Investigating the hows and whys of DNA endoreduplication

Abstract

Endoreduplication is a form of nuclear polyploidization that results in multiple, uniform copies of chromosomes. This process is common in plants and animals, especially in tissues with high metabolic activity, and it generally occurs in cells that are terminally differentiated. In plants, endoreduplication is well documented in the endosperm and cotyledons of developing seeds, but it also occurs in many tissues throughout the plant. It is thought that endoreduplication provides a mechanism to increase the level of gene expression, but the function of this process has not been thoroughly investigated. Numerous observations have been made of endoreduplication, or at least extra cycles of S‐phase, as a consequence of mutations in genes controlling several aspects of cell cycle regulation. However, until recently there were few studies directed at the molecular mechanisms responsible for this specialized cell cycle. It is suggested that endoreduplication requires nothing more elaborate than a loss of M‐phase cyclin‐dependent kinase activity and oscillations in the activity of S‐phase cyclin‐dependent kinase.

Endoreduplication, gene expression, cell cycle regulation, cyclin‐dependent kinase.

Introduction

Nuclear polyploidization occurs widely in metabolically active tissues of plants and animals, and the most common method by which it takes place is through endoreduplication (Brodsky and Uryvaeva, 1977; D'Amato, 1984). During this process, cells amplify their genome without chromatin condensation, segregation or cytokinesis, resulting in what appear to be multiple, uniform copies of the nuclear DNA. This process has been quantified in plants and animals (Bhatnagar and Sawhney, 1981; Kowles and Phillips, 1988), but its physiological significance is poorly understood. Endoreduplication could provide a mechanism to increase the availability of DNA templates and thus increase gene expression. Transcriptional and translational activities increase proportionately with each doubling of the genome, so the metabolic activity of a highly polyploid cell could be functionally equivalent to that of many diploid cells (D'Amato, 1984). It is also possible that endoreduplication maintains an optimum ratio between cell and nuclear size (Cavalier‐Smith, 1985). Typically, cells that have undergone endoreduplication are larger than comparable cells that have not. Being bigger and having a larger number of organelles (mitochondria and plastids) to service would potentially require greater transcription capacity. It was suggested that large cells have the capacity to increase their volume faster than smaller cells, and in the case of rapidly growing fruits and seeds, this could be advantageous (Grime and Mowforth, 1982). Because endoreduplication occurs frequently in seed storage tissues, such as cotyledons and endosperms, it has also been suggested that this process could provide a mechanism for storing nucleotides or nitrogen for the embryo. However, there is little experimental evidence showing that nucleotides are recycled, and this would be a very energy‐expensive way to store nitrogen. Consequently, the notion remains that endoreduplication more than likely occurs to provide higher than normal levels of gene expression when the constraints of time and space require it.

Endoreduplication is often described as a novel cell cycle in eukaryotes, because chromosomes replicate, but nuclei and cells do not divide. However, endoreduplication is so common in plants and animals that it would seem more appropriate to describe it simply as an ‘alternative’ cell cycle, or perhaps as a cell cycle of terminal differentiation, since endoreduplicated cells generally do not divide. Clearly, some of the mechanisms that direct sequential progression of the G1‐, S‐, G2‐ and M‐phases of the cell cycle are modified for endoreduplication to occur. Normally, chromosomes only replicate once per cell cycle, and progression through M‐phase is required to ‘license’ origins of replication for the next round of chromosome duplication. This review also considers how oscillating levels of cyclin‐dependent kinase (CDK) activity could allow multiple cycles of licensing and DNA replication without intervening mitoses.

Significant progress has been made identifying genes responsible for cell cycle regulation in plants (Mironov et al., 1999). Because of the frequency with which endoreduplication occurs in plants, and its presumed importance, there is a great deal of interest in understanding its regulation and role in development (Grafi, 1998). In Arabidopsis thaliana (L.) Heynh., as is true of mammals and Drosophila melanogaster (Zybina and Zybina, 1996; White‐Cooper and Glover, 1995), endoreduplication is under developmental control. Evidence was found of endoreduplication in most Arabidopsis tissues, including the leaves and stem of bolting plants, but not in inflorescence tissue (Galbraith et al., 1991). Endoreduplicated nuclei generally become more prevalent as the plant matures, but the mechanisms that trigger the process are unknown.

