Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Designer baby wikipedia , lookup
Gene therapy of the human retina wikipedia , lookup
History of genetic engineering wikipedia , lookup
Minimal genome wikipedia , lookup
Epigenetics of human development wikipedia , lookup
Site-specific recombinase technology wikipedia , lookup
Epigenetics in stem-cell differentiation wikipedia , lookup
Vectors in gene therapy wikipedia , lookup
Polycomb Group Proteins and Cancer wikipedia , lookup
Oncogene (2005) 24, 2756–2764 & 2005 Nature Publishing Group All rights reserved 0950-9232/05 $30.00 www.nature.com/onc Cell-cycle control in Caenorhabditis elegans: how the worm moves from G1 to S John Koreth1,2 and Sander van den Heuvel*,1 1 MGH Cancer Center, Harvard Medical School, Charlestown, MA 02129, USA; 2Dana Farber Cancer Institute, 44 Binney Street, Boston, MA 02115, USA The nematode Caenorhabditis elegans offers a powerful model system to study cell division control during animal development. Progress from the one-cell zygote to adult stage follows a nearly invariant pattern of divisions. This, combined with a transparent body and efficient genetics, allows for sensitive identification and quantitative analysis of cell-cycle mutants. Nearly all G1 control genes identified in C. elegans have mammalian homologs. Examples include a D-type cyclin and CDK4/6-related kinase, a member of the retinoblastoma protein family and CDK inhibitors of the Cip/Kip family. Genetic studies have placed the currently known G1 regulators into pathways similar to those in mammals. Together, this validates the use of C. elegans in identifying additional regulators of cell-cycle entry and exit. For instance, we recently found that the CDC-14 phosphatase promotes maintenance of the quiescent state. Here, we describe cellcycle control as an integral part of C. elegans development, summarize current knowledge of G1 control genes in the worm, compare the results with those obtained in other species, and discuss the possible implications of cellcycle studies in C. elegans for higher organisms, including humans. Oncogene (2005) 24, 2756–2764. doi:10.1038/sj.onc.1208607 Keywords: C. elegans cell cycle; G1–S; cyclin D; Cdk4/6; lin-35 Rb; cki-1 Cip/Kip Introduction The nematode Caenorhabditis elegans offers an attractive genetic system to study various aspects of development (Brenner, 1974), including the developmental regulation of cell proliferation. Mutants with altered patterns of cell divisions during larval development have been identified since the early days of C. elegans genetics. Several of the mutations are now known to affect genes that control cell-cycle progression. Cellcycle genes have also been identified in more recent studies focused on cell-cycle entry and withdrawal. *Correspondence: S van den Heuvel, MGH Cancer Center, Harvard Medical School, Charlestown, MA 02129, USA; E-mail: [email protected] Together, a pathway for G1 control in C. elegans has emerged, which includes positive regulators similar to the Cyclin D and Cdk4/6 oncogenes, negative regulators related to the retinoblastoma (Rb) tumor suppressor and cyclin-dependent kinase (CDK) inhibitors, and transcription factors of the E2F and DP families. Thus, the control of G1 progression involves similar players in C. elegans and mammals. Studies in C. elegans benefit from a variety of practical advantages: the animals are small (adults are B1 mm in length) and can be grown on agar plates or in liquid media. Its life cycle is short: in just over 3 days a fertilized egg develops into an adult hermaphrodite or male. Each hermaphrodite produces around 300 progeny by self-fertilization, while hermaphrodite-male crosses can generate more than a thousand progeny. One of its major strengths is efficient genetics. Traditionally, this meant forward genetics, which starts from mutagenesis. However, reverse genetics by RNA interference (RNAi) is also highly efficient in the worm (Fire et al., 1998), and genome-wide screens based on ‘feeding RNAi’ have become common (Kamath and Ahringer, 2003; Rual et al., 2004). Despite its simplicity, C. elegans completes an entire program of animal development and forms a broad variety of cell types and tissues. Its transparency, small size, nearly invariant cell lineage, and simple body plan have made it possible to trace the development of every somatic cell nucleus in the adult worm back to the single cell zygote (for accuracy we use ‘nuclei’ instead of ‘cells’, as some nuclei share a common cytoplasm) (Sulston and Horvitz, 1977; Sulston et al., 1983). This analysis allows for the precise prediction of the timing and localization of each cell division in the developing worm. Together, these characteristics offer a unique set of tools for the identification and characterization of cell-cycle mutants. Throughout C. elegans development, cells decide whether to remain quiescent or to enter another division cycle. These decisions form part of a larger program, which precisely coordinates cell division with cell growth, differentiation, tissue formation, and morphogenesis. Thus, the genes that regulate cell-cycle progression form an integral part of the network that controls formation of a well-defined organism. Major questions include how developmental signaling cascades, mediated, for instance, by insulin-related growth G1 progression in C. elegans J Koreth and S van den Heuvel 2757 factors, TGF-b family members, or Wnt ligands, impinge on the cell-cycle machinery to accomplish proper temporal and spatial control of cell division. As many of the regulators and pathways controlling cell fate, cell growth, cell death, and cell proliferation are conserved, C. elegans studies will likely help reveal how cell division is controlled