Oncogene (2008) 27, 1253–1262 & 2008 Nature Publishing Group All rights reserved 0950-9232/08 $30.00 www.nature.com/onc ORIGINAL ARTICLE Human and mouse splice variants: transformingactivity and subcellular localization

C Denicourt1, P Legault1, F-AC McNabb and E Rassart

Laboratoire de Biologie Mole´culaire, De´partement des Sciences Biologiques, Universite´ du Que´bec a` Montre´al, Que´bec, Canada

We have previously reported the identification of a novel characterized Gris1 as a novel common integration site 17 kDa truncated isoform of the cyclin D2 activated in in 13% of Graffi murine retrovirus-induced leukemias 13% of the leukemias induced by the Graffi murine (Denicourt et al., 2002). This novel locus was located on leukemia retrovirus. Retroviral integration in the Gris1 mouse chromosome 6, 75–85 kbp upstream of the cyclin locus causes an alternative splicingof the mouse cyclin D2 D2 gene. Integrations in Gris1 were shown to activate gene and expression of a truncated of 159 amino the expression of the 6.5 kb major transcript of the acids that is detected at high levels in the Gris1 tumors and cyclin D2 gene (CCND2)and also a smaller 1.1 kb also in normal mouse tissues mainly the brain and ovaries. transcript that we have shown to represent an alter- A truncated form of the cyclin D2 was also found in native transcript encoding a 17 kDa truncated cyclin D2 human. We show here that both mouse- and human- protein. The protein is detected at high level in the Gris1 truncated cyclin D2 are able to transform primary mouse tumors and also in smaller quantities in normal tissue embryo fibroblasts (MEF) when co-expressed with an mainly in the brain and ovaries (Denicourt et al., 2002). activated Ras protein. The truncated cyclin D2 localizes We also have reported the existence of several expressed only to the cytoplasm of transfected cells. It has retained sequence tags (ESTs)in the databases encoding a similar the ability to interact with cyclin-dependent truncated cyclin D2 in humans. Moreover, such a (CDKs), although it is a poor catalyst of pRb phosphor- transcript was also reported in Xenopus laevis cloned ylation. Interestingly, the presence of a similar, alterna- from an ovary cDNA library (Taı¨ eb and Jessus, 1996). tively spliced cyclin D2 mRNA was also detected in some Cyclin D2 is a G1 cyclin that belongs to the family of human brain tumors. three closely related D-type cyclins, namely, D1, D2 and Oncogene (2008) 27, 1253–1262; doi:10.1038/sj.onc.1210750; D3. These cyclins have been identified as positive published online 17 September 2007 regulators of the and have a well-established role in the progression through the of the cell Keywords: truncated cyclin D2; transformation; splice cycle. D-type cyclins assemble with their cyclin-depen- variant dent (CDK)partners CDK4 and CDK6 to phosphorylate target such as pRb (reviewed in Sherr and Roberts, 1999; Ortega et al., 2002). Mitogenic signaling ultimately leads to the upregulated expression Introduction of D-type cyclins as well as their kinase-associated activity. Cyclin D2 is a well-defined human proto-oncogene Slow-transforming murine retroviruses can induce and, when illegitimately activated or overexpressed, it leukemia in their host by insertional mutation of cellular can contribute to cellular transformation (reviewed in proto-oncogenes or tumor suppressor genes. Following Malumbres and Barbacid, 2001). Indeed, deregulated the retroviral integration, cellular genes may become expression of the cyclin D2 gene has been reported for aberrantly expressed through activation by viral pro- the first time in murine leukemia virus-induced T-cell moter or enhancer sequences. The abnormal expression leukemias by the identification of the common viral inte- of a certain proto-oncogene may then provide a growth gration site vin1 (Tremblay et al., 1992; Hanna et al., advantage to the target cell and contribute to malignant 1993). In collaboration with an activated Ras (H-RasV12), transformation. During the past years an important overexpressed cyclin D2 can transform primary number of critical cancer genes have been identified in rat embryo fibroblasts (REF)(Kerkhoff and Ziff, 1995). these leukemias by proviral tagging. We have recently In human, overexpression of cyclin D2 is associated with different types of lymphoid malignancy, testicular Correspondence: Dr E Rassart, De´ partement des Sciences Biologiques, carcinoma and ovarian cancer (Sicinski et al., 1996; Universite´ du Que´ bec a` Montre´ al, Case Postale 8888 Succ. Centre- Bartkova et al., 1999; Teramoto et al., 1999). ville, Montre´ al, Canada H3C-3P8. In this article, we present evidence that the over- E-mail: [email protected] or [email protected] 1These authors contributed equally to this work. expression of the truncated cyclin D2 can transform Received 4 June 2007; revised 25 July 2007; accepted 26 July 2007; primary MEF in conjunction with activated Ras. We published online 17 September 2007 also present evidence for a human homolog transcript of Characterization of mouse and human truncated cyclin D2 C Denicourt et al 1254 a truncated cyclin D2 that is overexpressed in some types of primary human brain tumors. As opposed to 12 3 the subcellular localization of the cyclin D2 protein, this truncated cyclin D2 was never observed in the nucleus of NIH/3T3 fibroblasts but was predominantly localized to the cytoplasm. Despite the fact that this truncated cyclin represents half of the full-length cyclin D2, this isoform still retains the ability to interact with CDK4, but the 45 6 complex is unable to phosphorylate its target pRb.

