JOURNAL TRANSCRIPT

---

Molecular Cell 24, 355–366, November 3, 2006 ª2006 Elsevier Inc.

DOI 10.1016/j.molcel.2006.09.007

Phosphorylation-Dependent Ubiquitination of Cyclin D1 by the SCFFBX4-aB Crystallin Complex Douglas I. Lin,1,2 Olena Barbash,1 K.G. Suresh Kumar,3 Jason D. Weber,4 J. Wade Harper,5 Andres J.P. Klein-Szanto,7 Anil Rustgi,1,6 Serge Y. Fuchs,3 and J. Alan Diehl1,2,* 1 The Leonard and Madlyn Abramson Family Cancer Research Institute and Cancer Center 2 Department of Cancer Biology 3 Department of Animal Biology University of Pennsylvania Philadelphia, Pennsylvania 19104 4 Department of Medicine Washington University School of Medicine St. Louis, Missouri 63110 5 Department of Pathology Harvard Medical School Boston, Massachusetts 02115 6 Department of Medicine University of Pennsylvania 7 Fox Chase Cancer Center Philadelphia, Pennsylvania 19111

Summary Growth factor-dependent accumulation of the cyclin D1 proto-oncogene is balanced by its rapid phosphorylation-dependent proteolysis. Degradation is triggered by threonine 286 phosphorylation, which promotes its ubiquitination by an unknown E3 ligase. We demonstrate that Thr286-phosphorylated cyclin D1 is recognized by a Skp1-Cul1-F box (SCF) ubiquitin ligase where FBX4 and aB crystallin govern substrate specificity. Overexpression of FBX4 and aB crystallin triggered cyclin D1 ubiquitination and increased cyclin D1 turnover. Impairment of SCFFBX4-aB crystallin function attenuated cyclin D1 ubiquitination, promoting cyclin D1 overexpression and accelerated cellcycle progression. Purified SCFFBX4-aB crystallin catalyzed polyubiquitination of cyclin D1 in vitro. Consistent with a putative role for a cyclin D1 E3 ligase in tumorigenesis, FBX4 and aB crystallin expression was reduced in tumor-derived cell lines and a subset of primary human cancers that overexpress cyclin D1. We conclude that SCFFBX4-aB crystallin is an E3 ubiquitin ligase that promotes ubiquitin-dependent degradation of Thr286-phosphorylated cyclin D1. Introduction Cyclin D1, the allosteric regulator of CDK4/6, is an integral mediator of growth factor-dependent G1 phase progression. Cyclin D1 expression, activation, and nuclear accumulation occur during mid-G1 phase (Marshall, 1999); nuclear cyclin D1/CDK complexes trigger G1 progression via inactivation of the retinoblastoma protein and related family members (Sherr, 1996). During S phase, cyclin D1 activation is opposed by Pro287directed phosphorylation of Thr286 by glycogen syn*Correspondence: [email protected]

thase kinase 3b (GSK3b); phosphorylation triggers cyclin D1 nuclear export and ubiquitination-mediated proteolysis (Alt et al., 2000). While phosphorylation of Thr286 directs CRM1 binding to cyclin D1, it is also an essential signal for ubiquitin-dependent destruction (Diehl et al., 1997). Ubiquitination is catalyzed through a pathway involving an E1 (ubiquitin-activating enzyme), E2 (ubiquitin-conjugating enzyme), and E3 (ubiquitin ligase). The latter enzyme is the most diverse, and it determines substrate specificity and the rate of ubiquitin conjugation. The E3 itself can be either a single protein or a multiprotein complex. An E3 that recognizes Thr286-phosphorylated cyclin D1 has not been identified. Regulated proteolysis is commonly utilized to maintain threshold levels of critical cell-cycle regulators. Two distinct classes of E3 ligases participate in the regulation of cell-cycle progression. The Skp1-Cul1-F box (SCF) family of E3 ubiquitin ligases promotes ubiquitination of phosphorylated substrates and typically targets mediators of the G1-S phase transition (Skowyra et al., 1997). While the F box component determines substrate specificity, additional cofactors may also facilitate substrate recognition (Ganoth et al., 2001; Hao et al., 2005; Spruck et al., 2001). The G2-M transition requires the E3 ligase activity of the anaphase promoting complex (APC/C) (Harper et al., 2002); CDC20 and CDH1 subunits of APC/C direct substrate recognition (Burton et al., 2005; Kraft et al., 2005; Pfleger et al., 2001). Cyclin D1 is frequently overexpressed in human cancer; overexpression is often associated with increased gene expression due to gene amplification or oncogene-induced signaling (Sherr, 1996). Accumulation of cyclin D1 in cancer can also result from disruption of cyclin proteolysis. Cyclin D1 proteolysis can be inhibited through several distinct mechanisms. First, recent work revealed that human cancer cells harbor mutations in cyclin D1 that disrupt Thr286 phosphorylation (Benzeno et al., 2006) and prevent ubiquitin-mediated proteolysis. Second, upstream oncogenic events, such as those targeting Wnt or Ras, inhibit GSK3b activity, thereby decreasing cyclin D1 turnover (Diehl et al., 1998; Rimerman et al., 2000). Finally, it remains plausible that loss of a D1-specific E3 ligase might contribute to D1 overexpression in human cancer. The latter mechanism emphasizes the importance of identifying the cyclin D1-specific E3 ligase. Given the necessity of Thr286 phosphorylation for cyclin D1 proteolysis (Diehl et al., 1997), an unidentified SCF complex is a likely cyclin D1 ubiquitin ligase. We have identified a SCF E3 ligase, wherein the F box protein, FBX4, targets Thr286-phosphorylated cyclin D1. FBX4 belongs to the Fbx class of F box proteins, which lack known structural motifs in their C terminus. Cyclin D1 recognition by FBX4 requires aB crystallin, which functions in a manner analogous to CKS1 in Skp2-dependent recognition of phosphorylated p27Kip1. The data presented reveal that both FBX4 and aB crystallin are required for rapid ubiquitination and degradation of cyclin D1 in vivo and that purified SCFFBX4-aB crystallin complexes can direct cyclin D1

Molecular Cell 356

Figure 1. Identification of aB Crystallin as a Cyclin D1-Associated Protein (A) Silver stain of affinity-purified cyclin D1 complexes prepared from parental NIH 3T3 cells or Flag-D1 3T3 cell lines treated with the proteasome inhibitor LLnL. Cyclin D1interacting proteins were identified by mass spectrometry. (B) Western blot of endogenous cyclin D1 precipitates prepared from proliferating NIH 3T3 cells treated with DMSO, N-acetylleucinyl-leucinyl-methioninal (LLM), or LLnL. Normal rabbit serum (NRS) precipitates served as a negative control. (C) The cyclin D1-aB crystallin interaction is dependent on cyclin D1 phosphorylation at Thr286. NIH 3T3 cells expressing either wild-type cyclin D1 or the D1T286A mutant were treated with vehicle, LLM, or LLnL for 6 hr, and cell lysates prepared from these cells were precipitated with a cyclin D1-specific antibody (13-17G) and probed for cyclin D1 and aB crystallin.