in a developmental context. The most common model systems, yeasts and mammalian cells in tissue culture, are ill suited to address this higher level of cell-cycle control. Therefore, developmental control of cell division is an area that remains poorly understood. Yet, the regulatory mechanisms involved are highly relevant to our understanding of, and potentially interfering with, the unrestrained proliferation of cancer cells. cycle profiles. Certain cells continue rapid divisions, others divide after an extended interphase of 2 h or more, yet other cells become post-mitotic or only resume divisions during larval development, after an arrest of over 18 h (Sulston et al., 1983). Thus, the regulation of cell-cycle events and the introduction of G1 and G2 phases are likely determined by the cell lineage. Maternal gene products, synthesized in the hermaphrodite and deposited into the egg, drive most embryonic divisions. Owing to perdurance of the maternal contribution of the wild-type gene product, homozygous cell-cycle mutants derived from a heterozygous mother tend to complete embryogenesis. Such homozygous mutants often display cell-cycle defects during larval development, usually during the first larval stage. Larval somatic cell cycles Cell-cycle regulation and C. elegans development As metazoans go through development, their cells progress through various types of cell cycles. These include the embryonic cell cycle, somatic cell cycle, endoreduplication cycle, and meiotic cell cycle. The cellcycle machinery used is tailored towards these individual cycles, and varies in the requirement for cell-cycle genes and the balance between their contributions. For instance, C. elegans cyd-1 cyclin D and cdk-4 Cdk4/6 are not required for most of embryogenesis and cdk-1 is not needed for the endoreduplication cycles (see below; Boxem et al., 1999). It is thus important to consider the type of cycle and developmental context when examining the function of cell-cycle components. For these reasons, we summarize briefly how the various cell division cycles follow each other during C. elegans development. Embryonic cell cycles As in other metazoans, early embryonic divisions in C. elegans are fast and cycle between S and M phases, apparently lacking the Gap phases G1 and G2 (Edgar and McGhee, 1988). The initial mitotic division cycles take approximately 15 min each and include cytoplasmic cleavage, in contrast to the extremely rapid nuclear divisions of early Drosophila embryos. Cells start to acquire different fates right from the asymmetric first mitotic division, and the division cycles of their descendents are asynchronous and progressively vary in length. Nearly all embryonic divisions occur during the first half of embryogenesis, within the proliferation phase that is completed approximately 7 h after fertilization. This phase is followed by a differentiation and tissue morphogenesis phase, during which the initially spherical embryo is stretched out into the larva that hatches from the egg. As C. elegans embryonic divisions are asynchronous, it is not possible to make generalized statements about the number of embryonic divisions that consist solely of S and M phases, or when the G1 and G2 phase are first introduced. Just a few hours into embryonic development, cells in different lineages diverge greatly in cell- After hatching in the presence of adequate food, the worm progresses through four larval stages (L1–L4) to become an adult. In total, 55 precursor cells continue to proliferate during larval development, thereby expanding the 558 nuclei of an early L1 larva to the 959 somatic nuclei and extensive germline of the adult hermaphrodite (recently reviewed in Lambie, 2002). The somatic nuclei appear to contain a 2n DNA content at the time of hatching and go through a DNA synthesis phase before initiating mitosis (Boxem et al., 1999; Boxem and van den Heuvel, 2001; our unpublished results). These divisions generally depend on the function of G1/S and G2/M control genes. Thus, the precursor cells of the post-embryonic lineages and their descendents follow canonical cell cycles in which the S and M phases are separated by Gap phases. As in embryogenesis, the length of the interphase varies greatly between different cell types. Divisions frequently follow each other within 1 h, but some cells remain quiescent for 20 h, before dividing again two larval stages later (Sulston and Horvitz, 1977). At least some divisions incorporate developmental information during the G1 phase. Extracellular signals can induce cell fate decisions with specific division patterns. For instance, heterochronic genes such as lin-4 and lin-14, which encode a micro-RNA and its target, respectively, control the larval stage-specific patterns of cell proliferation (Lee et al., 1993). These molecular timekeepers exercise organism-wide control over temporal identity and patterning, a role reminiscent of Hox gene functions in spatial control. Environmental factors can also induce a global arrest of the cell division program. L1 larvae that hatch in the absence of food arrest development, growth, and cell division until the food supply is restored. Starved larvae can also adopt an alternative specialized developmental arrest phase known as the ‘dauer’ (enduring) stage. Dauer larvae do not feed, and adapt their anatomy, behavior, and metabolism to long-term survival. Entry into the dauer stage is determined by multiple environmental inputs that include dauer pheromone, temperature, and food availability. These stimuli impact a complex network of interacting pathways that include TGF-b, insulin Oncogene G1 progression in C. elegans J Koreth and S van den Heuvel 2758 signaling, and cGMP coupled pathways (Kimura et al., 