R1R2R3-Myb proteins positively regulate cytokinesis through activation of KNOLLE transcription in Arabidopsis thaliana

Morphogenesis in plants is controlled by the rate and orientation of cell division and by cell expansion. Plant somatic cell division and in particular cytokinesis differ in some aspects from their counterparts in other organisms(Field et al., 1999). One major difference is that plant cells lack the contractile ring that assists in partitioning the cytoplasm from the surface in most animal cells, and instead construct a partitioning cell plate inside the cell. Two plant-specific microtubule structures, the preprophase band (PPB) and the phragmoplast, play central roles in the orientation and expansion of the cell plate as well as in the execution of cytokinesis (Otegui and Staehelin, 2000; Jürgens,2005). The cortical position of the transient PPB correlates with the site at which the outward growing cell plate will fuse with the parental plasma membrane at the end of cytokinesis. The phragmoplast assembles in the center of the division plane during anaphase-to-telophase transition. During cytokinesis, secretory vesicles carrying membrane and cell wall components are guided along the phragmoplast to the division plane and form, by fusion, a membraneous compartment enclosing the immature wall, the cell plate. Phragmoplast and cell plate expand laterally until the cell plate fuses with the parental plasma membrane and cell wall.

Relatively few genes involved in cytokinesis have been identified by mutation in plants, and they fall into two classes. Genes in the first class are required for proper orientation of the division plane, and mutants include fass/tonneau in Arabidopsis, and discordia and tangled in maize (Sylvester,2000; Smith,2001). Genes in the second class are required for the execution of cytokinesis. Mutations in these genes cause defects in cell plate formation,leading to a common phenotype that is characterized by the formation of multinucleate cells with gapped cell walls or cell wall stubs(Assaad et al., 1996; Yang et al., 1999; Strompen et al., 2002; Lukowitz et al., 1996; Falbel et al., 2003). Genes identified from these mutations encode proteins that are involved in cell plate membrane fusion (Assaad et al.,2001; Lukowitz et al.,1996), biogenesis of primary cell wall(Zuo et al., 2000; Lukowitz et al., 2001) or microtubule array formation or dynamics(Strompen et al., 2002; Soyano et al., 2003; Müller et al., 2004). KNOLLE (KN) and KEULE (KEU) were initially identified as mutants that are lethal at the seedling stage and which have an abnormal seedling body organization (Mayer et al., 1991). KN encodes a cytokinesis-specific,plant-specific syntaxin, whereas KEU encodes a Sec1/Munc18 protein that may regulate syntaxin function(Lukowitz et al., 1996; Assaad et al., 2001). HINKEL (HIK) is required for expansion of the phragmoplast during cell plate formation, and encodes a plant-specific kinesin-like protein(Strompen et al., 2002). HIK is the Arabidopsis ortholog of tobacco NACK1,which binds to the cell plate-associated mitogen-activated kinase (MAPK)kinase kinase, NPK1, and activates a MAPK pathway regulating phragmoplast expansion (Nishihama et al.,2002).

Transcripts of NACK1 and mitotic B1-type cyclin (CYCB1)accumulate specifically during late G2- and M-phases in synchronized cultures of tobacco BY-2 cells (Ito et al.,2001). Transcription of these genes is regulated by a common upstream cis-acting element, called MSA (mitosis-specific activator)(Ito et al., 1998; Ito, 2000). A group of Myb transcription factors in tobacco, NtmybA1, NtmybA2 and NtmybB, bind to the MSA motif in vitro and in yeast. NtmybA1 and NtmybA2 are structurally closely related transcriptional activators, whereas NtmybB acts as a competitive repressor in tobacco cells (Ito et al.,2001; Araki et al.,2004). Plants have a large number of Myb genes, most of which encode R2R3-Myb proteins containing two tandemly repeated sequences of ∼50 amino acids in the Myb domain and control diverse developmental processes(Stracke et al., 2001). By contrast, NtmybA1, NtmybA2 and NtmybB are R1R2R3-Myb proteins containing three tandemly repeated sequences and thus are structurally similar to Myb proteins in vertebrates and Drosophila(Ito, 2005). Mammalian Myb proteins were generally believed to play a role in cell-cycle regulation,particularly at the G1/S transition(Lipsick, 1996). According to more recent studies, however, Myb proteins may also play a role in the transcription of the cyclin B gene in Drosophila(Okada et al., 2002),zebrafish (Shepard et al.,2005), and human cells (Zhu et al., 2004). Thus, G2/M phase-specific transcription appears to be mediated by R1R2R3-Myb proteins in a wide range of evolutionarily distant organisms.