Results

Ectopic expression of the truncated cyclin D2 and activated Ras causes transformation of primary MEF Since the truncated cyclin D2 transcript was found over- expressed in the leukemias induced by Graffi retrovirus, b we hypothesized that the protein could probably function as an oncogene. MEF were transfected with a truncated cyclin D2 expressing construct and with an activated H-Ras to determine if it was able to trans- form in vitro. As controls, MEF were also transfected with H-Ras alone, H-Ras þ c-Myc and H-Ras þ cyclin D2. The truncated cyclin D2 was not able to induce transformation alone or with c-Myc (not shown and Figure 1a). Very few foci were obtained with H-Ras alone and no foci at all were seen in the mock-trans- fected MEF. As expected, the co-transfection of the cyclin D2 in combination with an activated Ras gave some foci. It has been shown that cyclin D2 is able to trans- form primary REF in co-expression with an activated Ras (Kerkhoff and Ziff, 1995). In that study, the oncogenic activity of cyclin D2 and H-Ras was much lower than the combination of c-Myc and H-Ras (Kerkhoff and Ziff, 1995). However, when the MEF Figure 1 Transformation properties of the truncated cyclin D2 on were co-transfected with an activated Ras and a primary mouse embryo fibroblasts (MEF). (a)For the focus truncated cyclin D2 expression vector, they presented formation assay, primary MEF were mock transfected (1), a large number of foci at a density comparable to that transfected with Ras EJ 6.6 construct (2), truncated cyclin D2 obtained with H-Ras þ c-Myc (Table 1). The density of and pSV2-Myc constructs (3), cyclin D2 and Ras EJ 6.6 constructs (4), truncated cyclin D2 and Ras EJ 6.6 constructs (5), pSV2-Myc foci induced with the truncated cyclin D2 was much and Ras EJ 6.6 constructs (6). Fourteen days after transfection, higher than that obtained with the full-length cyclin D2 cells were fixed with methanol and stained with methylene blue. suggesting that the truncated cyclin D2 is a more potent (b)MEF transfected with the truncated cyclin D2 and activated inducer of transformation. Also, the size of each foci Ras were also treated with G418 for selection of stable was larger with the truncated cyclin D2. Similar transfectants. The colonies showing a transformed phenotype were isolated and replated for observations. Upper panel (magnification transformation results were obtained with the human  40)shows typical foci obtained with truncated cyclin D2 and truncated cyclin D2 (results not shown). In contrast, the activated Ras. Arrows in lower panel (magnification  100) combination of the normal and the truncated cyclins D2 indicate transformed multinucleated cells. did not produce foci (results not shown). MEF transfected with the truncated cyclin D2 and activated Ras were also treated with G418 for selection of stable transfectants. The colonies showing a trans- with the combination of cyclin D2 and activated Ras formed phenotype were selected and replated. These stable transfectants (not shown; Kerkhoff and Ziff, cells were dependent on the presence of serum for their 1995). The majority of the truncated cyclin D2 and proliferation and showed no contact inhibition and H-Ras stable transfectants were attached very weakly to continued to proliferate even when confluency was the plates and only stable transfectants expressing low reached (Figure 1b). These cells were also morphologi- level of the truncated cyclin D2 were obtained (not cally different from the primary MEF as they were rod shown). It is possible that high levels of expression of the shaped and highly refractile. An important number of truncated cyclin D2 in transformed MEF interfere with cells presenting multinucleation were also observed cell adhesion. This property was also observed in the (Figure 1b, lower panel). Similar results were obtained cyclin D2 transformed REF (Kerkhoff and Ziff, 1995).

Oncogene Characterization of mouse and human truncated cyclin D2 C Denicourt et al 1255 Truncated cyclin D2 expression is detected in human to normal brain tissue, an increased expression of the brain tissues truncated human cyclin D2 was observed in certain We previously reported the identification of human tumor types (Figure 2). Oligodendrogliomas, gliomas ESTs encoding the homolog of the murine truncated and glioblastomas multiforme (GBM)showed several cyclin D2 (Denicourt et al., 2002). Interestingly, these fold increase in expression. In contrast, the expression ESTs were isolated from human brain tumor cDNA levels in meningiomas, schwannomas and ependymomas libraries. Also, in mouse, the truncated cyclin D2 is remained as in normal brain tissue. The variation of mainly expressed in the brain and ovaries (Denicourt expression observed in the GBMs could be due to some et al., 2002). To assess if the truncated cyclin D2 difference in the degree of malignancy in each sample or expression could be detected in primary brain tumors, different form of tumor. GBMs can be classified as we screened the RNA from a panel of different types of primary or secondary and there are evidences that brain tumors by semi-quantitative RT–PCR. Compared primary and secondary glioblastomas are distinct diseases that evolve through distinct genetic pathways Table 1 Transformation assays performed in this study and may require different therapeutic approaches (Watanabe et al., 1996). No of transformed colonies/dish (8 mg of DNA) The expression of the normal cyclin D2 was also Expressed protein Experiment 1 Experiment 2 Experiment 3 measured. No correlations were found between the expression levels of the two cyclins in the tumors nor Mock 0 0 0 between the expression levels of the full-length cyclin D2 H-Ras 3 6 11 and the different tumor types. c-Myc+mD2 Trc 0 0 0 H-Ras+mD2 34 49 55 H-Ras+mD2 Trc 92 167 195 Truncated cyclin D2 displays a unique cytoplasmic H-Ras+c-myc 89 156 201 localization Mock 0 0 To determine the subcellular localization of both the H-Ras 3 2 human and mouse truncated cyclin D2, we generated H-Ras+hD2 31 37 constructs containing, respectively, the truncated H-Ras+hD2 Trc 46 65 and the full-length cyclin D2 coding sequences fused H-Ras+c-Myc 52 87 to the EGFP or Myc epitope at the N-terminus. Mouse embryo fibroblast cells were transfected with the above-indicated A mouse cyclin D2 mutant (D2D157–289)was generated plasmids as described in Materials and methods. Numbers indicate by the deletion of amino acids 157–289 (Figure 3a). The the number of foci counted at 21 days post-transfection. constructs were transfected into NIH/3T3 cells and the

12

10

8

6 fold increase 4

2

0

NORMAL GBM 1573GBM 1595GBM 1673GBM 1725GBM 1726GLIO 1282GLIO 1485GLIO 1574 MEN 1711MEN 1772MEN 1801 OLIGOOLIGO 1376 OLIGO 1629 EPEN1630 SCHW1627 SCHW 1803 1802 tumor type Figure 2 Expression of the truncated cyclin D2 in human brain tumors. Semi-quantitative RT–PCR analysis of the human truncated cyclin D2 in different brain tissues: normal (normal brain), GBM (glioblastoma multiforme), glio (mixed glioma), oligo (oligodendroglioma), epen (ependymoma), schw (schwannoma), men (meningioma). Hypoxanthine-guanine phosphoribosyltransfer- ase (HPRT)was used as a control for the integrity of RNA samples. The fold increase indicates the relative level of the truncated human cyclin D2 amplification product over the HPRT internal control after normalization to the control sample (normal brain).