1997; Riddle and Albert, 1997). Further cell divisions are arrested until environmental conditions improve. Upon feeding, dauer larvae complete cell divisions characteristic of the L3 larval stage, molt, and continue fourth-stage and adult development. Endoreduplication cycles Cells in intestinal and hypodermal tissues (skin) undergo endoreduplication cycles during C. elegans larval development (Hedgecock and White, 1985). Such cycles are characterized by a DNA synthesis phase that is not followed by M phase, thus doubling the DNA ploidy with each additional cycle. Of the 20 intestinal cells, 14 undergo a final nuclear division at the end of the L1 stage. Subsequently, all intestinal nuclei go through an endoreduplication cycle during each larval stage, which results in intestinal nuclei with a 32n DNA content in adult animals. A major part of the hypodermis is formed by fusion of individual cells. Cells that are added to this syncytium during larval development usually undergo a round of DNA synthesis just before fusion (Hedgecock and White, 1985). Consequently, the skin contains a mixture of nuclei with 2n and 4n DNA contents. The meiotic cell cycle Late larval and adult animals contain a large number of precursor germ cells in various stages of meiotic progression. Hermaphrodites temporarily produce male gametes during the third larval stage, before switching to an oogenesis program (reviewed in Schedl, 1997). A mitotically active ‘stem cell’ population at the distal end of the germ line forms the precursor germ cells. These divisions are the only mitotic divisions that continue in adult animals. The precursor cells enter a prolonged meiotic prophase, in which homologous chromosomes pair, synapse and undergo recombination. Subsequently, oocytes mature and arrest during diakinesis until fertilization. Upon fertilization, the oocyte pronucleus completes meiosis I and II, after which the female and male pronuclei meet to form a zygote (reviewed in Kemphues and Strome, 1997). cell-cycle entry or withdrawal show defects in mobility (Uncoordinated phenotype), fertility (Sterile phenotype), and formation of vulval structures (Vulvaless, protruding-Vulva or Multi-Vulva phenotypes). Screens for mutants with abnormal cell lineages (Lin mutants) have defined several cell-cycle components (Horvitz and Sulston, 1980; Sulston and Horvitz, 1981). Examples include the Cullin LIN-19/CUL-1, the bTRCP-related F box WD protein LIN-23, the Rb family member LIN-35, and the DP1-related transcription factor LIN-55/DPL-1 (see below). Further, a screen for mutants defective in vulval formation revealed cye-1 cyclin E mutations (Fay and Han, 2000), and mutations in cyd-1 cyclin D and cdk-4 Cdk4/6 were isolated in a screen for mutants with G1-arrested postembryonic precursor cells (Boxem and van den Heuvel, 2001). Finally, a cdc-14 allele was isolated in a screen for mutants with vulval precursor cells that fail to become quiescent (Saito et al., 2004). The specific function of these G1 regulators in C. elegans is discussed below. Although this list is incomplete, strong similarities with mammalian G1 control mechanisms are apparent (Figure 1). However, there are also striking omissions, for instance, orthologs of the INK4 gene family remain to be identified in C. elegans. Also, the C. elegans p53 family member (cep-1) promotes apoptosis in response to DNA damage in the germ line, but has not been implicated in cell-cycle arrest (Derry et al., 2001; Schumacher et al., 2001). C. elegans cyclin D and Cdk4/6 The C. elegans genome encodes a single D-type cyclin, CYD-1, and single kinase of the Cdk4/6 subfamily, CDK-4. A reverse genetic approach revealed that both cyd-1 and cdk-4 are required for postembryonic cell divisions (Park and Krause, 1999), and cyd-1 and cdk-4 mutants were isolated in a forward genetic screen for G1 control genes (Boxem and van den Heuvel, 2001). Homozygous cyd-1 and cdk-4 null mutants go through embryogenesis, complete only a few larval cell divisions Extracellular Signals Cyclin D1,2,3 CDK4,6 CYD-1/CDK-4 C. elegans G1 control genes Cell-cycle genes have been identified in relatively unbiased genetic screens and through examination of candidate genes based on homology with other species. Owing to the persistence of maternal product and the nature of embryonic cell cycles, homozygous G1 control mutants tend to complete embryogenesis and show cell division defects during larval development. Dividing cells contribute extensively to the nervous system (in particular, the ventral nerve cord), the reproductive system (both the somatic gonad and germline), and the egg-laying system (the vulva and associated egg laying muscles) during normal postembryonic development (Sulston and Horvitz, 1977). Hence, mutants defective in Oncogene pINK family M G1 G2 S pRb,p107,p130 Cip/Kip family LIN-35 CKI-1,2 Cyclin E /CDK2 CYE-1/K03E5.3? Figure 1 Generalized model for regulation of G1 progression in mammalian cells. For most classes of regulators, a single family member is expressed in C. elegans. The mammalian genes are indicated in gray above the putative C. elegans orthologs, indicated in bold. (-) and (–|) indicate positive regulation and negative regulation, respectively. See text for further details G1 progression in C. elegans J Koreth and S van den Heuvel 2759 of the gonad precursor cells, and remain thin, uncoordinated, small, and sterile larvae. The precursor nuclei of the postembryonic lineages arrest with a 2n DNA content and without expression of an S-phase reporter, the ribonucleotide reductase (rnr) promoter driving green fluorescent protein (GFP) expression (rnrHgfp), indicating that cyd-1 and cdk-4 are required for progression from G1 to S phase. The close similarity of the cyd-1 and cdk-4 