Studies of a variety of plant cells and tissues suggest that endoreduplication is modulated by the levels of auxin, cytokinin, abscisic acid, and gibberellin (Meyers et al., 1990; Artlip et al., 1995; Valente et al., 1998). Indeed, there is mounting evidence that these growth regulators directly affect the expression of key cell cycle regulatory genes (Frank and Schmülling, 1999). For example, cytokinin has been implicated in the activation of both the M‐phase (Zhang et al., 1996) and the S‐phase CDKs (Riou‐Khamlichi et al., 1999). Auxin has been linked to the regulation of protein turnover via the ubiquitin‐proteosome pathway (Leyser et al., 1993; del Pozo and Estelle, 1999), which controls the stability of a number of cell cycle regulatory proteins (King et al., 1996). Recently, a gene that is highly expressed in differentiating alfalfa root nodules, ccs52 (Cebolla et al., 1999), was shown to be related to proteins involved in the destruction of cell cycle regulators during anaphase. Inhibition of ccs52 expression in transgenic Medicago truncatula plants reduced endoreduplication in petioles, hypocotyls and roots, suggesting that endoreduplication requires proteolysis of certain cell cycle regulators.

Endoreduplication in maize endosperm

Zea mays (L.) endosperm provides a useful tissue to investigate the mechanism of endoreduplication and its functional significance in a seed storage organ. This tissue develops from a fertilization event involving two polar nuclei in the female gametophyte and a sperm nucleus from the male gametophyte. Subsequently, this triploid nucleus undergoes multiple, synchronous divisions, giving rise to a syncytium of several hundred nuclei. Cell walls begin to form around the nuclei by the third day after pollination (DAP), and by the fourth or fifth day the tissue is fully cellularized and uninucleate (Kowles and Phillips, 1988). The most rapid period of endosperm growth is between 8 and 12 DAP, when cell division and enlargement are occurring. Usually by 12 DAP, the endosperm completely fills the central cavity of the kernel, and at this stage the outermost layer of cells, the aleurone, is clearly differentiated. Cell divisions cease in the central region of the endosperm after 12 DAP, and the nuclei commence to endoreduplicate their DNA. Cell divisions continue in the sub‐aleurone cells until approximately 20 DAP, but the mitotic index of the endosperm drops to less than 1% after 14 DAP.

Figure 1 illustrates the rapid rate of growth and development of the maize kernel between 5 and 15 DAP. At 5 DAP, the caryopsis is primarily filled with nucellus tissue; the endosperm is cellular, and it has enlarged much more rapidly than the embryo following fertilization. The nucellus and pericarp consist of small cells with uniformly small nuclei. At this stage the triploid endosperm nuclei are approximately the same size as the diploid nuclei of the pericarp and nucellus. The tissue morphology of the endosperm is not well preserved in the section shown in Fig. 1, probably because the primary walls are fragile at this stage of development. By 15 DAP, the nucellus has mostly disintegrated, and the endosperm occupies the largest volume of the kernel. The central endosperm cells, which compose the so‐called starchy endosperm region, are significantly larger than those at the periphery, especially the aleurone and sub‐aleurone layers. Figure 1 provides clear evidence of endoreduplication extending from the crown to the basal transfer cells in the central endosperm nuclei. The nuclei decrease in size from the central endosperm to the aleurone layer, suggesting a developmental gradient in the progress of endoreduplication.

