Cell-Specific Responses to Loss of Cyclin-Dependent Kinases

Home , Cyclin A, Cyclin A1, Cyclin A2, Cyclin B2, G1 phase, Prophase

C Berthet and P Kaldis

Mouse Cancer Genetics Program, Center for Cancer Research, National Cancer Institute-Frederick, Frederick, MD, USA

Inactivation of cyclin-dependent kinases (Cdks) and/or one or more components of the cell cycle machinery cyclins in mice has changed our viewof cell cycle (reviewed by Pagano and Jackson (2004); Aleem and regulation. In general, cells are far more resistant to the Kaldis (2006); Santamaria and Ortega (2006)), and in loss of Cdks than originally anticipated, suggesting wide- this review, we will outline how the lack of Cdk/cyclin spread compensation among the Cdks. Early embryonic complexes can affect cells (and organism) that rely on cells are, so far, not sensitive to the lack of multiple Cdks their proliferation status. We will report new insights or cyclins. In contrast, differentiated cells are more that these mouse models provide to understand pro- dependent on Cdk/cyclin complexes and the functional liferation control in stem cells, differentiated cells and redundancy is more limited. Our challenge is to better transformed cells. To a certain degree, these mouse understand these cell-type specific differences in cell cycle models mimic drugs that target these cell cycle proteins regulation that can be used to design efficient cancer and interpretation of the observed mutant phenotypes therapy. Indeed, tumor cells seem to respond to inhibition can improve our strategies for cancer therapy. of Cdk activities, however, with different outcome depend- Cell proliferation is a tightly regulated process and ing on the tumor cell type. Tumor cells share some most eukaryotic cells undergo cell division only in the proliferation features with stem cells, but appear more presence of mitogenic factors. The cell cycle machinery sensitive to loss of Cdk activity, somewhat resembling includes several mechanisms to sense the presence of differentiated cells. We summarize the current knowledge these growth factors, inhibitory factors or damage that of cell cycle regulation in different cell types and highlight prevents cell cycle progression. Different checkpoints their sensitivity to the lack of Cdk activities. along the cell division cycle determine progression or Oncogene (2007) 26, 4469–4477; doi:10.1038/sj.onc.1210243; arrest and eventually may trigger cell death. At the onset published online 5 February 2007 of the cell cycle, mitogenic factors stimulate the expression of proto-oncogenes, such as Ras or Myc, Keywords: cyclin; cyclin-dependent kinase; cell cycle; which results in transcriptional regulation leading to embryonic stem cells; tumor cells; differentiated cells increased proliferation. The family of D-type cyclins is the first to be induced through these mitogenic pathways, followed by the expression of cyclins E1 and E2. The D-type cyclins stimulate the activity of Cdk4/Cdk6, Introduction whereas E-type cyclins enhance Cdk2 activity. These complexes lead to passage through the restriction (R) Cell proliferation occurs at different rates depending on point at the boundary of the G1 and S phase, where extracellular factors and the genetic profile of cells, after the cell cycle progresses independently of growth which determines their pluripotency, differentiation factors. The key event in the G1 phase is the status and eventually their transformation grade. The phosphorylation of the pocket proteins (Retinoblastoma proliferation rate is regulated by proteins involved in the protein (Rb), p107, p130) by Cdk/cyclin D and E control of the cell cycle, which is driven by cyclins and complexes. Each Cdk complex phosphorylates presum- their associated kinases (Cdk for cyclin-dependent ably distinct residues of Rb (and the other pocket kinase) (Morgan, 1997). The fluctuating activities of proteins), and it has been postulated that this specificity several Cdk/cyclin complexes establish the progression provides a mechanism for temporal control of Rb of the cell cycle through each phase (G1, S, G2 and M). during G1 (reviewed by Mittnacht (1998); Knudsen and The composition and regulation of Cdks vary according Knudsen (2006)). Phosphorylation of Rb leads to the to cell type and environment. In the past few years, release of E2F proteins, resulting in the transcription of numerous mouse models have been generated targeting genes essential for entry into S phase. When mitogenic signals are not strong enough to enhance Cdk activity and inactivate Rb, cells cannot enter into S phase but Correspondence: Dr P Kaldis, Mouse Cancer Genetics Program, eventually will exit the cell cycle and acquire a reversible Center for Cancer Research, National Cancer Institute-Frederick, non-replicative state, quiescence or G0. The G1/S Bldg. 560/22-56, 1050 Boyles Street, Frederick, MD 21702-1201, USA. transition can also be affected by inhibitory signals E-mail: [email protected] Received 30 November 2006; accepted 30 November 2006; published (e.g., cell contact inhibition or by different stresses, such online 5 February 2007 as oxidative stress, DNA damage, etc) inducing the Cdk functions in different cell types C Berthet and P Kaldis 4470 expression of anti-proliferative signals. Tumor suppressor genes constitute these signaling pathways, which result in the inhibition of Cdk activity. The INK4 (p15Ink4b, p16Ink4a, p18Ink4c and p19Ink4d) and the CIP/KIP (p21Cip1,p27Kip1 and p57Kip2) families inhibit cyclin D and E complexes, respectively, and are among these tumor- suppressor genes (reviewed by Sherr and Roberts (1999)). Consistently, mice with defects in Cdk inhibitors (CKIs) are cancer prone (reviewed by Santamaria and Ortega (2006)). Moreover, expression of some of these CKIs is directly or indirectly regulated by p53, one of the major tumor-suppressor genes (reviewed by Vousden and Lu (2002)). The activation of p53 or Rb under stress conditions can irreversibly affect cells that will enter senescence, a permanent non-replicative state (reviewed by Campisi (2005)). Cell-cycle progression can also be blocked at other checkpoints (intra S phase, G2/M, mitosis) following cellular damage. These pathways usually activate p53 and/or p21, which will then inhibit Cdk2 and/or Cdk1 activity. At all the checkpoints, cells remain arrested until damage is repaired or cells undergo apoptosis. If cell death is avoided and cell cycle arrest is bypassed, it might lead to cell transformation. The balance between mitogenic, anti-mitogenic, apoptotic and stress response signals determines the fate of the cells and their ability to proliferate. Embryonic stem (ES) cells proliferate at a fast rate (cell cycle length of B8 h) and exhibit a short G1 phase (Figure 1a). Mouse ES cells are deficient in cyclin D-associated activity (Savatier et al., 1996; Burdon et al., 2002; Jirmanova et al., 2002) and rely on Cdk2 activity to drive the G1/S transition (Stead et al., 2002). Figure 1 Cell cycle regulators in different cell types. Stem cells (a), Components of the G1 checkpoint, such as Rb are differentiated cells (b) and tumor cells (c) regulate the cell cycle permanently inactivated (reviewed by Burdon et al. according to their specific environment. The G1-phase regulation is (2002)) and despite high levels of p53 expression in particularly modified in each cell type. The RB/E2F and p53 pathways are not active in stem cells and Cdk1 seems to be the mouse ES cells, cell cycle progression is not affected. major kinase regulating all the phases. During differentiation, the It appears that the p53-mediated response is inactive cells proliferate under the strict control of multiple Cdk/cyclin because p53 is sequestered into the cytoplasm (Aladjem complexes and their inhibitors. These inhibitory pathways are often et al., 1998; Prost et al., 1998). Indeed, ES cells do not inactivated in tumor cells, which lead to hyper-activation of the undergo G1/S cell-cycle arrest under stimuli that would Cdk/cyclin complexes. normally activate the p53-mediated pathway in somatic cells, whereas the cell-cycle arrest response is restored upon differentiation of ES cells (Aladjem et al., 1998; mouse resulting in reduced longevity and an early onset Prost et al., 1998). In contrast, the G2/M checkpoint of an aging phenotype (Tyner et al., 2002). Senescence remains active and knockouts of p53 targets, like Btg2, will occur in differentiated cells when they reach a affect the DNA damage-inducible G2/M arrest limited number of cell divisions (controlled by the length (Rouault et al., 1996). At the onset of differentiation, of telomeres) or upon activation of stress signals. Five the proliferation rate decreases with an extension of the major pathways have been identified that enable G1 phase, associated with the establishment of the transformed cells to escape senescence. These include Rb/E2F pathway regulating the G1/S checkpoint the telomerase, the p53, the Rb, the insulin-like growth (Figure 1b). Lack of Rb and p53 regulation affects cell factor (IGF)/Akt and the mitochondrial/oxidative stress differentiation and embryogenesis at several stages of pathways (reviewed by Miura et al. (2004); Campisi development, suggesting an essential role of these (2005)). Inappropriate inhibition or activation of these proteins in the maintenance of the proper balance pathways can lead to immortalization and are found between self-renewal and differentiation (reviewed by individually or in combination in all tumors. Prolifera- Miura et al. (2004)). During the differentiation process, tion of most tumor cells mimics the mode of regulation cells will encounter phases of proliferation and quies- described for ES cells: activation of proto-oncogenes cence that require a perfect balance between exogenous replacing mitogenic signal requirements, high telomer- signals and the cell cycle machinery. Disturbance of this ase activity allowing the maintenance of telomeres and delicate equilibrium can induce premature senescence, as permanent self-renewal, loss of tumor-suppressor gene suggested by the study of a hypermorphic p53 mutant activity that affects cell-cycle arrest and the apoptosis

