TO CYCLE OR NOT TO CYCLE: A CRITICAL DECISION IN CANCER

Transcription

1 TO CYCLE OR NOT TO CYCLE: A CRITICAL DECISION IN CANCER Marcos Malumbres and Mariano Barbacid Tumour cells undergo uncontrolled proliferation, yet tumours most often originate from adult tissues, in which most cells are quiescent. So, the proliferative advantage of tumour cells arises from their ability to bypass quiescence. This can be due to increased mitogenic signalling and/or alterations that lower the threshold required for cell-cycle commitment. Understanding the molecular mechanisms that underlie this commitment should provide important insights into how normal cells become tumorigenic and how new anticancer strategies can be devised. MILESTONES METAZOANS The kingdom Animalia (animals) that consists of roughly 35 phyla of multicellular organisms. Molecular Oncology rogram, Centro Nacional de Investigaciones Oncológicas, Melchor Fernández Almagro 3, Madrid, Spain. s: [email protected]; [email protected] In METAZOANS, one of the most crucial decisions that every proliferating cell must make is whether to continue another round of cell division or to exit the cell cycle and reach a quiescent state. Similarly, quiescent cells must decide whether to remain in their non-proliferative state or to re-enter the cell cycle. During early development, few cells abandon the cycle, but most of the cells in an adult organism are quiescent. Only specialized cells, such as those that populate the haematopoietic system or line the gut epithelium, maintain active proliferation. All cells have the capacity to enter quiescence and all quiescent cells except those that have reached a state of terminal differentiation have the capacity to re-enter the cycle, so what are the basic mechanisms that control these important decisions and how do small imbalances in these regulatory pathways lead to cancer? Cancer cells often show alterations in the signaltransduction pathways that lead to proliferation in response to external signals. Indeed, many growth factors and their receptors, as well as their membrane, cytoplasmic and nuclear downstream effectors have been identified as oncogenes or tumour-suppressor genes. Other genes mutated in cancer include those that inactivate apoptotic pathways, induce genomic instability and promote angiogenesis. Tumour-associated mutations in many of these molecules result in the alteration of the basic regulatory mechanisms that control the mammalian cell cycle. The basic cell cycle is divided into four phases (reviewed in REF. 1). During two of these phases, cells execute the two basic events in cell division: generation of a single and faithful copy of its genetic material (the synthetic or S phase) and partitioning of all the cellular components between two identical daughter cells (mitosis or M phase). The two other phases of the cycle G1 and G2 represent gap periods, during which cells prepare themselves for the successful completion of the S and M phases, respectively. When cells cease proliferation, either due to specific antimitogenic signals or to the absence of proper mitogenic signalling, they exit the cycle and enter a non-dividing, quiescent state known as G0. To ensure proper progression through the cell cycle, cells have developed a series of checkpoints that prevent them from entering into a new phase until they have successfully completed the previous one 2. It is likely that newly divided or quiescent cells must also pass certain checkpoints before they can enter the cycle. For instance, cells must make sure that they have reached their homeostatic size, otherwise cells will become smaller with each round of division. Metazoans must also control the number of cells in every organ, which coupled with cellular size determines the size of the organism. The genetic elements that control these parameters remain largely unknown. However, they can be influenced by external elements, such as the amount of available nutrients or the intensity of the mitogenic information that cells receive at any given time (see BOX 1). Indeed, these two parameters are key determinants for cell division. For instance, if cells have mitogenic requirements that are too stringent, they might not be able to proliferate at crucial times, such as during 222 DECEMBER 2001 VOLUME 1

2 Box 1 Cell growth versus cell division Cell growth (the increase in cell size and protein mass) is a term that has frequently been misused to mean cell proliferation. In fact, both processes are highly coordinated. Only in certain biological systems such as oocytes, neurons and muscle cells, where cell growth might exist without cell division, and in fertilized eggs, where cell divisions might occur without cell growth can these processes function in an independent, or even complementary, fashion. In most cells, however, cell division without concurrent cell growth would generate smaller daughter cells, which would affect their viability. Ribosome biosynthesis is a key process for cell growth. Before entering the cycle, cells need to accumulate sufficient translational machinery, mainly ribosomes, to ensure the rapid processing of transcripts through the cycle. This is accomplished, at least in part, by phosphorylation of the ribosomal S6 protein by S6 kinase (S6K) (see REF. 85 for a review). Once the appropriate pool of ribosomes has been achieved, the system is desensitized, either by negative regulators of S6K or by the size of the ribosomal pool (see figure). S6K is regulated by mitogenic stimuli mediated through the insulin receptor (IR)/IR substrate (IRS)/phosphatidylinositol-3 kinase (I3K)/DK1 pathway. S6K is also regulated directly by TOR,a member ofthe I3K-related kinase family 86,87. TOR is thought to be important in cell growth and amino-acid sensing 87, but its upstream activators and mechanism of Growth factors activation are unknown. TOR controls several Nutrients growth-related readouts, including actin organization, transcription and ribosome biosynthesis. TOR also affects translation of key regulators of DK1 IR cell proliferation, such as cyclin D and MYC, by IRS RAS phosphorylating 4E B1 (a translational inhibitor S6K TOR I3K that is also targeted by AKT/KB) and causing its Ribosome dissociation from the initiation factor eif4e. biosynthesis AKT Mitogen-activated protein kinases such as ERK S6 phosphorylate and activate MNK1, which in turn G0 4E-B1 40S is able to phosphorylate eif4e (REFS 88,89).The 3 60S MNK1 RAS/ERK cascade is also known to signal to cellcycle ERK regulators such as cyclin D or to induce elf4e progression through G1 (REF. 12). Several of these G1 5 proteins, such as I3K, AKT, MYC and RAS, can be Translation activated as oncogenes, which illustrates the Cell-cycle intimate connections between cell growth and progression Cyclin D cell proliferation. wound healing or during an infection. Conversely, loosening these controls might lead to unscheduled proliferation and possibly neoplastic growth. In culture, cells undergo a period of mitogen dependence before they enter the cycle. This transition the Restriction oint (R) represents a point of no return that commits cells to a new round of cell division (see BOX 2). Understanding the molecular mechanisms that allow cells to enter the cycle should provide important clues as to the determinants of normal versus abnormal proliferation in multicellular organisms. The molecular players Cyclin-dependent kinases. Some of the molecules that control the early events of the cell cycle have been extensively characterized. The central players are the cyclin-dependent kinases (CDKs), a group of serine/threonine kinases that form active heterodimeric complexes following binding to cyclins, their regulatory subunits (reviewed in REFS 3,4; see ONLINE TABLE 1). Several CDKs mainly CDK4, CDK6, CDK2 and possibly CDK3 cooperate to drive cells through G1 into S phase. CDK4 and CDK6 are thought to be involved in early G1 (see below), whereas CDK2 is required to complete G1 and initiate S phase. CDK4 and CDK6 form active complexes with the D-type cyclins (cyclins, D2 and D3). These kinases are highly related and, so far, have resisted functional differentiation, except for their distinct pattern of activation 5 6. Characterization of mice that are defective for the genes encoding these CDKs should provide information regarding their distinct roles in vivo 7 8. CDK2 is sequentially activated by the E-type cyclins cyclin and E2 during the G1/S transition, and the A-type cyclins cyclin A1 and A2 during S phase 4. The role of CDK3 is still obscure, mainly due to its low expression levels. Moreover, this enzyme is truncated, and presumably inactive, in several strains of inbred mice, indicating that it is dispensable for the cell cycle 9. CDK regulators. By controlling the level of CDK activity, CDK regulators can also control cell-cycle commitment (ONLINE TABLE 1). They include activators, mainly the cyclins, and inhibitors, generically known as CKIs (CDK kinase inhibitors; see below). CDKs are also regulated by phosphorylation (reviewed in REFS 3,10,11): they must be NATURE REVIEWS CANCER VOLUME 1 DECEMBER

