The plant cell cycle − 15 years on


Author for correspondence: Dennis Francis Tel: +44 29 20875086 Fax: +44 29 20874305 Email:



  • Summary 261

  • I. Introduction 262
  • II. Plant CDKs and cyclins 263
  • III. G0 → G1 → S – the dawn awakening 263
  • IV. S-phase – life gets replicated 264
  • V. G2/M – let's dance 266
  • VI. Mitosis – strictly come dancing 268
  • VII. Cytokinesis – partitioning life 269
  • VIII. M/G1 – party over 270
  • IX. Endocycles – a curious life of their own 270
  • X. Cell size 270
  • XI. Root branching 271
  • XII. Conclusions 271
  • Acknowledgements 273

  • References 273


The basic components of the plant cell cycle are G1 (postmitotic interphase), S-phase (DNA synthesis phase), G2 (premitotic interphase) and mitosis/cytokinesis. Proliferating cells are phosphoregulated by cyclin-dependent protein kinases (CDKs). Plant D-type cyclins are sensors of the G0 to G1 transition, and are also important for G2/M. At G1/S, the S-phase transcription factor, ‘E2F’, is released from inhibitory retinoblastoma protein. Negative regulation of G1 events is through KRPs (Kip-related proteins). Plant S-phase genes are similar to animal ones, but timing of expression can be different (e.g. CDC6 at the start of S-phase) and functional evidence is limited. At G2/M, A-type and the unique B-type CDKs when bound to A, B and D cyclins, drive cells into division; they are negatively regulated by ICK1/2 and perhaps also by WEE1 kinase. In Arabidopsis, a putative CDC25 lacks a regulatory domain. Mitosis depends on correct temporal activity of CDKs, Aurora kinases and anaphase promotion complex; CDK-cyclin B activity beyond metaphase is catastrophic. Endoreduplication (re-replication of DNA in the absence of mitosis) is characterized by E2F expression and down-regulation of mitotic cyclins. Some cell size data support, whilst others negate, the idea of cell size having an impact on development.


The great Vaudeville entertainer, George Burns, when asked in his late 70 s why he was chosen for the Warner Brothers film The Sunshine Boys (1975), said: ‘You know, my only other Warner Brothers movie was in 1934 ... I must have made a good impression!’ I wrote my first Tansley Review in 1991. Francis (1992) had much to say about yeast/animal cell cycle genes but not that much about plant cell cycle genes. Now, many essential plant cell cycle genes have been cloned, but only for a few is there functional evidence of their role in planta. There are approx. 11 000 genes regulated in the cell cycle (Menges et al., 2002, 2003) and approx. 80 genes that regulate the cell cycle (Menges et al., 2005). Also, a microarray analysis categorized into constitutively expressed, proliferative and inhibitory genes resolved a subset of 131 proliferation genes (Beemster et al., 2005).

Cell cycle nomenclature differentiates between fission yeast and budding yeast cell cycle genes/proteins (Table 1). A group of plant cell cyclists (Renaudin et al., 1996) organized an even newer nomenclature for plant cell cycle genes: the first three letters of the genus and the first two letters of the species followed by a semicolon(s) and then, in turn, followed by the family and family member. So the Arabidopsis homologue to cdc2 is abbreviated to Arath;CDKA;1 (Table 1).

Table 1.  Nomenclature for cell cycle genes and encoded proteins
 Fission yeastBudding yeastArabidopsis

Recently, the plant cell cycle has been reviewed in relation to plant growth regulators (del Pozo et al., 2005), development (de Jager et al., 2005) and checkpoints (Francis, 2003); these interfaces are not explored in detail here.

1. Plant cell cycle overture nuts and bolts

The cell cycle is a temporal regulator of cell division in proliferative cells (Fig. 1). It comprises mitosis (M), cytokinesis, postmitotic interphase (G1), DNA synthetic phase (S) and postsynthetic interphase (G2). During the journey the proliferative cell grows and is made ready to divide.

Figure 1.

The plant cell cycle. Top panel: BY-2 cells transformed with fission yeast cdc25 and stained with Hoechst: prophase (P), anaphase (A), cytokinesis (CK). Bar, 100 µm. Middle panel: mitosis (M), postmitotic interphase (G1), DNA synthetic phase (S) and premitotic interphase (G2). Bottom panel: a DNA fibre-autoradiograph of rye DNA. Root tips were pulsed with high- followed by low-specific-activity methyl 3H thymidine. Arrowheads, replication initiation points; and on either side replication forks are merging (arrows). Bar, 20 µm. Adapted from Francis & Bennett (1982).

Plant cells arrest in G1 or G2 if starved of sucrose (Van't Hof, 1966) or if treated with puromycin in the presence of sucrose (Webster & Van't Hof, 1969). These data supported the G1/S and G2/M ‘principal control point’ hypothesis and identified protein synthesis as critical at both transitions (Van't Hof, 1973). Phosphoregulation of proteins is a consistent biochemical feature of the G1/S and G2/M transitions, and cyclin-dependent kinases (CDKs) are the central players, but each is dependent on a noncatalytic partner, a cyclin. In fission yeast, cdc2 drives both the G1/S and G2/M transitions.

2. Early discoveries

The first reported plant homologue to cdc2 was in Chlamydomonas and wheat (John et al., 1989), and then in Pisum sativum (Feiler & Jacobs, 1990), Arabidopsis thaliana (Ferreira et al., 1991) and Medicago sativum (Magyar et al., 1993). A plant cyclin was first identified by Hata et al. (1991), followed by a flurry of papers reporting more and more plant cyclins (Day et al., 1996).

II. Plant CDKs and cyclins

Cyclin-dependent protein kinases are in seven classes: A–G. Of these, As, Bs, Ds and Fs are known to be linked to regulatory mechanisms of the cell cycle (for a review, see Umeda et al., 2005) (Table 2).

Table 2. Arabidopsis cyclin-dependent protein kinases (CDKs) and cyclins, where they function in the cell cycle and, where appropriate, the putative role they play outside of the cell cycle
 Family membersCell cycle phase or putative function outside of cell cycle (italics if latter)
A;1G1/S and G2
C;1Regulation of transcription
E Regulates RNA polymerase II
G ?
A1;1G1/S (G2/M)
A1;2G1/S (G2/M)
A2;1G1/S (G2/M)
A2;2G1/S (G2/M)
A2;3G1/S (G2/M)
A2;4G1/S (G2/M)
A3;1G1/S (G2/M)
A3;2G1/S (G2/M)
A3;3G1/S (G2/M)
A3;4G1/S (G2/M)
B1;1G2 or G2/M
B1;2G2 or G2/M
B1;3G2 or G2/M
B1;4G2 or G2/M
B2;1G2 or G2/M
B2;2G2 or G2/M
B2;3G2 or G2/M
B2;4G2 or G2/M
B3;1G2 or G2/M

Cyclins were first identified in sea urchins as proteins that increase in concentration during interphase, peak at G2/M and suddenly disappear before telophase (Evans et al., 1983), but not all oscillate (Dewitte & Murray, 2003). Cyclin binding with CDKs is normally a prerequisite for CDK activity. Most cyclins have a sequence of 100 amino acids, the cyclin box, which is required for binding to CDKs (De Veylder et al., 1997), and a destruction box, which is susceptible to ubiquitination that leads to rapid proteolytic destruction. G1 cyclins have proline, glutamate/aspartate, serine/threonine-rich sequences (PEST) in the C-terminal, which are markers of short-lived proteins.