Once endoreduplication begins in the endosperm, it can affect as many as 90% of the nuclei. However, the number of endocycles varies among endosperm cells and between maize genotypes. Based on a survey of maize genotypes (Kowles and Phillips, 1985), four to five cycles of endoreduplication are common in most inbreds, although individual nuclei with DNA content up to 690C (1C is the DNA content of a haploid nucleus during G1) were measured in some genotypes. The degree of endoreduplication in maize ‘flint’, ‘dent’ and ‘popcorn’ inbred lines has recently been analysed and its mode of inheritance has been investigated (Dilkes et al., unpublished data). Figure 2 shows flow cytometric analyses of nuclei from the endosperm of several inbred lines and illustrates the range of phenotypic variation in endoreduplication that was observed. Measurements were made at several stages of development to determine when the maximum level of endoreduplication was detected. In B73, a mid‐western dent, the DNA content in endosperm nuclei reached a maximum level of 96C with a mean ploidy of around 13C. SG18, a popcorn, undergoes a similar maximum number of endoreduplication cycles; however, a higher percentage of nuclei undergo multiple rounds of endoreduplication. In SG1533 and A1‐6, also popcorns, the highest mean ploidy levels that it was possible to measure was 26C, i.e. twice the value of B73. Additional cycles of endoreduplication may occur after this stage of development, but the accumulation of starch in the endosperm makes it very difficult to isolate intact nuclei and measure their DNA content by flow cytometry.

The significance of the variation in genome copy number in these different types of maize inbreds is not known. As there are significant morphological and genotypic differences between dent and popcorns, it could be misleading to draw conclusions regarding the phenotypic variation associated with high and low mean endoreduplication values in the endosperm. To investigate this question, it is necessary to introgress the high and low endoreduplication phenotypes into a common genetic background. This task is in progress.

A maternal effect influencing the degree of endoreduplication in maize endosperm has been observed (Cavallini et al., 1995), and this was described recently (Kowles et al., 1997). The pattern of endoreduplication in F₁ crosses resembles the phenotype of the female parent, and endosperms of F₂ families show little difference in the degree of endoreduplication. Fundamentally, there is no more variation than is observed in the F₁ endosperms. Only in the F₃ generation was significant variation observed in the degree of endoreduplication. Although the genetic basis of the maternal influence is poorly understood, it appears to be a common regulatory mechanism in endosperm (Haig and Westoby, 1991) and cotyledon (Lemontey et al., 2000) development. Several maternal‐effect genes influencing endosperm development have been identified (Ohad et al., 1996; Grossniklaus et al., 1998; Chaudhury et al., 1997), and because of the active research on this topic, this may be better understood in the future.

Maize defective kernel (dek) mutants manifest small or incompletely developed endosperms or embryos (Sheridan and Neuffer, 1980). Such mutants have often been considered good candidates for mutations affecting genes involved in the mitotic and endoreduplication cell cycles. However, to date the evidence for this has not been compelling. This question was investigated by examining the phenotypes of a group of 35 dek mutants with regard to cell size, cell number and the extent of endoreduplication (Kowles et al., 1992). All of these mutants showed reduced mitotic activity, and all except one had a reduced level of endoreduplication. These researchers interpreted their results to suggest that the endoreduplication cell cycle can be uncoupled from the mitotic process, but they did not identify a cell cycle defect per se. Consequently, it was not possible to distinguish between developmental effects resulting from defects in cell cycle regulation or mutations in general house‐keeping genes. Nevertheless, the maize dek mutants present interesting phenotypes that can eventually be understood through transposon tagging and reverse genetic approaches. Some of these mutants may turn out to result from defects in mechanisms controlling endoreduplication.

Fig. 1.

Median longitudinal sections of maize (W64A) developing kernels at 5 d and 15 d after pollination (DAP) showing nuclei labelled with propidium iodide (PI). The fluorescent PI signal is mainly confined to nuclei, though moderate fluorescence is present in the cytoplasm of subaleurone endosperm cells at 15 DAP due to labelling of RNA and some autofluorescence. The fluorescent signal is exceedingly high in regions with small, compact cells, i.e. embryo, peripheral nucellus, phloem parenchyma, and aleurone. Al, aleurone; CSEn, central starchy endosperm; Em, embryo; En, endosperm; Nu, nucellus; Pe, pericarp; Pl, placentochalaza; TC, transfer cells.

Fig. 2.