3 Box 2 The Restriction oint The term Restriction oint was coined in 1974 by Arthur ardee 90 to define a specific event in G1 after which cultured cells could proliferate independently of mitogenic stimuli. Briefly, cultured mammalian cells that had undergone mitosis within the previous 3 hours could be prevented from progressing through the cell cycle by growth-factor starvation or moderate inhibition of protein synthesis. These cells then re-entered the cell cycle after re-stimulation with growth factors. However, if the cells had undergone cell division more than 4 hours before, they did not respond to mitogen deprivation and advanced through the cell cycle with the same kinetics as unstarved cells. It was postulated that the latter cells had passed the restriction point (R). Today, R is often used to divide the early and late G1 phases. R does not represent a checkpoint as originally defined in yeast 2,91. In culture cells, R occurs 3 4 hours after mitosis (see figure). However, entry into S phase is usually initiated 5 13 hours after mitosis. This variability is characteristic of the late G1 phase and accounts for most of the observed differences in the length of the cell cycle. Indeed, the differential kinetics of these two transitions indicates distinct control mechanisms. The molecular Remove growth factors events that allow cells to pass R have not been well defined. However, G0 Mitogens G0 members of the family are likely to be important, as ablation of this gene family eliminates R 92,93. M Early G1 M It has been postulated that loss of regulation of R is critical in cancer. Early G1 R normally prevents cells from entering the cycle until they have Growth-factor accumulated a certain threshold of mitogen-induced events, so loosening R R removal of R control due to mutations in G1 regulators or other, as yet unidentified, Late G1 genes would allow cells to enter the cycle even in the absence of adequate G2 Late G1 G2 mitogenic signalling, leading to unscheduled proliferation. Validation of S S this model will require definition of the molecular players that regulate R. T LOO A short peptide sequence located between subdomains VII and VIII of protein kinases that folds as a loop. hosphorylation might alter its conformation and, hence, the catalytic activity of the protein kinases. RAS/RAF/MAK ATHWAY Signal-transduction pathway that is crucial for the integration of mitogenic signals mediated by a variety of growth factors. Activation of this pathway is involved in many cellular processes, including cell-cycle progression. SCF UBIQUITIN LIGASE Multiprotein complex that ligates ubiquitin to proteins that will be degraded by the proteasome. SCF complexes consist of three core subunits (SK1, CUL1 and X1) coupled to one of the several F- box proteins that recognize the protein to be ubiquitylated. RAS/I3K/AKT ATHWAY Signal-transduction pathway that is involved in many cellular processes, such as cell proliferation and apoptosis. Several members of this pathway have oncogenic or tumour-suppressor activities (for example, RAS, I3K and TEN). phosphorylated on a threonine residue (T172 in CDK4 and T160 in CDK2) located in their T LOO for proper catalytic activity. This is carried out by the CDK7 cyclin-h complex (also known as CAK; CDKactivating kinase), a serine/threonine kinase that is also involved in transcription and DNA repair 10. Inhibitory phosphorylation of adjacent threonine and tyrosine residues (T14/Y15 in CDK1) is mediated by dualspecificity kinases (such as WE and MYT1). This inhibition is relieved when the CDC25 phosphatases (CDC25A, CDC25B and CDC25C) dephosphorylate these residues, which triggers entry into mitosis 3,11. D-type cyclins. The D-type cyclins are important integrators of mitogenic signalling, as their synthesis is one of the main end points of the RAS/RAF/MAK ATHWAY (reviewed in REF. 12). Whether availability of sufficient amounts of D-type cyclins makes cells independent of mitogenic stimuli, and so allows them to pass through the R point, remains to be examined. Cyclin is a rather unstable molecule that is transported from the nucleus to the cytoplasm where it is targeted by the SCF UBIQUITIN LIGASE for proteolysis. Nuclear export is mediated by glycogen synthase kinase 3-β (GSK-3β), a kinase that is inhibited by the RAS/I3K/AKT ATHWAY 13,14. So, cyclin- availability is controlled by a balance between the RAS/RAF/MAK (cyclin- synthesis) and RAS/I3K/AKT (cyclin- stability) mitogenic pathways, as well as by GSK-3β and SCF activity (cyclin- degradation). Interestingly, cells that lack cyclin can proliferate, indicating that this molecule is not strictly necessary for G1 (REFS 15,16). As the three D-type cyclins show high sequence homology and are coexpressed in many tissues, it is possible that they have at least some redundant roles. It is, however, evident (see below) that overexpression of cyclin facilitates passage through G1 and promotes cell proliferation. CKIs. CKIs are of two types(reviewed in REF. 17). The four members of the INK4 family (also known as p16), INK4B(also known as p15), INK4C (also known as p18) and INK4D (also known as p19) exert their inhibitory activity by binding to the CDK4 and CDK6 kinases and preventing their association with D-type cyclins. Although INK4 proteins are biochemically indistinguishable from each other in vitro, the generation of gene-targeted mice that are deficient for each of these proteins has revealed significant functional differences (see ONLINE TABLE 1). The three members of the WAF/KI family WAF1 (also known as p21), (also known as p27) and KI2 (also known as p57) form heterotrimeric complexes with the G1/S CDKs. However, in stoichiometric amounts, they only inhibit the kinase activity of CDK2 cyclin-e complexes (see below). CKIs are induced in response to different cellular processes 4,17. For instance, WAF1 is one of the effectors of p53, a tumour suppressor that is important in the DNA-damage checkpoint a crucial process for human cancer that has been extensively reviewed elsewhere and will not be considered here. CDK substrates. The primary substrates of CDK4/6 and CDK2 in G1 progression are the members of the retinoblastoma protein family, p107 and p130 (reviewed in REFS 3,18). These molecules function as docking sites for a series of proteins that must be tightly regulated throughout the cell cycle. For instance, proteins bind to the E2F family of transcription factors to ensure that they remain inactive during M and G0. In addition, E2F complexes participate in active repression of some promoters: a mechanism that involves other protein families, such as histone deacetylases and chromosomal remodelling SWI/SNF complexes 19. So far, more than 100 -binding proteins have been identified 20, but the physiological relevance of most of these 224 DECEMBER 2001 VOLUME 1