There are 13 classes of cyclins in animals (A–L and T) (Pines, 1995) but only seven in plants: A, B, C, D, H, P and (T) within which approx. 60 plant cyclin genes belong. In the cell cycle context, we know most about the A, B and D classes.

III. G0 ‡ G1 ‡ S – the dawn awakening

Animal stem cells arrested in G0 of the cell cycle have potential for proliferation, are capable of self-maintenance and can regenerate tissues (Loeffler & Potten, 1997). In plants, the quiescent centre of the root apical meristem (RAM) is a reservoir of stem cells that can be stimulated to divide and repair damaged root caps (Clowes, 1958). More generally, there are G0 arrested cells in plant tissues, but not in all cases do they conform to the animal stem cell definition, and perhaps they shouldn't.

The first plant D-type cyclins were discovered in Arabidopsis: ArathCYC D1;1, D2;1 and D3;1 (Soni et al., 1995), joined later by D4 and D5 types (Riou-Khamlichi et al., 2000). In Arabidopsis, plant D cyclins are uniquely responsive to both exogenous cytokinins and sucrose (Riou-Khamlichi et al., 1999).

In G0, the retinoblastoma protein (Rb) is a negative regulator of the transcription factor, E2F, first identified as a DNA-binding protein essential for E1A-dependent activation of the adenovirus E2 promoter (Nevins, 1992; Dyson, 1998). A plant Rb was hypothesized firstly, because plant D-type cyclins exhibit an LXCXE motif that is critical for Rb-binding in humans (Soni et al., 1995). Secondly, the same motif was identified in Gemini viral proteins. Gemini viruses depend on their host proteins to replicate their single-stranded DNA by way of a double-stranded DNA intermediate (Gutierrez, 1998; Settlage et al., 2001). An Rb homologue was discovered through a maize EST used to isolate cDNAs encoding Rb-like proteins (Shen et al., 1994; Grafi et al., 1996). Expression studies revealed Zeama;Rb (Xie et al., 1996), two Rb-related genes, Zeama;RBR1 and Zeama;ZmRBR2 (Ach et al., 1997; Huntley et al., 1998), and more recently, Zeama; RBR3 (Sabelli et al., 2005). Arabidopsis only has a single Rb (deWitte & Murray, 2003).

In the past five years, many papers have appeared on plant E2F. First identified in wheat (Ramirez-Parra et al., 1999), there are three E2F-like genes (a, b, c) in Arabidopsis and two that encode dimerization proteins (DPa and DPb) (Kosugi & Ohashi, 2002; He et al., 2004). Neither E2Fs nor DPs have much affinity with DNA (at least in vitro), although paradoxically E2F was discovered as a DNA binding protein (discussed earlier). However, yeast two-hybrid assays indicated that E2Fa and E2Fb can interact with either DPa, whilst E2Fc interacts with DPb, but all three E2Fs can bind with either (Fig. 2). DPa promotes nuclear localization of E2Fa and E2Fb. Similarly, DPb acts on E2Fc (Kosugi & Ohashi, 2002). E2F a–c and DP genes are strongly expressed at S-phase. When E2Fa and DPa were expressed ectopically in leaves, differentiated cells re-entered S-phase (Rossignol et al., 2002). In tobacco, Rb binds and suppresses E2F but cyclin D is part of a mechanism that releases E2F (Uemukai et al., 2005). G1/S is triggered by a phosphorylation cascade that culminates in the release of E2Fs (Fig. 2).

Figure 2.

The plant G0/G1 model. In G0, the E2F transcription factors and dimerization proteins are inactive through binding with the retinoblastoma protein (Rb). At G0/1: CDKD – cyclin D phosphorylates CDKA on T 160/167, enabling CycD3;1 to bind. A cytokinin-induced transcellular induction culminates in CyCD3;1, which activates CDKA that, in turn, phosphorylates Rb1 and 2 (small circles). This releases the E2F complex, which then induces the expression of S-phase-specific genes (e.g. CDK6), driving cells into S-phase. Rectangles, inactive or inactivated protein; ellipses, activated protein; small circles, phosphorylations.

CycD3oe in leaf explants resulted in green calli on medium deprived of cytokinin, whereas wild-type cells were cytokinin-dependent (Riou-Khamlichi et al., 1999). Hence cytokinins regulate G1/S, at least partially, by inducing CycD-type transcription; CycD3;1 is rate-limiting for G1/S (Menges et al., 2006) (Fig. 2).

In mammals, two different suppressors of G1/S fall into two families: INK4 and Kip/Cip families, p21Cip1, p27Kip1 and p57Kip2 (Toyoshima & Hunter, 1994; Lee et al., 1995). Arabidopsis homologues show sequence identity in their C-terminal domains to mammalian p27Kip1 and were named Kip-related proteins (KRPs) 1–7 (De Veylder et al., 2001a; Verkest et al., 2005). KRPs exert a strong negative regulation on G1/S in unfavourable conditions. KRP 1 and 2 correspond to the earlier discovered interactors with CDKs, ICK1 and ICK2 that suppress CDK activity in G2 (Wang et al., 1997). They can inhibit both CYCD2-/CDKA and CYCD2/CDKB (Nakai et al., 2006).

Genes expressed in G1/S have sites on their promoters that can bind consensus E2F-sites (Mariconti et al., 2002). For example, an Arabidopsis E2F binds to the promoter of the S-phase gene, Arath;CDC6 (de Jager et al., 2001), and to the promoter of one of the minichromosome maintenance genes (MCM) which are essential for DNA replication (Stevens et al., 2002). Also, Arabidopsis, RNR1b, which is maximally expressed in mid S-phase, has an E2F binding site in its promoter (Chaboute et al., 2002). However, somewhat staggeringly, nearly 6000 Arabidopsis genes and approx. 10 000 sites within their promoters are potential E2F-binding sites, of which about a third are cell cycle genes (Ramirez-Parra et al., 2003).

Cytokinins and auxins regulate both the G1/S and G2/M transitions (DeWitte & Murray, 2003). For example, PROPORZ1, is a transcriptional adaptor protein involved in auxin and cytokinin signalling. In prz1 mutants grown on medium supplemented with auxin and cytokinin, E2Fc, CDKB;1 and KRPs were all down-regulated (Sieberer et al., 2003). These types of data illustrate rather well that the balance between cytokinins with auxins are key to both the cell cycle and development. In other words, don't bank on one of these regulators per se to provide complete answers about cell cycle regulation.

IV. S-phase – life gets replicated

Currently, S-phase is better understood in budding yeast, but what follows is by no means encyclopaedic. Readers are encouraged to consult Yabuuchi et al. (2006), who report an order in which the cofactors Sld3, GINS and Cdc45 regulate replication origins in fission yeast; they also provide an-up-to-date introduction to DNA replication in yeasts.