ubiquitination in vitro. Strikingly, analysis of breast cancer-derived cells revealed loss of aB crystallin in a subset of cell lines. Loss of aB crystallin resulted in cyclin D1 stabilization, revealing the requirement of the SCFFBX4-aB crystallin ligase for cyclin D1 proteolysis. Results Identification of aB Crystallin as a Major Component of Cyclin D1 Complexes To identify a putative cyclin D1 E3 ubiquitin ligase, we purified cyclin D1 complexes by affinity chromatography. Due to the short half-life of phosphorylated cyclin D1, cells were treated with proteasome inhibitors to enrich for phospho-T286 cyclin D1 and stabilize interactions with a putative E3 ligase. A major protein of w22 to 23 kDa (identified by mass spectrometric analysis as aB crystallin) copurified with Flag-cyclin D1, but not from parental NIH 3T3 cells (Figure 1A) or with a nondegradable cyclin D1 mutant (data not shown). We also noted a protein of w46 kDa that copurified with cyclin D1; the identity of this protein could not be confirmed due to contaminating actin peptides. Association between endogenous cyclin D1 and aB crystallin under conditions of proteasome inhibition was confirmed by precipitation with a cyclin D1-specific antibody followed by immunoblot with aB crystallin antibodies (Figure 1B). Binding of cyclin D1-aB crystallin depended upon the integrity of Thr286 since the cyclin D1T286A mutant did not coprecipitate with aB crystallin (Figure 1C). Phospho-T286-Dependent Interaction between Cyclin D1 and the SCFFBX4-aB Crystallin Ligase aB crystallin associates with a 46 kDa protein, FBX4, in vivo and increases ubiquitin conjugation of unknown proteins (den Engelsman et al., 2003). We therefore considered whether aB crystallin might facilitate targeting of FBX4 to phosphorylated cyclin D1. Indeed, endogenous cyclin D1 and aB crystallin do coprecipitate with FBX4 (Figure 2A and see Figure S1A in the Supplemental Data available with this article online). The binding of cyclin D1 to FBX4 and aB crystallin was observed only

under conditions of proteasome inhibition, likely reflecting the stabilization of labile complexes. To determine whether Thr286 phosphorylation is required for recognition by the SCFFBX4-aB crystallin complex, wild-type cyclin D1 or phosphorylationdeficient cyclin D1 mutants produced in SF9 insect cells (cyclin D1 is readily phosphorylated at Thr286 in Sf9 cells [Diehl et al., 1997]) were mixed with immobilized SCFFBX4-aB crystallin complexes produced in 293T cells. Wild-type cyclin D1 bound to the SCFFBX4-aB crystallin complex in vitro; omission of aB crystallin inhibited cyclin D1 binding to FBX4 (Figure 2B). Binding of SCFFBX4-aB crystallin to two phosphorylation-deficient D1T286A and D1P287A (Benzeno et al., 2006) mutants was not observed (Figure 2B). These results demonstrate that FBX4 binds to cyclin D1 in a phospho-Thr286-dependent manner and that aB crystallin increases the affinity of cyclin D1 for the SCFFBX4 complex. The role of Thr286 phosphorylation in mediating recognition by FBX4-aB crystallin, SKP1, and CUL1 complexes was determined using a phospho-T286 peptide corresponding to the C terminus of cyclin D1. FBX4-aB crystallin-SKP1-CUL1 complexes specifically associated with phospho-T286 peptides, but not with unphosphorylated peptides. Omission of either FBX4 or aB crystallin eliminated binding of SCFFBX4-aB crystallin to the phospho-D1 peptides (Figure 2C). FBX4 did not bind to other phosphodegrons (i.e., phospho-IkBa or phospho-MYC peptides), excluding the possibility that the SCFFBX4-aB crystallin complex simply has an affinity for phosphorylated serine/threonine residues (Figure 2D). Additionally, phosphorylated cyclin D1 peptides did not recruit other F box proteins such as FBW2 (Figure 2E) or SKP2 (data not shown). These results suggest that both aB crystallin and FBX4 are required for the recognition of phospho-T286 cyclin D1 by the SCFFBX4-aB crystallin complex. The SCFFBX4-aB Crystallin Complex Regulates Cyclin D1 Protein Stability In Vivo To determine whether the SCFFBX4-aB crystallin complex regulates cyclin D1 proteolysis, a dominant-negative

Proteolysis of Cyclin D1 by SCFFBX4-aB Crystallin 357

Figure 2. Phosphorylation-Dependent Interaction between Cyclin D1 and the SCFFBX4-aB Crystallin Complex (A) Endogenous FBX4, cyclin D1, and aB crystallin associate in vivo. Cell lysates prepared from proliferating NIH 3T3 cells treated with DMSO or MG132 were precipitated with antibodies directed against FBX4 or normal rabbit serum (NRS) and subjected to immunoblot with the indicated antibodies. (B) Phospho-T286-dependent binding of cyclin D1 to SCFFBX4-aB crystallin in vitro. Cyclin D1 or phosphorylation-deficient derivatives D1T286A and D1P287A were mixed with SCFFBX4-aB crystallin complexes purified by immunoaffinity chromatography from 293T cells and blotted for cyclin D1; where indicated, FBX4 and aB crystallin were omitted. (C) Cyclin D1-FBX4 binding is dependent on aB crystallin and Thr286 phosphorylation. 293T lysates containing the indicated proteins were incubated with beads coupled to the phosphorylated cyclin D1 peptide or to unphosphorylated cyclin D1 peptide. Bound proteins were detected by immunoblot. (D) Phospho-MYC and -IkKa do not interact with SCFFBX4-aB crystallin. Beads coupled to the indicated peptides were incubated with 293T lysates expressing SCFFBX4-aB crystallin complexes as in (C), and bound FBX4 was detected by immunoblot. (E) Beads coupled to the indicated peptides were incubated with 293T lysates harboring either SCFFBX4-aB crystallin or SCFFBW2 complexes, and bound FBX4 or FBW2 was detected by anti-FLAG western blot. Uncoupled beads served as negative (background) control.

FBX4 allele was created (DF)FBX4. F box deletion abrogates SKP1 binding (Zheng et al., 2002), but not substrate or aB crystallin recognition (Figure S1B; data not shown). Dominant-negative F box mutants have been used to assess substrate relationships (Carrano et al., 1999; Latres et al., 1999; Montagnoli et al., 1999). Expression of (DF)FBX4 increased the half-life of endogenous cyclin D1 from less than 30 min to greater than 60 min (Figure 3A). Expression of (DF)FBX4 also induced accumulation of cyclin D1 in an aB crystallin-dependent manner, but did not increase cyclin D1T286A levels (Figure 3B). The effect of aB crystallin likely reflects the capacity to promote (DF)FBX4 binding to phosphorylated cyclin D1 in cells. As an independent assessment of FBX4 and aB crystallin function, we utilized short-hairpin vectors to target either component. Knockdown of FBX4 or aB crystallin extended the half-life of cyclin D1 to greater than 60 min, whereas the half-life remained less than 30 min in cells expressing control shRNA (Figures 3C and 3D). Knockdown of FBX4 also stabilized cyclin D2 (data not shown), suggesting that SCFFBX4-aB crystallin may regulate the proteolysis of all D type cyclins. Knockdown of FBX4 in human U2OS cells also triggered accumulation

of cyclin D1 (Figure S1C) compared to cells expressing shRNAs directed against another F box protein, b-TrCP (Jin et al., 2003), demonstrating that FBX4 regulates cyclin D1 proteolysis in mouse and human cells. Conversely, knockdown of FBX4 did not affect accumulation of cyclins E or A (Figure 5C; Figure S1C), which are ubiquitinated by the SCFFBW7 and APC/C complexes, respectively (Harper et al., 2002; Koepp et al., 2001). The ability of FBX4 and aB crystallin to drive cyclin D1 proteolysis was also investigated. 293T cells were transfected with cyclin D1, FBX4, and aB crystallin, and cyclin D1 abundance was assessed by immunoblot. Ectopic expression of FBX4 and aB crystallin decreased levels of cyclin D1 in a dose-dependent manner (Figure 3E). Reduction of cyclin D1 was dependent on Thr286 phosphorylation (Figure 3E) and proteasome activity (Figure 3F). These data suggest the SCFFBX4-aB crystallin ligase regulates cyclin D1 stability and levels in vivo. The SCFFBX4-aB Crystallin Ubiquitin Ligase Catalyzes Cyclin D1 Ubiquitination Reduced cyclin D1 turnover in cells expressing either (DF)FBX4 or knockdown of FBX4 suggested that FBX4 directs cyclin D1 polyubiquitination. Consistent with