4 REVIEWS M G2 p130 D E2F G0 S R D E2F Early G1 Late G1 DOMINANT-NEGATIVE MUTANT A non-functional mutant protein that competes with the normal, non-mutated protein, blocking its activity. ROTEASOME A 26S multiprotein complex that catalyzes the breakdown of polyubiquitylated proteins. D E2F Cyc E WAF1/KI CDK4/6 CDK2 Cyc E Cyc D Mitogens INK4 Antimitogens WAF1/KI Figure 1 Regulation of G1 and the G1/S transition. In quiescent, G0 cells, E2F D transcription factors are bound to p130, the principal pocket protein in these cells, which keeps them inactive. In G1, however, E2F D complexes predominate. Mitogenic signalling results in cyclin D (Cyc) synthesis, formation of active CDK4/6 cyclin-d complexes and initial phosphorylation of. artially phosphorylated still binds to E2F D, but the transcription factor is still able to transcribe some genes, such as cyclin E, presumably due to impaired repression. Cyclin E binds to and activates CDK2. It is generally accepted that CDK2-dependent phosphorylation of results in its complete inactivation, which allows induction of the E2F-responsive genes that are needed to drive cells through the G1/S transition and to initiate DNA replication. INK4 and WAF1/KI proteins can inhibit CDK4/6 or CDK2 kinases, respectively, following specific antimitogenic signals. The CDK4/6 complexes can also bind WAF1/KI inhibitors, while remaining active. This sequesters them from CDK2, which facilitates its full activation. R represents the restriction point that separates the mitogendependent early G1 phase from the mitogen-independent late G1 phase. interactions has not been established. The activity of the proteins is modulated by sequential phosphorylation by CDK4/6 cyclin-d and CDK2 cyclin-e complexes can also be regulated by acetylation, mediated by histone acetylases associated to p300/cb. These acetylases are under cell-cycle control and prevent efficient phosphorylation by CDK2 cyclin E (REF. 24). Hyper-phosphorylated proteins release the molecules that bind to their hypophosphorylated isoforms, allowing them to carry out their specific tasks in cell-cycle progression. CDK2 also phosphorylates other substrates that are involved in DNA replication 4,17. G1 regulation in normal cells During early G1, cells integrate external information that is derived from mitogenic stimuli and nutrient availability to prepare themselves for passing through the various phases of the cell cycle (see BOX 1). Following generation of sufficient amounts of D-type cyclins and the subsequent CDK4/6 cyclin-d complexes, and possibly other members of the family, depending on the cell type becomes partially phosphorylated 21 23, which results in the partial activation of E2F D transcription factors. These participate in the generation of molecules required for the G1/S transition, such as cyclin E (reviewed in REF. 20; FIG. 1). can be dephosphorylated by the 1 phosphatase, which restores growth-suppressing function after mitosis (reviewed in REF. 25). Is CDK4/6-mediated phosphorylation of an irreversible process that releases cells from mitogenic dependence? This as well as other hypotheses which propose that CDK4/6 kinases are important in early G1 regulation should be re-examined in light of recent results indicating that CDK4 might be dispensable for cell-cycle commitment and proliferation (BOX 3). Cdk2 has not been ablated in mice or flies. However, expression of a Cdk2 DOMINANT-NEGATIVE MUTANT (unlike similar mutants for Cdk4 and Cdk6) blocks cultured cells in G1, which illustrates the need for Cdk2 activity at the G1/S transition 26. As indicated above, CDK2 is sequentially activated by E-type and A-type cyclins. CDK4 and presumably CDK6 contributes to CDK2 activation by at least two different mechanisms. First, cyclin- synthesis is mediated by E2F D following phosphorylation by CDK4/6 cyclin D 22,27. Second, CDK4/6 cyclin-d heterodimers can bind WAF1/KI inhibitors without losing their kinase activity This interaction titrates these inhibitors away from CDK2 cyclin-e complexes, which activates the CDK2 kinase 17 (FIG. 1). These mechanisms have been at least partially corroborated in vivo by experiments with gene-targeted mice Replacement of the gene that encodes cyclin by cyclin (REF. 31) restores most of the phenotypic defects of cyclin--defective animals 15,16, and as cyclin cannot activate CDK4 (or CDK6), the most probable explanation is that the primary role of cyclin in vivo is to activate CDK2 cyclin. Similarly, loss of Cdkn1b (the gene that encodes Kip1) in mice that lack cyclin restores some, but not all, of their phenotypic defects 32,33. This differential effect could be due to the presence of Waf1, a protein that might have a redundant function with Kip1. Cells do not rely exclusively on CDK4/6 activity to fully activate CDK2. Indeed, is phosphorylated by CDK2 as well as by other, as yet unknown, kinases an event that triggers its ROTEASOME-mediated degradation 34,35. The decrease in CDK2 activity at the G1/S transition is mainly controlled by degradation of the E-type cyclins, for which a specific ubiquitin ligase, CDC4 /FBW7/AGO, has been recently identified Cell-cycle mutations in human tumours Molecular analysis of human tumours has shown that cell-cycle regulators are frequently mutated in human neoplasias (FIG. 2), which underscores how important the maintenance of cell-cycle commitment is in the prevention of human cancer 4. Alterations include overexpression of cyclins (mainly and ) and CDKs (mainly CDK4 and CDK6), as well as loss of CKI (mainly, INK4B and ) and expression. Tumour-associated changes in the expression of these regulators frequently result from chromosome alterations (amplification of cyclin or CDK4, translocation of CDK6 and deletions of INK4 proteins or ) or epigenetic inactivation (methylation of INK4 or promoters). Miscoding mutations in CDK4 and CDK6, resulting in loss of INK4 binding, have also been identified, albeit with low frequency 39,40. Genetic or epigenetic alterations of CDK2 or its regulators have rarely been described in human tumours, yet NATURE REVIEWS CANCER VOLUME 1 DECEMBER

5 >80% Glioma/blastoma >80% Breast INK4B CDK6 CDK4 >90% Lung >80% ancreas D3 >80% Endometrium 70% Bladder CDK4 p130 INK4B CDK4 p130 >90% Gastrointestinal CDK4 p130 D2 CDK2 CDK4 >90% Bone marrow (leukaemia) INK4B CDK4 >80% ituitary D3 CDK6 CDK4 CDK4 p130 INK4B INK4B CDK6 CDK4 >90% Head and neck D2 D3 p130 p15 >90% Lymphoma >20% Melanoma CDK2 D2 p16 70% rostate >90% Liver >90% Testis/ovary >80% Bone (osteosarcoma) CDK4 90% Other sarcomas Figure 2 Mutation of G1/S regulators in human cancer. Only alterations that occur in more than 10% of primary tumours have been considered. Numbers represent the percentage of tumours with alterations in any of the listed cell-cycle regulators. The loci in which specific genetic or epigenetic alteration have been defined are in pink. The alterations for which no mechanistic explanation has been provided are in yellow. Alterations relevant for tumour prognosis are indicated by asterisks. loss of CDKN1B (the gene that encodes in humans) expression, and overexpression of cyclin frequently occur and correlate with poor prognosis (FIG. 2). How and cyclin- expression is altered in tumour cells is not yet well understood. degradation is mediated, at least in part, by SK2 an F-BOX ROTEIN of the SCF complex. SK2 has oncogenic properties and its expression is altered in some tumour cells 41,42. Similarly, Box 3 CDK4: a regulator of cell growth, cell number or cell proliferation? F-BOX ROTEIN Component of the machinery for the ubiquitin-dependent degradation of proteins. F-box proteins recognize specific substrates and present them to the SCF complex and the proteasome machinery. Deletion of Cdk4 in Caenorhabditis elegans, Drosophila melanogaster and mice has raised important issues regarding the role of this kinase in cell proliferation. In C. elegans, cyclin D or Cdk4 proteins are specifically dispensable for embryonic development, a process that involves cell division in the absence of growth. However, they are required for post-embryonic development, as both cell growth and division are needed 94.In D. melanogaster, Cdk4 ablation affects cell number 95. Cdk4 mutant flies develop