Oncogene Characterization of mouse and human truncated cyclin D2 C Denicourt et al 1256 56 137 156 289 59 D2 N EWELVVLGKLKWNLAAVTPH C LxCxE Cyclin Box 56 137 156 59 D2Trc N VMPPSLLLTLPFPITLRPPH C LxCxE Cyclin Box 56 137 156 59 D2 157-289 N EWELVVLGKLKWNLAAVTPH C LxCxE Cyclin Box

pEGFP-C2 mD2

mD2Trc mD2 157-289

hD2 hD2Trc

Figure 3 Cellular localization of human, mouse truncated and mutant cyclin D2. (a)Schematic representation of the normal, truncated and mutant mouse cyclin D2. The first 136 amino acids are identical. The letters in bold indicate conserved amino acids in the new C-terminal region of the cyclin D2Trc. (b)These three constructs were fused to the C-terminus of enhanced GFP using the pEGFP-C2 vector (mD2, mD2Trc, mD2D157–289). Similar constructions were made with the human normal (hD2) and truncated (hD2Trc)cyclin D2. The resulting constructs were transfected into NIH/3T3 cells, and the cells were examined by confocal microscopy.

localization of the proteins was examined by fluores- marker of the Golgi complex (BODIPY-Tr ceramids) cence confocal microscopy. As shown in Figure 3b, did not show significant co-localization with the EGFP alone shows a diffused generalized localization. EGFP-truncated cyclin D2 (not shown). Treatment of EGFP-cyclin D2 protein was localized to the nucleus as transfected cells with leptomycin B revealed that the expected. The mouse truncated cyclin D2, on the con- truncated cyclin D2 was not imported to the nucleus trary, localizes only to cytoplasmic region. The mutant (Supplementary data). Cells transfected with Myc- D2D157–289 who only differs from the D2Trc in the tagged truncated cyclin D2 construct followed by last 20 amino acids of the C-terminal end is observed in immunocytochemistry with a Myc epitope antibody the nucleus. The human homolog also shows the showed the same cytoplasmic distribution suggesting same cellular distribution (Figure 3b). Markers of the that the fused EGFP moiety does not alter the early and late endosomes ( 5a and Rab7)as well as a localization of the fusion protein (not shown).

Oncogene Characterization of mouse and human truncated cyclin D2 C Denicourt et al 1257 Truncated cyclin D2 retains the ability to interact with absence of IL-3. To arrest the cells in G0/G1, IL-3 was CDK4 in NIH/3T3 fibroblasts and forms an inactive removed from the culture medium for 16 h. After, IL-3 complex was reintroduced in the medium to induce re-entry in the In fibroblasts, CDK4 appears to be the most prominent cell cycle. Samples of cells were collected at different partner of cyclin D2 (Matsushime et al., 1992; Xiong time points and the expression of the cyclin D2 et al., 1992). To test whether the truncated cyclin D2 can transcripts was analysed by northern blot. Figure 5a also interact with CDK4, NIH/3T3 fibroblasts were shows that the truncated cyclin D2 transcript declined transfected with the Myc-tagged truncated and full- rapidly in the absence of IL-3. When the IL-3-deprived length cyclin D2 (positive interaction control)con- cells were retreated back with IL-3, the 1.1 kb transcript structs. Cell lysates were immunoprecipitated with an was rapidly induced within 3 h. We also observed the anti-CDK4 antibody. Western blot analysis with an same kinetic for the cyclin D2 6.5 kb transcript. anti-Myc epitope antibody showed that both cyclin D2 The presence of the proteins was also analysed and the and truncated cyclin D2 could be co-immunoprecipi- profiles of both the 17 kDa truncated and 32 kDa cyclin tated in complex with CDK4 (Figure 4a). To evaluate if D2 proteins coincide with the observed induction of the the truncated cyclin D2 is able to form an active mRNAs (Figure 5b). These results suggests that the complex with CDK4 or CDK6, we performed immu- induction of the truncated cyclin D2 synthesis is noprecipitation kinase assays using pRb C-terminus controlled in the same fashion as the induction of the protein as a substrate. Myc-tagged cyclin D2 and cyclin D2 protein, both expression being dependent on truncated cyclin D2 constructs were transfected in the presence of mitogenic factors. NIH/3T3 cells and assayed for kinase activity after immunoprecipitation with Myc epitope antibody. Inter- estingly, we were not able to detect any activity above background for the truncated cyclin D2 (Figure 4b). Discussion This suggests that the truncated cyclin D2 may not form complex with CDK under physiologic conditions or that The truncated cyclin D2 was initially identified as a the D2Trc–CDK4 complex that we detected may have proto-oncogene that becomes activated upon retroviral activity toward another substrate. Because phosphor- insertion of the Graffi murine leukemia virus (Denicourt ylation of CDK4 by CAK (CDK-activating kinase)is et al., 2002). The results presented in this study also required for the activation of the assembled complex support the notion that the truncated cyclin D2 may prior to the of pRb (Kato et al., 1994), play an important role in tumorigenesis. Indeed, we we verified if the assembled D2Trc–CDK4 complex was have shown that ectopic expression of the truncated phosphorylated by CAK. Myc-tagged cyclin D2 and cyclin D2, with an activated Ras, is able to induce a D2Trc constructs were transfected in NIH/3T3 cells, transformed phenotype to primary MEF. Moreover, the isolated by immunoprecipitation with Myc epitope truncated cyclin D2 seems to be a more potent oncogene antibodies, and the immunoprecipitates were suspended than the full-length cyclin D2 since the number of foci in a kinase reaction buffer containing recombinant CAK was much more important and comparable to that (CDK7//MAT1)and [ g32P]ATP. CDK4 asso- observed with the co-expression of c-Myc and activated ciated with full-length cyclin D2 was efficiently phos- Ras (Figure 1a). phorylated by CAK, but the CDK4 associated to D2Trc Cells derived from transformed foci after selection for was a very poor substrate (Figure 4c, upper panel). The the expression of the truncated cyclin D2 showed a high inability of CAK to phosphorylate the CDK4–D2Trc percentage of multinucleation, which is a characteristic complex is not due to a decreased association of CDK4 of abnormal . These cells also presented loss of with D2Trc compared to the full-length cyclin D2 since contact inhibition and grew as aggregates under equivalent amounts of CDK4 were assembled with the subconfluent conditions suggesting that the ability of two cyclin proteins, as confirmed by blotting of the D2 the truncated cyclin D2 to transform primary MEF may immunoprecipitates with anti-CDK4 antibodies be due to alterations in cell adhesion (Figure 1b). It was (Figure 4c, lower panel). Moreover, in vitro kinase assay suggested that primary REF transformed by both using the total cellular lysate as substrate was also activated Ras and full-length cyclin D2 possibly performed and no significant phosphorylation from the presented an altered expression of cell adhesion proteins D2Trc–CDK4 complex was observed (results not since these cells attached very weakly to tissue culture shown). plate and that this property necessitated their passage on collagen-coated plates (Kerkhoff and Ziff, 1995). We have subcloned cell lines from individual foci of primary Expression of the truncated cyclin D2 in 32Dcl3 cells MEF and found that the cyclin D2 was poorly In the hematopoietic cell line 32Dcl3, cyclin D2 is expressed. This is probably due to a selection against expressed in proliferating cells and its expression is the establishment of transformed cell lines expressing dependent on interleukin-3 (IL-3). The expression high levels of the truncated cyclin D2. It is also possible declines rapidly in the absence of this growth factor that the high expression of the truncated cyclin D2 (Ando et al., 1993). To examine if the mouse truncated completely inhibits the adhesion of these cells. Consis- cyclin D2 presents the same profile of expression in this tent to our observations, rodent fibroblasts were found cell line, we analysed its expression in the presence and to have a strong selection against the establishment of