Molecular Cell 358

Figure 3. The SCFFBX4-aB Crystallin Ubiquitin Ligase Regulates Cyclin D1 Stability and Protein Levels In Vivo (A) NIH 3T3 cells expressing empty vector or (DF)FBX4 were pulse labeled with 35S methionine/cysteine and ‘‘chased’’ for the indicated time periods. Cyclin D1 was precipitated and visualized by autoradiography. (B) Western analysis of whole-cell extracts prepared from 293T cells transfected with cyclin D1, CDK4, (DF)FBX4, and aB crystallin. (C and D) NIH 3T3 cells expressing either control, aB crystallin (C), or FBX4 shRNAs (D) were treated with 100 mg/mL of cycloheximide for the indicated time periods, and cyclin D1 half-life was determined by western blotting. (E) Expression of FBX4 and aB crystallin promotes cyclin D1 proteolysis. 293T cells were transfected with cyclin D1, CDK4, aB crystallin, FBX4, and GFP. Cell lysates were prepared and immunoblotted for cyclin D1, FBX4, and GFP. (F) Proteasome inhibition inhibits FBX4-dependent cyclin D1 proteolysis.

this notion, (DF)FBX4 expression decreased cyclin D1 polyubiquitination in vivo and this effect was potentiated by coexpression of aB crystallin (Figures 4A and 4B). Similar results were obtained when these reactions were analyzed by immunoblot with either a HA antibody (to detect HA-ubiquitin) or a cyclin D1 antibody (Figures 4A and 4B), excluding the possibility that these results reflected ubiquitination of coprecipitating proteins. Reduced cyclin D1 ubiquitination was also observed in cells expressing shRNAs directed against FBX4 (Figure 4C). Thus, impairment of SCFFBX4-aB crystallin function via either (DF)FBX4 or FBX4 knockdown stabilizes cyclin D1 as a result of decreased cyclin D1 polyubiquitination. In vitro-transcribed and -translated 35S-labeled cyclin D1 was mixed with purified SCFFBX4-aB crystallin complexes to determine whether cyclin D1 was a direct substrate. After addition of ATP, ubiquitin, E1, E2, and purified SCFFBX4-aB crystallin, we detected higher molecular weight cyclin D1 species consistent with cyclin D1 polyubiquitination (Figure 4D, lane 2). In the absence of FBX4 (lane 3), Cul1/Skp1/Roc1 (lane 4), or E1/E2 (lane 5), cyclin D1 ubiquitination was eliminated. In addition, a monoubiquitinated cyclin D1 form was also seen in our in vitro ubiquitination reactions (Figure 4D, lane 2, lower panel), but not in control reactions lacking FBX4 (Figure 4D, lanes 1 and 3, lower panel), supporting the role

of SCFFBX4-aB crystallin as an E3 ligase for cyclin D1. Replacement of wild-type FBX4 with (DF)FBX4 impaired polyubiquitination of cyclin D1 (Figure 4E, lane 2), demonstrating that ubiquitination depended on Cul1/Skp1/ Roc1 and excluding the possibility that ubiquitination was mediated by another contaminating F box protein. Similarly, omission of aB crystallin inhibited cyclin D1 ubiquitination (Figure 4E, lane 3), demonstrating that aB crystallin is an essential cofactor. Ubiquitinating activity was also dependent on Thr286 phosphorylation of cyclin D1 (Figure 4F). Similar results were obtained using SCFFBX4-aB crystallin produced in Sf9 cells (data not shown). These results provide direct biochemical evidence that cyclin D1 is a bona fide substrate for SCFFBX4-aB crystallin. Cytoplasmic Localization of FBX4 and aB Crystallin Cyclin D1 ubiquitination and proteolysis are postulated cytoplasmic events (Diehl et al., 1998), suggesting the involvement of a cytoplasmic E3 ligase. Immunofluorescence staining, utilizing two distinct FBX4 antibodies, revealed cytoplasmic FBX4 in both NIH 3T3 and U2OS cells (Figure 5A). Knockdown of FBX4 dramatically reduced staining with either antibody, confirming antibody specificity (Figures 5B and 5C). The cytoplasmic localization was confirmed by biochemical fractionation

Proteolysis of Cyclin D1 by SCFFBX4-aB Crystallin 359

Figure 4. The SCFFBX4-aB Crystallin Complex Catalyzes Cyclin D1 Ubiquitination In Vivo and In Vitro (A) NIH 3T3 cells expressing Flag-cyclin D1 were transfected with HA-ubiquitin, (DF)FBX4, and aB crystallin and treated with 20 mM MG132 for 0, 2, and 4 hr. Cell lysates were subjected to precipitation with either normal rabbit serum (NRS) or with a cyclin D1-specific antibody, and ubiquitinated proteins were visualized by anti-HA immunoblot. (B) Same as in (A) except that cyclin D1 immunoprecipitates were probed with a cyclin D1 antibody. (C) Human U2OS cells were cotransfected with HA-tagged ubiquitin, cyclin D1, and shRNAs specific for firefly luciferase or FBX4. Cyclin D1 ubiquitination was detected by immunoprecipitation with an anti-Flag antibody followed by anti-HA western blotting. (D) The SCFFBX4-aB crystallin complex catalyzes ubiquitination of cyclin D1 in vitro. Ubiquitination of in vitro-transcribed and -translated cyclin D1 was assessed by mixing cyclin D1 with GSK3b, ATP, ubiquitin, E1, E2, and a complete SCFFBX4-aB crystallin complex (lane 2). The reactions were also carried out omitting the following: the entire E3 ligase (lane 1); FBX4 (lane 3); Cul1, Skp1, and Roc1 (lane 4); or E1 and E2 (lane 5). (E) The F box of FBX4 and aB crystallin are required for cyclin D1 polyubiquitination in vitro. Ubiquitination reactions were initiated by addition of ATP, ubiquitin, and E1 and E2 enzymes. Where indicated, wild-type FBX4 was replaced with (DF)FBX4 (lane 2), and aB crystallin was omitted from the reaction (lane 3). (F) Phosphorylation-deficient cyclin D1T286A is refractory to SCFFBX4-aB crystallin-dependent ubiquitination. Cyclin D1 ubiquitination in (D)–(F) was assessed by autoradiography (upper panels), and input is shown (lower panels).