normally and eclose at the same time as normal flies. However, they are smaller in size due to the presence of fewer cells, rather than to a decrease in cell size. The slower growth rate could mean that cells take longer to reach the size requirement for division. So, Cdk4 seems to regulate homeotic cell numbers, but is not necessary for cell proliferation. In mice, Cdk4 is also dispensable for proliferation in most cell types 7,8. Cdk4-null mice are also significantly smaller. Whether this phenotype is due to the generation of fewer cells or to hormonal or other physiological abnormalities is not known. roliferating mouse embryo fibroblasts (MEFs) that lack Cdk4 have normal (or almost normal) cell-cycle profiles 7,8, which could be a consequence of the redundant roles of Cdk4 and Cdk6 in mice. Conversely, gain-of-function experiments indicate that Cdk4 and cyclin stimulate cell proliferation. For instance, fruitflies that overexpress Cdk4 cyclin D display hyperplastic proliferation of normal-size cells in some tissues 95,96. Cyclin- overexpression also induces tumours 60, and is a key mediator of breast-cancer induction by Ras or Neu oncogenes 97. Similarly, mice that express a knocked-in Cdk4 mutant that cannot be downregulated by INK4 inhibitors develop tumours with almost complete penetrance 58 and have a high susceptibility to develop melanomas 59. These results indicate that although Cdk4 is dispensable for cell proliferation, its misregulation facilitates cell proliferation and contributes to neoplastic development (see FIG. 2). 226 DECEMBER 2001 VOLUME 1

6 Box 4 Other cell-cycle mutations in cancer Losing control of the commitment to cell division seems to be an important step in the progression of human cancer. However, it is possible that misregulation of other cell-cycle events might also contribute to neoplasia. After all, chromosomal instability and aneuploidy, two prominent consequences of defective M-phase checkpoints, are hallmarks of cancer cells. The proper distribution of chromosomes is controlled at the metaphaseto-anaphase transition by the spindle-assembly checkpoint (see figure, a; and REFS 98,99 for recent reviews). The gatekeeper of this checkpoint is the anaphase-promoting complex (AC), a multisubunit ubiquitin ligase that, with CDC20, targets proteins such as securin for degradation. If the kinetochores are correctly attached to the spindle, AC is activated and securin is degraded. This activates the separin protease, which cleaves the cohesin molecules between sister chromatids, allowing them to separate. However, if the kinetochores are not correctly attached to the spindle, AC is kept in an inactive state and cells arrest in metaphase (see figure, b). AC activity is primarily regulated by MAD2, which senses microtubule attachment to the kinetochores. Although the mechanisms that regulate MAD2 activity are not fully understood, they have implicated, among other proteins, the BUB family of kinases 99. BUB1 has been found mutated in colon cancer cell lines 100. Similarly, mice heterozygous for Mad2 develop lung tumours after a long latency 101. Yet, a search for genetic mutations of these molecules in human tumours has failed to yield positive results 102. Additional molecules involved in the spindleassembly checkpoint have been implicated in a Metaphase anaphase b Metaphase arrest human cancer. They include polo-like kinase 1 transition LK1 (LK1) and two members of the Aurora family of Aurora protein kinases Aurora-A and -B 99, Finally, misregulation of CDK1, an essential Sister chromatids Sister chromatids MAD2 regulator of mitosis, could also be involved in CDC20 BUB CDC20 tumour development. Lats1, a protein that directly AC AC binds and negatively regulates Cdk1 (REF. 107), is a tumour suppressor in mice 108. Lats1-deficient mice Ub develop soft-tissue sarcomas, ovarian tumours and Normal spindle Separin show pituitary dysplasia. These results indicate Improper spindle that Lats1 could act as a Cdk1 inhibitor and that its Separin disappearance might lead to uncontrolled proliferation. Whether human tumours harbour LATS1 mutations has not been addressed as yet. In summary, evidence indicating that malfunction of mitotic events contributes to human cancer is still scarce. However, the likelihood that errors in critical M-phase checkpoints might result in chromosomal abnormalities is keeping cancer researchers alert. Securin Securin ORGANOMEGALY A phenotype characterized by organs of abnormally large size. HALO-INSUFFICIENCY A phenotype that arises in diploid organisms due to the loss of only one allele. the F-box protein responsible for cyclin- degradation CDC4/FBW7/AGO is inactivated in some cancer cell lines 36,37. Alternatively, cancer-associated cyclin-e overexpression could be explained by inactivation, which is mediated by deregulated CDK4/6 activity. Recently, a new cell-cycle inhibitory protein that enhances WE-dependent phosphorylation of CDK2, named Cables, has been reported to be inactivated in 50 60% of primary colon and head and neck tumours 43. Loss or inactivation of is a frequent event in human tumours. However, p130 is less frequently lost and p107 inactivation has not been reported (reviewed in REF. 44). These observations indicate that p107 and p130 might be important for differentiation rather than proliferation, a concept supported by results obtained with gene-targeted mice (see below). Mutations in the E2F family of transcription factors have not been observed in human tumours. Recently, mutations in cell-cycle regulators that do not affect cell growth or cell proliferation have been identified (see BOX 4). These mutations appear in molecules that are involved in the control of sister-chromatid separation during mitosis, and are likely to induce aneuploidy as well as other chromosomal alterations, that will contribute to the transformed phenotype. Animal models Genetically modified mice have provided new insights into the roles of cell-cycle regulators in both normal and pathological processes. Early gene-targeting studies showed that Rb deficiency is lethal in mouse embryos, whereas Rb +/ mice develop tumours of endocrine origin (pituitary, thyroid and adrenal medulla) 45,46. By contrast, deletion of the other members of this family p107 or p130 has few or no effects on mouse embryonic development or tumour susceptibility (reviewed in REF. 47). Combined deficiency of Rb +/ and p107, however, causes retinoblastoma in the mouse, showing that these proteins partially overlap in tumour suppression 48. Ablation of results in ORGANOMEGALY and a high penetrance of pituitary tumours In addition,this protein is HALO-INSUFFICIENT for tumour suppression 52, indicating a functional effect for the low levels of expression that are frequently observed in human tumours (FIG. 2). Mice deficient fo Cdkn1a (which encodes Waf1) also develop a variety of tumours (mainly sarcomas and B-cell lymphomas), although these tumours have a long latency period 53.Genetic studies have also shown that deletion of Cdkn2a (the gene that encodes Ink4a in mice) 54,55 or Cdkn2b (the gene that encodes Ink4b in mice) 56 results in low, NATURE REVIEWS CANCER VOLUME 1 DECEMBER