All eukaryote cells replicate their nuclear DNA in a semiconserved mechanism, whereby the parent molecule is unwound and each DNA strand becomes the template for nascent DNA synthesis. Structurally, the cell enters S-phase with single arm chromosomes but exits with double arm chromosomes.

In S-phase, DNA unwinding is transient and occurs discretely at initiation points/replication origins. An initiation site and the span of DNA replicated by its forks constitute a replicon. Initiation points are spaced at regular intervals throughout each chromosome, but even within one stretch of DNA, initiation points can fire asynchronously (Blow et al., 2001). Replication forks initiate at replication origins where DNA is transiently unwound. Factors determining the position of replication origins on chromosomal DNA are only poorly understood (Machida et al., 2005).

What defines a replication origin is as unclear in plants as it is in animals. For example, in Xenopus embryos, DNA replication can begin from anywhere, regardless of sequence, in a partially periodic pattern (Blow et al., 2001). However, in budding yeast, initiation points occur at well-defined DNA sequences (Diffley & Labib, 2002).

Clusters of adjacent origins typically initiate together, but different clusters initiate at different times in S-phase. In Arabidopsis, there are 30 000 replicons that function in two overlapping replication families (Van't Hof et al., 1978). However, little is known about how plant replicons are regulated, although we know something of their plasticity. For example, replication origins in diploid rye (2n = 2x = 14) are c. 60 kb apart (Francis & Bennett, 1982), whilst in bread wheat (2n = 6x = 42) they are at 15 kb intervals (Francis et al., 1985a). In the allohexaploid triticale (2n rye × 4n wheat) initiation sites were detected every 15 kb (Kidd et al., 1992). Thus replicons spaced at 60 kb intervals in rye chromosomes in a rye background are spaced at the same frequency as those of wheat chromosomes in the triticale background (15 kb intervals). We regarded initiation sites at 60 kb intervals as strong sites of initiation, and those separated by 15 kb as weak sites of initiation.

There is also considerable plasticity of plant replication origins. Secondary initiation points could be induced by exposing cells to UVB and a DNA cross-linker, psoralen (Francis et al., 1985b). Conversely, in the garden pea, trigonelline treatment rendered some initiation sites dormant (Mazzuca et al., 2000). How such secondary initiation sites are switched on/off is unknown.

In Saccharomyces cerevisiae, replication initiation sites were identified as autonomous replicating sequences (ARSs), which can initiate DNA replication in plasmids that lack a replication origin. ARS1 comprises four functionally essential domains: A, B1, B2 and B3. The core sequence of A is A/TTTTATG/ATTTA/T and provides a binding site for the origin recognition complex (ORC) (Bell & Stillman, 1992; Cocker et al., 1996). It remains a mystery why an ARS assay has not proved useful in identifying replication origins in other eukaryotes.

Origin recognition complexes (ORCs) can comprise up to six polypetides that bind strongly to the replication origin (Bell & Stillman, 1992; Lygerou & Nurse, 1999) and remain in place from cell cycle to cell cycle (Fig. 3). However, persistence of ORCs during mitosis has been questioned (Gilbert, 1998). In Drosophila, the ORC seems to show a preference for a 440 bp tract within a specific gene (Austin et al., 1999).

Figure 3.

The G1/S model. The CDC6 and CTD1 enable the origin recognition complex (ORC) to become receptive to minichromosome maintenance (MCM) proteins. The MCMs form a ring that is opened by ORC-/CDC6-CDT1. MCMs are then clamped around the DNA (licensing). At G1/S, a wave of phosphorylation by DBF-CDC7 dislodges the ORC, exposing the replication initiation site to the replisome complex, the DNA molecule unwinds and replication begins. Cyclin-dependent protein kinases (CDKs) also operate in the phosphorylation cascade triggered by DBF-CDC7. However, the CDK substrates are not completely resolved. Other putative cofactor proteins are omitted from this model. Rectangles, inactive or inactivated protein; ellipses, activated protein; small circles, phosphorylations; tick marks, cloned plant genes.

In budding yeast, CDC6p primes the prereplicative complex (PRC) at replication origins (Cocker et al., 1996). CDC6p is a member of the AAA+ family (ATPases associated with a variety of cellular activities). In a mutant cdc6p, loading of the minichromosome maintenance (MCM) complex was perturbed. In fission yeast, Cdt1 can bind both ORC and MCM2-7, suggesting that it may function in bringing MCM2-7 to origin DNA (reviewed by Diffley, 2004).

At S-phase, CDC7-DBF4 kinase (dbf mutants exhibit a dumbbell former phenotype; Jackson et al., 1993) is recruited to origins and phosphorylates the MCM2-7 complex. As replication forks initiate at origins, CDC45 and the GINS complex are recruited to the MCM2-7, which probably forms the major helicase activity that unwinds DNA ahead of the replication fork (Moyer et al., 2006). Another protein SLD3 is an important cofactor for CDC45 recruitment (Yabuuchi et al., 2006). RPA proteins bind to new unwound single-stranded DNA exposed by origin DNA unwinding, and this in turns allows the recruitment of DNA polymerase alpha – primase. DNA primase primes the exposed single-stranded DNA with a short RNA molecule (11 bp) that has an exposed 3′OH group with which replicative DNA polymerases can bind. This then allows assembly of the replisome, a complex of priming and replicative enzymes mentioned earlier alongside DNA ligases, proliferative cell nuclear antigen (PCNA) and several DNA polymerases (Matsumoto et al., 1994).

In Pisum sativum, polymerase α has primase activity and is partly homologous to primases from other systems (Bryant et al., 1992). Arabidopsis has ORCs (Gavin et al., 1995), CDT1 (Lin et al., 1999), CDC6 (Ramos et al., 2001), MCMs (Springer et al., 1995; Sabelli et al., 1996) and CDC45 (Stevens et al., 2004). In Pisum sativum, Pissa;MCM3 shows c. 50% homology with human MCM3 (Bryant et al., 2001).

Arath;CDC6 is expressed maximally in early S-phase, and its promoter contains an E2F consensus site (Drambruskas et al., 2003). However, note the difference in the timing of Arath;CDC6 expression in early S compared with G1 in budding yeast.

ORC, CDC6/Cdc18 and Cdt1 promote the loading of MCM2-7 complexes onto replication origins during late mitosis and G1, thereby ‘licensing’ the origin for use in the subsequent S-phase (Blow & Dutta, 2005). As initiation occurs at each origin, the MCM2-7 moves ahead of the replication fork, thereby ensuring that it fires only once in each cell cycle (Blow et al., 2001; Machida et al., 2005). In plants, CDC6oe may counter licensing factors in inducing endoreduplication (nuclear DNA replication in the absence of mitosis; Castellano et al., 2001; see Section IX).