(Figure 5D). While our results reveal that FBX4 and aB crystallin are largely cytoplasmic, they do not exclude the possibility that a small fraction of FBX4 and aB crystallin transiently shuttles between nuclear and cytoplasmic compartments (den Engelsman et al., 2004). Cell-Cycle-Dependent Binding of FBX4 to Cyclin D1 Following reentry from G0, cyclin D1 induction is limited by growth factor-dependent gene expression (Matsushime et al., 1991). At the G1/S phase transition, cyclin D1 is phosphorylated at Thr286 by GSK3b, exported from the nucleus, and degraded in the cytoplasm (Diehl et al., 1998). We hypothesized that, following G0 release, FBX4 would not limit cyclin D1 accumulation until the G1/S phase transition when GSK3b phosphorylates cyclin D1, inducing its cytoplasmic redistribution. To test this hypothesis, we assessed the cell-cycle dependence of cyclin D1-FBX4 binding following G0 release. NIH 3T3 fibroblasts were cultured for 24 hr in medium containing

0.1% FBS and released in medium containing 10% FBS. Lysates were prepared from NIH 3T3 cells synchronized at G0 (0 hr), G1 (8 hr), and S (18 hr) phases of the cell cycle. Four hours prior to harvesting, cells were treated with MG132 to enrich for cyclin D1-FBX4 complexes. Immunoblot of FBX4 precipitates revealed enrichment of cyclin D1-FBX4 complexes during S phase following cell-cycle reentry (Figure 6A). To assess the dependence of cyclin D1-FBX4 interaction on GSK3b, we transfected NIH 3T3 cells with dominant-negative GSK3b or empty vector and assessed binding. Expression of this GSK3b mutant dramatically reduced cyclin D1-FBX4 binding (Figure 6B) as well as cyclin D1 phosphorylation (data not shown). Knockdown of the SCFFBX4-aB Crystallin Ligase Accelerates G1 Phase Progression Because overexpression of cyclin D1 correlates with increased rates of cell-cycle progression (Ohtsubo and

Molecular Cell 360

Figure 5. Cytoplasmic Localization of FBX4 and aB Crystallin (A) Immunofluorescence of asynchronously growing NIH 3T3 and U2OS cells with N-terminal (top panel) or C-terminal (middle and bottom panels) FBX4-specific antibodies. (B) Immunofluorescence of asynchronously growing mouse fibroblasts stably expressing FBX4-specific hairpins (F-3T3-2, F-3T3-4). (C) Western analysis confirming knockdown of FBX4 in NIH 3T3 cells. Asterisk indicates a nonspecific band detected by the secondary antibody. (D) Stable NIH 3T3 cells expressing empty vector or Flag-tagged FBX4 were fractionated into cytoplasmic (C), nuclear (N), and total (T) extracts. Fractions were used for immunoprecipitation with an anti-Flag antibody or for direct western blots as indicated.

Roberts, 1993; Quelle et al., 1993; Resnitzky et al., 1994), we expected that compromised cyclin D1 proteolysis would accelerate G1 transition. However, because G0G1-S phase movement is largely dependent upon cyclin D1 gene expression, we reasoned that reentry rates would be refractory to FBX4 and aB crystallin levels. To test this, cell lines wherein either aB crystallin or FBX4 is knocked down were synchronized by serum starvation. Cells were released in complete media containing BrdU, and S phase entry was assessed by FACS. Knockdown of aB crystallin or FBX4 did not accelerate the G0 to S phase transition (Figure 6C). Similarly, expression of (DF)FBX4 did not accelerate S phase entry following a serum starvation (G0) block (data not shown). We next assessed whether loss of FBX4 or aB crystallin would accelerate progression from G2 to S phase. Cells were arrested at the G2/M transition by nocodazole treatment and released once again in medium containing BrdU. Whereas control NIH 3T3 cell lines entered S phase between 9 and 10 hr postnocodazole (G2/M) release, knockdown of aB crystallin or FBX4 reproducibly accelerated G1 phase traverse, resulting in S phase entry 4–6 hr post-nocodazole release (Figure 6D and Table S1). Cyclin D1 levels remained elevated in knockdown cells relative to controls following nocodazole release (Figure 6E). Consistent with D1 as the primary target,

knockdown of FBX4 or aB crystallin did not accelerate cell-cycle progression in D12/2 fibroblasts (Figures 6F and 6G) or D1T286A-expressing 3T3 cells (data not shown). FBX4 and aB Crystallin Levels Are Altered in Human Cancers Overexpression of cyclin D1 in various human malignancies suggests that impairment of SCFFBX4-aB crystallin function might contribute to cyclin D1 overexpression in cancer. Indeed, the aB crystallin locus maps to chromosome 11q22.3-q23.1, a region deleted in human cancer (Dohner et al., 1999). To quantify FBX4 and aB crystallin levels in human malignancies, we assessed expression in a matched tumor versus normal mRNA array (Experimental Procedures). aB crystallin mRNA levels were low in 40% of tumors examined as compared to corresponding normal controls (Figure 7A); in contrast, FBX4 was downregulated in 24% of tumors (Figure 7A). mRNA levels of FBX4 and aB crystallin mRNA were downregulated in a broad spectrum of tumors, including prostate, thyroid, and breast adenocarcinomas and lymphoma. The reduced frequency of FBX4 loss may reflect a necessity for FBX4-mediated degradation of additional substrates (Lee et al., 2006). We also assessed protein levels of FBX4, aB crystallin, and cyclin D1 by immunohistochemical analysis of

Proteolysis of Cyclin D1 by SCFFBX4-aB Crystallin 361

Figure 6. The SCFFBX4-aB Crystallin Ubiquitin Ligase Regulates Cell-Cycle Progression (A) Cell-cycle-dependent binding between FBX4 and cyclin D1. NIH 3T3 cells were released from a G0 (serum starvation) block, and binding between cyclin D1 and FBX4 was assessed by IP-western. (B) Lysates prepared from NIH 3T3 cells transfected with GFP or GSK3b kinase-dead (KD) constructs and treated with MG132 were precipitated with FBX4 antiserum or rabbit anti-mouse IgG and subjected to immunoblot with indicated antibodies. (C) Kinetics of cell-cycle reentry from G0 in NIH 3T3 cells expressing shRNAs against FBX4 or aB crystallin. (D) NIH 3T3 cells expressing shRNAs against FBX4 and aB crystallin were released from a nocodazole (G2/M) block in media containing BrdU, and percentage of cells in S phase was assessed by FACS. (E) Immunoblot (cyclin D1) of NIH 3T3 cells expressing shRNAs against FBX4 and aB crystallin released from a nocodazole (G2/M) block for the indicated time. (F) Knockdown of FBX4 and aB crystallin in 3T3 immortalized D12/2 fibroblasts (right panel); control blot of D1 and Skp1 (left panel). (G) D12/2 cells expressing shRNAs against FBX4 and aB crystallin were released from a nocodazole block, and S phase entry was assessed as in (D).

primary esophageal tumors (n = 36), a cancer in which cyclin D1 is frequently deregulated (Sherr, 1996). In normal esophageal tissue, accumulation of cyclin D1 was mutually exclusive with FBX4 and aB crystallin (Figures S2A and S2B); cyclin D1 was highest in proliferative basal cells. In tumors, aB crystallin and FBX4 levels

were negative or very weak in 64% and 52% of adenocarcinomas and in 54% and 64% of squamous cell carcinomas, respectively (Figures S2A and S2B). A subset of these tumors (24%, 6/25 adenocarcinomas; 54%, 6/11 squamous cell carcinomas) exhibited negative or low aB crystallin or FBX4 protein levels with

Molecular Cell 362

Figure 7. FBX4 and aB Crystallin Levels Are Altered in Human Cancers (A) Quantification of FBX4 and aB crystallin mRNA levels in a tumor versus matching normal tissue in an mRNA array. (B) Expression of aB crystallin, cyclin D1, and FBX4 in normal human mammary epithelial cells (HMEC), nontumorigenic breast MCF10A cells, and tumorigenic breast cancer cell lines (MCF7, MDA-MB-231, and BT474). (C) MCF10A, MCF7, or MCF7 cells infected with a retrovirus encoding aB crystallin were treated with cycloheximide for the indicated intervals, and levels of cyclin D1, aB crystallin, and actin were assessed by immunoblot.