7 Table 1 Small-molecule CDK inhibitors Compound Family Specificity Features References Olomoucine urines CDK1, CDK2 and CDK5 >> CDK4 Growth arrest in G1/S and G2/M. Induces apoptosis. 110 Roscovitine urines CDK1, CDK2 and CDK5 >> CDK4 Growth arrest in G1/S and G2/M. Induces 110 apoptosis and differentiation in some cases. urvalanol urines CDK1, CDK2, CDK5 Growth arrest in G1/S and G2/M. 75 NU2058 and urine/ pyrimidine CDK1, CDK2 Growth inhibition of tumour cells. 111 NU6027 UCN-01 Alkaloids CDK1, CDK2, CDK4 Growth inhibition and apoptosis. Induces G1/S and 80 G2/M arrest. Some activity in clinical trials against melanoma, lymphoma and sarcoma. Indirubin-5- Indirubins CDK1 > CDK5 > CDK2 >> CDK4 The founder of this class, indirubin, is the active 79 sulphonic acid component of a traditional Chinese recipe against leukaemia. Flavopiridol Flavonoid CDK1, CDK2, CDK4 Growth arrest in G1/S and G2/M. Also 80 induces apoptosis and differentiation. Clinical trials have shown some activity against lymphoma, and colon and renal cancer. Kenpaullone aullones CDK1 > CDK2 >> CDK4 Delayed cell-cycle progression. Inhibits growth of 112 tumour cells in vitro. Butyrolactone I Natural product CDK1 > CDK2 Isolated from Aspergillus. Growth arrest in 113 G1/S and G2/M. Also induces apoptosis and differentiation is some systems. Hymenialdisine Natural product CDK1, CDK5 > CDK2 >> CDK4 Isolated from a marine sponge. 114 SU9516 Indolinone CDK2 Decrease in phosphorylation and caspase activation. Causes G0/G1 and G2/M block. CINK4 Triamino- CDK4/6 cyclin Causes dephosphorylation and growth arrest 116 pyrimidine in culture and smaller tumours in a xenograft model. D yrido-pyrimidine CDK4, CDK6 > CDK2 G1 arrest in -positive cells, in correlation with 117 hypophosphorylation. Fascaplysin Natural product CDK4 G1 arrest and dephosphorylation at sites that are 118 specific for CDK4 kinase. Asterisks refer to compounds tested in clinical trials. CDK, cyclin-dependent kinase;, retinoblastoma. although significant, susceptibility to tumour development. Conversely, mice that are null for Cdkn2c (the gene that encodes Ink4c in mice) develop pituitary tumours with high penetrance 56,57. Importantly, knockin mice that express a Cdk4 mutant originally found in human melanoma 39 and that is insensitive to INK4 inhibitors (Cdk4 R24C) develop several tumours in a variety of tissues 58 and display high susceptibility to melanoma development 59. Similarly, overexpression of cyclin or cyclin in transgenic mice also leads to tumour development 60,61. As observed in human cancer (FIG. 2), alterations in different families of cell-cycle regulators cooperate in tumour development. So, deficiency in cooperates with ablation of WAF1/KI proteins 62,63. In addition, lack of WAF1 or synergizes with the absence of INK4 proteins 57,64 or the CDK4 R24C mutation 58. Therapeutic strategies The frequent loss of G1 regulation in human cancer has revealed targets for possible therapeutic intervention. Indeed, restoring proper R-point control to cancer cells might allow them to return to a quiescent state. Alternatively, we could take advantage of their uncontrolled proliferation to facilitate apoptotic death 65 or to specifically expose cancer cells to cytotoxic treatments 66. However, the best strategies to accomplish these goals have not been determined, and will require proper experimentation in adequate animal models (see above) and, ultimately, in clinical trials. CDKs are actively being targeted, due to their central role in the control of cell-cycle progression. The complexity of CDK regulation offers a number of possible routes to their therapeutic inhibition (FIG. 3). The most direct and promising strategies involve designing inhibitors that block CDK activity (TABLE 1). The clinical success of STI-571 an inhibitor of the ABL kinase in the treatment of chronic myelogenous leukaemia (CML) has provided hope that this type of strategy can work. Other strategies that have good probabilities of success include preventing cyclin CDK interaction with peptidomimetics or multifunctional chemical structures for example, derived from a structure activity relationship by nuclear magnetic resonance analysis 67, restoring CKI function by gene therapy 68, or by using peptidomimetics 69 and preventing CKI degradation. In those cases in which CKI loci have been silenced by epigenetic events, CKI expression could be restored by demethylating agents such as 5 - deoxyazacytidine (5-dAZA) that prevent promoter methylation 5-dAZA has antitumour activity in leukaemia and other neoplasias 70.Alternative strategies to prevent degradation of CKIs such as could involve 228 DECEMBER 2001 VOLUME 1

8 AN-CDK INHIBITORS CDK inhibitors that do not discriminate between different CDKs. These drugs can frequently inhibit other protein kinases. Restore CKI activity CKI Cyclin c revent CKI degradation e d Inhibit cyclin synthesis Inhibit kinase activity a CDK AT WE either blocking phosphorylation, which triggers its degradation, or its interaction with SK2, one of the members of the SCF complex. Some proteasomal inhibitors such as S-341 have shown interesting results in clinical trials, in spite of their obvious lack of specificity 71. Other strategies that have been less explored and/or are less likely to succeed include decreasing levels of cyclin or cyclin E by reducing specific transcription of these genes by antisense techniques 72 or increasing degradation of these proteins by stimulating signals (phosphorylation) recognized by their specific proteasomes. Downregulating positive regulators, such as the CDK-activating kinase (CAK) or the CDC25 dualspecificity phosphatases, are also viable strategies 73. Finally, activating negative regulators, such as the dualspecificity kinases WE or MYT1, might also yield anti-proliferating or pro-apoptotic benefits. Resolution of the tertiary structure of CDKs bound to their cognate cyclins or to CKIs (reviewed in REF. 74) is facilitating the design of small-molecule inhibitors 75 (TABLE 1). Yet current strategies tend to target the ATbinding site that is necessary for their kinase activity. Unfortunately, catalytic residues are well conserved across eukaryotic protein kinases, which makes adequate specificity a big challenge as there are more than 500 protein kinases in the human genome 76.For instance, staurosporines, suramin and 6-dimethylaminopurine are relatively nonspecific protein kinase h CDC25 Cyclin b Inhibit CDK cyclin interaction f i Activate WE CDK Cyclin Downregulate CDC25 g CDK7 Cyc H Downregulate CAK Activate cyclin degradation Figure 3 CDK regulation and opportunities for therapeutic intervention. CDKs are regulated by the availability of positive (cyclins) and negative (CDK kinase inhibitors; CKIs) partners, and by phosphorylation at critical residues. These regulatory events are, in principle, susceptible to therapeutic intervention. They include: direct inhibition of CDK kinase activity (a); prevention of cyclin CDK interaction (b); regulation of CKI function by restoring their levels (c) or inhibiting their degradation (d); inhibition of cyclin synthesis (e) or promoting cyclin degradation (f); inhibition of CDK activating kinases (CAKs) (g) or CDC25 phosphatases (h); and stimulation of inhibitory kinases (i). Strategies a c are currently favoured. inhibitors that also inhibit CDKs. Conversely, flavopiridol, olomucine, butyrolactone-1 and paullones are more selective for CDK inhibition (see TABLE 1 and REFS 77,78 for reviews). Assuming that the specificity issue could be resolved, therapeutic strategies must also take into account which CDK is the best target. Although most mutations in human cancer affect CDK4 and CDK6 or their regulators (FIG. 2), current drug-discovery efforts favour the design of either AN-CDK INHIBITORS or inhibitors that are specific for CDK2. This is not surprising as clinical investigators are more concerned with achieving potent cell-killing activity than target specificity. Indirubin an active ingredient of a traditional Chinese recipe given to patients with CML was the first example of a CDK inhibitor that was used to treat human cancer 79, and flavopiridol a flavonoid derived from an indigenous plant from India was the first CDK modulator tested in clinical trials. Early-phase trials have shown activity in some patients with non-hodgkin s lymphoma, and renal, prostate, colon and gastric carcinomas (reviewed in REF. 80). Another CDK inhibitor tested in clinical trials, UCN-01, has shown activity in patients with melanoma, lymphoma and sarcomas 80. UCN-01 is able to block the cell cycle and to abrogate checkpoints that are induced by genotoxic stress, due to modulation of the CHK1 kinase. In addition, UCN- 01, along with other CDK inhibitors, induces apoptosis (TABLE 1). Although much more work needs to be done in this area, it is probable that this activity is mediated by E2F. As CDK2 activity is required for the dissociation of E2F from