V.G2/M – let's dance

In higher plants that are two classes of CDK, A and B, which drive cells into mitosis, B-type CDKs are unique to plants and help regulate the G2/M transition (Table 2). CDKB genes fall into two subgroups, CDKB1 (PPTALRE) and CDKB2 (P(P/S)TTLRE) (Joubes et al., 2000; Dewitte & Murray, 2003). CDKB genes are unable to complement yeast cdc2 mutants. In Arabidopsis there is a peak of CDKB1;1 kinase activity at G2/M, whereas CDKA1 activity peaks at G1/S and G2/M (Joubes et al., 2000). In tobacco BY-2 cells, CDKA exhibits kinase activity from S-phase to G2/M whilst CDKB kinase activity peaks in mid-to-late G2 (Porceddu et al., 2001; Sorrell et al., 2001). In fission yeast, Suc 1 (suppressor of Cdc2) over expression could destabilise the Cdc2 complex but in WT it is a catalytic sub unit of Cdc2 (Brizuela et al., 1987). More recently it was renamed, CKS1 (catalytic subunit of CDK) that can interact with CDKs that, in animals, degrade the G1 inhibitor, p27 (Ganoth et al., 2001). In Arabidopsis the exact role of CKS1 is unknown. On the one hand, Arath;CKS1oe inhibits growth (De Veylder et al., 2001b), while on the other, it does not suppress CDKs. Indeed, it may be promotive in directing the CDK towards its substrate (Fig. 4) (De Veylder et al., 1997). Arath;CKS1, a protein used as bait in a two-hybrid screen, interacted with Arath;CDKB1;1, Arath;CDKB1;2 and Arath;CDKB2;1, Arath;CDKB2;2 (see Section II.2 and De Veylder et al., 2001a). Their expression profiles are mainly confined to meristems.

Figure 4.

The tobacco BY-2 G2/M model. In G2, ICK1 and ICK2 inhibit cyclin-dependent protein kinase (CDK) activity. WEE1 kinase negatively regulates CDKA/B through phosphorylation of threonine14 and tyrosine15 residues. CDK4-cyclin F is a CDK-activating kinase that phosphorylates CDKA/B at or near threonine 160 and this is followed by binding of cyclin A/B and possibly CycD4;1. CKS1 may function to direct CDKA/B to their substrates. At G2/M, a cytokinin transduction signal culminates in a putative CDC25, which dephosphorylates CDK at the T 14 and Y 15 residues. CDKA/B is now fully activated and, through phosphorylation, deactivates WEE1 and activates CDC25, driving cells into division. Rectangles, inactive or inactivated protein; ellipses, activated protein; small closed circles, phosphorylations.

ICK1oe suppresses CDK activity (Lui et al., 2000; Zhou et al., 2002; Weinl et al., 2005). ICK1/2 can be induced by abscisic acid (ABA), a plant growth regulator that is often overproduced in stress situations (Wang et al., 1998). Thus, an ABA-mediated signal transduction chain culminating in ICK expression would be a neat way of preventing cells from entering mitosis in unfavourable conditions. ABA-signalling might also culminate with KRPs (homologues to ICKs) that are important for G0/G1/S-phase transitions (De Veylder et al., 2002). ICK1/KRPs can also block cells at G2/M but allow S-phase, which in turn can lead to endoreduplication. However, endoreduplicating cells could re-enter mitosis. ICK1oe in Arabidopsis resulted in dwarf plants comprising large cells (Wang et al., 2000). The tobacco Nicta;KIS1a, a homologue to Arath;ICK1, inhibits CDK/cyclin activity in vitro. Nicta;KIS1aoe in Arabidopsis also resulted in small organs with large cells. (Jasinski et al., 2002).

Yeast 2 hybrids identified a D-type cyclin, CYCD4, that binds to CDKB2;1 (Kono et al., 2003). Whilst some B-types are expressed from G2 to M-phase (see Section II.2), CYCD4;1 is expressed throughout the cell cycle. B-type CDKB and CYCD4;1 may form an active kinase complex (Sorrell et al., 1999). At G2/M, mitogenic complexes comprise a catalytic CDKA-cyclinA, CDKB-B cyclin or CDK-cyclin D4, or possibly all three (Sorrell et al., 2001; Kono et al., 2003). I assign a roving role for plant D-type cyclins in the cell cycle as part of the plant's armour that enables flexibility as environmental conditions fluctuate.

Another negative regulator is WEE1 kinase. In fission yeast, it phosphorylates Cdc2 on threonine14 and tyrosine15 (Den Haese et al., 1995). The wee1 homologue in budding yeast (SWE1) acts only on threonine19 of CDC28 (Booher et al., 1993; Sia et al., 1996). In fission yeast, Cdc2 kinase regulation is also imposed by Mik1 (mitotic inhibitory kinase) and, in animals, by WEE1 kinase (tyrosine15) and MYT1 (membrane-bound tyrosine threonine kinase) kinase that phosphorylates both threonine14 and tyrosine15 (Mueller et al., 1995). Neither mik1 nor MYT1 have been identified in Arabidopsis. Together with ICK1/2, WEE1 is the only protein kinase to be a putative negative regulator of A- and B-type CDKs. Plant WEE1 was first identified in a monocot (Zea mays; Sun et al., 1999), while the first full-length dicot WEE1 was cloned in Arabidopsis (Sorrell et al., 2002).

In Arabidopsis, WEE1 kinase can phosphorylate CDKA;1 at tyr15, and CDKD;1, ;2 and ;3 at tyrosine 23/24 in vitro ( Shimotohno et al., 2006) and, most recently, CDKB types (M. Umeda, pers. comm.). These data are in accordance with a WEE1 kinase regulating CDKA and Bs at G2/M and regulating a plant CAK kinase (encoded by CDKDs (Shimotohno et al., 2006). However, in Arabidopsis, a lack of distinctive mutant phenotype in WEE1 knockouts, but hypersensitivity of these KOs in hydroxyurea-treated cells, led to the conclusion that WEE1's major role is in the DNA repair checkpoint in plants exposed to stress (Boudolf et al., 2006; De Schutter et al., 2007). A very recent model of G2/M features CDKB phosphorylating ICK2/KRP2, which then dissociates from CDKA, enabling the latter, which is presumed to be dephosphorylated on threonine14 and tyrosineY15, to drive cells into mitosis (Boudolf et al., 2006). The model now requires a demonstration that ICK2 is the natural substrate for CDKB; seemingly both CDKA and CDKB can phosphorylate ICK2/KRP2 to a similar extent in vitro (Verkest et al., 2005a).

Arath;WEE1 expression is confined to proliferative tissues (Sorrell et al., 2002), as is CDKB2;2 (Porceddu et al., 2001). In tobacco BY-2 cells, Nicta;WEE1 transcription peaks in mid-S-phase (Gonzalez et al., 2004). Nicta;CDCKB is only expressed in G2, resulting in a peak of kinase activity in late G2, whereas A-type CDK expression and kinase activity are comparatively stable from G1/S through to mitosis (Sorrell et al., 2001; Orchard et al., 2005). Interestingly, in proliferative cells of tobacco, plants transformed with a dominant negative version of CDKB exhibited a prolonged G2 (Porceddu et al., 2001). Given their confinement to proliferative cells there might be a differential inhibitory effect of WEE1 kinase on CDKB compared with CDKA-types. G2/M may be regulated rather differently in BY-2 cells compared with Arabidopsis.