loss-of-function phenotypes combined with binding of the proteins in vitro strongly indicate that C. elegans CYD-1 and CDK-4 function together in a complex. Based on reporter gene expression, transcription of cyd-1 and cdk-4 initiates before the 300-cell stage (Park and Krause, 1999). Yet, only a few very late embryonic divisions depend on cyd-1 and cdk-4 function. This is likely not a consequence of maternal product, as RNAi, which usually blocks the maternal contribution efficiently, mimicked the mutant phenotype. Even when combined with the homozygous null mutations, RNAi of cyd-1 and cdk-4 did not cause embryonic lethality or enhance the mutant phenotype (Boxem and van den Heuvel, unpublished). Thus, phosphorylation of target proteins by the CDK-4/CYD-1 kinase is not essential during the embryonic proliferation phase (see below: ‘Comparison with other systems’). In addition to arrest of cell divisions, homozygous cyd-1 and cdk-4 mutants show strongly reduced larval growth (Park and Krause, 1999; Boxem and van den Heuvel, 2001). The growth retardation follows the arrest of cell division, and undivided cells in the mutants are initially larger than their divided counterparts in wildtype animals (Figure 2). Therefore, lack of cell division does not result from reduced cell growth. Interestingly, Drosophila Cyclin D and Cdk4 also contribute to cell growth, while their role in G1 progression is less critical (see below; Datar et al., 2000; Meyer et al., 2000, 2002). The essential role of the Cyclin D/Cdk4 kinase in C. elegans can help define in vivo substrates that regulate G1–S progression downstream of cyd-1 and cdk-4. Loss of function of negative regulators that act downstream may be anticipated to rescue the cell proliferation defects and larval arrest of cyd-1 and cdk-4 mutant animals. Indeed, the retinoblastoma and Cip/Kip inhibitor pathways have been found to fulfill such functions. The retinoblastoma and E2F protein families The lin-35 gene was initially defined in a forward screen for genes that redundantly regulate vulval cell fates in C. elegans (Ferguson and Horvitz, 1989). Vulval cell fates are controlled by at least four regulatory pathways, including inductive signaling through an EGF-like growth factor receptor/tyrosine kinase/Ras/MAP kinase cascade. Mutations that increase inductive signaling cause extra cells to adopt vulval cell fates, resulting in animals that display a multivulva (Muv) phenotype. The inductive pathway is antagonized by a general repressive mechanism that involves the ‘synthetic multivulva’ (synMuv) genes. The synMuv genes form three different Figure 2 Retardation of cell growth in cyd-1 and cdk-4 mutants follows arrest of cell division. Cell boundaries of the hypodermal seam cells are visualized by the junctional marker AJM-1HGFP in a first-stage wild-type larva (Top), L3 stage wild-type larva (Middle) and L3 stage cyd-1(he112) mutant animal (bottom). The hypodermal seam cells in the mutants grow substantially, indicating that the lack of division is not caused by a growth defect. The size of the animals and individual cells are identical between wild-type and mutants until approximately 16 h of larval development, at which point cyd-1 and cdk-4 mutant animals start to lag behind. In contrast, cell divisions fail starting as early as late embryogenesis (Boxem and van den Heuvel, 2001; unpublished data) classes, A, B, and C, that act redundantly (Ceol and Horvitz, 2004). Only animals that contain mutations in two different classes show an Muv phenotype. Importantly, the synMuv class B gene lin-35 was found to encode a member of the retinoblastoma tumorsuppressor protein family (Lu and Horvitz, 1998). The predicted LIN-35 protein is not particularly close to a specific mammalian family member (pRb, p107, and p130). Although the spacer that separates the pocket A and B domains of LIN-35 is short, like that of pRb, the protein is somewhat more closely related to p130 and p107 (LIN-35 shares 20, 19, and 15% overall aminoacid identity with p130, p107, and pRb, respectively, and residues 744–839 of the LIN-35 pocket B region are 34, 34, and 30% identical to these respective proteins). Thus, lin-35 Rb likely evolved from a single common ancestor before it diverged into three distinct genes. The synMuv class B genes not only include lin-35 Rb, but also homologs of other genes known to interact with Rb in other species. These include, efl-1 (E2F family like) and dpl-1 (DP family like) (Ceol and Horvitz, 2001), as well as lin-53 (RbAP46/48) and hda-1 (HDAC1), which encode components of the Nucleosome Remodeling and Deacetylase (NURD) complex (Solari and Ahringer, 2000). The combined genetic and molecular data suggest Oncogene G1 progression in C. elegans J Koreth and S van den Heuvel 2760 that class B synMuv genes inhibit vulval cell-fate determination and antagonize Ras signaling through transcriptional repression mediated by an E2F/pRb/ NURD complex. These studies did not reveal a cell-cycle role for lin-35 Rb. Animals homozygous for lin-35 Rb null alleles have a reduced brood size, but are otherwise viable and have mostly normal cell division patterns. More direct measurements of DNA content demonstrated that lin35 Rb is not rate limiting for S-phase entry (Boxem and van den Heuvel, 2001). However, a role for lin-35 Rb in G1 control became apparent in sensitized genetic backgrounds. Firstly, the cyd-1 and cdk-4 loss-offunction phenotypes are rescued to a large extent by concomitant inactivation of lin-35 Rb: the double mutant animals show expression of the rnrHgfp S-phase reporter, restore postembryonic DNA synthesis, go through endoreduplication cycles, and show substantial postembryonic cell division and growth. In addition, lin35 Rb inactivation greatly