DNA, CDK2 inhibition might result in E2F activity persisting beyond S phase, which could lead to apoptosis. This circumstance could also be exploited to specifically kill transformed cells in which E2F activity is upregulated 65. Trials of CDK inhibitors in combination with other chemotherapy agents are being considered. In addition, chemical inhibition of CDKs might spread to other areas of medicine. For instance, CDK inhibitors are being studied in nephrology, cardiovascular diseases, neurodegenerative disorders and even in the fight against unicellular parasites and viral diseases (for reviews, see REFS 81,82). Apart from these considerations, we must not forget that some of the most widely used antitumour drugs the taxanes and the vinca alkaloids inhibit the cell cycle by altering the mitotic spindle 83. Whereas the vinca alkaloids depolymerize microtubules thereby increasing the soluble tubulin pool the taxanes (and also a new class of promising antitumour drugs, the epothilones 83,84 ) stabilize microtubules, thereby increasing the microtubular mass. In spite of these obvious mechanistic differences, they might have a similar end-point activity the induction of apoptotic death due to activation of the spindle checkpoint. Understanding the molecular bases for the differential effect of these compounds on tumour versus normal cells should open new windows of opportunity to improve current therapeutic strategies. NATURE REVIEWS CANCER VOLUME 1 DECEMBER

9 Conclusions Cancer is frequently considered to be a disease of the cell cycle. Although many specific cell-cycle regulatory mechanisms have been studied in depth in vitro, it is still unclear how diverse cellular processes are coordinately deregulated with cell division in human cancer. An integrated view from biochemical studies, experimental animal models and the analysis of cancer mutations will undoubtedly help to exploit this knowledge in therapeutic approaches. Future research should focus on crucial aspects of cell-cycle regulation in vivo, such as the pathways of synthesis and proteolysis of essential proteins, post-translational regulations and, importantly, the functional specificity or redundancy among family members. Similarly, explaining how cells coordinate cell growth and cell proliferation, and how they decide whether to cycle or not to cycle, is a greater challenge. Ultimately, the design of highly specific and effective anticancer drugs should come from multidisciplinary approaches that combine, among other strategies: molecular pathology, to identify all the mutations/alterations present in human tumours; functional genomics, to select the most important targets; structural and computational biology, to predict protein structure and target inhibitor interactions; computer-assisted combinatorial chemistry, to design the best possible inhibitors; and animal models with high predictability value, to increase the chances of success of leading drug candidates in the clinic the ultimate test. 1. Norbury, C. & Nurse,. Animal cell cycles and their control. Annu. Rev. Biochem. 61, (1992). 2. Hartwell, L. & Weinert, T. Checkpoints: controls that ensure the order of cell cycle events. Science 246, (1989). 3. Morgan, D. O. Cyclin-dependent kinases: engines, clocks, and microprocessors. Annu. Rev. Cell Dev. Biol. 13, (1997). 4. Sherr, C. J. Cancer cell cycles revisited. Cancer Res. 60, (2000). 5. Meyerson, M. & Harlow, E. Identification of G 1 kinase activity for cdk6, a novel cyclin D partner. Mol. Cell. Biol. 14, (1994). 6. Jinno, S. et al. Oncogenic stimulation recruits cyclindependent kinase in the cell cycle start in rat fibroblast. roc. Natl Acad. Sci. USA 96, (1999). 7. Rane, S. G. et al. Loss of Cdk4 expression causes insulindeficient diabetes and Cdk4 activation results in B-cell hyperplasia. Nature Genet. 22, (1999). 8. Tsutsui, T. et al. Targeted disruption of Cdk4 delays cell cycle entry with enhanced p27kip1 activity. Mol. Cell. Biol. 19, (1999). References 7 and 8 describe the generation of mice that are deficient for Cdk4. Cdk4 seems to be dispensable for the cell cycle in most cell types. Ablation of Cdk4, however, severely affects proliferation of specific cell types, such as pancreatic β cells, which leads to diabetes. 9. Ye, X., Zhu, C. & Harper, W. A premature-termination mutation in the Mus musculus cyclin-dependent kinase 3 gene. roc. Natl Acad. Sci. USA 13, (2001). 10. Nigg, E. A. Cyclin-dependent kinase 7: at the cross-roads of transcription, DNA repair and cell cycle control? Curr. Opin. Cell Biol. 8, (1996). 11. Ekholm, S. V. & Reed, S. I. Regulation of G 1 cyclindependent kinases in the mammalian cell cycle. Curr. Opin. Cell Biol. 12, (2000). 12. Malumbres, M. & ellicer, A. Ras pathways to cell cycle control and cell transformation. Front. Biosci. 3, D887 D912 (1998). 13. Diehl, J. A., Cheng, M., Roussel, M. F. & Sherr, C. J. Glycogen synthase kinase-3β regulates cyclin proteolysis and subcellular localization. Genes Dev. 12, (1998). 14. Alt, J. R., Cleveland, J. L., Hannink, M. & Diehl, J. A. hosphorylation-dependent regulation of cyclin nuclear export and cyclin -dependent cellular transformation. Genes Dev. 14, (2000). 15. Fantl, V., Stamp, G., Andrews, A., Rosewell, I. & Dickson, C. Mice lacking cyclin are small and show defects in eye and mammary gland development. Genes Dev. 9, (1995). 16. Sicinski,. et al. Cyclin provides a link between development and oncogenesis in the retina and breast. Cell 82, (1995). 17. Sherr, C. J. & Roberts, J. M. Cdk inhibitors: positive and negative regulators of G1-phase progression. Genes Dev. 13, (1999). 18. Adams,. D. Regulation of the retinoblastoma tumor suppressor protein by cyclin/cdks. Biochim. Biophys. Acta 1471, (2001). 19. Harbour, J. W. & Dean, D. C. The prb/e2f pathway: expanding roles and emerging paradigms. Genes Dev. 14, (2000). 20. Morris, E. J. & Dyson, N. J. Retinoblastoma protein partners. Adv. Cancer Res. 82, 1 54 (2001). 21. Lundberg, A. S. & Weinberg, R. A. Functional inactivation of the retinoblastoma protein requires sequential modification by at least two distinct cyclin CDK complexes. Mol. Cell. Biol. 18, (1998). 22. Harbour, J. W., Luo, R. X., Dei Santi, A., ostigo, A. A. & Dean, D. C. Cdk phosphorylation triggers sequential intramolecular interactions that progressively block Rb functions as cells move through G1. Cell 98, (1999). 23. Ezhevsky, S. A., Ho, A., Becker-Hapak, M., Davis,. K. & Dowdy, S. F. Differential regulation of retinoblastoma tumor suppressor protein by G1 cyclin-dependent kinase complexes in vivo. Mol. Cell. Biol. 21, (2001). References culminate a series of reports that describe the distinct effects of cyclin-d-dependent and cyclin-e-dependent CDKs on the protein. The different role of these kinases in regulation provides us with interesting insights regarding the regulation of the early G1 and late G1 phase. 24. Chan, H. M., Krstic-Demonacos, M., Smith, L., Demonacos, C. & La Thangue, N. B. Acetylation control of the retinoblastoma tumour-suppressor protein. Nature Cell Biol. 3, (2001). 25. Tamrakar, S., Rubin, E. & Ludlow, J. W. Role of prb dephosphorylation in cell cycle regulation. Front. Biosci. 5, (2000). 26. Van den Heuvel, M. & Harlow, E. Distinct roles for cyclindependent kinases in cell cycle control. Science 262, (1993). This study provided evidence for a distinct role of the different CDKs in G1/S progression. Whereas dominant-negative forms of CDK2 or CDK3 can arrest cultured cells in G1, similar mutant forms of CDK4 or CDK6 cannot. 27. Zhang, H. S., ostigo, A. A. & Dean, D. C. Active transcriptional repression by the Rb E2F complex mediates G1 arrest triggered by p16ink4a, TGFβ, and contact inhibition. Cell 97, (1999). 28. Blain, S. W., Montalvo, E. & Massagué, J. Differential interaction of the cyclin-dependent kinase (Cdk) inhibitor p27 Kip1 with cyclin-a Cdk2 and cyclin-d2 Cdk4. J. Biol. Chem. 272, (1997). 29. LaBaer, J. et al. New functional activities for the p21 family of CDK inhibitors. Genes Dev. 11, (1997). 