CDC25 is the final part of the G2/M jigsaw puzzle. It dephosphorylates CDKs at G2/M and the fully active kinase drives cells into mitosis. Identification of a plant homologue to cdc25 has been a long and winding road that culminated with the first plant CDC25 in the alga, Ostreococcus tauri; this full-length phosphatase rescued a fission yeast ts cdc25 mutant (Khadaroo et al., 2004). In Arabidopsis, a small CDC25 was identified but it lacks a regulatory domain. Its encoded protein can dephosphorylate a protein analogue in vitro (Landrieu et al., 2004) and it can induce a short cell length when overexpressed in fission yeast (Sorrell et al., 2005). However, for at least two reasons, the jury is still out on a higher plant CDC25. Firstly, it lacks the regulatory domain common to all other cdc25s. Secondly, it is expressed in low quantities throughout Arabidopsis, mirroring weak expression patterns for Arath;CDKA1;1 outside of proliferative regions (Sorrell et al., 2005).

Arath;CDC25 is homologous to Arabidopsis arsenate reductase (ACR2) (Bleeker et al., 2006) and ACR2 knockdowns make Arabidopsis hypersensitive to arsenate (Dhankher et al., 2006). Arath;CDC25 is also closely related to the fission yeast phosphatase gene stp1, which rescues the cdc25−22 (Mondesert et al., 1994). pyp3 can also dephosphorylate fission yeast Cdc2 (Millar et al., 1992) but pyp3 has not been detected in Arabidopsis.

Because CDC25 and WEE1 knockouts grow more or less normally, this suggests functional redundancy with other protein kinases and phosphatases able to phosphoregulate plant CDKs at the G2/M transition. Importantly, other phosphatases can regulate cyclinB-cdc2. In alfalfa, phosphatases PP1 and PP2A form an activating complex that dephosphorylates the mitotic F-type CDKs (Meszaros et al., 1997, 2000).

In Nicotiana plumbaginifolia, cells depleted of benzyl adenine arrested in G2, but replenishment with this cytokinin restored mitotic competence. Either cytokinin treatment or Schizosaccharomyces pombe (Sp) cdc25 could dephosphorylate plant CDK (Zhang et al., 1996, 2005). Regarding other hormonal controls at G2/M, auxin could induce a plant like cdc2 but cytokinin was necessary for its activation (John et al., 1993). It was later hypothesized that a cytokinin signal transduction chain culminates in CDC25 (Zhang et al., 2005).

The John hypothesis gained further credence when Spcdc25oe in tobacco BY-2 cells induced premature CDKB- but not CDKA- activity, a shortened G2 phase and premature division at a small cell size. Wild-type cells were blocked by lovastatin (LVS), an inhibitor of cytokinin synthesis (Redig et al., 1996; Laureys et al., 1998) but Spcdc25oe cells bypass this block because they are cytokinin-independent (Orchard et al., 2005).

VI. Mitosis – strictly come dancing

Mitosis features the spectacular separation of chromatids into two new cells with exactly the same number of chromosomes, although, by then, half the amount of nuclear DNA. Maximum condensation is at metaphase, where each chromosome is ‘joined’ to spindle proteins that will deliver chromatids to opposite ends of the cell. In prophase, the two chromatids of each chromosome are glued together by cohesin (Fig. 5), which is synthesized and attached to each chromosome (chromatid) in the telophase of the previous mitosis (Nasmyth, 2001). A physical link between spindles and centromeric DNA is provided by the kinetochore, which, via its inner face, slots into centromeric DNA and, on the outer, connects to mitotic spindles (Connelly & Hieter, 1996). In budding yeast, DAM1P is also a binding protein that can regulate kinetochore-spindles (Cheeseman et al., 2002). DAM1P is a substrate for IpI1p (increases in ploidy), a kinase initially identified as a member of a serine/threonine kinase family responsible for regulating ploidy level in budding yeast (Chan & Botstein, 1993). In multicellular organisms it is a member of the Aurora/IpI1p kinase family (Bischoff & Plowman, 1999; Lin et al., 2006). Arath;Aurora 1, 2 and 3 (Demidov et al., 2005) can phosphorylate histone H3 in vitro and their expression is confined to dividing regions of the plant. GFP-tagged Arath;AURORAS locate to spindle microtubules, centromeres and the presumptive cell plate (Demidov et al., 2005). This fits neatly with a hypothesized AURORA-mediated regulation of ‘centromere-kinetochore-microtubule’ linkage in mitotic cells.

Figure 5.

Mitosis. At prophase (P), the chromosome is configured as two arms glued by cohesin, a multiprotein complex. At metaphase (M), the chromosome is connected to mitotic tubules (MT) by the kinetochore (KC). This is regulated by the AURORA kinase complex and separase, which is released from its chaperone, securin. At anaphase (A) the separation of chromosomes is regulated by separase and the auxin-linked anaphase promotion complex (APC). Cyclin and various other proteins are ubiquitinated (circles) by the APC and delivered to the 26S proteasome. Note, cyclin-dependent protein kinase (CDK) is inactive from this point on. Rectangles, inactive or inactivated protein; ellipses, activated protein; circles, ubiquitinations; tick marks, plant clones.

Anaphase is triggered by B-type cyclin degradation and separase activity that, during interphase and prophase of mitosis, is suppressed by the chaperone, securin (Funabiki et al., 1996; Ciosk et al., 1998). Securin is degraded, separase becomes active and chromosomes held at metaphase begin to separate (Nasmyth, 2001). The anaphase-promoting complex (APC) regulates both anaphase to telophase and M/G1 (Irniger et al., 1995; McGarry & Kirschner, 1998; Grossberger et al., 1999). APC ubiquitinates PDS1 in budding yeast (Cohen-Fix et al., 1996) and Cut 1 and 2 in fission yeast (Funabiki et al., 1996). Their enticement into the 26S proteasome is a prerequisite for sister chromatid separation in early anaphase. Later, the APC ubiquitinates cyclin B, and its destruction suspends Cdc2 kinase.

In Arabidopsis the APC comprises various components (Fulop et al., 2005b). For example, Arabidopsis, HOBBIT (HBT) is a homologue of the budding yeast CDC27 subunit of the APC. HBT can complement a nuc2/cdc27 budding yeast mutant and its transcription is cell cycle-regulated (Blilou et al., 2002). It may have dual functions in the APC and in IAA-regulated differentiation (Blilou et al., 2002). In anaphase, mitogenic CDK is inactivated by destruction of its cyclin partner (Pines, 1995). However, in fission yeast, WEE1 activity may persist into cytokinesis, activated by the so-called clp1/flp1 complex which has a role in CDK inactivation (Baldin & Ducommin, 1995; Bardin & Amon, 2001). During G2, B1-type cyclins were detected both in the cytoplasm and in the nucleus, but when cells entered mitosis, B1-type cyclins were seen on chromosomes until metaphase, but neither during anaphase nor during telophase (Criqui et al., 2001). Moreover, HsWEE1 locates to the cell plate during cytokinesis (Baldin & Ducommin, 1995). Is this an example of a dual life of a cell cycle regulator, phosphorylating CDKs in G2 but other proteins at cytokinesis?

VII. Cytokinesis – partitioning life

In plants, the preprophase band (p.p.b.) is an elliptical whorl of tubulin that forms a Saturn-like ring around the nucleus in late G2. Its remnants remain either side of the division plane, persisting through to prometaphase but then disappear (Ito, 1998).