enhances the number of extra cell divisions in a cki-1 Kip1 loss of function background (Boxem and van den Heuvel, 2001; see below). These results are consistent with LIN-35 Rb functioning as a major downstream target of the Cyclin-D/CDK-4 complex (Figure 3). Further, lin-35 Rb and cki-1 Cip/ Kip apparently provide parallel levels of control and cooperate to inhibit G1-to-S phase progression. Both of these conclusions agree with numerous results from studies in mammalian systems. The same sensitized backgrounds have been used to examine cell-cycle regulatory functions of synMuv cydcyd-1 Cyclin D cdkcdk-4 CDK4/6 The CIP/KIP family of Cdk inhibitors cdccdc-14 Cdc14 linlin-36 linlin-35 Rb eflefl-1 dpldpl-1 E2F DP linlin-9 ckicki-1 Cip/Kip cyecye-1 Cyclin E ? K03E5.3 CDK2 G1 S Figure 3 Model for G1 regulation during post-embryonic development of C. elegans. Based on mutant phenotypes and genetic interactions, the genes indicated have been placed in a branched pathway. The results indicate both that G1 control is conserved between C. elegans and mammals, and that novel control elements can be found through genetic studies in C. elegans. (-) and (–|) indicate positive regulation and negative regulation, respectively. See text for further details Oncogene genes other than lin-35 Rb (Boxem and van den Heuvel, 2002). While none of the class A genes examined scored in these assays, a subset of the synMuv class B genes was found to act in G1 control. Specifically, efl-1 E2F4/5, dpl-1 DP, lin-9 Mip130/TWIT, lin-36, and lin-15B contribute to inhibition of G1 progression. In each assay, efl-1 and lin-36 behaved most similar to lin-35 Rb: their inactivation rescued cyd-1 and cdk-4 mutants substantially and increased hyperproliferation when combined with cki-1 Kip1 RNAi. Thus, EFL-1 E2F4/5 and the predicted Zn-finger protein LIN-36 likely act in a pathway or complex with LIN-35 Rb to repress G1/S control genes (Figure 3). Inactivation of dpl-1 DP and efl-1 E2F4/5 caused similar effects in these assays; however, dpl-1 inactivation on its own reduced cell division and expression of the rnrHgfp S-phase reporter. Thus, DPL-1 DP acts both to promote and inhibit G1/S progression. In accordance with observations in other systems, an EFL-1/DPL-1 complex likely recruits LIN-35 Rb to repress transcription, while DPL-1 may act with another E2F family member to activate S-phase genes. As yet, a transcriptional activating E2F member has not been identified; efl-2 E2F is a good candidate but an RNAi phenotype has not been observed (Ceol and Horvitz, 2001; Boxem and van den Heuvel, 2002). The role of LIN-9 and the predicted novel protein LIN-15B remain poorly defined. Unlike lin-35 Rb, inactivation of these genes did not enhance the cki-1 Kip1 overproliferation phenotype. In fact, lin-15B appeared to act in parallel to lin-35 Rb in G1 control (Boxem and van den Heuvel, 2002). The C. elegans genome contains two closely linked loci on chromosome II that are predicted to encode members of the CIP/KIP family of Cdk inhibitors. The CKI-1 and CKI-2 proteins share approximately 30% aminoacid identity and are each about equally close to p21Cip1 and p27Kip1 (Hong et al., 1998; Feng et al., 1999). However, a function has been described only for cki-1. CKI-1 is considered orthologous to mammalian p27Kip1, as it acts as a developmental regulator of cell-cycle entry and apparently not in DNA damage response. Overexpression of cki-1 causes cell-cycle arrest in G1, while loss of function results in extra divisions in multiple cell lineages, including the intestinal and hypodermal lineages (Hong et al., 1998; Boxem and van den Heuvel, 2001; Fukuyama et al., 2003). A deletion that spans cki-1 and cki-2 causes embryonic lethality, which can be rescued by cki-1 expression (Fukuyama et al., 2003). Inactivation of cki-1 by RNAi results in partial embryonic lethality as well as sterile adults with hyperproliferation during larval development. Postembryonic precursor cells in cki-1(RNAi) animals fail to arrest in G1 and ectopically express the Sphase marker rnrHgfp. Thus, cki-1 Kip1 function is rate limiting for S-phase entry, in particular in cells that enter a prolonged quiescent state before dividing again. Interestingly, loss of cki-1 Kip1 also affects aspects of G1 progression in C. elegans J Koreth and S van den Heuvel 2761 cell differentiation, cell fate, and cell death (Fukuyama et al., 2003; Kostic et al., 2003). Multiple levels of control have been shown to affect cki-1 promoter activity (Hong et al., 1998). This involves regulators that are cell-type dependent, larval stage dependent-such as heterochronic genes, or environmentally determined–such as L1 arrest in response of food starvation and induction of the dauer stage. Multiple 50 upstream sequences in the cki-1 promoter region mediate the different signals (Hong et al., 1998). Thus, proper temporal regulation of cki-1 expression appears to control lineage-specific and environmentally induced arrest of cell division. Inactivation of cki-1 significantly rescued the postembryonic arrest of cell proliferation in cyd-1 or cdk-4 loss-of-function mutants. This indicates that the worm CDK inhibitors act downstream of or in parallel to the Cyclin-D/CDK-4 complex. Thus, as outlined above, lin35 Rb and cki-1 Kip1 each act downstream of cyd-1/ cdk-4. In contrast to lin-35 Rb, cki-1 inactivation allows just one round of endoreduplication in cyd-1 mutants. As expected, CKI-1 was found to bind CYE-1 Cyclin E in the yeast two-hybrid system (Boxem and van den Heuvel, unpublished data). These results are consistent with a model in which CKI-1 Kip1 inhibits the Cyclin E kinase by direct association. As proposed for other systems, in the absence of CKI-1 Kip1, residual Cyclin E kinase