concomitant cyclin D1 overexpression (Figure S2A, arrows, and Figure S2B). In addition, screening of ten tumor-derived cell lines revealed loss of aB crystallin in breast cancer-derived, MCF7, MDA-MB-231, and BT474 cells (Figure 7B). aB crystallin loss in MCF7 cells results from loss of chromosome 11 (http://www.path.cam.ac.uk/wpawefish/ BreastCellLineDescriptions/mcf7.htm). The absence of aB crystallin correlated with decreased cyclin D1 proteolysis and increased D1 levels relative to control breast epithelial cells (Figures 7B and 7C and Figure S3). Reintroduction of aB crystallin into MCF7 cells (Figure S1D) restored the rapid kinetics of cyclin D1 proteolysis (Figure 7C), suggesting that stabilization of cyclin D1 in these cells resulted from loss of aB crystallin. Thus, impairment of SCFFBX4-aB crystallin function is a potential mechanism underlying cyclin D1 overexpression in human cancer. Discussion Cyclin D1 Is a Substrate for the SCFFBX4-aB Crystallin Complex Herein, we describe experiments revealing the following: (1) aB crystallin and FBX4 are components of phospho-T286 cyclin D1-containing protein complexes, (2) aB crystallin is necessary for recognition of phosphorylated cyclin D1 by the SCFFBX4 ubiquitin ligase, (3) the SCFFBX4-aB crystallin complex ubiquitinates phosphorylated cyclin D1 and targets it for proteasomal degradation, and (4) the aB crystallin accessory component is lost in multiple breast cancer cell lines, resulting in cyclin D1 stabilization and overexpression. These data sup-

port the notion that reduced cyclin D1 proteolysis likely contributes to cyclin D1 overexpression in certain malignancies. Several lines of evidence suggest that cyclin D1 is a true substrate for the SCFFBX4-aB crystallin ubiquitin ligase. First, cyclin D1, FBX4, and aB crystallin are present in endogenous complexes in vivo. Second, the interaction between cyclin D1 and the SCFFBX4-aB crystallin complex is dependent on cyclin D1 phosphorylation (a characteristic of SCF-substrate interactions). Third, overexpression of FBX4 and aB crystallin triggers the destruction of wild-type cyclin D1, but not of the phosphorylation-deficient cyclin D1 mutant, D1T286A. Fourth, cyclin D1 is stabilized in vivo by the expression of either a dominant-negative FBX4 or by shRNAs against FBX4 and aB crystallin mRNAs. Finally, purified SCFFBX4-aB crystallin complexes catalyze ubiquitination of cyclin D1 in vitro, providing direct biochemical evidence that cyclin D1 is a bona fide substrate for the SCFFBX4-aB crystallin complex. The SCFSKP2 ligase has also been implicated in the ubiquitination of cyclin D1 (Ganiatsas et al., 2001; Yu et al., 1998) . Knockdown of SKP2 promoted accumulation of cyclin D1(Yu et al., 1998). However, evidence for direct ubiquitination by the SCFSKP2 is absent. Indeed, a physical association between SKP2 and either cyclin D1 or phospho-T286 cyclin D1 has not been reported. The stabilization of cyclin D1 observed in cells expressing antisense oligos against SKP2 likely reflects stabilization of other substrates such as p21Cip1 or p27Kip1 (Bornstein et al., 2003; Carrano et al., 1999), both of which can decrease cyclin D1 proteolysis (Alt et al., 2002; Cheng et al., 1999; Coleman et al., 2003).

Proteolysis of Cyclin D1 by SCFFBX4-aB Crystallin 363

Possible Roles of aB Crystallin within the SCFFBX4-aB Crystallin Complex Our data suggest that aB crystallin acts as a specificity determinant that facilitates FBX4 recognition of Thr286-phosphorylated cyclin D1. It is possible that it is the chaperone-like activity of aB crystallin that provides this function. However, there is no data to suggest that the chaperoning function of aB crystallin depends upon substrate phosphorylation. Thus, we favor an explanation wherein targeting of phosphorylated cyclin D1 to SCFFBX4-aB crystallin represents a distinct function for aB crystallin. Three distinct mechanisms for aB crystallin function in this process can be envisioned. In the first, aB crystallin might directly bind to phospho-Thr286 of cyclin D1 and bridge D1 to SCFFBX4. Such a mechanism is analogous to the function provided by CKS1 in the SCFSKP2-CKS1 complex where CKS1 binds to Thr187-phosphorylated p27Kip1 (Bashir et al., 2004; Bornstein et al., 2003; Ganoth et al., 2001; Hao et al., 2005). In an alternative model, the phosphoacceptor site within SCFFBX4-aB crystallin is cooperatively recognized by both FBX4 and aB crystallin. Finally, it remains formally possible that aB crystallin may not participate in the phosphoacceptor recognition site. Rather, aB crystallin might function as an allosteric regulator of FBX4 by inducing a conformational change of FBX4, thereby increasing the affinity of FBX4 for phosphorylated cyclin D1.

Spatial Control of Cyclin D1 Ubiquitination Previous work suggested that phosphorylation-dependent ubiquitination of cyclin D1 occurs in the cytoplasm (Diehl et al., 1998). Indeed, our analysis revealed that both FBX4 and aB crystallin are exclusively cytoplasmic. Based on the steady-state cytoplasmic localization of FBX4 and the fact that, following a G0 release, cyclin D1 binds to FBX4 during S phase (a cell-cycle phase wherein cyclin D1 is cytoplasmic [Diehl et al., 1998]), we suggest that FBX4-dependent ubiquitination of cyclin D1 is a cytoplasmic event. Although our data suggest that FBX4 is the major E3 ligase, it remains possible that alternative cytoplasmic adaptor mechanisms exist that target cyclin D1 for destruction. Consistent with this, cyclin D1T286A exhibits reduced turnover relative to wild-type cyclin D1 in aB crystalline-deficient MCF7 cells (data not shown). In contrast to cytoplasmic ubiquitination machinery, there may also exist a nuclear, phosphorylationindependent cyclin D1 destruction pathway that is not mediated by the SCFFBX4-aB crystallin complex. Evidence for this pathway comes from the discovery of cyclin D1b, an alternative splice variant of canonical cyclin D1 (Betticher et al., 1995; Howe and Lynas, 2001; Lu et al., 2003; Solomon et al., 2003). Cyclin D1b lacks the last exon of canonical cyclin D1, and therefore Thr286. Since cyclin D1b cannot be phosphorylated at Thr286, it cannot be exported to the cytoplasm. Although the half-life of cyclin D1b is longer than canonical cyclin D1, cyclin D1b degradation remains proteasome dependent (data not shown), suggesting the existence of phosphorylation-independent machinery that earmarks nuclear cyclin D1b for destruction. Phosphorylationindependent destruction of cyclin D1 has been pro-