Oncogene Cdk functions in different cell types C Berthet and P Kaldis 4471 activities by different means (e.g., dominant-negative proteins, overexpression of inhibitors, drugs). These approaches taught us that the activity of Cdk/cyclin complexes coincides with cell cycle progression and its inhibition results in cell-cycle arrest or apoptosis. However, these experiments did not target a single cyclin or Cdk, and even rarely a single family, which resulted in off-target effects. Studies using knockout mouse models were needed to deepen our knowledge of speciﬁc in vivo functions associated with each cyclin or Cdk. Ultimately, we should be able to choose and target the best pathways in the context of cancer therapy. Such work started in 1995, when two groups reported knockout mice of cyclin D1, with only tissue-speciﬁc defects observed in particular cell types (Fantl et al., 1995; Sicinski et al., 1995). It was the beginning of ongoing research programs, aiming to inactivate Cdks or cyclins in mice and to explore non-redundant functions among these complexes (reviewed by Pagano and Jackson (2004); Aleem and Kaldis (2006); Santamaria and Ortega (2006)).

Mouse knockout models for cell cycle regulators Similar to cyclin D1 ablation (and other cyclins), inactivation of Cdk4 or Cdk6 did not prevent cell cycle progression (Rane et al., 1999; Tsutsui et al., 1999; Figure 2 Pathways affecting proliferation in different cell types. Malumbres et al., 2004). Although these two kinases The control of the cell cycle is essential to maintain a proper share binding partners (D-type cyclins) and substrates, it balance between proliferation and differentiation in most cells. appears from these mouse studies that Cdk4 and Cdk6 Each cell type exhibits different proliferation characteristics and their response to the loss of Cdk activities varies. ES cells are more different than anticipated and might not proliferate very rapidly and are independent of D-type cyclin completely compensate for each other. Inactivation of activity. Moreover, they are not sensitive to the loss of other Cdks Cdk6 does not impair cell cycle progression, although or cyclins (at least the ones that have been knockout), suggesting a the number of certain types of hematopoietic cells is high plasticity of the embryonic cell cycle. In contrast, differentiated cells have acquired a G1/S checkpoint, do not proliferate reduced (Malumbres et al., 2004). Lack of Cdk4 has indefinitely and can enter quiescence (G0). The loss of Cdks or a mild effect on cell cycle progression with a 4 h delay in cyclins affects specifically certain types of differentiated cells. Cdk4/ S phase entry in mouse embryo fibroblasts (MEFs) cyclin D appears to be involved in a broad range of cell types. from quiescence, and the cell number of several Combined inactivation of Cdk2 and Cdk4 affects cell proliferation differentiated cell types (pancreatic b cells, anterior during embryogenesis. Tumor cells are sometimes susceptible to Cdk inhibition, even if their proliferation features resemble the pituitary cells, Leydig cells) is decreased (Rane et al., ones in stem cells (no G1/S checkpoint, no senescence, high 1999; Tsutsui et al., 1999). The S phase delay observed in telomerase activity, activation of proto-oncogenes, inactivation of Cdk4À/À MEFs is indirectly related to inhibition of Cdk2 tumor suppressor genes). Studying mouse models with inactivated activity. Indeed, Cdk4À/À MEFs display decreased Cdk2 Cdks or cyclins help us to better understand the specificity of the cell-cycle regulation in different cell types. activity owing to the redistribution of p27, and combined inactivation of Cdk4- and p27-restored Cdk2 activity and S phase entry (Tsutsui et al., 1999). Overall, pathways (Figure 1c). The loss of tumor-suppressor none of the components of the cyclin D-associated gene activity is often permanent in tumor cells (LOH, complexes are required in early embryogenesis. In mutations), which distinguishes them from stem cells. In contrast, specific inhibition of cyclin D1 in cancer cell Figure 2, we have summarized the regulatory pathways lines resulted in cell-cycle arrest (Zhou et al., 1995; that need to be modified for the transition from one cell Arber et al., 1997; Kornmann et al., 1998; Sauter et al., type to another. These pathways change the prolifera- 1999; Lee et al., 2000). Moreover, cyclin D1 or Cdk4- tion characteristics of cells and regulate Cdk/cyclin null mice are resistant to breast cancers triggered by the activities. We will discuss results derived from the ErbB2 oncogene (Yu et al., 2001, 2006; Reddy et al., analysis of ‘cell cycle’ mouse models and how these 2005; Landis et al., 2006), skin tumor induced by 7,12- pathways are affected in these contexts. dimethylbenz[2]nthracene/12-O-tetradecanoylphorbol-13- acetate treatment (Robles et al., 1998; Rodriguez-Puebla Coupling of Cdk/cyclin activities and cell cycle et al., 2002) or c-myc overexpression (Miliani de Marval progression et al., 2004). These results highlight a major difference between stem cell and cancer cell proliferation. For many years, our understanding of cell cycle pro- Although Cdk4 activity is not detectable in stem gression was based on studies that inhibited Cdk/cyclin cells (Savatier et al., 1994), cancer cells often maintain