Oncogene Characterization of mouse and human truncated cyclin D2 C Denicourt et al 1258

rc IP α CDK4 yc WB α myc NIH-3T3pCMV-m myc-D2Trcmyc-D2 Myc Myc-D2 Myc-D2T

αMyc IP 32 PO4GST-RB Kinase

34 KDa myc-D2 GST-RB Coomassie 17 KDa myc-D2trc

Lysate: αCDK4 blot CDK4

Myc-cyclin D2 αMyc blot Myc-Truncated cyclin D2

αMyc IP Myc-D2 Myc-D2Trc

CDK4

Myc-D2 Myc-D2Trc αMyc IP CDK4 CDK4

CDK4

Figure 4 Analysis of the Myc-tagged truncated cyclin D2 binding with cyclin-dependent kinase 4 (CDK4)and phosphorylation of pRb. (a)For co-immunoprecipitations, NIH/3T3 mock transfected (NIH/3T3),transfected with the pCMV-Myc vector, the Myc- truncated cyclin D2 construct (Myc-D2Trc)and the Myc-cyclin D2 construct (Myc-D2)were used. Immunoprecipitation (IP)was performed with anti-CDK4 antibodies and 1 mg of protein from the whole-cell lysate. The upper panel shows the precipitated Myc- tagged truncated cyclin D2 and the Myc-tagged cyclin D2 detected by western blot analysis with a Myc epitope antibody. The lower panel shows the endogenous expression levels of CDK4 in the different transfected cells used for IP. (b)For kinase assays, NIH/3T3 cells were transfected with pCMV-Myc vector, Myc-cyclin D2 and Myc-truncated cyclin D2. Lysates were prepared and subjected to immune precipitation using anti-Myc antibody. Precipitates were tested for in vitro kinase activity using glutathione S- (GST)-Rb C0-terminus as substrate. Kinase reactions were analysed by SDS–PAGE followed by exposure of the gel on X-ray film (top panel). The gel was stained with Coomassie Brilliant Blue to visualize the substrate (GST-Rb) in each lane. Whole-cell lysates corresponding to each lane were analysed by immunoblot with anti-CDK4 and c-Myc antibody. (c) In vitro CAK phosphorylation of cyclin D2–CDK4 complexes. NIH/3T3 cell lysates containing CDK4 assembled with either Myc-cyclin D2 or Myc-D2Trc were precipitated with anti-Myc antibodies. Suspended immune complexes were used as substrates for recombinant CAK (CDK7/cyclin H/MAT1 produced in Sf21 cells), and 32P-labeled complexes were denaturated and separated on polyacrylamide gels. The position of phosphorylated CDK4 is indicated on the left. Autoradiographic exposure time was 3 h (top panel). A portion of each cell lysate used in the CAK assay was precipitated with anti-Myc antibodies, separated on gel, blotted to PVDF membrane, probed with anti-CDK4 antibodies and visualized by enhanced chemiluminescence.

cell lines that overexpress D-type cyclins (Quelle et al., cycle progression by regulating the expression and 1993). Integrin-mediated cell adhesion to the extracel- activities of D-type cyclins (Assoian, 1997; Schwartz lular matrix is required for normal cell growth. It was and Assoian, 2001). Roovers et al. (1999)have demon- shown that integrin signaling through focal adhesion strated that integrin-mediated cell adhesion is required kinase (FAK)plays a role in the regulation of the cell for sustained ERK activation and induction of cyclin