Variability in endoreduplication among selected maize inbred lines. The histograms show the C values of endosperm nuclei at the peak of endoreduplication in each genotype: Sg18, 26 DAP; Sg1533, 29 DAP; A1‐6, 27 DAP; Kp58k, 25 DAP; B73, 23 DAP.

Modifying the cell cycle to create endoreduplication

The normal mitotic cell cycle consists of periods of DNA synthesis (S‐phase) and chromosome separation (M‐phase) preceded by gaps: G1 and G2, respectively. During this cycle, the orderly progression of events that cause chromosomes to duplicate and separate is controlled by CDKs, which consist of two essential components: a serine/threonine protein kinase and a cyclin regulatory subunit. Temporal activity of CDKs is controlled by changes in cyclin components, multiple reactions that affect phosphorylation of the protein kinase and its association with inhibitors (for reviews see Nurse, 1994; Elledge, 1996; Nasmyth, 1996; Mironov et al., 1999).

CDKs have been detected in all eukaryotes examined, and they are encoded by a variable number of genes. In Schizosaccharomyces pombe (fission yeast), there is only one CDK, p34^Cdc2, which is homologous to p34^Cdc28 in Saccharomyces cerevisiae (budding yeast). In humans, there is a family of CDKs with molecular masses ranging from 33–40 kDa, which are related to yeast p34^Cdc2/Cdc28. Human CDK1 interacts with mitotic cyclins A and B and promotes the transition through G2 to M‐phase (called MPF, for mitosis promoting factor), while CDK2 interacts with cyclins D, E and A and functions at G1 and S‐phase. A number of plant CDK genes have been isolated (John et al., 1989; Colasanti et al., 1991; Hirt et al., 1993), and they group in families of at least two different types (Dudits et al., 1998). For example, two types of p34^Cdc2/Cdc28 sequences were found in Arabidopsis, one which is consistently correlated with competence to divide and one which is cell cycle regulated (Shaul et al., 1996). In maize, two p34^Cdc2/Cdc28‐related cDNAs were described that differ by only seven amino acids and are related to the former type of Arabidopsis p34^Cdc2/Cdc28 (Colasanti et al., 1991). One of the two mRNAs was more prevalent among the clones isolated, but the role of these genes in maize cell cycle regulation has not been determined.

The oscillation in CDK activity during the cell cycle is controlled by sophisticated mechanisms involving transcriptional and post‐transcriptional regulation (Morgan, 1997). A principal means of regulation is an alteration in cyclin abundance. Mitotic cyclins reach maximum levels at the G2/M‐phase transition, while G1 cyclins peak at G1/S phase. Oscillations in cyclin abundance involve transcriptional modulation of specific genes (Cross and Tinkelenderg, 1991; Dirick and Nasmyth, 1991), translational regulation of mRNAs (Swenson et al., 1986) and proteolytic degradation (Glotzer et al., 1991; King et al., 1996). Cyclin genes have been isolated from a number of plant species, e.g. soybean and carrot (Hata et al., 1991), Arabidopsis (Hemerly et al., 1992; Soni et al., 1995), alfalfa (Hirt et al., 1992), and maize (Renaudin et al., 1994; Sun et al., 1997), although their biological activities have not been well documented. Sequences corresponding to four different A‐ and B‐type mitotic cyclins were isolated from immature maize ears (Renaudin et al., 1994), and cDNA clones corresponding to B‐type cyclins were isolated from developing maize endosperm (Sun et al., 1997).

The activity of CDKs is also regulated by phosphorylation/de‐phosphorylation of the protein kinase component and by inhibitory proteins (Fig. 3A). Three phosphorylation sites have been documented to be involved in switching CDK activity on and off. Phosphorylation at threonine 161 of human CDK1 (Thr 160 of CDK2) activates the complex (Krek and Nigg, 1991), while phosphorylation at threonine 14 and tyrosine 15 of CDK1 and CDK2 inactivates kinase activity. Wee1, a kinase originally identified in S. pombe, phosphorylates tyrosine 15 (Parker et al., 1992). A membrane‐associated protein kinase, p68^Myt1 was described in Xenopus oocytes that phosphorylates both Thr 14 and Tyr 15 of CDK1 (Mueller et al., 1995). To ensure CDK1 is activated during mitosis, a phosphatase, Cdc25, dephosphorylates both threonine 14 and tyrosine 15 (Kugmagai and Dunphy, 1991). Wee1 and Cdc25 are also regulated by phosphorylation (Morgan, 1997).