Oncogene Cdk functions in different cell types C Berthet and P Kaldis 4472 high Cdk4 activity and are able to proliferate in the Cdk2-inhibitors in cancer therapy. Depletion of Cdk1 absence of mitogenic signals. Inhibition of cyclin D1 in NCI-H1299 or U2OS cells resulted in an accumula- renders cancer cells dependent again on mitogenic tion in G2/M and a moderate slowing of G2/M signals. progression after release from synchronization (Cai So far, ablation of cyclins A2 and B1 are the only D/ et al., 2006). Moreover, U2OS cells have a greater E/A/B-type cyclin family members that result in early propensity to undergo endoreplication, suggesting that embryonic lethality, suggesting an essential role in cell these cells lack an efficient postmitotic checkpoint. Cdk1 proliferation or early development (Murphy et al., 1997; silencing does not induce cytotoxicity in NCI-H1299 or Brandeis et al., 1998). However, cyclin B1-null embryos U2OS cells and it might be compensated by Cdk2, as survive until mid-gestation (Brandeis et al., 1998) and Cdk2/cyclin B complexes have been observed in Cdk1 cyclin A2-deficient embryos reach the blastocyst stage knock down cells (Cai et al., 2006). (Murphy et al., 1997). This implies that at least early Taken together, these results suggest that none of the embryonic cycles can occur in the absence of cyclin B1 cyclins or Cdks that have been mutated so far, affect and A2. Cyclin A2 is apparently the most critical non- early embryonic cell cycles and proliferation of stem redundant cyclin in cultured somatic cells, as it is a rate- cells does not seem to be dependent on a specific Cdk/ limiting component required for S- and M-phase cyclin complex (the jury is still out for Cdk1). In progression (Furuno et al., 1999). None the less, cyclin contrast, proliferation of differentiated cells is affected A2 activity was not detected in adult spleen or specifically by the lack of at least one Cdk/cyclin immortalized MEFs harvested from Cdk2-null mice complex. Cdk2 is important for germ cells, Cdk6/cyclin (Berthet et al., 2003), possibly uncovering another D complexes in hematopoietic cells, and Cdk4/cyclin D parallel pathway in cell cycle progression. The genera- complexes appear to be important in a larger panel of tion of conditional cyclin A2 knockout mice would help cells. Inhibition of a single Cdk affects cancer cells to further investigate these observations. Similarly, the moderately, but to a higher extent than it affects ES cell-cycle community is waiting for the characterization cells. Responses to Cdk inhibition varies in each cell of Cdk1 mouse models to uncover the mystery of a line, which suggests that tumor cells exhibit specific possible ‘central’ Cdk, comparable with Cdk1 that sensitivities as has been seen in differentiated cells. In drives the yeast cell cycle by itself. The idea of a single both cancer cells and differentiated cells, but not ES essential Cdk in mammalian cells, like in yeast, started cells, Cdk4/cyclin D1 complexes seem to be required in a to make sense when it was demonstrated that Cdk2, the broad range of cell types. S phase regulating kinase, was not required for mitotic cell cycle (Berthet et al., 2003; Ortega et al., 2003). Cdk2- null mice are viable and cell proliferation is barely Compound mouse mutants affected with a 4 h delay observed in S phase entry in MEFs released from quiescence. We have shown using Uncovering the compensatory mechanisms led to the Cdk2À/Àp27À/À mice that Cdk1 compensates for Cdk2, development of complex mouse models to discriminate binds cyclin E and can promote the G1/S transition the functional overlap of cyclins or Cdks (reviewed by (Aleem et al., 2005). Nevertheless, Cdk2 is essential Pagano and Jackson (2004); Aleem and Kaldis (2006); for meiosis, indicating that Cdk1 cannot compensate Santamaria and Ortega (2006)). Inactivation of D-type for Cdk2 in germ cells (Berthet et al., 2003; Ortega (D1, D2, D3) or E-type (E1, E2) cyclins has been et al., 2003). reported, and surprisingly cell proliferation was not abolished (Geng et al., 2003; Parisi et al., 2003; Kozar et al., 2004). The lack of cyclin E affects trophoblast Silencing of Cdks in cell lines endoreplication and cell cycle reentry of quiescent MEFs, but embryos reach birth when the placental In cancer cell lines, it appears that inhibition of Cdk1 or defect is rescued (Geng et al., 2003). Triple cyclin D Cdk2, using siRNAs, has different effects depending on knockout leads to late embryonic lethality owing to the cell lines tested. Cdk2 siRNA does not interfere with hematopoietic and cardiac defects, and in vitro cyclin proliferation of several colon cancer cell lines or U2OS D-deficient MEFs exhibit higher dependence on serum osteosarcoma cells (Tetsu and McCormick, 2003; Cai than wild-type cells (Kozar et al., 2004). These et al., 2006; Payton et al., 2006). In contrast, Cdk2 phenotypes suggest that proliferation of ES cells is not siRNA delays G1 progression in NCI-H1299 non-small dependent on cyclin D or cyclin E. This is not too cell lung cancer cells (Cai et al., 2006), induces G1 arrest surprising for the D-type cyclins, as Cdk4/cyclin D in melanoma cells (Du et al., 2004) and results in activity was not detected in ES cells and the unlimited accumulation of HT29 colon cancer cells in S and G2/M abundance of growth factors during the early stages of phases (Moffat et al., 2006). Taken together with the embryonic development. The fact that E-type cyclins are fact that Cdk1 compensates for Cdk2 in mice (Aleem also not essential for the entry into S phase corroborates et al., 2005), it is likely that in a wide variety of cell lines, that only cyclin A2 cannot be fully compensated in early Cdk1 can compensate for Cdk2. Nevertheless, there are embryogenesis. Double knockouts for Cdk2/Cdk4 and some discrepancies, suggesting that compensatory me- Cdk4/Cdk6 corroborate this view, as no early embryo- chanisms could be affected by genetic alterations in nic defects have been observed (Malumbres et al., 2004; some cell lines and open the opportunity to use Berthet et al., 2006). However (and maybe finally), a