The timing of cytokinesis is, in part, dependent on the degradation of a B-type cyclin. In a very elegant study, a cyclin gene with a mutation in its destruction box was transformed into tobacco (Weingartner et al., 2004). The destruction box is normally primed by ubiquitin for proteolytic degradation (Genschik et al., 1998). Nondegradable cyclin B had very little effect on S-phase, G2/M, prophase, metaphase or early anaphase, but thereafter mitotic spindles of the phragmoplast were perturbed (Weingartner et al., 2004). CyclinB1 was able to bind/activate both CDKA and CDKB in vitro. Most probably, CDK A and B kinase activity is essential for G2/M, prophase and metaphase. However, unless inactivated through cyclinB degradation, continued kinase activity creates chaos by perturbing anaphase through to cytokinesis (Weingartner et al., 2004). Timing is everything.

With its unique structures, plant cytokinesis is becoming well characterized (Fig. 6). Just before cytokinesis, a bunch of microtubules, the phragmoplast, remains at the division plane and it acts as conduit for regulatory proteins of cytokinesis. KNOLLE (also known as SYP11) is expressed only during M phase (Lauber et al., 1997; Völker et al., 2001); knolle mutant cells cannot fuse vesicles in what should be cytokinesis (Lukowitz et al., 1996). However, the knolle mutation is not lethal and KNOLLE expression is insufficient for cytokinesis (Völker et al., 2001). However, a KNOLLE-interacting-SNARE complex localizes to the cell plate during cytokinesis (Heese et al., 2001). Also, a syntaxin, SYP31, but not KNOLLE, interacts with the ATPase, CDC48, at the division plane (Feiler et al., 1995) and with the AAA-type ATPase, N-methylmaleide-sensitive fusion protein (NSF) (Rancour et al., 2002) (Fig. 6).

Figure 6.

Plant genes that associate with the cell plate or plus end of phragmoplast microtubules during cytokinesis. A KNOLLE-SNARE complex interacts with/regulates the cell plate (dashed vertical); SNARE but not KNOLLE localizes to the cell plate, as does SYP31. The latter interacts with both CDC48 and N-methylmaleide-sensitive fusion protein (NSF). HINKEL and NACK1 colocalize to the phragmoplast plus end beneath the cell plate and via MAP3K regulates depolymerization of plus ends of the phragmoplast. Subsequently, remnant phragmoplasts reorientate and assist in the widening of the cell plate. (This model is adapted from an excellent review of cytokinesis by Jürgens (2005).)

This complex enables vesicle fusion across the cell plate. In budding yeast, CDC48 is involved in endoplasmic reticulum (ER) and Golgi membrane fusion and in ubiquitin-degradation pathways (Ye et al., 2001). Presumably, yeast and plant CDC48 proteins share the same biochemical properties but have adapted to catalyse substrates in different structures/locations (Fig. 6).

Phragmoplast MTs stabilize beneath the expanding cell plate (Fig. 6) and this may be regulated by HINKEL, which encodes a kinesin-related protein (Strompen et al., 2002). The cytokinesis protein NACK1 colocalizes with the plus ends of phragmoplast microtubules and via MAP kinase helps to regulate depolymerization of MTs beneath the cell plate (Nishihama et al., 2002). Associated with this complex is NPK1 kinase (nucleus and phragmoplast NACK1-like localized protein kinase). NACK1 protein accumulates at anaphase and telophase when NPK1 is activated (Ishikawa et al., 2002).

VIII. M/G1 – party over

Upon completion of mitosis, proteolytic enzymes degrade ubiquitin-charged cyclins and other redundant M-phase proteins via a large (26S) proteasome (Glotzer et al., 1991; Varshavsky, 2000; Genschik et al., 1998; Plesse et al., 1998). Here, proteins of the anaphase-promoting complex (APC) are also vital (Irniger et al., 1995; King et al., 1995), as is SCF complex (Skp1-Cullin-1/Cdc53-F-box) – ubiquitin ligase (Feldman et al., 1997; Skowyra et al., 1997).

In Arabidopsis thaliana > 1400 genes encode components of, or interactors with, the 26S proteasome. Ninety per cent of them encode subunits of the E3 ubiquitin ligases, which confer substrate specificity to the 26S pathway (reviewed by Moon et al., 2004). Plant SCF is strongly linked to auxin-induced degradation of AUX/IAA proteins (Leyser et al., 1993; Gray et al., 2001). In Arabidopsis the so-called SKP1 CUL1 complex, colocalizes with mitotic spindles in metaphase, and the phragmoplast at telophase (Farras et al., 2001).

In plants, regulated protein degradation by the ubiquitin/26S proteasome is crucial to the cell cycle. For example, down-regulation of AtRBX1, a component of the SCF, inhibits growth and results in developmental perturbations (Lechner et al., 2002). SKP1 has been cloned in Arabidopsis and its expression is confined to meristems (Porat et al., 1998); it has a regulatory role in the 20S subunit of plant 26S proteasome (Farras et al., 2001), much like fission yeast skp1 (Lehman et al., 2004). In Arabidopsis, the auxin-regulated transport inhibitor response 1 (TIR1) protein is related to human SKP2 (Ruegger et al., 1998), which has a role in 26S proteasome-dependent degradative pathways. Also TIR1 is related to the budding yeast GRR1, a protein that through SKP1 mediates ubiquitin-dependent proteolysis (Li & Johnston, 1997). Interestingly, in Arabidopsis, the so-called SKP1 CULLIN complex colocalizes with mitotic spindles in metaphase, and the phragmoplast at telophase (Farras et al., 2001). It may function during mitosis and in M/G1. Either way, it is very sensitive to auxin-induced signals.

IX. Endocycles – a curious life of their own

Some cells replicate their nuclear DNA, bypass mitosis and re-replicate (Joubes & Chevalier, 2000). For example, in the suspensor of Phaseolus coccineus there is a basipetal gradient of endoreduplication culminating in a huge basal cell which can have anywhere between 2000 and 4000 DNA C amounts (Nagl et al., 1985). Re-replication of chromosomes is intense, and can result in micronuclei that may be necessary to generate specific enzymes in high concentrations (e.g. alpha amylase, beta amylase). Starch mobilization may occur through the suspensor which contains high concentrations of gibberellic acid (Alphi et al., 1979). Endoreduplication is so extensive that the chromosomal structure can be resolved microscopically as multi-‘banded’ structures. They are polytene chromosomes and, to my knowledge, were first observed and characterized by Tschermark-Woess (1956) in her study of antipodal cells in ovules. Other sightings include leaves, roots and some associated tissues of the pollen sacs and endosymbiotic N2-fixing roots (Kondorosi & Kondorosi, 2004). Endoreduplication also occurs in other parts of the plant, although nothing as extensive as the polyteny in basal suspensor cells.

In Arabidopsis, endoreduplication in leaf cells is regulated by the CDK inhibitors, KRP2 (Verkest et al., 2005a) and Rb (Park et al., 1995), together with DEL1 (DP-E2F like protein; Viieghe et al., 2005). In Arabidopsis, E2Fa-DPaoe caused extra DNA replication that, in turn, led to endoreduplication. (De Veylder et al., 2002), whilst Arath;CKS1 is expressed in both somatic and endocycles (Jacqmard et al., 1999). In maize endosperm, Zeama;CYCB1;3 is expressed in proliferating cells but not in those undergoing endoreduplication, whereas PCNA (a good marker of S-phase) and Zeama;CDC2 mRNA remained comparatively constant (Sun et al., 1999).