posed to occur through the activity of a protein called antizyme (Newman et al., 2004). Other SCFFBX4 Substrates FBX4 also targets the telomeric DNA repeat binding protein Pin2/Trf1 for degradation (Lee et al., 2006), potentially contributing to telomere maintenance. In contrast to D1, recognition does not depend upon aB crystallin. At first, it seems somewhat counterintuitive that impairment of FBX4 function could promote both a proliferative signal, via cyclin D1 stabilization, and a growth-inhibitory one by inducing telomere shortening via Pin2/Trf1 stabilization. However, the dominant FBX4 downstream effector substrate in vivo may depend on cellular context. For instance, a cancer cell may have acquired the expression of telomerase or additional factors to evade checkpoint responses induced by the effects of Pin2/ Trf1 stabilization. In addition, stabilization of one FBX4 substrate may compensate for the effect of stabilizing another. Cyclin D1, for example, can induce the maintenance of telomere length, via a telomerase-independent mechanism (Opitz et al., 2001). Additional work should focus on establishing substrate hierarchy. Role of FBX4 and aB Crystallin in Cancer and Cell-Cycle Progression Cyclin D1 is overexpressed in a variety of human tumors that do not exhibit cyclin D1 gene amplification or structural abnormalities of the D1 locus, which suggests that increased cyclin D1 stability is a potential mechanism whereby cyclin D1 overexpression occurs in human cancer. Indeed, mutations of cyclin D1 at Thr286 and Pro287 have been found in human tumors (Benzeno et al., 2006), thereby rendering cyclin D1 more stable. Impairment of SCFFBX4-aB crystallin function may also account for cyclin D1 overexpression in a subset of human tumors. In fact, the human aB crystallin locus is located on chromosome 11q22.3-q23.1, a region frequently deleted in human cancer (Dohner et al., 1999). In a preliminary screen of several human cancers, we have observed that both FBX4 and aB crystallin are downregulated at the protein and mRNA levels, suggesting that SCFFBX4-aB crystallin function might be impaired in human cancer. Our results do not exclude the possibility that variations in FBX4 and aB crystallin levels in the tumors examined are secondary to differences in tumor proliferative fractions. However, reintroduction of aB crystallin in MCF7 cells, a breast cancer cell line that lacks aB crystallin, restored the rate of cyclin D1 proteolysis to that of wild-type. Future studies will by necessity address whether mutations in FBX4 or aB crystallin that specifically abolish SCFFBX4-aB crystallin binding to cyclin D1 occur in human cancer in the absence of significant alteration in cyclin D1 mRNA levels. Future studies with higher power will also address whether the loss of FBX4 and the loss of aB crystallin are mutually exclusive or coordinate events and will address the relation of such losses to cyclin D1 overexpression. Our results suggest that impairment of SCFFBX4-aB crystallin function may contribute to cyclin D1 overexpression in human cancer via decreased ubiquitin-mediated cyclin D1 degradation. While the function of the SCFFBX4-aB crystallin complex in cancer remains speculative, our data reveal a role of

Molecular Cell 364

this E3 ubiquitin ligase as regulator of cell-cycle progression. Knockdown of either FBX4 or aB crystallin reproducibly accelerated G1 phase progression from a G2/M (nocodazole) block, but not from a G0 (serum starvation) block, suggesting that decreased cyclin D1 proteolysis increases the rate of cell-cycle progression. The failure of FBX4 and aB crystallin knockdown to accelerate the G0 to S phase transition likely reflects the fact that the G0 to S transition heavily relies on the induction of cyclin D1 transcription (Matsushime et al., 1991). Indeed, our results are consistent with the observation that aB crystallin2/2 lens epithelial cells exhibit hyperproliferation and genomic instability (Bai et al., 2003). Future efforts are needed to establish the potential tumor suppressive action of both FBX4 and aB crystallin in relation to cyclin D1 as a substrate. Experimental Procedures Cell Culture Conditions and Transfections NIH 3T3, 293T, and U20S cells were maintained in DMEM medium containing glutamine, antibiotics, and 10% FBS (Gemini). NIH 3T3 cell derivatives engineered to overexpress Flag-FBX4 and Flag(DF)FBX4 were generated by transfection using Lipofectamine Plus (Invitrogen) with pcDNA3 vectors encoding the appropriate cDNA. Transfected cell lines were selected and maintained in 400 mg/mL G418. Stable NIH 3T3 cell lines expressing cyclin D1 have been described previously (Alt et al., 2000). Insect Sf9 cells have been described previously (Summers and Smith, 1987). Transient expression of vectors encoding respective cDNAs was achieved by using Lipofectamine Plus according to the instructions from the manufacturer. Metabolic labeling and protein half-life measurements were performed as previously described (Diehl and Sherr, 1997; Rimerman et al., 2000). Gene Silencing pSM2c shRNA vectors targeting mouse FBX4 and mouse aB crystallin were purchased from Open Biosystems CSHL library (Mouse aB crystallin target sequence: TATTAGCTTAATAATCTGGGCC; Mouse FBX4 target sequence: ATTCATCTTGCCATTCATGACGT). NIH 3T3 cells were infected with retroviruses encoding aB crystallin or FBX4 shRNAs; pools and clones of infected cells were selected with puromycin. Knockdown of human FBX4 was achieved by transfection with shRNA vectors (Paddison et al., 2002; Paddison and Hannon, 2002) containing either of the following target sequences: GCCGGTACAGTGTGATTCCACAGATTCAA; GGCATTGAGTGGATT CTTGAAGAAGTGGA. Construction of FBX4 Mutants pcDNA3 Flag-tagged FBX4 was a gift from Michelle Pagano. Flag(DF)FBX4 mutant was generated using the following primers by mutagenesis using the QuikChange mutagenesis kit (Stratagene): forward, 50 -GAGGAGGTGGATGAGAGGGATCTTCCTTCTTGG-30 ; reverse, 50 -CCAAGAAGGAAGATCCCTCTCATCCACCTCCTC-30 . Immunoprecipitation and Immunoblotting NIH 3T3 cells were treated with 10 mM MG132 or 50 mM leucyl-leucylnorleucinal (LLnL) for 4 hr. Cells were lysed in buffer containing 50 mM HEPES (pH 8.0), 150 mM NaCl, 2.5 mM EGTA, 1 mM EDTA, 0.1% Tween 20, protease, and phosphatase inhibitors (1 mM PMSF, 20 U/ml aprotinin, 5 mg/ml leupeptin, 1 mM DTT, 0.4 mM NaF, 10 mM b-glycerophosphate, and 100 nM okadaic acid). Cyclin D1 was precipitated with the mouse monoclonal antibody D1-7213G. Alternatively, FBX4 was precipitated with a rabbit polyclonal antibody generated by immunizing rabbits with FBX4-derived peptides. Immune complexes were captured with protein A Sepharose, and resolved by SDS-PAGE. The following antibodies were used for westerns: cyclin D1 (72-13G), aB crystallin (SPA 223) (Stressgen), HA (12CA5), SKP1 (Transduction Labs), and b-actin (AC15) (Sigma). FBX4 antiserum was generated by immunizing rabbits with peptides