Oncogene Cdk functions in different cell types C Berthet and P Kaldis 4473 pronounced impairment of cell cycle progression has appropriate cell-cycle arrest and eventually trigger a been observed in double knockouts of Cdk2 and Cdk4 cytotoxic effect exclusively to the tumor cells. (Berthet et al., 2006). Combined loss of Cdk2 and Cdk4 leads to a progressive decrease of Rb phosphorylation after midgestation or in primary MEFs. Proliferation is Control of senescence by Cdk/cyclin complexes reduced in vivo and the ability of cells to proliferate during embryogenesis is probably related to their Normal cells undergo senescence in response to various dependence on the Rb/E2F pathway. Rb is always stress signals, such as telomere shortening, oxidative phosphorylated in stem cells (Savatier et al., 1994) and stress, DNA damage or aberrant oncogene activation. pocket protein functions are not required in stem cells Two major tumor-suppressor cascades, the p16INK4a–Rb (Dannenberg et al., 2000; Sage et al., 2000). In double and the ARF–p53–p21 pathways, are involved in knockouts of Cdk2 and Cdk4, Rb is initially phos- senescence signaling, although differences exist between phorylated but a negative loop leads to concomitant human and mouse cells (reviewed by Blasco (2005); decreases of Rb phosphorylation and transcription of Dimri (2005); Kiyokawa (2006)). Mainly, mouse cells E2F target genes (Berthet et al., 2006). Among these express telomerase and do not exhibit shortened genes, Cdk1 expression declines, which further decreases telomeres. These pathways involve inhibition of Cdk4 Rb phosphorylation. Several possibilities could explain by p16INK4a or Cdk2 by p21, which suggests a central role this scenario, but overall it is associated with the number for Cdks in senescence. Under regular culture condi- of divisions and/or the state of differentiation (Berthet tions, MEFs express p16INK4a and p19ARF, which leads to and Kaldis, 2006). In contrast, when Cdk6 is deleted in a Cdk inhibition and the hypo-phosphorylation of pocket Cdk2-null background, mice are viable and exhibit proteins. The lack of the three pocket proteins or the similar phenotypes that were observed in respective acute inactivation of Rb is enough to bypass senescence single knockouts (Malumbres et al., 2004). Further- (Dannenberg et al., 2000; Sage et al., 2000). Similarly, more, combined inactivation of Cdk4 and Cdk6 affects p19 is essential and its absence leads to bypass of hematopoietic differentiation, but not significantly cell senescence and cells become immortal (Kamijo et al., proliferation (Malumbres et al., 2004). These models 1997). On the other hand, MEFs lacking p16INK4a emphasize the difference between Cdk4 and Cdk6 and undergo senescence apparently normally, suggesting demonstrate that Cdk2 and Cdk4 share a specific task in that p19ARF–p53–p21 pathway might be more important the control of Rb that can be compensated only in early in these cells (Sharpless et al., 2001). Premature embryogenesis. This indicates that mechanisms regulat- senescence has not been reported in MEFs deficient of ing the proliferation of stem cells and differentiated cells Cdk4 or Cdk4/Cdk6, corroborating that p16INK4a–Rb are different, especially in the G1/S transition. In pathway is functional or not essential under these differentiated cells, Cdk2 and Cdk4 contribute to conditions. In contrast, Cdk2À/ÀCdk4À/À MEFs enter coupling of the G1 phase to mitosis (Berthet and senescence prematurely, which contributes to their lack Kaldis, 2006). of proliferation at passage 4 (Berthet et al., 2006). The The Rb/E2F pathway is often inactivated in tumor proliferation defect and premature senescence can cells, through hyper activation of G1 Cdk/cyclin be rescued by human papillomavirus-E7 infection, complexes. In contrast to stem cells, combined inhibi- which inhibits all pocket proteins. A similar rescue of tion of G1 Cdks could impact proliferation of tumor Cdk2À/ÀCdk4À/À MEFs is not observed in the absence cells. To our knowledge, siRNAs or drugs that have of p27 (Cdk2À/ÀCdk4À/Àp27À/À MEFs), which enhances been developed to target Cdk4/6 or Cdk1/2 specifically, Cdk1 activity. These results suggest that inhibition have not yet been combined to inhibit both Cdk2 and of Cdk4 plays an important role in the regulation of Cdk4 in tumor models. On the other hand, combined senescence (see also Zou et al. (2002)) and it highlights depletion of Cdk1 and Cdk2 had a higher impact than the cooperation of the signals through the p16–Cdk4– single knockdown, inducing apoptosis in U2OS and a Rb and p19ARF–p53–p21–Cdk2–Rb pathways. It is strong G2/M block in NCI-H1299 cells (Cai et al., worth to note that p21 and p27 transcription is 2006). This confirms the mutual compensation between decreased in Cdk2À/ÀCdk4À/À MEFs, similar to known Cdk1 and Cdk2. Using Roscovitine, or even more E2F-target genes. It will be interesting to test if the specific Cdk1/2 inhibitors, also leads to cell-cycle arrest inactivation of p19ARF, p53 or p21 could modulate entry or apoptosis in several cell lines (Payton et al., 2006; into senescence of Cdk2À/ÀCdk4À/À MEFs. Ribas et al., 2006; Wesierska-Gadek et al., 2006). MEFs are convenient to study senescence, but other Surprisingly, Roscovitine can also protect thymocytes differentiated cell-types need to be investigated to better against apoptosis, and several groups suggest that Cdk2 understand this signaling pathway and eventually target might be required in these cells for programmed cell it in tumor cells. Similar to proliferation, behavior of death (Gil-Gomez et al., 1998; Hakem et al., 1999). tumor cells in terms of senescence might be different Further investigations will be needed to determine the than ES cells. Both express telomerase and can divide specific role of each Cdk depending on cell type; unlimitedly. However, it has been shown by cell fusion however, it is encouraging for tumor therapy to notice experiments that senescence was dominant over im- that combined inactivation of Cdks might impair tumor mortality (Wadhwa et al., 1994; Bertram et al., 1999) cell proliferation. Depending on tumor type, we might and senescent cells can exist in premalignant tumors be able to modulate Cdk activity and initiate the most (Collado et al., 2005). It has been suggested that ES cells

