53. Nat related to the E6-AP ubiquitin-protein ligase. Proc Natl Genet 1: 45–49. Acad Sci USA 92: 2563–2567.

Oncogene CUL7 binds to p53 P Andrews et al 4548 Jenkins CW, Xiong Y. (1995). Immunoprecipitation and Scheffner M, Nuber U, Huibregtse JM. (1995). Protein immunoblotting in cell cycle studies. In Pagano M (ed.). ubiquitination involving an E1–E2–E3 enzymes ubiquitin Cell Cycle: Material and Methods. Springer-Verlag: New thioester cascade. Nature 373: 81–83. York, pp 250–263. Skowyra D, Craig K, Tyers M, Elledge SJ, Harper JW. (1997). Joazeiro CA, Weissman AM. (2000). RING finger proteins: F-box proteins are receptors that recruit phosphorylated mediators of ubiquitin ligase activity. Cell 102: 549–552. substrates to the SCF ubiquitin-ligase complex. Cell 91: 209– Kipreos ET, Lander LE, Wing JP, He W-W, Hedgecock EM. 219. (1996). cul-1 is required for cell cycle exit in C. elegans and Tan P, Fuches SY, Angus A, Wu K, Gomez C, Ronai Z et al. identifies a novel gene family. Cell 85: 829–839. (1999). Recruitment of a ROC1-CUL1 ubiquitin ligase by Ko LJ, Prives C. (1996). p53: puzzle and paradigm. Genes Dev Skp1 and HOS to catalyze the ubiquitination of IkBa. Mol 10: 1054–1072. Cell 3: 527–533. Levine AJ. (1997). p53, the cellular gatekeeper for growth and Tang Z, Li B, BharadwajR, Zhu H, Ozkan E, Hakala K division. Cell 88: 323–331. et al. (2001). aAPC2 Cullin protein and APC11 RING Lin J, Chen J, Levine AJ. (1994). Several hydrophobic amino protein comprise the minimal ubiquitin ligase module of acids in the p53 amino-terminal domain are required for the -promoting complex. Mol Biol Cell 12: transcriptional activation, binding to MDM-2 and the 3839–3851. adenovirus 5 E1B 55-kD protein. Genes Dev 8: 1235–1246. Tao W, Levine AJ. (1999). Nucleocytoplasmic shuttling of Manke IA, Nguyen A, Lim D, Stewart MQ, Elia AE, Yaffe oncoprotein Hdm2 is required for Hdm2-mediated degrada- MB. (2005). MAPKAP kinase-2 is a tion of p53. Proc Natl Acad Sci USA 96: 3077–3080. kinase that regulates the G2/M transition and Tsai SC, Pasumarthi KB, Pajak L, Franklin M, Patton B, progression in response to UV irradiation. Mol Cell 17: 37–48. Wang H et al. (2000). Simian virus 40 large T antigen binds a Marin I, Lucas JI, Gradilla AC, Ferrus A. (2004). Parkin and novel Bcl-2 homology domain 3-containing proapoptosis relatives: the RBR family of ubiquitin ligases. Physiol protein in the cytoplasm. J Biol Chem 275: 3239–3246. Genomics 17: 253–263. Xu L, Wei Y, Reboul J, Vaglio P, Shin TH, Vidal M et al. Nikolaev AY, Li M, Puskas N, Qin J, Gu W. (2003). Parc: a (2003). BTB proteins are substrate-specific adaptors in an cytoplasmic anchor for p53. Cell 112: 29–40. SCF-like modular ubiquitin ligase containing CUL-3. Ohta T, Michel JJ, Schottelius AJ, Xiong Y. (1999). ROC1, a Nature 425: 316–321. homolog of APC11, represents a family of cullin partners Yu H, Peters J-M, King RW, Page AM, Hieter P, Kirschner with an associated ubiquitin ligase activity. Mol Cell 3: 535– MW. (1998). Identification of a cullin homology region in a 541. subunit of the anaphase-promoting complex. Science 279: O’Keefe K, Li H, Zhang Y. (2003). Nucleocytoplasmic 1219–1222. shuttling of p53 is essential for MDM2-mediated cytoplas- Zachariae W, Shevchenko A, Andrews PD, Ciosk R, Galova mic degradation but not ubiquitination. Mol Cell Biol 23: M, Stark MJR et al. (1998). Mass spectrometric analysis of 6396–6405. the anaphase-promoting complex from yeast: identification Pickart CM. (2001). Mechanisms underlying ubiquitination. of a subunit related to cullins. Science 279: 1216–1219. Ann Rev Biochem 70: 503–533. Zhang Y, Xiong Y. (2001a). Control of p53 ubiquitination and Pintard L, Willis JH, Willems A, Johnson JL, Srayko M, Kurz nuclear export by MDM2 and ARF. Cell Growth Diff 12: T et al. (2003). The BTB protein MEL-26 is a substrate- 175–186. specific adaptor of the CUL-3 ubiquitin-ligase. Nature 425: Zhang Y, Xiong Y. (2001b). A p53 amino-terminal nuclear 311–316. export signal inhibited by DNA damage- induced phosphor- Roth J, Dobbelstein M, Freedman DA, Shenk T, Levine AJ. ylation. Science 292: 1910–1915. (1998). Nucleo-cytoplasmic shuttling of the hdm2 oncopro- Zhang Y, Xiong Y, Yarbrough WG. (1998). ARF promotes tein regulates the levels of the p53 protein via a pathway MDM2 degradation and stabilizes p53: ARF-INK4a locus used by the human immunodeficiency virus rev protein. deletion impairs both the Rb and p53 tumor suppression EMBO J 17: 554–564. pathways. Cell 92: 725–734.

Oncogene