generated based on the N or C terminus of FBX4 (Rockland Immunochemicals, PA). In Vitro Peptide Pull-Down 293T cells transfected as indicated were lysed in Tween 20 buffer plus 1 mM PMSF, 20 U/ml aprotinin, 5 mg/ml leupeptin, 1 mM DTT, 0.4 mM NaF, 10 mM b-glycerophosphate, and 100 nM okadaic acid. Whole-cell extracts were incubated with 100 mg of biotinylated peptides (synthesized by the Baylor College of Medicine Protein Chemistry Core), and bound proteins were captured on avidin agarose beads. Beads were boiled in SDS sample buffer prior to separation by SDS-PAGE. Mass Spectrophotometry Cyclin D1 complexes were separated by SDS-PAGE and visualized with Coomassie Brilliant Blue staining. Excised bands were destained with 50% acetonitrile (v/v)-25 mM NH4HCO3, dehydrated with 100% acetonitrile, dried under vacuum, and incubated with bovine trypsin (1 ug/ml, Sigma) overnight at 37 C. Peptides were extracted with 50% acetonitrile and 5% trifluoroacetic acid, dried under vacuum, and reconstituted in 50% acetonitrile and 0.1% trifluoroacetic acid. Peptide solutions were mixed with an equal volume of a-Cyano-4-hydroxycinnamic acid matrix and spotted onto MALDI plates. Peptides were calibrated with Sequazyme peptide mass standards kit (PE Biosystem) and analyzed by MALDI-TOF mass spectrometry (Voyager DE Pro, Applied Biosystems). Identification of proteins was performed using MS-Fit software (http:// prospector.ucsf.edu/ucsfhtml4.0/msfit.htm). In Vivo Ubiquitination Cells expressing HA-ubiquitin were lysed in buffer containing 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1% NP-40, 1% deoxycholate, 1 mM PMSF, 20 U/ml of aprotinin, 5 mg/ml of leupeptin, 1 mM DTT, 0.4 mM NaF, 5 mM NEM, and 20 mM MG132. Cyclin D1 or Flag-D1 derivatives were precipitated with either the cyclin D1 antibody or with the M2 monoclonal antibody. Proteins were separated by SDSPAGE, and ubiquitinated intermediates were detected by immunoblot. In Vitro Ubiquitination 293T cells were transfected with vectors encoding Flag-FBX4, aB crystallin, Cul1, Skp1, and ROC1. The SCF complexes were purified from the lysates of these cells using M2 agarose (Sigma) and incubated with in vitro-translated 35S-labeled cyclin D1 and unlabeled CDK4 in the presence of recombinant GSK3b (5 U), E1, E2 (UbcH5A), ATP, and ubiquitin for 10 min at 30 C, followed by incubation for another 50 min at 37 C. Proteins were resolved on 10% SDS-PAGE and visualized by autoradiography. Subcellular Fractionation Cells were collected by centrifugation at 300 3 g for 4 min at 4 C and suspended in 400 ml buffer A (10 mM HEPES [pH 7.5], 10 mM KCl, 1 mM DTT, 1 mM PMSF, 20 U/ml aprotinin, 5 mg/ml leupeptin, 0.4 mM NaF, and 10 mM b-glycerophosphate). After incubation on ice for 15 min, 12.5 ml of 10% NP-40 was added and incubated on ice for 10 min. Nuclei were pelleted at 1500 3 g for 5 min, and the supernatant (cytoplasmic extracts) was centrifuged at 13,000 3 g. Nuclei were washed with 1 ml of buffer A twice and suspended in 50 ml of buffer B (20 mM HEPES [pH 7.5], 0.4 M NaCl, 1 mM DTT, 1 mM PMSF, 20 U/ml aprotinin, 5 mg/ml leupeptin, 0.4 mM NaF, and 10 mM b-glycerophosphate). Nuclear extract was collected by centrifugation at 18,000 3 g for 5 min. Immunofluorescence Cells were fixed in 70% methanol, 30% acetone for 15 min at 220 C and blocked for 1 hr in 4% BSA/PBS, followed by incubation in primary antibody (6702, 1:100; affinity-purified 7099, 1:2) for 1 hr at room temperature. Cells were washed in PBS and incubated in secondary FITC-conjugated antibody (1:200) for 30 min at room temperature. Cells were washed with PBS and counterstained with Hoechst (1:500). Tumor versus Normal mRNA Array mRNA array representing the matched malignant/normal tissues was purchased from Biochain Institute (Hayward, CA).

Proteolysis of Cyclin D1 by SCFFBX4-aB Crystallin 365

Hybridizations with cDNA probes specific for FBX4 and aB crystallin were performed according to instructions. mRNA levels were normalized to b-actin. Supplemental Data Supplemental Data include three figures and one table and can be found with this article online at http://www.molecule.org/cgi/ content/full/24/3/355/DC1/. Acknowledgments We thank M. Pagano for the FBX4, FBW2, CUL1, and SKP1 constructs, P. Sicinski for immortalized D12/2 fibroblasts, S. Elledge for Fbx4 shRNA vectors, and M. Lessie for technical assistance. We thank members of the Diehl, Fuchs, Bassing, and Kushner labs for critical insight. This work was supported by grants from the NIH (CA93237 and CA111360), a Leukemia & Lymphoma Scholar award (J.A.D.), P01-CA098101 (A.R.), and the Abramson Cancer Center (J.A.D. and S.Y.F.). Received: May 11, 2006 Revised: July 11, 2006 Accepted: September 12, 2006 Published: November 2, 2006 References Alt, J.R., Cleveland, J.L., Hannink, M., and Diehl, J.A. (2000). Phosphorylation-dependent regulation of cyclin D1 nuclear export and cyclin D1-dependent cellular transformation. Genes Dev. 14, 3102– 3114. Alt, J.R., Gladden, A.B., and Diehl, J.A. (2002). p21(Cip1) promotes cyclin D1 nuclear accumulation via direct inhibition of nuclear export. J. Biol. Chem. 277, 8517–8523. Bai, F., Xi, J.H., Wawrousek, E.F., Fleming, T.P., and Andley, U.P. (2003). Hyperproliferation and p53 status of lens epithelial cells derived from alphaB-crystallin knockout mice. J. Biol. Chem. 278, 36876–36886. Bashir, T., Dorrello, N.V., Amador, V., Guardavaccaro, D., and Pagano, M. (2004). Control of the SCF(Skp2-Cks1) ubiquitin ligase by the APC/C(Cdh1) ubiquitin ligase. Nature 428, 190–193. Benzeno, S., Lu, F., Guo, M., Barbash, O., Zhang, F., Herman, J.G., Klein, P.S., Rustgi, A., and Diehl, J.A. (2006). Identification of mutations that disrupt phosphorylation-dependent nuclear export of cyclin D1. Oncogene, in press. Betticher, D.C., Thatcher, N., Altermatt, H.J., Hoban, P., Ryder, W.D., and Heighway, J. (1995). Alternate splicing produces a novel cyclin D1 transcript. Oncogene 11, 1005–1011. Bornstein, G., Bloom, J., Sitry-Shevah, D., Nakayama, K., Pagano, M., and Hershko, A. (2003). Role of the SCFSkp2 ubiquitin ligase in the degradation of p21Cip1 in S phase. J. Biol. Chem. 278, 25752– 25757.

alphaB-crystallin recruits the F-box protein FBX4 to nuclear SC35 speckles. Eur. J. Biochem. 271, 4195–4203. Diehl, J.A., and Sherr, C.J. (1997). A dominant-negative cyclin D1 mutant prevents nuclear import of cyclin-dependent kinase 4 (CDK4) and its phosphorylation by CDK-activating kinase. Mol. Cell. Biol. 17, 7362–7374. Diehl, J.A., Zindy, F., and Sherr, C.J. (1997). Inhibition of cyclin D1 phosphorylation on threonine-286 prevents its rapid degradation via the ubiquitin-proteasome pathway. Genes Dev. 11, 957–972. Diehl, J.A., Cheng, M., Roussel, M.F., and Sherr, C.J. (1998). Glycogen synthase kinase-3beta regulates cyclin D1 proteolysis and subcellular localization. Genes Dev. 12, 3499–3511. Dohner, H., Stilgenbauer, S., Dohner, K., Bentz, M., and Lichter, P. (1999). Chromosome aberrations in B-cell chronic lymphocytic leukemia: reassessment based on molecular cytogenetic analysis. J. Mol. Med. 77, 266–281. Ganiatsas, S., Dow, R., Thompson, A., Schulman, B., and Germain, D. (2001). A splice variant of Skp2 is retained in the cytoplasm and fails to direct cyclin D1 ubiquitination in the uterine cancer cell line SK-UT. Oncogene 20, 3641–3650. Ganoth, D., Bornstein, G., Ko, T.K., Larsen, B., Tyers, M., Pagano, M., and Hershko, A. (2001). The cell-cycle regulatory protein Cks1 is required for SCF(Skp2)-mediated ubiquitinylation of p27. Nat. Cell Biol. 3, 321–324. Hao, B., Zheng, N., Schulman, B.A., Wu, G., Miller, J.J., Pagano, M., and Pavletich, N.P. (2005). Structural basis of the Cks1-dependent recognition of p27(Kip1) by the SCF(Skp2) ubiquitin ligase. Mol. Cell 20, 9–19. Harper, J.W., Burton, J.L., and Solomon, M.J. (2002). The anaphasepromoting complex: it’s not just for mitosis any more. Genes Dev. 16, 2179–2206. Howe, D., and Lynas, C. (2001). The cyclin D1 alternative transcripts [a] and [b] are expressed in normal and malignant lymphocytes and their relative levels are influenced by the polymorphism at codon 241. Haematologica 86, 563–569. Jin, J., Shirogane, T., Xu, L., Nalepa, G., Qin, J., Elledge, S.J., and Harper, J.W. (2003). SCFbeta-TRCP links Chk1 signaling to degradation of the Cdc25A protein phosphatase. Genes Dev. 17, 3062– 3074. Koepp, D.M., Schaefer, L.K., Ye, X., Keyomarsi, K., Chu, C., Harper, J.W., and Elledge, S.J. (2001). Phosphorylation-dependent ubiquitination of cyclin E by the SCFFbw7 ubiquitin ligase. Science 294, 173–177. Kraft, C., Vodermaier, H.C., Maurer-Stroh, S., Eisenhaber, F., and Peters, J.M. (2005). The WD40 propeller domain of Cdh1 functions as a destruction box receptor for APC/C substrates. Mol. Cell 18, 543–553. Latres, E., Chiaur, D.S., and Pagano, M. (1999). The human F box protein beta-Trcp associates with the Cul1/Skp1 complex and regulates the stability of beta-catenin. Oncogene 18, 849–854.