Oncogene Cdk functions in different cell types C Berthet and P Kaldis 4474 are spontaneously immortal because telomerase, p53, inhibition of a specific Cdk. The cytotoxic effect on Rb and IGF/Akt pathways are under strict regulation, tumors might depend on the overlapping role of the whereas in tumor cells they are usually deregulated or Cdks to maintain survival signals. lost (Miura et al., 2004). We could envision that in some tumor cells (e.g., tumor associated with loss of p27), combined inhibition of Cdk2 and Cdk4 could lead to Cdk/cyclin activity, checkpoints and tumor development senescence, which could be used as a possible therapeu- tic approach. Checkpoints are molecular events that manage the surveillance of cellular damage and eventually induce arrest of cell cycle progression. Each phase of the Cdk/cyclin activity and apoptosis cell cycle and each type of damage is controlled by specific mechanisms, however, the end point remains Cdk activity has been suggested to contribute to the inhibition of Cdk/cyclin activity. This goal is apoptosis in different cell types. So far, this hypothesis achieved by stimulation of CKI expression, depho- has been proposed based on indirect evidence, and sphorylation of Cdks, cyclin degradation or cytoplasmic mouse knockouts of Cdks could provide new insights. retention of Cdk/cyclin complexes (reviewed by Li and The chemical inhibitor, Roscovitine, has been used in Zou (2005)). several models to investigate the potential role of Cdks. Mouse knockouts represent good models to study So far, it has been shown that Roscovitine can block these molecular events in vivo and determine which Cdk apoptosis in thymocytes after radiation, dexamethasone, activity is required or dispensable to maintain a normal heat-shock or anti-CD3 treatment (Gil-Gomez et al., checkpoint. Martin et al. (2005) have shown, for 1998; Hakem et al., 1999). An apparent link between example, that Cdk2 was not required for p21-induced apoptosis and increase of Cdk2 activity has been cell-cycle arrest after DNA damage. This suggests that observed, suggesting that Cdk2 is required for apoptosis compensatory mechanisms that contribute to maintain associated with antigen-induced negative selection of normal cell-cycle progression are also effective under thymocytes. Roscovitine blocks this apoptotic pathway stress conditions. However, lack of Cdks and cyclins independent of T-cell antigen-receptor rearrangements could confer some resistance to tumors, indicating a (Williams et al., 2000), suggesting a role independent of difference between tumors and normal tissue. We DNA repair, where Cdk2 could also be involved mentioned that lack of cyclin D1 or Cdk4 confers (Muller-Tidow et al., 2004). Similarly, inhibition of resistance to breast or skin tumors (Robles et al., 1998; Cdks attenuates apoptosis in terminally differentiated Yu et al., 2001, 2006; Rodriguez-Puebla et al., 2002; neurons (Monaco and Vallano, 2003; Appert-Collin Reddy et al., 2005; Landis et al., 2006). Moreover, et al., 2006) and prevents cisplatin-induced cytotoxicity ablation of CKIs leads to multiple tumors, so it is in the kidney (Price et al., 2006). In contrast, Roscovi- possible to analyse the consequences of the lack of Cdk tine induces apoptosis in neutrophiles, which also are activity in the context of non-effective checkpoints (e.g., terminally differentiated cells (Rossi et al., 2006). These loss of CKIs, p53 or Rb). In this regard, Cdk2À/À mice non-proliferating cells apparently express Cdks, suggest- have been crossed with p27-null mice, which develop ing that these kinases may play a role unrelated to cell pituitary tumors (Fero et al., 1996; Kiyokawa et al., cycle. Their role as regulators of E2F/Rb pathway might 1996; Nakayama et al., 1996). Surprisingly, similar correlate with the apoptotic function of E2F in such tumors were observed in Cdk2À/Àp27À/À mice and this cells (reviewed by Stanelle and Putzer (2006)). Cdks lead to uncover that Cdk1 was an in vivo target of p27 could potentially be involved in survival signals, but it is and compensates for Cdk2 (Aleem et al., 2005). On the essential that we better understand this function to other hand, inactivation of Cdk4 rescues tumor-free reconcile these contradictory results. survival of p18INK4cÀ/À mice (Pei et al., 2004). These Several apoptosis-inducing agents enhance cyclin results highlight that CKIs, and generally tumor- A-associated activity in HeLa cells and overexpression suppressor genes, could have several targets in vivo of Bcl2-suppressed apoptosis and reduced the amount of and we must try to understand the crosstalk between the Cdk1 and Cdk2 in the nucleus (Meikrantz et al., 1994). different pathways to be able to affect the proper Moreover, combined silencing of Cdk1 and Cdk2 by target(s), if we want to re-establish cell-cycle arrest and siRNAs induces apoptosis in U2OS cell lines, but not in eventually inhibit tumor growth. NCI-H1299 (Cai et al., 2006). The limited cell death Loss of proper G1/S control and loss of p53 are in NCI-H1299 cells was significantly increased by major events in tumor progression, directly affecting co-depletion of Cdk9, involved in transcriptional cell-cycle checkpoints. With the development of more regulation. In fact, in U2OS cells, inhibition of Cdk1 complex mouse models, it would be interesting to test and Cdk2 results in reduced expression of RNA whether inhibition of multiple Cdks will affect tumor polymerase II and its associated phosphorylated forms. growth. This question should be addressed in the These data suggest that combined effects on cell cycle context of p53 or Rb deficiency. The crosstalk between progression and transcription occur in doubly depleted the Rb/E2F and p53 pathways is sometimes difficult to U2OS cells that could contribute to the apoptotic address, and we wonder what are the consequences of response (Cai et al., 2006). The sensitivity of cell lines p53 mutation in combination with the loss of one or is probably related to their plasticity to adjust the more Cdks.