30. Cheng, M. et al. The p21 Cip1 and p27 Kip1 Cdk inhibitors are essential activators of cyclin D-dependent kinases in murine fibroblasts. EMBO J. 18, (1999). References illustrate the differential consequences of the binding of WAF1/KI inhibitors to CDK4/6 cyclin-d and CDK2 cyclin-e/a complexes. 31. Geng, Y. et al. Rescue of cyclin deficiency by knockin cyclin E. Cell 97, (1999). Replacement of cyclin by a human cyclin cdna restores most of the deficiencies of cyclin- knockout mice. As cyclin E cannot activate CDK4 or CDK6, these observations indicate that cyclin is a main downstream target of cyclin. 32. Geng, Y. et al. Deletion of the p27 Kip1 gene restores normal development in cyclin -deficient mice. roc. Natl Acad. Sci. USA, 98, (2001). 33. Tong, W. & ollard, J. W. Genetic evidence for the interactions of cyclin and p27 Kip1 in mice. Mol. Cell. Biol. 21, (2001). 34. Sheaff, R. J., Groudine, M., Gordon, M., Roberts, J. M. & Clurman, B. E. Cyclin E-CDK2 is a regulator of p27 Kip1. Genes Dev. 11, (1997). 35. Malek, N.. et al. A mouse knock-in model exposes sequential proteolytic pathways that regulate p27 Kip1 in G1 and S phase. Nature 413, (2001). Cells that express a knock-in Kip1 T187A mutant are unable to downregulate Kip1 during S and G2. However, this mutation does not affect Kip1 downregulation in G1, indicating the existence of an alternative degradation process. 36. Moberg, K. H., Bell, D. W., Wahrer, D. C. R., Haber, D. A. & Hariharan, I. K. Archipielago regulates cyclin E levels in Drosophila and is mutated in human cancer cell lines. Nature 413, (2001). 37. Strohmaier, H. et al. Human F-box protein hcdc4 targets cyclin E for proteolysis and is mutated in a breast cancer cell line. Nature 413, (2001). 38. Koepp, D. M. et al. hosphorylation-dependent ubiquitination of cyclin E by the SCF Fbw7 ubiquitin ligase. Science 294, (2001). 39. Wolfel, T. et al. A p16 Ink4a -insensitive Cdk4 mutant targeted by cytolytic T lymphocytes in a human melanoma. Science 269, (1995). First description of a tumour-associated point mutation in a CDK molecule. Interestingly, this CDK4 mutation (R24C) disrupts the binding with the INK4 family of inhibitors, highlighting the importance of this interaction in tumour development. 40. Easton, J., Wei, T., Lahti, J. M. & Kidd, V. J. Disruption of the cyclin D/cyclin-dependent kinase/ink4/retinoblastoma protein regulatory pathway in human neuroblastoma. Cancer Res. 58, (1998). 41. Latres, E. et al. Role of the F-box protein Skp2 in lymphomagenesis. roc. Natl Acad. Sci. USA 98, (2001). 42. Gstaiger, M. et al. Skp2 is oncogenic and overexpressed in human cancers. roc. Natl Acad. Sci. USA 98, (2001). 43. Wu, C.-L. et al. Cables enhances Cdk2 tyrosine 15 phosphorylation by Wee1, inhibits cell growth, and is lost in many human colon and squamous cancers. Cancer Res. 61, (2001). 44. aggi, M. G. & Giordano, A. Who is the boss in the retinoblastoma family? The point of view of Rb2/p130, the little brother. Cancer Res. 61, (2001). 45. Lee, E. Y. H.. et al. Mice deficient for Rb are nonviable and show defects in neurogenesis and hematopoiesis. Nature 359, (1992). 46. Jacks, T. et al. Effects of an Rb mutation in the mouse. Nature 359, (1992). 47. Mulligan, G. & Jacks, T. The retinoblastoma gene family: cousins with overlapping interests. Trends Genet. 14, (1998). 48. Robanus-Maandag, E. et al. p107 is a suppressor of retinoblastoma development in prb-deficient mice. Genes Dev. 12, (1998). 49. Fero, M. L. et al. A syndrome of multiorgan hyperplasia with features of gigantism, tumorigenesis, and female sterility in p27 Kip1 -deficient mice. Cell 85, (1996). 50. Kiyokawa, H. et al. Enhanced growth of mice lacking the cyclin-dependent kinase inhibitor function of p27 Kip1. Cell 85, (1996). 51. Nakayama, K. et al. Mice lacking p27 Kip1 display increased body size, multiple organ hyperplasia, retinal dysplasia, and pituitary tumors. Cell 85, (1996). 52. Fero, M. L., Randel, E., Gurley, K. E., Roberts, J. M. & Kemp, C. J. The murine gene p27 Kip1 is haplo-insufficient for tumour suppression. Nature 396, (1998). 230 DECEMBER 2001 VOLUME 1

10 53. Martin-Caballero, J., Flores, J. M., Garcia-alencia,. & Serrano, M. Tumor susceptibility of p21(waf1/cip1)- deficient mice. Cancer Res. 61, (2001). 54. Krimpenfort,., Quon, K. C., Mooi, W. J., Loonstra, A. & Berns, A. Loss of p16 Ink4a confers susceptibility to metastatic melanoma in mice. Nature 413, (2001). 55. Sharpless, N. E. et al. Loss of p16 INK4a with retention of p19 ARF predisposes to tumorigenesis. Nature 413, (2001). References 54 and 55 define the specific role of as a tumour suppressor. They also highlight that loss of increases predisposition for melanoma, a neoplasia in which the CDKN2A (which encodes and ARF) region was first described as a susceptibility locus (see also reference 59). 56. Latres, E. et al. Limited overlapping roles of p15 INK4b and p18 INK4c cell cycle inhibitors in proliferation and tumorigenesis. EMBO J. 19, (2000). 57. Franklin, D. S. et al. Cdk inhibitors p18 Ink4c and p27 Kip1 mediate two separate pathways to collaborative suppress pituitary tumorigenesis. Genes Dev. 12, (1998). 58. Sotillo, R. et al. Wide spectrum of tumors in knock-in mice carrying a Cdk4 protein insensitive to INK4 inhibitors. EMBO J. (in the press). 59. Sotillo, R. et al. Invasive melanoma in Cdk4 targeted mice. roc. Natl Acad. Sci. USA 98, (2001). 60. Wang, T. C. et al. Mammary hyperplasia and carcinoma in MMTV-cyclin transgenic mice. Nature 369, (1994). These studies (references 58 60) illustrate the role of CDK4 cyclin-d misregulation in tumour development and tumour susceptibility. 61. Bortner, D. M. & Rosenberg, M.. Induction of mammary gland hyperplasia and carcinomas in transgenic mice expressing human cyclin E. Mol. Cell. Biol. 17, (1997). 62. Brugarolas, J., Bronson, R. T. & Jacks, T. p21 is a critical regulator essential for proliferation control in Rb-deficient cells. J. Cell Biol. 141, (1998). 63. ark, M. S. et al. p27 and Rb are on overlapping pathways suppressing tumorigenesis in mice. roc. Natl Acad. Sci. USA 96, (1999). 64. Franklin, D. S., Godfrey, V. L., O Brien, D. A., Deng, C. & Xiong, Y. Functional collaboration between different cyclindependent kinase inhibitors suppresses tumor growth with distinct tissue specificity. Mol. Cell. Biol. 20, (2000). 65. Chen, Y.-N.. et al. Selective killing of transformed cells by cyclin/cyclin-dependent kinase 2 antagonists. roc. Natl Acad. Sci. USA 96, (1999). 66. Blagosklonny, M. V. & ardee, A. B. Exploiting cancer cell cycling for selective protection of normal cells. Cancer Res. 61, (2001). 67. Noble, M. E. M. & Endicott, J. A. Chemical inhibitors of cyclin-dependent kinases: insights into design from X-ray crystallographic studies. harmacol. Ther. 82, (1999). 68. Sandig, V. et al. Adenovirally transferred p16 INK4a /CDKN2 and p53 genes cooperate to induce apoptotic tumor cell death. Nature Med. 3, (1997). 69. Fahraeus, R., Lain, S., Ball, K. L. & Lane, D.. Characterization of the cyclin-dependent kinase inhibitory domain of the INK4 family as a model for a synthetic tumour suppressor molecule. Oncogene 16, (1998). 70. Lubbert, M. DNA methylation inhibitors in the treatment of leukemias, myelodysplastic syndromes and hemoglobinopathies: clinical results and possible mechanisms of action. Curr. Top. Microbiol. Immunol. 249, (2000). 71. Adams, J. et al. roteasome inhibitors: a novel class of potent and effective antitumor agents. Cancer Res. 59, (1999). 72. Arber, N. et al. Antisense to cyclin inhibits the growth and tumorigenicity of human colon cancer cells. Cancer Res. 57, (1997). 73. estell, K. E., Ducruet, A.., Wipf,. & Lazo, J. S. Small molecule inhibitors of dual specificity protein phosphatases. Oncogene 19, (2000). 74. avletich, N.. Mechanisms of cyclin-dependent kinase regulation: structures of Cdks, their cyclin activators and Cip and Ink inhibitors. J. Mol. Biol. 287, (1999). 75. Gray, N. S. et al. Exploiting chemical libraries, structure, and genomics in the search for kinase inhibitors. Science 281, (1998). An example of the synergistic combination of in vitro and in silico approaches in the search for CDK inhibitors of therapeutic value. 