CDK inhibitors (CKIs) control the activity of CDKs by binding tightly with them or CDK‐associated proteins (Fig. 3b). For example, p40^Sic1 in S. cerevisiae binds to p34^Cdc28 and reduces its phosphorylation of protein substrates (Mendenhall, 1993). p40^Sic1 is phosphorylated by p34^Cdc28 in a cell cycle‐dependent manner, and this may promote its degradation. Multiple CKIs have been isolated from mammalian systems, and these fall into two classes: the INK4 ankyrin‐containing proteins (p15^INK4b, p16^INK4a, p18^INK4c, and p19^INK4d) bind specifically to CDK4 and CDK6 complexes, while p21^{cip1/waf1/sdi1}, p27^kip1, and p57^kip2 interact with cyclin D‐, E‐ and A‐containing CDKs, suggesting a role at a specific phase of the cell cycle (Pines, 1994; Peter and Herskowitz, 1994; Morgan, 1997; Sherr and Roberts, 1999). Thus far, two CDK inhibitors, ICK1 (Wang et al., 1998) and ICK2 (Lui et al., 2000), have been described in Arabidopsis. These inhibitors are related to the p21 and p27 inhibitors of mammals. Transcription of ICK1 appears to be regulated by abscisic acid.

For S‐phase to occur, chromatin must be primed to initiate DNA replication at sequences called origins of replication (Blow and Laskey, 1988). The mechanism by which this occurs is not completely understood, but the following is an emerging consensus (Leatherwood, 1998; Donaldson and Blow, 1999). Origins of replication occur in different states, each of which is associated with the binding of specific proteins. The origin replication complex (ORC) consists of six subunits that appear to remain associated with origins of replication throughout the cell cycle (Rowles et al., 1996). Proteins called cdc6/cdc18 and Cdt1 bind ORCs during G1 (Malorano et al., 2000; Nishitani et al., 2000) and promote loading of MCM proteins (Donovan et al., 1997; Tanaka et al., 1997). The latter are required to recruit the replication machinery to the origins, and they are thought to provide helicase activity to the replication complex. Once the MCM complex is fully assembled, the DNA is ‘licensed’ for replication. When DNA synthesis begins, MCMs become phosphorylated, preventing the origins of replication from re‐initiating DNA synthesis until M‐phase is completed (Coleman et al., 1996). Active CDKs block the association of licensing factors with origins of replication during S‐phase and G2 (Aparicio et al., 1997; Donovan et al., 1997) and licensing can not re‐occur until there is a loss of CDK activity late in mitosis.

In the standard cell cycle, progression to S‐phase requires completion of M‐phase, but with endoreduplication this dependency is uncoupled and chromatin is re‐licensed in the absence of mitosis. In yeast, CDK activity determines the dependency of S‐phase on M‐phase (Nurse, 1994). Deletion of p56^Cdc13, the B‐type cyclin in S. pombe, causes cells to undergo repeated rounds of S‐phase, without progressing to M‐phase. As a result, the ploidy can be increased to greater than 32C (Hayles et al., 1994). Likewise, overexpression of Rum1, a CDK inhibitor, leads to increased DNA content and nuclear enlargement (Moreno and Nurse, 1994; Correa‐Bordes and Nurse, 1995). Mathematical models of cell cycle control, based on fission yeast genetics and biochemistry, predict that an oscillation of CDK activity in the absence of MPF is sufficient to induce endoreduplication (Novak and Tyson, 1997). More qualitative studies based on observations from synchronized and unsynchronized Drosophila and mammalian cells are also consistent with this model (Sauer et al., 1995; Duronio and O'Farrell, 1995; Lilly and Spradling, 1996; Sigrist and Lehner, 1997; Su and O'Farrell, 1998; MacAuley et al., 1998).