Oncogene Cdk functions in different cell types C Berthet and P Kaldis 4475 Conclusions discrepancy, as adult stem cells exit the ‘embryonic’ cell cycle and enter quiescence (Figure 2). Such tumor cells Inhibition of Cdk activity has been a target for many can acquire characteristics of the cell cycle of differ- years aiming to decrease proliferation in tumor cells. We entiated cells, which renders them dependent on Cdks are hoping that the development of mouse models (see Figure 1). Our challenge now is to identify which lacking Cdks would help us to determine which cell loss of Cdk activity can lead to growth arrest, senescence types are dependent critically on a specific Cdk. It seems or apoptosis, in the context of specific human tumors. that redundancy is more abundant than initially Using mouse models lacking Cdks or cyclins, an in vivo suspected. However, it remains true that some differ- tumor models will help to further investigate the entiated cells are dependent on a defined Cdk/cyclin potential of Cdk inhibition in cancer therapy. It is likely complex. In contrast, ES cells seem to be able to fully that each tumor type will have its own specificity, similar compensate for the lack of multiple Cdks. These mouse to what we observed in our mouse models and in models highlight the fact that cell cycle regulation is very differentiated cell types. different in ES cells compared with differentiated cells. When proliferation in tumor cells is investigated, it appears that they behave like stem cells. However, when Acknowledgements analysing data that was accumulated from tumor cell We thank the Kaldis laboratory and MCGP for support and lines, it appears that proliferation can be affected by the discussion. We are also thankful to Satya Ande, Shuhei inhibition of Cdk activity and tumor cells behave more Kotoshiba, Weimin Li, Kasim Diril and Padmakumar VC for like specific differentiated cell types. It has been even comments on the manuscript. This research was supported by suggested that, in some case, tumor cells are derived the Intramural Research Program of the NIH, National from adult stem cells. This possibility could explain this Cancer Institute, Center for Cancer Research.

References

Aladjem MI, Spike BT, Rodewald LW, Hope TJ, Klemm M, Burdon T, Smith A, Savatier P. (2002). Signalling, cell cycle Jaenisch R et al. (1998). ES cells do not activate p53- and pluripotency in embryonic stem cells. Trends Cell Biol dependent stress responses and undergo p53-independent 12: 432–438. apoptosis in response to DNA damage. Curr Biol 8: 145–155. Cai D, Latham Jr VM, Zhang X, Shapiro GI. (2006). Aleem E, Kaldis P. (2006). Mouse models of cell cycle Combined depletion of cell cycle and transcriptional regulators: new paradigms. Results Probl Cell Differ 42: cyclin-dependent kinase activities induces apoptosis in 271–328. cancer cells. Cancer Res 66: 9270–9280. Aleem E, Kiyokawa H, Kaldis P. (2005). Cdc2-cyclin E Campisi J. (2005). Senescent cells, tumor suppression, and complexes regulate the G1/S phase transition. Nat Cell Biol organismal aging: good citizens, bad neighbors. Cell 120: 7: 831–836. 513–522. Appert-Collin A, Hugel B, Levy R, Niederhoffer N, Coupin G, Collado M, Gil J, Efeyan A, Guerra C, Schuhmacher AJ, Lombard Y et al. (2006). Cyclin-dependent kinase inhibitors Barradas M et al. (2005). Tumour biology: senescence in prevent apoptosis of postmitotic mouse motoneurons. Life premalignant tumours. Nature 436: 642. Sci 79: 484–490. Dannenberg JH, van Rossum A, Schuijff L, te Riele H. (2000). Arber N, Doki Y, Han EK, Sgambato A, Zhou P, Kim NH Ablation of the retinoblastoma gene family deregulates et al. (1997). Antisense to cyclin D1 inhibits the growth and G1 control causing immortalization and increased cell tumorigenicity of human colon cancer cells. Cancer Res 57: turnover under growth-restricting conditions. Genes Dev 1569–1574. 14: 3051–3064. Berthet C, Aleem E, Coppola V, Tessarollo L, Kaldis P. Dimri GP. (2005). What has senescence got to do with cancer? (2003). Cdk2 knockout mice are viable. Curr Biol 13: Cancer Cell 7: 505–512. 1775–1785. Du L, Lyle CS, Obey TB, Gaarde WA, Muir JA, Bennett BL Berthet C, Kaldis P. (2006). Cdk2 and Cdk4 cooperatively et al. (2004). Inhibition of cell proliferation and cell cycle control the expression of Cdc2. Cell Div 1: 10. progression by specific inhibition of basal JNK activity: Berthet C, Klarmann KD, Hilton MB, Suh HC, Keller JR, evidence that mitotic Bcl-2 phosphorylation is JNK- Kiyokawa H et al. (2006). Combined loss of Cdk2 and Cdk4 independent. J Biol Chem 279: 11957–11966. results in embryonic lethality and Rb hypophosphorylation. Fantl V, Stamp G, Andrews A, Rosewell I, Dickson C. Dev Cell 10: 563–573. (1995). Mice lacking cyclin D1 are small and show defects Bertram MJ, Berube NG, Hang-Swanson X, Ran Q, Leung JK, in eye and mammary gland development. Genes Dev 9: Bryce S et al. (1999). Identification of a gene that reverses 2364–2372. the immortal phenotype of a subset of cells and is a member Fero ML, Rivkin M, Tasch M, Porter P, Carow CE, Firpo E of a novel family of transcription factor-like genes. Mol Cell et al. (1996). A syndrome of multiorgan hyperplasia with Biol 19: 1479–1485. features of gigantism, tumorigenesis, and female sterility in Blasco MA. (2005). Mice with bad ends: mouse models for the p27Kip1-deficient mice. Cell 85: 733–744. study of telomeres and telomerase in cancer and aging. Furuno N, den Elzen N, Pines J. (1999). Human cyclin A is EMBO J 24: 1095–1103. required for mitosis until mid prophase. J Cell Biol 147: Brandeis M, Rosewell I, Carrington M, Crompton T, 295–306. Jacobs AM, Kirk J et al. (1998). Cyclin B2-null mice Geng Y, Yu Q, Sicinska E, Das M, Schneider JE, develop normally and are fertile whereas cyclin B1-null mice Bhattacharya S et al. (2003). Cyclin E ablation in the die in utero. Proc Natl Acad Sci USA 95: 4344–4349. mouse. Cell 114: 431–443.