76. Lander et al. Initial sequencing and analysis of the human genome. Nature 409, (2001). 77. Garrett, M. D. & Fattaey, A. CDK inhibition and cancer therapy. Curr. Opin. Genet. Dev. 9, (1999). 78. Crews, C. M. & Mohan, R. Small-molecule inhibitors of the cell cycle. Curr. Opin. Chem. Biol. 4, (2000). 79. Hoessel, R. et al. Indirubin, the active constituent of a Chinese antileukaemia medicine, inhibits cyclin-dependent kinases. Nature Cell Biol. 1, (1999). Indirubin can specifically inhibit CDKs and arrest cell-cycle progression. An example of how classical Chinese medicine might help to identify inhibitors of molecular targets. 80. Senderowicz, A. M. & Sausville, E. A. reclinical and clinical development of cyclin-dependent kinase modulators. J. Natl Cancer Inst. 92, (2000). 81. Brooks, G. & La Thangue, N. B. The cell cycle and drug discovery: the promise and the hope. Drug Discovery Today 4, (1999). 82. Meijer, L., Leclerc, S. & Leost, M. roperties and potential applications of chemical inhibitors of cyclin-dependent kinases. harmacol. Ther. 82, (1999). 83. Wood, K. W., Cornwell, W. D. & Jackson, J. R. ast and future of the mitotic spindle as an oncology target. Curr. Opin. harmacol. 1, (2001). 84. Altmann, K.-H. Microtubule-stabilizing agents: a growing class of important anticancer drugs. Curr. Opin. Chem. Biol. 5, (2001). 85. Thomas, G. An encore for ribosome biogenesis in the control of cell proliferation. Nature Cell. Biol. 2, E71 E72 (2000). 86. Schmelzle, T. & Hall, M. N. TOR, a central controller of cell growth. Cell 103, (2000). 87. Rohde, J., Heitman, J. & Cardenas, M. E. The TOR kinases link nutrient sensing to cell growth. J. Biol. Chem. 276, (2001). 88. Fukunaga, R. & Hunter, T. MNK1, a new MA kinaseactivated protein kinase, isolated by a novel expression screening method for identifying protein kinase substrates. EMBO J. 16, (1997). 89. Waskiewicz, A. J., Flynn, A., roud, C. G. & Cooper, J. A. Mitogen-activated protein kinases activate the serine/threonine kinases Mnk1 and Mnk2. EMBO J. 16, (1997). 90. ardee, A. B. A restriction point for control of normal animal cell proliferation. roc. Natl Acad. Sci. USA 71, (1974). 91. lanas-silva, M. D. & Weinberg, R. A. The restriction point and control of cell proliferation. Curr. Opin. Cell Biol. 9, (1997). 92. Dannenberg, J.-H., van Rossum, A., Schuijff, L. & te Riele, H. Ablation of the retinoblastoma gene family deregulates G1 control causing immortalization and increased cell turnover under growth-restricting conditions. Genes Dev. 14, (2000). 93. Sage, J. et al. Targeted disruption of the three Rb-related genes leads to loss of G1 control and immortalization. Genes Dev. 14, (2000). References 92 and 93 illustrate the specific and redundant roles of the protein family members in regulating the cell cycle. 94. ark, M. & Krause, M. W. Regulation of postembryonic G1 cell cycle progression in Caenorhabditis elegans by a cyclin D/CDK-like complex. Development 126, (1999). 95. Meyer, C. A. et al. Drosophila Cdk4 is required for normal growth and is dispensable for cell cycle progression. EMBO J. 19, (2000). Shows that in Drosophila, the single cyclin-ddependent Cdk (equivalent to mammalian CDK4 and CDK6) is not required for cell proliferation but affects homeotic cell number. 96. Datar, S. A., Jacobs, H. W., de la Cruz, A. F., Lehner, C. F. & Edgar, B. A. The Drosophila cyclin-d Cdk4 complex promotes cellular growth. EMBO J. 19, (2000). 97. Yu, Q., Geng, Y. & Sicinski,. Specific protection against breast cancers by cyclin ablation. Nature 411, (2001). Genetic analysis of the relationship between the tumorigenic properties of specific oncogenes and cyclin in breast cancer. This report exemplifies the use of genetic models to shape specific pathways for therapeutic intervention in human cancer. 98. Shah, J. V. & Cleveland, D. W. Waiting for anaphase: Mad2 and the spindle assembly checkpoint. Cell 103, (2000). 99. Nigg, E. A. Cell division mitotic kinases as regulators of cell division and its checkpoints. Nature Rev. Mol. Cell. Biol. 2, (2001) Cahill, D.. et al. Mutations of mitotic checkpoint genes in human cancers. Nature 392, (1998) Michel, L. S. et al. MAD2 haplo-insufficiency causes premature anaphase and chromosome instability in mammalian cells. Nature 409, (2001). First genetic demonstration that defects in the mitotic checkpoint can lead to tumour development Hernando, E. et al. Molecular analyses of the mitotic checkpoint components hsmad2, hbub1 and hbub3 in human cancer. Int. J. Cancer 95, (2001) Smith, M. R. et al. Malignant transformation of mammalian cells initiated by constitutive expression of the polo-like kinase. Biochem. Biophys. Res. Commun. 234, (1997) Simizu, S. & Osada, H. Mutations in the lk gene lead to instability of lk protein in human tumour cell lines. Nature Cell. Biol. 2, (2000) Giet, R. & rigent, C. Aurora/Ipl1p-related kinases, a new oncogenic family of mitotic serine-threonine kinases. J. Cell. Sci. 112, (1999) Zou, H., McGarry, T. J., Bernal, T. & Kirschner, M. W. Identification of a vertebrate sister-chromatid separation inhibitor involved in transformation and tumorigenesis. Science 285, (1999) Tao, W. et al. Human homologue of the Drosophila melanogaster lats tumour suppressor modulates CDC2 activity. Nature Genet. 21, (1999) St John, M. A. R. et al. Mice deficient of Lats1 develop soft-tissue sarcomas, ovarian tumours and pituitary dysfunction. Nature Genet. 21, (1999). References 107 and 108 provide the first link between CDK1 deregulation and tumour development. These reports raise important questions regarding a possible role of CDK1 in cell transformation Vesely, J. et al. Inhibition of cyclin-dependent kinases by purine analogues. Eur. J. Biochem. 224, (1994) De Azevedo, W. F., Leclerc, S., Meijer, L., Strand, M. & Kim, S. H. Inhibition of cyclin-dependent kinases by purine analogues: crystal structure of human cdk2 complexed with roscovitine. Eur. J. Biochem. 243, (1997) Arris, C. E. et al. Identification of novel purine and pyrimidine cyclin-dependent kinase inhibitors with distinct molecular interactions and tumor cell growth inhibition profiles. J. Med. Chem. 43, (2000) Zaharevitz, D. W. et al. Discovery and initial characterization of the paullones, a novel class of small-molecule inhibitors of cyclin-dependent kinases. Cancer Res. 59, (1999) Kitagawa et al. A cyclin-dependent kinase inhibitor, butyrolactone I, inhibits phosphorylation of protein and cell cycle progression. Oncogene 9, (1994) Meijer, L. et al. Inhibition of cyclin-dependent kinases, GSK-3β and CK1 by hymenialdisine, a marine sponge constituent. Chem. Biol. 7, (1999) Lane, M. E. et al. A novel cdk2-inhibitor, SU9516, induces apoptosis in colon carcinoma cells. Cancer Res. 61, (2001) Soni, R. et al. Selective in vivo and in vitro effects of a small molecule inhibitor of cyclin-dependent kinase 4. J. Natl Cancer Inst. 93, (2001) Fry, D. W. et al. Cell cycle and biochemical effects of D A potent inhibitor of the cyclin D-dependent kinases CDK4 and CDK6. J. Biol. Chem. 276, (2001) Soni, R. et al. Inhibition of cyclin dependent kinase 4 (Cdk4) by fascaplysin, a marine natural product. Biochem. Biophys. Res. Commun. 275, (2000). Online links DATABASES The following terms in this article are linked online to: CancerNet: chronic myelogenous leukaemia colon tumours gastric carcinoma head and neck tumours melanoma non- Hodgkin s lymphoma prostate carcinoma renal carcinoma LocusLink: AKT BUB1 Cables CB CDC20 CDC25A CDC25B CDC25C CDC4 CDK2 CDK3 Cdk4 CDK4 Cdk6 CDK6 CDK7 Cdkn2a Cdkn2b Cdkn2c cyclin A1 cyclin A2 cyclin cyclin D2 cyclin D3 cyclin cyclin E2 cyclin H D e1f4e ERK GSK-3β INK4B INK4C INK4D IR IRS KI2 Lats1 LATS1 MAD2 MAK MNK1 MYT1 Neu p107 p130 p300 p53 I3K KB LK1 1 RAF RAS SK2 TOR Waf1 WAF1 WE FURTHER INFORMATION BioMedNet Mouse Knockout & Mutation Database: CancerNet: CenterWatch: Database of prb-binding proteins: TBASE (The Transgenic/Targeted Mutation Database): Access to this interactive links box is free online. NATURE REVIEWS CANCER VOLUME 1 DECEMBER