activity may drive one more round of DNA synthesis. Additional cycles depend on transcription of S-phase genes, which requires inactivation of lin-35 Rb. The C. elegans CDC-14 phosphatase A putative null allele of cdc-14 was identified in a screen for mutants with inappropriate divisions of the vulval precursor cells at a time of normal quiescence (Saito et al., 2004). Subsequent studies showed that cdc-14 is required for developmental arrest of cell division in multiple lineages, acting similar to cki-1 but not as strongly. A variety of genetic experiments and transgene expression studies indicated that cdc-14 acts upstream of cki-1 and promotes accumulation of CKI-1 protein in the nucleus. As mammalian p27Kip1 is degraded in a phosphorylation- and ubiquitin-dependent fashion, CDC-14 might counteract CKI-1 phosphorylation and promote a stable form of CKI-1 that can accumulate to the levels required for developmental arrest of cell division. Functional GFPHCDC-14 fusion proteins showed a highly dynamic pattern of localization. In cells that temporally arrested, GFPHCDC-14 accumulated in the cytoplasm. In contrast, the fusion protein was recruited into the nucleus as soon as cells entered a terminally differentiated state. Interestingly, only cells that undergo prolonged developmental G0/G1 phase arrest are affected by cdc-14 loss of function, indicating that cytoplasmic localization is critical for CDC-14 to arrest cell division. The lack of a mitotic Cdc-14 phenotype in these studies was surprising. For one, cytokinesis defects have been observed to result from cdc-14 RNAi, which can be explained, at least in part, by the use of a sensitized zen-4(ts) background (Gruneberg et al., 2002; Mishima et al., 2004). Moreover, Cdc14 is required for mitotic exit in the budding yeast S. cerevisiae (Visintin et al., 1998). In addition, the worm GFPHCDC-14 fusion protein showed strong association with the spindle apparatus in mitosis (Saito et al., 2004). Thus, although cdc-14 deletion mutants do not display defects in mitosis or cytokinesis, a nonessential mitotic role is likely. Yeast Cdc14p dephosphorylates the Cdk inhibitor Sic1p, which acts in mitosis as well as G1 phase, yet Cdc14p is kept inactive for most of the cell cycle by sequestration into the nucleolus (Mendenhall, 1993; Schwob et al., 1994; Shou et al., 1999; Visintin et al., 1999). In contrast, Cip/Kip family members act predominantly in G1 phase, and Cdc14 appears to act during multiple cell-cycle phases in animal cells. These and other differences between yeast and animal cells could shift the balance in Cdc14 requirement. Cyclin-E/Cdk-2 The cye-1 gene encodes the only C. elegans E-type cyclin. Null mutations in cye-1 were found to cause a surprisingly mild phenotype in the worm, displaying late larval cell division defects and sterility (Seydoux et al., 1993; Fay and Han, 2000). In contrast, cye-1 RNAi causes embryonic arrest at the B100 cell stage, thus the late phenotype of homozygous mutants appears to reflect a long-lived maternal contribution of cye-1 products. CYE-1 protein is present in early embryos and cye-1 transcription initiates as early as the 28-cell stage (Brodigan et al., 2003). Why cye-1 RNAi does not cause defects at an even earlier time of embryonic development is unclear at present. Currently, an unequivocal Cdk2 ortholog has not yet been identified in C. elegans. A candidate gene, K03E5.3, shares B43% amino-acid identity with human Cdk2. Depletion of its gene product by RNAi results in highly variable defects, ranging from sterile adults, arrest at early and late larval stages and embryonic lethality (Boxem et al., 1999). This range of phenotypes could reflect incomplete inactivation by RNAi, or nonessential gene functions. It is possible that redundancy or developmental flexibility allows some cell divisions to continue in the absence of Cyclin E and Cdk2 function. This would be in accordance with the surprisingly limited phenotypes of double knockout Cyclin E1 and E2 mice, as well as the viability of Cdk2 knockout mice (Berthet et al., 2003; Geng et al., 2003; Ortega et al., 2003). Only limited data are available to place cye-1 in a G1 control pathway, but its genetic interactions are clearly different from cyd-1. Inactivation of lin-35 Rb does not affect the cye-1 null-mutant phenotype and, conversely, cye-1 RNAi is equally severe in a wild-type or lin-35 Rb mutant background (Boxem and van den Heuvel, 2001). However, incomplete inactivation of cye-1 by feeding RNAi is suppressed by lin-35 Rb loss of function (Ceron and van den Heuvel, unpublished). These results are Oncogene G1 progression in C. elegans J Koreth and S van den Heuvel 2762 consistent with lin-35 Rb acting upstream and negatively regulating cye-1, a hypothesis that is also supported by the presence of multiple E2F binding sites in the cye-1 promoter (Brodigan et al., 2003). SCF components Among the first cloned C. elegans cell-cycle genes were cul-1 (previously named lin-19). Both lin-19/cul-1 and lin-23 mutants were identified in early screens and found to display excessive cell divisions with accelerated G1–S transitions in all post-embryonic lineages (Kipreos et al., 1996, 2000). CUL-1 was the founding member of the cullin family in metazoans, and LIN-23 is an F-box WD repeat protein, most similar to MET30 in yeast, human b-TRCP, and the slmb F-box protein in Drosophila (Kipreos et al., 2000). Thus, based on similar phenotypes, both proteins are likely components of an Skp1Cul1-F box (SCF) protein complex that acts as an E3 ubiquitin ligase and targets proteins for degradation. The critical target proteins of the CUL-1/LIN-23 SCF are expected to be positive regulators of the G1–S transition. In analogy to a vertebrate CUL-1 containing