Oncogene Cdk functions in different cell types C Berthet and P Kaldis 4476 Gil-Gomez G, Berns A, Brady HJ. (1998). A link between and mouse genes applied to an arrayed viral high-content cell cycle and cell death: Bax and Bcl-2 modulate screen. Cell 124: 1283–1298. Cdk2 activation during thymocyte apoptosis. EMBO J 17: Monaco III EA, Vallano ML. (2003). Cyclin-dependent kinase 7209–7218. inhibitors: cancer killers to neuronal guardians. Curr Med Hakem A, Sasaki T, Kozieradzki I, Penninger JM. (1999). Chem 10: 367–379. The cyclin-dependent kinase Cdk2 regulates thymocyte Morgan DO. (1997). Cyclin-dependent kinases: engines, apoptosis. J Exp Med 189: 957–968. clocks, and microprocessors. Annu Rev Cell Dev Biol 13: Jirmanova L, Afanassieff M, Gobert-Gosse S, Markossian S, 261–291. Savatier P. (2002). Differential contributions of ERK and Muller-Tidow C, Ji P, Diederichs S, Potratz J, Baumer N, PI3-kinase to the regulation of cyclin D1 expression and to Kohler G et al. (2004). The cyclin A1–CDK2 complex the control of the G1/S transition in mouse embryonic stem regulates DNA double-strand break repair. Mol Cell Biol cells. Oncogene 21: 5515–5528. 24: 8917–8928. Kamijo T, Zindy F, Roussel MF, Quelle DE, Downing JR, Murphy M, Stinnakre M-G, Senamaud-Beaufort C, Winston NJ, Ashmun RA et al. (1997). Tumor suppression at the mouse Sweeney C, Kubelka M et al. (1997). Delayed early INK4a locus mediated by the alternative reading frame embryonic lethality following disruption of the murine product p19ARF. Cell 91: 649–659. cyclin A2 gene. Nat Genet 15: 83–86. Kiyokawa H. (2006). Senescence and cell cycle control. Results Nakayama K, Ishida N, Shirane M, Inomata A, Inoue T, Probl Cell Differ 42: 257–270. Shishido N et al. (1996). Mice lacking p27Kip1 display Kiyokawa H, Kineman RD, Manova-Todorova KO, increased body size, multiple organ hyperplasia, retinal Soares VC, Hoffman ES, Ono M et al. (1996). Enhanced dysplasia, and pituitary tumors. Cell 85: 707–720. growth of mice lacking the cyclin-dependent kinase inhibitor Ortega S, Prieto I, Odajima J, Martin A, Dubus P, Sotillo R function of p27Kip1. Cell 85: 721–732. et al. (2003). Cyclin-dependent kinase 2 is essential for Knudsen ES, Knudsen KE. (2006). Retinoblastoma tumor meiosis but not for mitotic cell division in mice. Nat Genet suppressor: where cancer meets the cell cycle. Exp Biol Med 35: 25–31. 231: 1271–1281. Pagano M, Jackson PK. (2004). Wagging the dogma; tissue- Kornmann M, Arber N, Korc M. (1998). Inhibition of basal specific cell cycle control in the mouse embryo. Cell 118: and mitogen-stimulated pancreatic cancer cell growth by 535–538. cyclin D1 antisense is associated with loss of tumorigenicity Parisi T, Beck AR, Rougier N, McNeil T, Lucian L, Werb Z and potentiation of cytotoxicity to cisplatinum. J Clin Invest et al. (2003). Cyclins E1 and E2 are required for 101: 344–352. endoreplication in placental trophoblast giant cells. EMBO Kozar K, Ciemerych MA, Rebel VI, Shigematsu H, J 22: 4794–4803. Zagozdzon A, Sicinska E et al. (2004). Mouse development Payton M, Chung G, Yakowec P, Wong A, Powers D, and cell proliferation in the absence of D-cyclins. Cell 118: Xiong L et al. (2006). Discovery and evaluation of dual 477–491. CDK1 and CDK2 inhibitors. Cancer Res 66: 4299–4308. Landis MW, Pawlyk BS, Li T, Sicinski P, Hinds PW. (2006). Pei XH, Bai F, Tsutsui T, Kiyokawa H, Xiong Y. (2004). Cyclin D1-dependent kinase activity in murine development Genetic evidence for functional dependency of p18Ink4c on and mammary tumorigenesis. Cancer Cell 9: 13–22. Cdk4. Mol Cell Biol 24: 6653–6664. Lee RJ, Albanese C, Fu M, D’Amico M, Lin B, Watanabe G Price PM, Yu F, Kaldis P, Aleem E, Nowak G, Safirstein RL et al. (2000). Cyclin D1 is required for transformation by et al. (2006). Dependence of cisplatin-induced cell death in activated Neu and is induced through an E2F-dependent vitro and in vivo on cyclin-dependent kinase 2. J Am Soc signaling pathway. Mol Cell Biol 20: 672–683. Nephrol 17: 2434–2442. Li L, Zou L. (2005). Sensing, signaling, and responding to Prost S, Bellamy CO, Clarke AR, Wyllie AH, Harrison DJ. DNA damage: organization of the checkpoint pathways in (1998). p53-independent DNA repair and cell cycle arrest in mammalian cells. J Cell Biochem 94: 298–306. embryonic stem cells. FEBS Lett 425: 499–504. Malumbres M, Sotillo R, Santamaria D, Galan J, Cerezo A, Rane SG, Dubus P, Mettus RV, Galbreath EJ, Boden G, Ortega S et al. (2004). Mammalian cells cycle without the D- Reddy EP et al. (1999). Loss of Cdk4 expression type cyclin-dependent kinases Cdk4 and Cdk6. Cell 118: causes insulin-deficient diabetes and Cdk4 activation results 493–504. in b-islet cell hyperplasia. Nat Genet 22: 44–52. Martin A, Odajima J, Hunt SL, Dubus P, Ortega S, Reddy HK, Mettus RV, Rane SG, Grana X, Litvin J, Malumbres M et al. (2005). Cdk2 is dispensable for cell Reddy EP. (2005). Cyclin-dependent kinase 4 expression is cycle inhibition and tumor suppression mediated by p27Kip1 essential for neu-induced breast tumorigenesis. Cancer Res and p21Cip1. Cancer Cell 7: 591–598. 65: 10174–10178. Meikrantz W, Gisselbrecht S, Tam SW, Schlegel R. (1994). Ribas J, Boix J, Meijer L. (2006). (R)-roscovitine (CYC202, Activation of cyclin A-dependent protein kinases during Seliciclib) sensitizes SH-SY5Y neuroblastoma cells to nutlin- apoptosis. Proc Natl Acad Sci USA 91: 3754–3758. 3-induced apoptosis. Exp Cell Res 312: 2394–2400. Miliani de Marval PL, Macias E, Rounbehler R, Sicinski P, Robles AI, Rodriguez-Puebla ML, Glick AB, Trempus C, Kiyokawa H, Johnson DG et al. (2004). Lack of cyclin- Hansen L, Sicinski P et al. (1998). Reduced skin tumor dependent kinase 4 inhibits c-myc tumorigenic activities in development in cyclin D1-deficient mice highlights the epithelial tissues. Mol Cell Biol 24: 7538–7547. oncogenic ras pathway in vivo. Genes Dev 12: 2469–2474. Mittnacht S. (1998). Control of pRB phosphorylation. Current Rodriguez-Puebla ML, Miliani de Marval PL, LaCava M, Opinion in Genetics & Development 8: 21–27. Moons DS, Kiyokawa H, Conti CJ. (2002). Cdk4 Miura T, Mattson MP, Rao MS. (2004). Cellular lifespan and deficiency inhibits skin tumor development but does not senescence signaling in embryonic stem cells. Aging Cell 3: affect normal keratinocyte proliferation. Am J Pathol 161: 333–343. 405–411. Moffat J, Grueneberg DA, Yang X, Kim SY, Kloepfer AM, Rossi AG, Sawatzky DA, Walker A, Ward C, Sheldrake TA, Hinkle G et al. (2006). A lentiviral RNAi library for human Riley NA et al. (2006). Cyclin-dependent kinase inhibitors