Phosphorylation of ‘pocket proteins’, e.g. retinoblastoma (pRb) (Krek et al., 1995; Weinberg, 1996), is also likely to play a role in endoreduplication. In mammals, progression through G1/S‐phase requires inactivation of Rb, which is believed to mediate growth inhibition during the G0 and G1 phases of the cell cycle and functions as the gatekeeper for S‐phase (Riley et al., 1994). Pocket proteins are phosphorylated during G1 by cyclin D‐ and E‐mediated CDK activity (Matsushime et al., 1992). Recent evidence suggests that cyclin D‐mediated phosphorylation is not sufficient to release the cell from G1, but it may be required for the G0/G1 transition (Ezhevsky et al., 1997). Maize appears to contain multiple pocket proteins, and the conservation of their interactions with transcriptional regulators suggests the cell cycle regulatory functions they control are also conserved (Grafi et al., 1996; Xie et al., 1996; Ach et al., 1997; Huntley et al., 1998).

Recent studies in animal and plant systems suggest that endoreduplication requires nothing more elaborate than a loss of M‐phase CDK (MPF) activity and oscillations in the activity of S‐phase CDKs (Fig. 4). This permits origins of replication to become re‐licensed for an additional cycle of replication, while preventing the induction of M‐phase. Up‐regulation of S‐phase CDKs and down‐regulation of MPF activity has been observed in all endoreduplicating systems examined in detail, for example, Drosophila embryos (Sauer et al., 1995), rat placental trophoblasts (MacAuley et al., 1998) maize endosperm (Grafi and Larkins, 1995), and tomato fruit (Joubès et al., 1999). The cycling of S‐phase CDK activity could be either auto‐regulatory or negatively feedback‐regulated. As described below, there are increasing data that support this view of the series of events that lead to DNA endoreduplication in plant and animal systems.

MPF inactivation in maize endosperm is accompanied by a reduction in the level and activity of cyclin B (Sun et al., 1997). This occurs by transcriptional down‐regulation and perhaps alternative splicing of cyclin B RNA; it also likely involves destruction of cyclin B protein, but this has not been documented. It is unclear which of these events might be the inductive signal for MPF inactivation, but it is possible that a CKI and/or the Wee1 kinase may be involved. Maize endosperm accumulates a CDK inhibitor coincident with the onset of endoreduplication, although the nature of the CKI remains to be clarified (Grafi and Larkins, 1995). CDK inhibition might also result from phosphorylation via the Wee1 kinase. Maize Wee1 (ZmWee1) mRNA accumulates during the period of endoreduplication (Sun et al., 1999); the enzyme could phosphorylate the M‐phase CDK, the S‐phase CDK, or both. S‐phase induction could be controlled by competition between an up‐regulated S‐phase kinase and its inhibitors, perhaps Wee1 and/or a CKI. A gap phase, during which CDK activity is low, is promoted and maintained by pocket proteins, e.g. ZmRb1, which should operate as negative regulators of genes promoting DNA replication. Phosphorylation of ZmRb1 appears to increase during endoreduplication (Grafi et al., 1996). The period of low CDK activity would allow chromatin licensing, and the gap‐phase would end when sufficient S‐phase CDK accumulates and escapes inhibition by Wee1 and the CKI.