Burton, J.L., Tsakraklides, V., and Solomon, M.J. (2005). Assembly of an APC-Cdh1-substrate complex is stimulated by engagement of a destruction box. Mol. Cell 18, 533–542.

Lee, T.H., Perrem, K., Harper, J.W., Lu, K.P., and Zhou, X.Z. (2006). The F-box protein FBX4 targets PIN2/TRF1 for ubiquitin-mediated degradation and regulates telomere maintenance. J. Biol. Chem. 281, 759–768.

Carrano, A.C., Eytan, E., Hershko, A., and Pagano, M. (1999). SKP2 is required for ubiquitin-mediated degradation of the CDK inhibitor p27. Nat. Cell Biol. 1, 193–199.

Lu, F., Gladden, A.B., and Diehl, J.A. (2003). An alternatively spliced cyclin D1 isoform, cyclin D1b, is a nuclear oncogene. Cancer Res. 63, 7056–7061.

Cheng, M., Olivier, P., Diehl, J.A., Fero, M., Roussel, M.F., Roberts, J.M., and Sherr, C.J. (1999). The p21(Cip1) and p27(Kip1) CDK ‘inhibitors’ are essential activators of cyclin D-dependent kinases in murine fibroblasts. EMBO J. 18, 1571–1583.

Marshall, C. (1999). How do small GTPase signal transduction pathways regulate cell cycle entry? Curr. Opin. Cell Biol. 11, 732–736.

Coleman, M.L., Marshall, C.J., and Olson, M.F. (2003). Ras promotes p21(Waf1/Cip1) protein stability via a cyclin D1-imposed block in proteasome-mediated degradation. EMBO J. 22, 2036–2046. den Engelsman, J., Keijsers, V., de Jong, W.W., and Boelens, W.C. (2003). The small heat-shock protein alpha B-crystallin promotes FBX4-dependent ubiquitination. J. Biol. Chem. 278, 4699–4704. den Engelsman, J., Bennink, E.J., Doerwald, L., Onnekink, C., Wunderink, L., Andley, U.P., Kato, K., de Jong, W.W., and Boelens, W.C. (2004). Mimicking phosphorylation of the small heat-shock protein

Matsushime, H., Roussel, M.F., Ashmun, R.A., and Sherr, C.J. (1991). Colony-stimulating factor 1 regulates novel cyclins during the G1 phase of the cell cycle. Cell 65, 701–713. Montagnoli, A., Fiore, F., Eytan, E., Carrano, A.C., Draetta, G.F., Hershko, A., and Pagano, M. (1999). Ubiquitination of p27 is regulated by Cdk-dependent phosphorylation and trimeric complex formation. Genes Dev. 13, 1181–1189. Newman, R.M., Mobascher, A., Mangold, U., Koike, C., Diah, S., Schmidt, M., Finley, D., and Zetter, B.R. (2004). Antizyme targets cyclin D1 for degradation. A novel mechanism for cell growth repression. J. Biol. Chem. 279, 41504–41511.

Molecular Cell 366

Ohtsubo, M., and Roberts, J.M. (1993). Cyclin-dependent regulation of G1 in mammalian fibroblasts. Science 259, 1908–1912. Opitz, O.G., Suliman, Y., Hahn, W.C., Harada, H., Blum, H.E., and Rustgi, A.K. (2001). Cyclin D1 overexpression and p53 inactivation immortalize primary oral keratinocytes by a telomerase-independent mechanism. J. Clin. Invest. 108, 725–732. Paddison, P.J., and Hannon, G.J. (2002). RNA interference: the new somatic cell genetics? Cancer Cell 2, 17–23. Paddison, P.J., Caudy, A.A., and Hannon, G.J. (2002). Stable suppression of gene expression by RNAi in mammalian cells. Proc. Natl. Acad. Sci. USA 99, 1443–1448. Pfleger, C.M., Lee, E., and Kirschner, M.W. (2001). Substrate recognition by the Cdc20 and Cdh1 components of the anaphase-promoting complex. Genes Dev. 15, 2396–2407. Quelle, D.E., Ashmun, R.A., Shurtleff, S.A., Kato, J.Y., Bar-Sagi, D., Roussel, M.F., and Sherr, C.J. (1993). Overexpression of mouse D-type cyclins accelerates G1 phase in rodent fibroblasts. Genes Dev. 7, 1559–1571. Resnitzky, D., Gossen, M., Bujard, H., and Reed, S.I. (1994). Acceleration of the G1/S phase transition by expression of cyclins D1 and E with an inducible system. Mol. Cell. Biol. 14, 1669–1679. Rimerman, R.A., Gellert-Randleman, A., and Diehl, J.A. (2000). Wnt1 and MEK1 cooperate to promote cyclin D1 accumulation and cellular transformation. J. Biol. Chem. 275, 14736–14742. Sherr, C.J. (1996). Cancer cell cycles. Science 274, 1672–1677. Skowyra, D., Craig, K.L., Tyers, M., Elledge, S.J., and Harper, J.W. (1997). F-box proteins are receptors that recruit phosphorylated substrates to the SCF ubiquitin-ligase complex. Cell 91, 209–219. Solomon, D.A., Wang, Y., Fox, S.R., Lambeck, T.C., Giesting, S., Lan, Z., Senderowicz, A.M., and Knudsen, E.S. (2003). Cyclin D1 splice variants. Differential effects on localization, RB phosphorylation, and cellular transformation. J. Biol. Chem. 278, 30339–30347. Spruck, C., Strohmaier, H., Watson, M., Smith, A.P., Ryan, A., Krek, T.W., and Reed, S.I. (2001). A CDK-independent function of mammalian Cks1: targeting of SCF(Skp2) to the CDK inhibitor p27Kip1. Mol. Cell 7, 639–650. Summers, M.D., and Smith, G.E. (1987). A manual of methods for baculovirus vectors and insect cell culture procedures. Tex. Agric. Exp. St. Bull. 1555. Yu, Z.K., Gervais, J.L., and Zhang, H. (1998). Human CUL-1 associates with the SKP1/SKP2 complex and regulates p21(CIP1/WAF1) and cyclin D proteins. Proc. Natl. Acad. Sci. USA 95, 11324–11329. Zheng, N., Schulman, B.A., Song, L., Miller, J.J., Jeffrey, P.D., Wang, P., Chu, C., Koepp, D.M., Elledge, S.J., Pagano, M., et al. (2002). Structure of the Cul1-Rbx1-Skp1-F boxSkp2 SCF ubiquitin ligase complex. Nature 416, 703–709.