Oncogene Cdk functions in different cell types C Berthet and P Kaldis 4477 enhance the resolution of inflammation by promoting Stead E, White J, Faast R, Conn S, Goldstone S, Rathjen J inflammatory cell apoptosis. Nat Med 12: 1056–1064. et al. (2002). Pluripotent cell division cycles are driven by Rouault JP, Falette N, Guehenneux F, Guillot C, Rimokh R, ectopic Cdk2, cyclin A/E and E2F activities. Oncogene 21: Wang Q et al. (1996). Identification of BTG2, an anti- 8320–8333. proliferative p53-dependent component of the DNA damage Tetsu O, McCormick F. (2003). Proliferation of cancer cells cellular response pathway. Nat Genet 14: 482–486. despite CDK2 inhibition. Cancer Cell 3: 233–245. Sage J, Mulligan GJ, Attardi LD, Miller A, Chen S, Williams B Tsutsui T, Hesabi B, Moons DS, Pandolfi PP, Hansel KS, et al. (2000). Targeted disruption of the three Rb-related Koff A et al. (1999). Targeted disruption of CDK4 delays genes leads to loss of G1 control and immortalization. Genes cell cycle entry with enhanced p27Kip1 activity. Mol Cell Biol Dev 14: 3037–3050. 19: 7011–7019. Santamaria D, Ortega S. (2006). Cyclins and CDKs in Tyner SD, Venkatachalam S, Choi J, Jones S, Ghebranious N, development and cancer: lessons from genetically modified Igelmann H et al. (2002). p53 mutant mice that display early mice. Front Biosci 11: 1164–1188. ageing-associated phenotypes. Nature 415: 45–53. Sauter ER, Nesbit M, Litwin S, Klein-Szanto AJ, Cheffetz S, Vousden KH, Lu X. (2002). Live or let die: the cell’s response Herlyn M. (1999). Antisense cyclin D1 induces apoptosis to p53. Nat Rev Cancer 2: 594–604. and tumor shrinkage in human squamous carcinomas. Wadhwa R, Kaul SC, Mitsui Y. (1994). Cellular mortality to Cancer Res 59: 4876–4881. immortalization: mortalin. Cell Struct Funct 19: 1–10. Savatier P, Huang S, Szekely L, Wiman KG, Samarut J. Wesierska-Gadek J, Schreiner T, Gueorguieva M, Ranftler C. (1994). Contrasting patterns of retinoblastoma protein (2006). Phenol red reduces ROSC-mediated cell cycle arrest expression in mouse embryonic stem cells and embryonic and apoptosis in human MCF-7 cells. J Cell Biochem 98: fibroblasts. Oncogene 9: 809–818. 1367–1379. Savatier P, Lapillonne H, van Grunsven LA, Rudkin BB, Williams O, Gil-Gomez G, Norton T, Kioussis D, Brady HJ. Samarut J. (1996). Withdrawal of differentiation inhibitory (2000). Activation of Cdk2 is a requirement for antigen- activity/leukemia inhibitory factor up-regulates D-type mediated thymic negative selection. Eur J Immunol 30: cyclins and cyclin-dependent kinase inhibitors in mouse 709–713. embryonic stem cells. Oncogene 12: 309–322. Yu Q, Geng Y, Sicinski P. (2001). Specific protection Sharpless NE, Bardeesy N, Lee KH, Carrasco D, against breast cancers by cyclin D1 ablation. Nature 411: Castrillon DH, Aguirre AJ et al. (2001). Loss of p16Ink4a 1017–1021. with retention of p19Arf predisposes mice to tumorigenesis. Yu Q, Sicinska E, Geng Y, Ahnstrom M, Zagozdzon A, Nature 413: 86–91. Kong Y et al. (2006). Requirement for CDK4 kinase Sherr CJ, Roberts JM. (1999). CDK inhibitors: positive and function in breast cancer. Cancer Cell 9: 23–32. negative regulators of G1-phase progression. Genes Dev 13: Zhou P, Jiang W, Zhang YJ, Kahn SM, Schieren I, 1501–1512. Santella RM et al. (1995). Antisense to cyclin D1 inhibits Sicinski P, Donaher JL, Parker SB, Li T, Fazeli A, Gardner H growth and reverses the transformed phenotype of human et al. (1995). Cyclin D1 provides a link between development esophageal cancer cells. Oncogene 11: 571–580. and oncogenesis in the retina and breast. Cell 82: 621–630. Zou X, Ray D, Aziyu A, Christov K, Boiko AD, Gudkov AV Stanelle J, Putzer BM. (2006). E2F1-induced apoptosis: et al. (2002). Cdk4 disruption renders primary mouse cells turning killers into therapeutics. Trends Mol Med 12: resistant to oncogenic transformation, leading to Arf/p53- 177–185. independent senescence. Genes Dev 16: 2923–2934.

Oncogene