Oncogene Characterization of mouse and human truncated cyclin D2 C Denicourt et al 1259 +Il-3: 2412 10 8 3 1 0 However, in our study, the truncated cyclin D2 was never observed in the cell nucleus of NIH/3T3 cells grown at different density. This was confirmed by leptomycin B treatment of cells transfected with an 6 kb EGFP-truncated cyclin D2 construct (Supplementary data). It was shown that a T156A mutant, which retains its ability to interact with CDK4, localizes uniquely to the cytoplasm of NIH/3T3 cells (Diehl and 1.1 kb Sherr, 1997). On the cyclin D2 protein, the T156 residue of cyclin D1 corresponds to T154. This residue is not conserved on the truncated cyclin D2 and this could explain its cytoplasmic localization. This observation is +Il-3: 2412 10 8 3 1 0 also supported by the nuclear localization of mutant D2D157–289 who still retains the T154 residue. It is interesting to note that mutant D2D157–289 that Cyclin D2 contains the first 156 amino acids of the normal cyclin D2 remains to the nucleus albeit the D2Trc is cytoplasmic. This indicates that the last 20 amino acids Cyclin D2 trc are responsible of the localization. We have also demonstrated that the truncated cyclin D2 can interact with CDK4 when expressed ectopically in NIH/3T3 cells. The cyclins present a globular GAPDH structure containing two sequential 90 amino acid repeats, known as the cyclin folds, each consisting of Figure 5 Expression analysis of the truncated cyclin D2 in 32Dcl3 five super imposable a-helical bundles (Jeffrey et al., cells. (a)Northern blot analysis of the full-length cyclin D2 (6.5 kb 1995; Kim et al., 1996; Russo et al., 1996; Andersen transcript upper panel)and the truncated cyclin D2 (1.1 kb et al., 1997). The two cyclin folds are connected via a transcript lower panel)in 32Dcl3 cells. Cells were starved of interleukin-3 (IL-3)for 16 h (0)and restimulated by adding back short linker peptide of 10–15 amino acids. The first IL-3. Samples of cells were removed at 1, 3, 8, 10, 12 and 24 h after cyclin fold corresponds to what is defined as the cyclin IL-3 stimulation. RNA (10 mg)was loaded in each lane. The box that makes contact with the interacting CDK. The exposure period for the lower panel was twice as much as that for second repeat on the cyclin does not contribute to the the top panel. (b)The 32Dcl3 cells were treated as in ( a)and samples of cells were analysed by western blot with specific cyclin–CDK interface. The truncated cyclin D2 contains antibodies against the cyclin D2 (top panel), the truncated cyclin an almost intact cyclin box that could be responsible D2 (middle panel)or GAPDH (bottom panel).Thirty micrograms and sufficient for the interaction with CDK4. We of protein was loaded per lane. examined if the truncated cyclin D2–CDK4 complex was catalytically active. It appears that the complex was D1 expression by growth factors. Also, ectopic over- unable to phosphorylate recombinant pRb protein expression of cyclin D1 induced anchorage-independent in vitro (Figure 4b). This lack of kinase activity is due to cell cycle progression of NIH/3T3 and Rat1 cells the inability of CAK to recognize and phosphorylate the (Schulze et al., 1996; Zhu et al., 1996; Resnitzky, D2Trc–CDK4 complex, which thus remains catalyti- 1997). These observations suggest that the truncated cally inactive (Figure 4c). binding must induce cyclin D2 is able to influence cellular adhesion. a conformational change in the structure of CDK4 to We also showed that the truncated cyclin D2 is make it available for phosphorylation. The fact that expressed in a subset of primary brain tumors (oligo- CAK could not phosphorylate CDK4 when complexed dendrogliomas, mixed gliomas and glioblastomas multi- to the D2Trc suggests that the truncated form could not forme)supporting again its role in tumorigenesis. It is induce the conformational change but instead give rise interesting to note that the truncated cyclin D2 is to abortive complexes that are no longer recognized by abundant in the most aggressive brain tumor types. This CAK. The integrity of Thr-156 on cyclin D1 is required overexpression does not seem to be correlated with the for CAK activity on the CDK4–cyclin complex (Diehl expression of the normal cyclin D2 (data not shown)but and Sherr, 1997). The loss of Thr-154 on the D2Trc could be related with the degree of malignancy of these could be responsible for the inability of the CDK4 specific tumors. Buschges et al. (1999)also reported that complex to be phosphorylated. Interestingly, a splice the expression of normal cyclin D2 was not statistically variant of cyclin D1 (cyclin D1b)harboring a different significant between different tumor grades. We have C0-terminus and lacking T286 was found to be a potent already shown that the truncated cyclin D2 is present in nuclear oncogene compared to the normal cyclin D1 (Lu the brain of normal mouse (Denicourt et al., 2002). et al., 2003). This isoform has retained its ability to bind We have shown that the truncated cyclin D2 localizes CDK4, but was unable to catalyse efficiently pRb phos- to a distinct cytoplasmic substructure in NIH/3T3 cells. phorylation and inactivation (Solomon et al., 2003). D-types cyclins are known to accumulate in the cell This suggests that critical residues in the C-terminus nucleus during G1 phase and disappear from the nuclei of cyclin D proteins are important for substrate of cells undergoing DNA synthesis (Baldin et al., 1993). specificity. Despite its inability to phosphorylate Rb,