SCF complex, Cyclin E is considered a candidate critical target. Not only SCF, but also the APCCdh1 E3 ubiquitin ligase, contributes to G1 control. This is indicated by the increased number of larval cell divisions in mutants that simultaneously lack lin-35 Rb and fzr-1 Cdh1 function (Fay et al., 2002). Comparison with other systems As indicated in the preceding discussion, cell-cycle effectors and their regulatory modules tend to be conserved between C. elegans and mammals. Consequently, many of the genetic observations made in C. elegans support existing models of G1 control. This includes the observed roles for a Cyclin D–Cdk4/6 related kinase in promoting G1 progression, and cooperation between pRb and Cip/Kip family members in inhibiting this process. Furthermore, C. elegans Rb acts in concert with E2F/DP transcription factors, and likely represses cyclin E transcription. Despite these global similarities, a number of specific aspects of the mutant phenotypes are surprising and warrant further discussion. CYD-1 and CDK-4 are required late in embryogenesis and throughout larval development, but why not at earlier times? It is important to consider that S-phase entry during embryonic divisions is controlled largely by maternal mRNA and protein products. Our observations suggest that the most critical function of the Cyclin D kinase is inactivation of LIN-35 Rb. If cyd-1/cdk-4 are inactive, LIN-35 Rb likely continues to act as a transcriptional repressor. While Rb activity can prevent S-phase entry during larval development, a transcriptional repressor cannot stop embryonic expression of maternally provided products. Thus, cyd-1 and cdk-4 likely have limited functions during embryogenesis Oncogene because cell divisions can continue without zygotic transcription of cell-cycle genes. Mice that completely lack Cyclin D kinase activity also proceed through development until late embryogenesis (Kozar et al., 2004; Malumbres et al., 2004). Such embryos complete extensive proliferation as well as cell differentiation, morphogenesis, and organogenesis. Careful characterizations have indicated roles for compensatory mechanisms in the triple D-type cyclin KO mice (Kozar et al., 2004). Thus, rather than persistent maternal gene product, developmental flexibility and genetic redundancy likely explain how more complex animals manage to alleviate the requirement for Cyclin D kinase activity. The fact that cell divisions arrest abruptly in cyd-1 and cdk-4 larvae may point to a lower degree of redundancy and flexibility during worm development. The situation is somewhat different again in Drosophila, as the contribution of fly Cyclin D/Cdk4 in G1 progression appears to be limited. Interestingly, DmCycD/Cdk4 complexes do not associate with the fly Cip/Kip inhibitor dacapo (Meyer et al., 2000). Association of mammalian Cyclin D–Cdk4/6 complexes with p21Cip1 and p27Kip1 has been suggested to contribute to activation of the Cyclin E–Cdk2 kinase acting downstream, by sequestering its inhibitor. C. elegans cki-1 Kip1 loss of function partly suppresses the requirement for cyd-1 and cdk-4 in vivo, and CKI-1 associates with worm Cyclins D and E in the yeast twohybrid system (Boxem and van den Heuvel, 2001). Thus, C. elegans CYD-1/CDK-4 may contribute to CKI-1 inactivation, through sequestration or phosphorylation. As Drosophila Cyclin D–Cdk4/6 fails to interact with dacapo, it may have a less prominent role in G1 progression. Vertebrate Cip/Kip inhibitors have also been suggested to act as assembly factors for Cyclin D–Cdk4/6 complexes (LaBaer et al., 1997; Sherr and Roberts, 1999). Such a role has not been detected for the C. elegans orthologs. Inactivation of cki-1 by RNAi, or deletion of cki-1 and cki-2 together, results in strongly increased numbers of cell divisions. However, this number is substantially reduced when cki-1(RNAi) is combined with cyd-1 or cdk-4 loss of function. The difference indicates that the CYD-1/CDK-4 kinase is not inactivated by cki-1 loss of function, and that the worm Cip/Kip inhibitor plays only a negative role in cell-cycle regulation. Another surprise is that the single Rb family member is not essential in the worm. Hardly any extra divisions have been observed in lin-35 Rb null mutants, although a low number of extra intestinal nuclei are formed. Examining the timing of DNA replication also did not indicate premature entry into S phase in lin-35 Rb mutants, in contrast to cki-1 (RNAi) animals in which cells prematurely enter the division cycle (Boxem and van den Heuvel, 2001). Thus, lin-35 Rb acts redundantly with cki-1 Kip1 in G1 control, and cki-1 is more rate limiting. This shift in balance compared to more complex animals may be explained by the fact that more than half of the somatic cells in the worm are G1 progression in C. elegans J Koreth and S van den Heuvel 2763 formed during embryogenesis. Given the importance of maternal product during these embryonic divisions, post-transcriptional levels of regulation must be more prominent in this animal. Conclusion From worm to man, the genes that control G1 progression and the hierarchies or pathways in which they act are well conserved. C. elegans can be used to clarify the in vivo interactions of various known G1 regulators. These data form a solid base on which to start adding additional genes and pathways involved in the regulation of cell-cycle entry. Such genes could be identified in forward and reverse genetic screens using, for instance, a sensitized genetic background or specific cell-cycle reporters. Areas that can be addressed include: the molecular mechanisms that couple normal development and the cell-autonomous cell-cycle machinery and the connection between the external milieu and developmental cell- cycle