The most detailed description to date for how endoreduplication occurs in mammals comes from studies of trophoblast giant cells in rodent placenta. Giant cells undergo endoreduplication during terminal differentiation, reaching ploidies of more than 1000C (Varmuza et al., 1988). During the transition from a mitotic cell cycle to endoreduplication, trophoblast cells reduce transcription of cyclin B, and eventually the protein, as well as MPF activity, is not detectable (MacAuley et al., 1998). Coincidentally, there is an increase in cyclin A‐ and cyclin E‐associated CDKs, the levels of which oscillate prior to and during S‐phase. Generally, endoreduplication and the mitotic cell cycle involve the development of cyclin E/CDK activity (Sherr, 1994; Edgar and Lehner, 1996). Modulation of G1‐ and S‐phase CDK activity in endoreduplicating trophoblast cells appears to involve the cyclic synthesis and destruction of p57^Kip2. p57^Kip2 transcripts are not present in trophoblast cells until the onset of endoreduplication, and the level of the protein oscillates with each endocycle (Hattori et al., 2000). While a direct association of p57^Kip2 with a G1/S‐phase CDK has not been demonstrated, ectopic expression of p57^Kip2 promoted giant cell differentiation in precursor cells and expression of a stable mutant form of p57^Kip2 blocked endoreduplication. The level of p27^Kip1 is regulated by the ubiquitin‐proteosome pathway via CDK‐mediated phosphorylation of the CKI, and a similar mechanism may control the level of p57^Kip2. Consequently, it was postulated that p57^Kip2 regulates the level of cyclin E/CDK activity, creating two distinct gap phases: a G1‐like period when p57^Kip2 levels are low and there is high CDK activity, and a G2‐like period when p57^Kip2 levels are high and CDK activity drops (Hattori et al., 2000). This would allow cycles of S‐phase and chromosome licensing.

Other evidence implicating CKIs as important regulators of the endoreduplication cell cycle comes from studies in which p21^Cip1/Waf1 was expressed in transfected mammalian cells (Niculescu III et al., 1998). If the cells possessed a functional pRb mechanism, p21^Cip1/Waf1 inhibited cell cycle progression in G1. However, inactivation or loss of pRb activity was associated with G2 arrest and cycles of endoreduplication. The state of pRb in trophoblast giant cells was not examined in the previously mentioned studies, but these cells progress to the endoreduplication cell cycle from G2 (MacAuley et al., 1998; Hattori et al., 2000), a period during which pRb is hyperphosphorylated. Inactivation of the Rb pathway during endoreduplication in maize endosperm would be consistent with the observed increase in phosphorylation of ZmRb1 (Grafi et al., 1996).

Fig. 3.

Model depicting the regulation of cyclin‐dependent kinase (CDK). (A) Cyclin binding and phosphorylation at T‐161 by Cdc2‐activating kinase (CAK) are required for CDK activation; phosphorylation of the T‐14 or Y‐15 residues by the inhibitory Wee1 kinase abolishes ATP binding. (B) Association with a CDK inhibitor (CKI) blocks substrate recognition.

Fig. 4.

CDK oscillations drive S‐phase induction.

Future prospects

Although research on cell cycle regulation in plants lags behind that in yeast and animals, rapid progress is being made identifying genes responsible for key components of plant cell cycle regulation (Mironov et al., 1999). This is largely a consequence of genomics projects, which have led to the identification of the genes encoding these generally rare transcripts. The early results of cell cycle regulation studies in flowering plants suggest control of these processes is fundamentally very similar to that in mammals (Murray, 1997; Mironov et al., 1999; Sun et al., 1999). Since there are efficient methods to create transgenic plants that over‐ or under‐express these genes, over the next few years a great deal can be expected to be learned about the mechanisms responsible for cell cycle control and their importance in plant development. In some respects, this research will be easier with transgenic plants than animals; it is certainly less expensive.

Progress in understanding the function of endoreduplication in plants has been limited by the ability to regulate this process genetically or biochemically in vivo. Consequently, many of the data describing endoreduplication and its effects on plant growth and development are primarily correlative. However, this situation is changing rapidly. Several recent studies have provided insight regarding conserved changes in plant and animal cell cycle regulation that accompany the process of endoreduplication (Niculescu et al., 1998; MacAuley et al., 1998; Hattori et al., 2000). Whether there is a standard mechanism by which endoreduplication occurs in plants and animals remains to be seen; however, preliminary findings suggest this may be the case.

Our research is supported by a grant from the Department of Energy (DE‐96ER20242) to BAL. RAD and CMC are supported by graduate fellowships from the Conselho Nacional de Desenvolvimento Cientifico e Tecnologico‐Brazil.