Oncogene Characterization of mouse and human truncated cyclin D2 C Denicourt et al 1260 the truncated cyclin D2 isoform demonstrated enhanced Alan Friedman (Johns Hopkins University, Baltimore, MD, transforming activity compared with cyclin D2. The USA). These cells were maintained in RPMI 1640 medium ability of the truncated cyclin D2 to promote cellular (GIBCO, Burlington, ON, Canada)supplemented with 10% transformation is probably attributed to its constitu- of heat-inactivated fetal calf serum (GIBCO)and 10% of tively cytoplasmic localization. Truncated cyclin D2 WEHI-3B cells conditioned medium (WEHI–CM)as a source of IL-3. NIH/3T3 murine fibroblasts (ATCC, Manassas, VA, could sequester p27 (a cyclin D/CDK inhibitor)in the USA)were cultured in Dulbecco’s modified Eagle’s medium cytoplasm where it is degraded by the cytoplasmic (GIBCO)supplemented with 10% of calf serum (GIBCO).All ligase complex KPC, this step being essential transfections were made with Polyfect regent (QIAGEN, for the progression of the cell cycle into (Susaki Mississauga, ON, Canada)according to the manufacturer’s et al., 2007). An upregulated cell cycle caused by the protocol. presence of the truncated cyclin D2 in the cytoplasm could be the cause of the enhanced transformation in Mouse embryo fibroblasts transformation assay presence of an activated Ras. Primary fibroblasts were isolated from 15 days BALB/c mouse Expression of D-type cyclins is dependent on con- embryos. The cells were transfected with 4 mg of the pCMV- tinuous mitogenic stimulation. In the hematopoietic cell D2Trc construct or the pCMV-D2 construct made by cloning line 32Dcl3, cyclin D2 is expressed in proliferating cells the complete coding sequence of the truncated cyclin D2 or and this expression is dependent on the presence of IL-3. cyclin D2 obtained by PCR. The fragments were cloned in the pRcCMV vector (Invitrogen, Burlington, ON, Canada). The It was also shown in the same cell line that the human truncated cyclin D2, obtained by PCR, and normal expression and the protein levels of cyclin D2 declined cyclin D2, Image clone ID: 4133621 (ATCC), were cloned in rapidly upon withdrawal of IL-3 (Ando et al., 1993). To the pCMV5 vector. The cells were also transfected with 4 mgof investigate if the expression of the truncated cyclin D2 the RAS EJ 6.6 construct (Tabin et al., 1982)and the pSV2- was also regulated in the same fashion in 32Dcl3 cells, Myc construct. Transfections were made with Polyfect we have reproduced some of the experiments by Ando (QIAGEN)according to the manufacturer’s instructions. et al. (1993). Our results show that the truncated cyclin D2 expression and protein stability are dependent on the Primary human brain tumors presence of mitogenic stimuli (IL-3)as already shown Samples of primary human brain tumors were obtained from for the cyclin D2. This suggests that the expression of The Arthur and Sonia Labatt Brain Tumour Research Center both transcripts (6.5 and 1.1 kb)of the cyclin D2 gene is (Toronto, Canada). controlled via the same regulatory elements, which are activated by growth factors. Indeed, the two transcripts RNA extraction and northern blot analysis most likely depend on the same promoter as they share Total RNA from frozen human brain tumor tissue or 32Dcl3 0 cells was extracted with the TRIzol reagent (Invitrogen) the same 5 end and the first two exons (Denicourt et al., according to the manufacturer’s instructions. 2002). Site-specific phosphorylation by GSK-3B of cyclins D1 and D2 on a single threonine residue Reverse transcription–PCR (T286)is critical for maintaining their rapid turnover Total RNA (2 mg)was reverse transcribed into cDNA by using rates throughout the cell cycle and also upon an oligo-dT primer (10 mM), dNTP mix (5 mM each dNTP), signaling withdrawal (Diehl and Sherr, 1997; Diehl 10 Â reverse transcription buffer (500 mM Tris–HCl pH 8.3, et al., 1998). Although the truncated cyclin D2 does not 250 mM KCl, 50 mM MgCl2 and 50 mM DTT), 10 U RNase harbor this T286 residue, our results showed that the inhibitor (GE Healthcare, Baie d’Urfe, Quebec, Canada)and protein level declined rapidly in the absence of IL-3. 4 U of Omniscript Reverse Transcriptase (QIAGEN). The mix Also, we have demonstrated that the stability of the was incubated at 37 1C for 60 min. PCR was performed in truncated cyclin D2 was dependent on proteasomal 100 ml with 5 ml of the cDNA template, 0.2 mM dNTP, 2 mM degradation (data not shown). It is possible that the MgCl2, 0.25 mM of each primer and 2.5 U of Taq DNA 1 turnover of the truncated cyclin D2 is dependent on the polymerase. The cycling conditions were 94 C for 3 min, followed by 94 1C for 30 s, 60 1C for 30 s and 72 1C for 1 min phosphorylation of another residue. for 25 cycles and a final extension at 72 1C for 10 min. Taken together, these data demonstrate the activity of Hypoxanthine-guanine phosphoribosyltransferase (HPRT) the truncated cyclin D2 in cellular transformation and a primers were used as an internal control under the same role in tumor formation. Because of its cytoplasmic conditions. The primer sequences were as follows: normal localization, we propose that the truncated cyclin D2 cyclin D2, cyc1 (50-GGCTGGGGTCCCGACTCCGAAG-30) may contribute to transformation by sequestering p27 in and cyc2 (50-GCAGCTCAGTCAGGGCATCACA-30), trun- the cytoplasm, causing its degradation and, therefore, cated cyclin D2, cyc1 (50-GGCTGGGGTCCCGACTCCGA 0 0 0 upregulating the cell cycle. It will, therefore, be impor- AG-3 )and tcyc (5 -TGGCCAGGGGAAAGCGGGAATC-3 ). The HPRT primers were HPS1 (50-GTGATGAAGGAGATG tant to identify the protein partners of this truncated 0 0 cyclin D2. GGAGGCC-3 )and SPH1 (5 -CTTCGTGGGGTCCTTTTC ACC-30). The resulting products were separated on a 1% agarose gel and the individual bands were quantified using the volume Rect tool from the Quantity one 4.4.0 software Materials and methods (Bio-Rad, Mississauga, ON, Canada).