Arabidopsis SIAMESE (SIM) might be a regulator of endopolyploidy. A recessive mutation in the sim locus results in clusters of identical twin multicellular trichomes; trichomes are normally single celled (see Walker et al., 2000; Weinl et al., 2005). But again, is this a repression of mitosis per se or is it specific for endocycles?

Functional significance of endopolyplody in other plant tissues?

What function(s) can we ascribe these polyploid cells since there are higher plant species that do not exhibit endopolyploidy (Evans & Van't Hof, 1975)? Perhaps endopolyploidy is a dead end with cells taking longer to wind down from proliferation to differentiation. These tardy cells exhibit one or two rounds of endoreduplication and then arrest. However, plants expend energy to make endopolyploid cells, so there must be a point to it. Perhaps in response to environmental stress, endoreduplication is a phase in which mitotic chromosomes shelter until conditions improve. However, central to this hypothesis would be competence of these endoreduplicated cells to re-enter mitosis upon return of favourable conditions (Woll et al., 2005).

X.Cell size

In budding yeast, coordination of division with growth occurs at Start (a budding yeast term meaning that the mother cell has started to bud), where cells must reach a critical cell size to enter the cell cycle (Hartwell et al., 1974; Jorgensen et al., 2002). In fission yeast, a size control operates where large cells at birth required less growth to achieve critical size (Fantes, 1977). In newly formed small cells, size control functioned by shifting the time to division rather than being dependent on growth rate per se (Fantes & Nurse, 1977). When mitotic size was measured in wee1-50 Δcdc25 double mutants, mitotic size control was lost (Sveiczer et al., 1996). These authors concluded that wee1 was the main genetic element regulating cell size because, in wee1 mutants, G2/M size control was replaced by a size control at G1/S (Sveiczer et al., 1996).

In fission yeast, cell size is regulated more by wee1 than cdc25 at G2/M (Sveiczer et al., 1996). Somehow, the cell senses critical mitotic cell size and at this juncture Wee1 is down-regulated/degraded/partitioned, and Cdc25 phosphatase activates Cdc2. In Schizosaccharomyces pombe, the importance of Wee1 and Cdc25 to size regulation was demonstrated when Spcdc25oe and Spwee1oe resulted in short and long mitotic cells, respectively (Russell & Nurse, 1986, 1987). Arath;WEE1oe induced a long mitotic cell length in fission yeast (Sorrel et al., 2002). Conversely, Spcdc25oe, induces a small mitotic cell size in tobacco and Arabidopsis (Bell et al., 1993; Orchard et al., 2005; S. Li, unpublished).

Is cell size important for development?

Not only does Spcdc25oe induce a reduced cell size, but it also induces an increased frequency of lateral roots (Bell et al., 1993; McKibbin et al., 1998; S. Li, et al., unpublished). Spcdc25oe also induced more lateral roots in Arabidopsis, whilst Arath;WEE1oe in Arabidopsis has the converse effect: long epidermal cells, slow root growth and a reduced frequency of lateral roots (I. Siciliano et al., unpublished). In my view, these data are the strongest hint yet that genes regulating mitotic cell size can influence plant development, at least indirectly.

In an elegant study, Mizuhami & Fischer (2000) concluded that organ size is determined by internal developmental factors and cell number but not cell size. This was a conclusion reached through studying the effects of the Arabidopsis transcription factor, AINTEGUMENTA (Elliot et al., 1996). During organogenesis, ant-1 organs were smaller with fewer cells, but 35S::AINT organs were larger with more cells (Mizuhami & Fischer, 2000). Ectopic AINT expression allowed petal cells to proliferate for a longer period than normal without altering cell cycle time (Mizuhami & Fischer, 2000). CYCD3;1oe also dramatically increased leaf cell number but did not affect the expression of AINT (Dewitte et al., 2003), suggesting that AINT may regulate cell number in organ development by controlling CYCD3;1 expression.

XI. Root branching

Lateral root formation begins when mother pericycle cells are induced to divide (Himanen et al., 2002). The pericycle retains its mitotic competence as evidenced by spatial expression of CDKA therein (Hemerly et al., 1993). Adjacent to protoxylem tips, the pericycle is partitioned by transverse divisions (Dubrovsky et al., 2001). A D-type cyclin is also expressed at the start of lateral root morphogenesis (De Veylder et al., 1999), which could be the activating cyclin for CDKA activity. In a recent transcriptomic analysis, wild-type maize was compared with rootless with undetectable meristems (rum1) (not to be confused with fission yeast rum1!). Maize rum1 could initiate neither seminal roots nor postembryogenic laterals. Four putative cell cycle genes were preferentially up-regulated in wild-type compared with rum1: HD2 type histone acetylase, Ran binding protein, a cyclin and WEE1 (Woll et al., 2005).

There is an important auxin angle to lateral root morphogenesis. In brief, mutants deficient in auxin are characterized by long primary roots, few lateral roots and short hypocotyls (Estelle & Somerville, 1987). The aberrant lateral root forming (alf)4-1 mutation blocks the initiation of lateral roots. The encoded ALF4 protein is conserved, is nuclear localized and expressed in most tissues. Indeed, ALF4 functions independently of auxin signalling but maintains the pericycle in a mitotically competent state (DiDonato et al., 2004). In this case there ought to be some tie-up between ALF4 and CDKA that could also contribute to mitotic competence.

In the primary root apical meristem, there are far more transverse than longitudinal divisions. In maize, the rare longitudinal ones occur in small cells at the very tip of the root apical meristem. Such a division creates two files that divide transversely from this point on to increase cell number in each column (see Fig. 1 of Lück et al., 1994). Thus, as plant cells lengthen, the more likely it is there will be transverse divisions, whereas in isodiametric cells, the frequency of periclinal to transverse divisions is more equal. Given a reduced frequency of lateral roots and an increased epidermal cell size in an Arath;WEE1oe line, I hypothesize that WEE1-induced cell lengthening reduces the pericycle's potential to form lateral roots and that in roots expressing Spcdc25, reduction in pericycle cell length is consistent with an increased frequency with which this tissue initiates laterals.

I have tabulated genes that affect cell size, and/or phenotype from papers quoted here (Table 3). The picture is far from clear. Data support and negate the idea that cell size is linked to development. Also simple separation of in vitro and in vivo effects fails to resolve the ongoing controversy.