progression decisions in vivo. In these areas many questions remain, for example, how does starvation prevent cell proliferation in the first larval stage? How do the pathways that induce dauer formation connect to the cell-cycle machinery? What is the interrelationship between cell growth, differentiation and G1 progression? C. elegans has numerous advantages that make it an excellent model to identify pathways regulating the organism-wide control of cell-cycle progression during development and environmental stress. Such pathways will extend our knowledge of cell-cycle control, central to both normal development and cancer. Acknowledgements We thank our many colleagues and collaborators for generously sharing reagents and ideas, and apologize to those whose valuable work could not be cited owing to space constraints. We thank Mike Boxem, John S. Satterlee and Inge The for their suggestions and critical review of the manuscript. This work was funded by grants from the Claudia Adams Barr Program (to JK) and the National Institutes of Health (to SvdH). References Berthet C, Aleem E, Coppola V, Tessarollo L and Kaldis P. (2003). Curr. Biol., 13, 1775–1785. Boxem M, Srinivasan DG and van den Heuvel S. (1999). Development, 126, 2227–2239. Boxem M and van den Heuvel S. (2001). Development, 128, 4349–4359. Boxem M and van den Heuvel S. (2002). Curr. Biol., 12, 906– 911. Brenner S. (1974). Genetics, 77, 71–94. Brodigan TM, Liu J, Park M, Kipreos ET and Krause M. (2003). Dev. Biol., 254, 102–115. Ceol CJ and Horvitz HR. (2001). Mol. Cell, 7, 461–473. Ceol CJ and Horvitz HR. (2004). Dev. Cell, 6, 563–576. Datar SA, Jacobs HW, de la Cruz AF, Lehner CF and Edgar BA. (2000). EMBO J., 19, 4543–4554. Derry WB, Putzke AP and Rothman JH. (2001). Science, 294, 591–595. Edgar LG and McGhee JD. (1988). Cell, 53, 589–599. Fay DS and Han M. (2000). Development, 127, 4049–4060. Fay DS, Keenan S and Han M. (2002). Genes Dev., 16, 503– 517. Feng H, Zhong W, Punkosdy G, Gu S, Zhou L, Seabolt EK and Kipreos ET. (1999). Nat. Cell Biol., 1, 486–492. Ferguson EL and Horvitz HR. (1989). Genetics, 123, 109–121. Fire A, Xu S, Montgomery MK, Kostas SA, Driver SE and Mello CC. (1998). Nature, 391, 806–811. Fukuyama M, Gendreau SB, Derry WB and Rothman JH. (2003). Dev. Biol., 260, 273–286. Geng Y, Yu Q, Sicinska E, Das M, Schneider JE, Bhattacharya S, Rideout WM, Bronson RT, Gardner H and Sicinski P. (2003). Cell, 114, 431–443. Gruneberg U, Glotzer M, Gartner A and Nigg EA. (2002). J. Cell Biol., 158, 901–914. Hedgecock EM and White JG. (1985). Dev. Biol., 107, 128– 133. Hong Y, Roy R and Ambros V. (1998). Development, 125, 3585–3597. Horvitz HR and Sulston JE. (1980). Genetics, 96, 435–454. Kamath RS and Ahringer J. (2003). Methods (Duluth), 30, 313–321. Kemphues KJ and Strome S. (1997). C. elegans II. Riddle DL, Blumenthal T, Meyer BJ and Priess JR (eds). Cold Spring Harbor Laboratory Press: New York, pp. 335–359. Kimura KD, Tissenbaum HA, Liu Y and Ruvkun G. (1997). Science, 277, 942–946. Kipreos ET, Gohel SP and Hedgecock EM. (2000). Development, 127, 5071–5082. Kipreos ET, Lander LE, Wing JP, He WW and Hedgecock EM. (1996). Cell, 85, 829–839. Kostic I, Li S and Roy R. (2003). Dev. Biol., 263, 242–252. Kozar K, Ciemerych MA, Rebel VI, Shigematsu H, Zagozdzon A, Sicinska E, Geng Y, Yu Q, Bhattacharya S, Bronson RT, Akashi K and Sicinski P. (2004). Cell, 118, 477–491. LaBaer J, Garrett MD, Stevenson LF, Slingerland JM, Sandhu C, Chou HS, Fattaey A and Harlow E. (1997). Genes Dev., 11, 847–862. Lambie EJ. (2002). BioEssays, 24, 38–53. Lee RC, Feinbaum RL and Ambros V. (1993). Cell, 75, 843– 854. Lu X and Horvitz HR. (1998). Cell, 95, 981–991. Malumbres M, Sotillo R, Santamaria D, Galan J, Cerezo A, Ortega S, Dubus P and Barbacid M. (2004). Cell, 118, 493–504. Mendenhall MD. (1993). Science, 259, 216–219. Meyer CA, Jacobs HW, Datar SA, Du W, Edgar BA and Lehner CF. (2000). EMBO J., 19, 4533–4542. Meyer CA, Jacobs HW and Lehner CF. (2002). Curr. Biol., 12, 661–666. Mishima M, Pavicic V, Gruneberg U, Nigg EA and Glotzer M. (2004). Nature, 430, 908–913. Ortega S, Prieto I, Odajima J, Martin A, Dubus P, Sotillo R, Barbero JL, Malumbres M and Barbacid M. (2003). Nat. Genet., 35, 25–31. Oncogene G1 progression in C. elegans J Koreth and S van den Heuvel 2764 Park M and Krause MW. (1999). Development, 126, 4849–4860. Riddle DL and Albert PS. (1997). C. elegans II. Riddle DL, Blumenthal T, Meyer BJ and Priess JR (eds). Cold Spring Harbor Laboratory Press: New York, pp. 739–768. Rual JF, Ceron J, Koreth J, Hao T, Nicot AS, HirozaneKishikawa T, Vandenhaute J, Orkin SH, Hill DE, van den Heuvel S and Vidal M. (2004). Genome Res., 14, 2162–2168. Saito RM, Perrault A, Peach B, Satterlee JS and van den Heuvel S. (2004). Nat. Cell Biol., 6, 693–695. Schedl T. (1997). C. elegans II. Riddle DL, Blumenthal T, Meyer BJ and Priess JR (eds). Cold Spring Harbor Laboratory Press: New York, pp. 241–269. Schumacher B, Hofmann K, Boulton S and Gartner A. (2001). Curr. Biol., 11, 1722–1727. Schwob E, Bohm T, Mendenhall MD and Nasmyth K. (1994). Cell, 79, 233–244. Oncogene Seydoux G, Savage C and Greenwald I. (1993). Dev. Biol., 157, 423–436. Sherr CJ and Roberts JM. (1999). Genes Dev., 13, 1501–1512. Shou W, Seol JH, Shevchenko A, Baskerville C, Moazed D, Chen ZW, Jang J, Charbonneau H and Deshaies RJ. (1999). Cell, 97, 233–244. Solari F and Ahringer J. (2000). Curr. Biol., 10, 223–226. Sulston JE and Horvitz HR. (1977). Dev. Biol., 56, 110–156. Sulston JE and Horvitz HR. (1981). Dev. Biol., 82, 41–55. Sulston JE, Schierenberg E, White JG and Thomson JN. (1983). Dev. Biol., 100, 64–119. Visintin R, Craig K, Hwang ES, Prinz S, Tyers M and Amon A. (1998). Mol. Cell, 2, 709–718. Visintin R, Hwang ES and Amon A. (1999). Nature, 398, 818–823.