Cell lines, cell culture conditions and transfections Immunoprecipitation and immunoblotting The 32Dcl3 diploid, non-leukemic, IL-3-dependent murine NIH/3T3 cells (1 Â 106 per 100 cm dish)transfected with myeloid cell line (Greenberger et al., 1983)was obtained from the Myc-tagged truncated cyclin D2 (Myc-D2Trc)and

Oncogene Characterization of mouse and human truncated cyclin D2 C Denicourt et al 1261 Myc-tagged full-length cyclin D2 (Myc-D2)constructs were (Matsuoka et al., 1994). The beads were resuspended in 50 ml lysed 24 h after transfections in immunoprecipitation buffer of CAK buffer containing protease, phosphatase inhibitors (IP buffer)containing 50 m M Tris–HCl (pH 7.5), 150 mM and 1.4 mg of recombinant CDK7–cyclin H–MAT1 complex NaCl, 1% Nonidet P-40 and complete protease inhibitor (Upstate, Charlottesville, VA, USA). After addition of 40 mCi cocktail (Roche, Laval, Quebec, Canada)and sonicated on ice of [g-32P] ATP in the presence of 50 mM ATP, the suspensions (Fisher cell dismembrator, 80% microtip power 10 s). Lysates were incubated at 30 1C for 30 min. The precipitates were were clarified by centrifugation at 10 000 g for 5 min. The washed six times in Tween 20 lysis buffer (50 mM HEPES (pH supernatants were precipitated for 4 h at 4 1C with protein 7.5), 150 mM NaCl, 1 mM EDTA, 2.5 mM EGTA, 1 mM A-Sepharose beads (GE Healthcare)alone or precoated with dithiothreitol, 0.1% Tween 20 and 10% glycerol), followed saturating amounts of the anti-CDK4 antibody (Santa Cruz by denaturation by boiling in electrophoresis sample buffer for sc-601, Santa Cruz, CA, USA). The proteins immunoprecipi- 5 min. Phosphorylated products were separated on denaturing tated on beads were washed five times with 1 ml of IP buffer. polyacrylamide gels and visualized by autoradiography. The beads were resuspended in 2 Â loading buffer (30% glycerol, 4% SDS, 160 mM Tris–HCl (pH 6.8), 10% IL-3 starvation, stimulation and cell cycle analysis B-mercaptoethanol and 0.02% bromophenol blue)and of 32Dcl3 cells boiled for 10 min at 95 1C. Immunoprecipitated samples IL-3 starvation was performed by washing the cells twice with were separated on 12% SDS–PAGE and electrophoretically PBS and reseeding them in culture medium without WEHI– transferred to Immobilon-P PVDF membranes (Millipore, CM for 16 h. After the IL-3 starvation, the cells were Billerica, MA, USA). centrifuged and reseeded in culture medium containing For immunoblotting analysis, proteins were extracted in WEHI–CM. Samples of cells were collected at different time RIPA buffer and the protein concentration in the different points after IL-3 stimulation for northern and western blot extracts was measured using the Bio-Rad protein assay. Equal analyses. amounts of lysate proteins (30–80 mg)were diluted in 2 Â sample buffer (30% glycerol, 4% SDS, 160 mM Tris–HCl, pH Immunofluorescence 6.8, 10% b-mercaptoethanol and 0.02% bromophenol blue) NIH/3T3 cells were seeded on coverslips 16 h prior to and boiled for 3 min. Protein were separated on 12% SDS– transfection. The cells were fixed 8–24 h after transfection in PAGE and electrophoretically transferred to Immobilon-P 4% paraformaldehyde for 10 min and permeabilized in 0.2% PVDF membranes (Millipore). Membranes were blocked in Triton X-100. Fixed cells were washed in PBS, and then PBS containing 0.2% Tween-20 and 4% skimmed milk blocked in 10% normal goat serum for 1 h at room temper- powder (PBS-Tween-milk)for 1 h at room temperature. The ature. Immunostaining was subsequently performed with an antibodies were diluted in PBS-Tween-milk at the following anti-Myc epitope antibody (Santa Cruz)(1:50)for 1 h at 25 1C. dilutions: anti-truncated cyclin D2 peptide antibody (1:100) The cells were then incubated with Cy3-conjugated anti-mouse (Denicourt et al., 2002), the anti-cyclin D2 (1:1000) (Cell IgG (Jackson Immunoresearch, West Grove, PA, USA)for 1 h Signaling, Danvers, MA, USA), the anti-Myc epitope anti- at 25 1C. The cells were washed extensively, mounted and body (1:5000)(Santa Cruz)or the anti-GAPDH antibody imaged using a Zeiss Axioskop fluorescence microscope. (1:300)(Chemicon, Temecula, CA, USA).The blots were incubated for 1 h at room temperature, washed in PBS-Tween and further incubated for 1 h with either anti-rabbit or anti- Construction of GFP fusion protein expression vectors mouse IgG-peroxidase (BD Pharmingen, Mississauga, ON, The cyclin D2 and the truncated cyclin D2 coding sequences Canada). The immune complexes were revealed using the ECL for both the human and mouse were PCR amplified and plus chemiluminescence regents (GE Healthcare). cloned into in pEGFP-C2 vector (BD Biosciences Clontech, Mississauga, ON, Canada). The constructs were transfected in NIH/3T3 cells with QIAGEN Polyfect transfection reagent as Kinase assay recommended by the supplier. Cells were plated on Permanox NIH 3T3 cells were transfected with Myc-tagged cyclin D2 and Plastic Chamber Slide System (Fisher Scientific, Ottawa, ON, truncated cyclin D2 using Lipofectamine 2000 (Invitrogen). Canada). Twenty-four hours later, cells were visualized by Twenty-four hours after transfection, cells were lysed in buffer confocal microscopy. For control, cells were transfected with containing 50 mM HEPES (pH 7.5), 10 mM MgCl2, 0.1% pEGFP-C2 vector alone. Tween 20, 1 mM dithiothreitol, 25 mM ATP and protease inhibitors. Lysates were precleared with protein G-Sepharose Construction of cyclin D2 mutant beads (Zymed, Burlington, ON, Canada)on a rotating wheel The cyclin D2 mutant (D2D157–289)was produced by a at 4 1C for 1 h. Two micrograms of anti-Myc epitope (9E10 PCR-based site-directed mutagenesis method as previously SantaCruz)antibodies was used for immunoprecipitation. described by Fisher and Pei (1997). The following oligonucleo- Kinase reactions were performed in 10 mM MgCl2 and 50 mM tide primers were used: 50-PTTAGTGAGGGGTGACTGCG HEPES (pH 7.2)with 2 mg of bacterially isolated glutathione GCCAGG-30 (stop codon is underlined)and 5 0-GCCACCGC 0 S-transferase (GST)-Rb C -terminus, 50 mM cold ATP and GGTGGAGCGCCAATTC-30. Cyclin D2 cloned into pBlue- 32 10 mCi of [g- P]ATP (Amersham, Baie d’Urfe, Quebec, script II KS þ was used as PCR template. Canada). Reactions were performed at 30 1C for 30 min. Acknowledgements CAK activation assay NIH 3T3 cells were transfected with Myc-tagged cyclin D2 and We thank the Arthur and Sonia Labatt Brain Tumour truncated cyclin D2 using Lipofectamine 2000 (Invitrogen). Research Center (http://www.sickkids.ca/BTRC/ Toronto, Twenty-four hours after transfection, cells were lysed and Canada)and Dr Abhijit Guha for the gift of human tumor immunoprecipitated as described above. The immunoprecipi- tissues. This work was supported by grant FRN 37994 from tates were washed three times with IP buffer and then three the Canadian Institutes of Health Research and by La Socie´ te´ times with CAK buffer (80 mM b-glycerophosphate (pH 7.3), de recherche sur le Cancer. CD is a recipient of a FCAR PhD 15 mM MgCl2,20mM EGTA and 5 mM dithiothreitol) scholarship.

Oncogene Characterization of mouse and human truncated cyclin D2 C Denicourt et al 1262 References