Table 3.  The cell size response to the over-expression of various cell cycle genes together with phenotypic changes compared with wild type
Gene over-expressedSpeciesCell size phenotypeFrequency of lateral rootsOther developmental changesReferences
Spcdc25Schizosaccharomyces pombesmall Russell & Nurse (1986)
Spwee1S. pombelong Russell & Nurse (1987)
Arath;CDC25S. pombesmall Sorrell et al. (2005)
Arath;WEE1S. pombelong Sorrell et al. (2002)
Spcdc25BY-2 cellssmall Orchard et al. (2005)
Arath;WEE1 in By-2 cellsBY-2 cellssmalll D. Francis & H. J. Rogers (unpublished)
Spcdc25Spcdc25Nicotiana tabacum internode segments of vegetative N. tabacum –IAA, –zeatinsmallnullincreasendPitted leaves and precious flowering Vegetative shootsBell et al. (1993)
Suchomelova et al. (2004)
Spcdc25Arabidopsis thalianasmallincreaseD. Francis & H. J. Rogers (unpublished)
Arath;WEE1A. thalianalargedecreaseD. Francis & H. J. Rogers (unpublished)
DN Arath;CDKAN. tabacumlargendnullHemerley et al. (1995)
Arath;ICK1/2A. thalianasmallndSerrated leaves, modified leaves, dwarfismWang et al. (2000)
Nicta;KIS1AN. tabacumlargend Jasinski et al. (2002)
CYCD2+3A. thalianasmallndPredominant plane of division in S3 layer of SAMs-periclinal.Dewitte et al. (2003)

XII. Conclusions

Reviewing 15 yr worth of cell cycle research led me to construct cell cycle models based on plant genes. I am intrigued about cell cycle genes that initially cropped up in genetic screens and mutational analysis but which were found to encode cell cycle-like genes. Here is my slant on three such genes: HBT, PRO and SIM.

HBT is strongly linked to the APC and has a perceived role in regulating planes of cell division and differentiation (Blilou et al., 2002; see Section VI). hbt mutants are perturbed postembryonically. Clearly HBT has pleiotropic effects but it is not a fundamental cell cycle gene since hbt mutants can form embryos.

PROLIFERA is a homologue of MCM7 (Stinchcomb et al., 1980). Its expression is confined to regions of proliferation, and PRL is maternally inherited and is expressed at G1/S of the cell cycle, all consistent with a role during DNA replication. Seemingly its essentiality for DNA replication is questionable in that mutations in PRL are not lethal. PRL is expressed strongly in early embryogenesis but decreases at later stages of endosperm development. Springer et al. (2000) suggested that PRL down-regulation is part of the mechanism that induces endoreduplication during late endosperm development – an interesting idea but one that runs counter to MCM7's participation in DNA replication in both endoreduplicative and somatic cell cycles. Maybe there is functional redundancy in the MCM2-7 complex.

SIAMESE may also regulate endoreduplication; sim mutants have multicelluar trichomes whilst they are unicellular in wild type (see Section IX). SIM is hypothesized to be a supressor of mitosis that is a prerequisite for endoreduplication. Clearly, there are many ways in which mitosis can be suppressed (see Section V). What makes SIM such a specific cell cycle regulator?

At least two plant CDKs, C and E, function outside of the cell cycle (Fulop et al., 2005a). CDKC is thought to be a regulator of transcription, whilst E regulates RNA polymerase II (Table 2). In my view, both would be requirements for cycling cells and for growing cells that are not cycling. Indeed, Hutchins et al. (2004) detected a CDKA-eIF4 complex that was abundant in proliferating and growing cells. Their interaction was only observed in proliferating cells, indicating a tie-up between post-translational mechanisms that link to the regulation of cell proliferation.

The flexibility for some ‘cell cycle’ genes could be an important property of plant cells to respond rapidly to stress conditions. The sessile nature of plants dictates that they must have flexibility to survive in the natural environment. Is this the reason for the multitude of cyclins that in plants are sensitive measures of extrinsic factors such as cold and drought and intrinsic factors such as plant growth regulators? Larry Fowke's group proposes that stress induces an increase in abscisic acid which, among other things, up-regulates ICKs; in between signalling will be interesting.

Mention of D-type cyclins under various subheadings was deliberate to illustrate their plasticity and widespread expression. The D-type cyclins, which are sensitive markers of nutrient and plant growth regulators, regulate G0/G1, and G2/M is partly regulated.

E2Fs have become a focus of plant cell cycle research being involved in the stimulation of S-phase-specific genes in both somatic and endoreduplicating cells. S-phase genes have been cloned in plants but functional evidence is lacking for many. Sites of replication are very flexible in plants, although we do not understand much about the regulation of their initiation.

We begin to see functional evidence for genes expressed at G2/M. We also know that a zeatin signal is essential for this transition but we know precious little about this signal transduction chain. Notably, B-type CDKs are unique to the plant cell cycle. At present, there are two generic schools of thought. Either CDKB is the driver for mitosis or CDKB phosphorylates ICK2, which releases CDKA as the driver. We need more proteomic data to better understand G2/M complexes in plants.

Only by visualizing and spatial tracking of genes during the cell cycle can we move forward significantly. In animals, cyclin B2 is localized to the ER whilst cyclin B1 shuttles between the nucleus and cytoplasm. B1 also associates with microtubules and the centrosome. Whilst WEE1 is mostly nuclear, MYT1 associates with the ER and Golgi. Cdc25C is a nuclear protein but Cdc25B accumulates in the cytoplasm (in humans there are CDC25s, A, B and C) (excellently reviewed by Pines, 1999). Comparable papers on the spatial expression studies of plant cell cycle genes are beginning to appear.

There are subtle differences in the timing of cytokinesis genes in plants compared with yeasts/animals (e.g. CDC48). I conclude that plant cell cycle proteins do not have different biochemical functions compared with animal ones, but they locate their substrates in different structures that can also alter timing of their activity compared with comparable animal cell cycle proteins.

I might be wrong that endoreduplication runs to a differentiated dead end. It may well have a role in the plant's response to stress, but to test this we ought to examine the mitotic competence of these endoreduplicated cells upon removal of stress.

Regulation of cell size in plants continues to be debated. There are so many conflicting data on the role of cell size in development or, indeed, lack of role. My view is that cells must achieve a critical size for division, but during development it is more difficult to interpret cell size data given the fixed position of plant cells in situ.

This review rests on 214 publications compared with approx. 180 in my first Tansley review, of which only approx. 15 are common to both. It has been a busy time for plant cell cyclists. However, I leave you with this sobering thought. Thomas Schmülling et al. undertook a transcriptomic analysis of Arabidopsis. Benzyl adenine was given to wild-type and the transcript changes were recorded after 15 min, and again after 120 min. In the 15 min treatment, 71 genes were up-regulated, including a high proportion of transcription factors. Two hours later, 36 out of the 71 genes were still up-regulated, but most of the transcription factors detected at 15 min were replaced by a new wave of transcription factors. Not one of the known up-regulated genes belonged to the cell cycle (Brenner et al., 2005). Time to get busy!


I thank Chris Norbury (Oxford University, UK) for his advice about certain aspects of yeast cell cycles, Julian Blow (Dundee University, UK) for advice about DNA replication, and Jim Murray (Cambridge University, UK) for steering me on the right D cyclin course. I also thank my co-investigator, Hilary Rogers (Cardiff University, UK), for her critical comments. However, I hereby confirm that I am solely responsible for the contents of this review and for the conclusions drawn. I apologise to anyone in the plant cell cycle field who I have not quoted; it was by accident rather than design. Funding has been received from BBSRC, Cardiff and Worcester universities. Finally, I thank another of my colleagues, Eshwar Mahenthiralingam (Cardiff University), for introducing me to the delights of Endnote®.