Andersen G, Busso D, Poterszman A, Hwang JR, Wurtz JM, Ripp R Matsushime H, Ewen ME, Strom DK, Kato JY, Hanks SK, Roussel et al. (1997). The structure of cyclin H: common mode of kinase MF et al. (1992). Identification and properties of an atypical activation and specific features. EMBO J 16: 958–967. catalytic subunit (p34PSK-J3/cdk4)for mammalian D type G1 Ando K, Ajchenbaum-Cymbalista F, Griffin JD. (1993). Regulation of cyclins. Cell 71: 323–334. G1/S transition by cyclins D2 and D3 in hematopoietic cells. Proc Ortega S, Malumbres M, Barbacid M. (2002). Cyclin D-dependent Natl Acad Sci USA 90: 9571–9575. kinases, INK4 inhibitors and cancer. Biochim Biophys Acta 1602: Assoian RK. (1997). Control of the G1 phase cyclin-dependent kinases 73–87. by mitogenic growth factors and the extracellular matrix. Cytokine Quelle DE, Ashmun RA, Shurtleff SA, Kato JY, Bar-Sagi D, Roussel MF Growth Factor Rev 8: 165–170. et al. (1993). Overexpression of mouse D-type cyclins accelerates Baldin V, Lukas J, Marcote MJ, Pagano M, Draetta G. (1993). Cyclin G1 phase in rodent fibroblasts. Genes Dev 7: 1559–1571. D1 is a nuclear protein required for cell cycle progression in G1. Resnitzky D. (1997). Ectopic expression of cyclin D1 but not Genes Dev 7: 812–821. induces anchorage-independent cell cycle progression. Mol Cell Biol Bartkova J, Rajpert-de Meyts E, Skakkebaek NE, Bartek J. (1999). 17: 5640–5647. D-type cyclins in adult human testis and testicular cancer: relation to Roovers K, Davey G, Zhu X, Bottazzi ME, Assoian RK. (1999). cell type, proliferation, differentiation, and malignancy. J Pathol Alpha5beta1 integrin controls cyclin D1 expression by sustaining 187: 573–581. mitogen-activated activity in growth factor-treated Buschges R, Weber RG, Actor B, Lichter P, Collins VP, Reifenberger cells. Mol Biol Cell 10: 3197–3204. G. (1999). Amplification and expression of cyclin D genes (CCND1, Russo AA, Jeffrey PD, Pavletich NP. (1996). Structural basis of cyclin- CCND2 and CCND3)in human maligant gliomas. Brain Pathol 9: dependent kinase activation by phosphorylation. Nat Struct Biol 3: 435–442. 696–700. Denicourt C, Kozak C, Rassart E. (2002). A new common integration Schulze A, Zerfass-Thome K, Berges J, Middendorp S, Jansen-Durr P, site in Graffi MuLV-induced leukemias, Gris1: overexpression of a Henglein B. (1996). Anchorage-dependent transcription of the cyclin truncated cyclin D2 due to alternative splicing. J Virol 77: 37–44. A gene. Mol Cell Biol 16: 4632–4638. Diehl JA, Sherr CJ. (1997). A dominant-negative cyclin D1 mutant Schwartz MA, Assoian RK. (2001). Integrins and cell proliferation: prevents nuclear import of cyclin-dependent kinase 4 (CDK4)and regulation of cyclin-dependent kinases via cytoplasmic signaling its phosphorylation by CDK-activating kinase. Mol Cell Biol 17: pathways. J Cell Sci 114: 2553–2560. 7362–7374. Sherr CJ, Roberts JM. (1999). CDK inhibitors: positive and negative Diehl JA, Cheng M, Roussel MF, Sherr CJ. (1998). Glycogen synthase regulators of G1-phase progression. Genes Dev 13: 1501–1512. kinase-3beta regulates cyclin D1 and subcellular Sicinski P, Donaher JL, Geng Y, Parker SB, Gardner H, Park MY localization. Genes Dev 12: 3499–3511. et al. (1996). Cyclin D2 is an FSH-responsive gene involved in Fisher CL, Pei GK. (1997). Modification of a PCR-Based site-directed gonadal cell proliferation and oncogenesis. Nature 384: 470–474. mutagenesis method. Biotechniques 23: 570–574. Solomon DA, Wang Y, Fox SR, Lambeck TC, Giesting S, Lan Z et al. Greenberger JS, Sakakeeny MA, Humphries RK, Eaves CJ, Eckner (2003). Cyclin D1 splice variants. Differential effects on localization, RJ. (1983). Demonstration of permanent factor-dependent multi- RB phosphorylation, and cellular transformation. J Biol Chem 278: potential (erythroid/neutrophil/basophil)hematopoietic progenitor 30339–30347. cell lines. Proc Natl Acad Sci USA 80: 2931–2935. Susaki E, Nakayama K, Nakayama KI. (2007). Cyclin D2 translocate Hanna Z, Jankowski M, Tremblay P, Jiang X, Milatovich A, Francke p27 out of the nucleus and promotes its degradation at the G0–G1 U et al. (1993). The Vin-1 gene, identified by provirus insertional transition. Mol Cell Biol 13: 4626–4640. mutagenesis, is the cyclin D2. Oncogene 8: 1661–1666. Tabin CJ, Bradley SM, Bargmann CI, Weinberg RA, Papageorge AG, Jeffrey PD, Russo AA, Polyak K, Gibbs E, Hurwitz J, Massague J Scolnick EM et al. (1982). Mechanism of activation of a human et al. (1995). Mechanism of CDK activation revealed by the oncogene. Nature 300: 143–149. structure of a cyclinA-CDK2 complex. Nature 376: 313–320. Taı¨ eb F, Jessus C. (1996). Xenopus cyclin D2: cloning and expression Kato J, Matsuoka M, Polyak K, Massague´ J, Sherr CJ. (1994). Cyclic in oocytes and during early development. Biol Cell 3: 99–111. AMP-induced G1 phase arrest mediated by an inhibitor (p27kip1)of Teramoto N, Pokrovskaja K, Szekely L, Polack A, Yoshino T, Akagi cyclin-dependent kinase-4 activation. Cell 79: 487–496. T et al. (1999). Expression of cyclin D2 and D3 in lymphoid lesions. Kerkhoff E, Ziff EB. (1995). Cyclin D2 and Ha-Ras transformed Int J Cancer 81: 543–550. rat embryo fibroblasts exhibit a novel deregulation of cell Tremblay PJ, Kozak CA, Jolicoeur P. (1992). Identification of a novel size control and early S phase arrest in low serum. EMBO J 14: gene, Vin-1, in murine leukemia virus-induced T-cell leukemias by 1892–1903. provirus insertional mutagenesis. J Virol 66: 1344–1353. Kim KK, Chamberlin HM, Morgan DO, Kim SH. (1996). Three- Watanabe K, Tachibana O, Sata K. (1996). Overexpression of the dimensional structure of human cyclin H, a positive regulator of the EGF receptor and mutations are mutually exclusive in the CDK-activating kinase. Nat Struct Biol 3: 849–855. evolution of primary and secondary glioblastomas. Brain Pathol 6: Lu F, Gladden AB, Diehl JA. (2003). An alternatively spliced Cyclin 217–223. D1 isoform, Cyclin D1b, is a nuclear oncogene. Cancer Res 63: Xiong Y, Zhang H, Beach D. (1992). D type cyclins associate with 7056–7061. multiple protein kinases and the DNA replication and repair factor Malumbres M, Barbacid M. (2001). To cycle or not to cycle: a critical PCNA. Cell 71: 505–514. decision in cancer. Nat Rev Cancer 1: 222–231. Zhu X, Ohtsubo M, Bohmer RM, Roberts JM, Assoian RK. (1996). Matsuoka M, Kato JY, Fisher RP, Morgan DO, Sherr CJ. (1994). Adhesion-dependent cell cycle progression linked to the expression Activation of cyclin-dependent kinase 4 (cdk4)by mouse MO15- of cyclin D1, activation of cyclin E-cdk2, and phosphorylation of the associated kinase. Mol Cell Biol 14: 7265–7275. retinoblastoma protein. J Cell Biol 133: 391–403.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).

Oncogene