The Arabidopsis genome contains five genes that encode R1R2R3-Myb proteins (Stracke et al.,2001). Genome-wide expression analysis of synchronized Arabdidopsis cells revealed G2/M phase-specific expression of 82 genes, which include CYCB1, CYCB2, CDC20.1, KN and AtNACK1/HIK (Menges et al.,2005). All but 17 of these genes contain at least one MSA element,suggesting that MSA-binding R1R2R3-Myb proteins may regulate many G2/M phase-specific genes and thereby control progression of mitosis and cytokinesis. Here, we analyze the in vivo function of two closely related Arabidopsis R1R2R3-Myb proteins, MYB3R1 and MYB3R4 (also known as MYB3R-1 and MYB3R-4), which are homologous to tobacco transcriptional activators NtmybA1 and NtmybA2. We isolated plants with mutations in the MYB3R1 and MYB3R4 genes and characterized the phenotype of single and double mutants. The myb3r1 myb3r4 double mutant showed characteristic cytokinesis defects during embryogenesis and postembryonic development. Our genetic and molecular studies suggest that MYB3R1and MYB3R4 have partially overlapping function and quantitatively promote cytokinesis, mainly through transcriptional activation of the KN gene.

The Arabidopsis ecotype Columbia (Col) was used as wild type. Seeds with T-DNA insertion mutations in MYB3R1 (At4g32730) and MYB3R4 (At5g11510) from the SALK collection(Alonso et al., 2003) were obtained from the Arabidopsis Biological Resource Center (ABRC):SALK_018482 (myb3r1-1), SALK_059819 (myb3r4-1) and SALK_034806 (myb3r4-2). The kn allele X37-2 (Ler background)(Lukowitz et al., 1996) was used for genetic crosses and for transformation. Plants were grown either on soil or agar medium at 22°C under continuous illumination (70-120μmol/m²/second). Standard agar growth medium contained 0.5×MS salts, 2% w/v sucrose and 0.7% w/v agar.

The MM2d cell line (Arabidopsis Ler ecotype)(Menges and Murray, 2002) was obtained from Bayer BioScience N.V. (Gent, Belgium). Maintenance and synchronization were performed essentially as described for tobacco BY-2 cells(Nagata et al., 1992; Ito et al., 1997). Briefly,cells were grown in modified Linsmaier and Skoog medium at 27°C in the dark, with rotation at 130 rpm. For subculturing, 3 ml of saturated culture were transferred into 95 ml fresh medium. For synchronization, 15 ml of 7-day-old culture were transferred into 95 ml fresh medium containing 5 mg/l aphidicolin (Wako, Japan). Cells were treated with aphidicolin for 24 hours,washed with fresh medium and then cultured in 100 ml medium. The mitotic index was determined as described (Ito et al.,1997).

Promoter regions of CDKA;1 (1.9 kb) and RPS5A (1.6 kb)were amplified by PCR from Col genomic DNA and cloned in the binary vector pPZP211 (Hajdukiewicz et al.,1994). Primers used for PCR were 5′-AAGAGCTCAATTCCTGAATAATAAAGCTGAAG-3′ and 5′-AACTGCAGTTACAACTGATAACCGTATAGCTC-3′ for CDKA;1, and 5′-AACTGCAGTTGATTCGCTATTTGCAGTGCAC-3′ and 5′-AAACAGAGCGTGAGCTCAAATAC-3′ for RPS5A. For construction of the CDKA;1::KN and RPS5A::KN vectors, KNfull-length cDNA sequence was placed downstream of the promoters. These constructs were introduced by floral-dip transformation(Clough and Bent, 1998) into Arabidopsis lines that were homozygous for myb3r4-1 and heterozygous for myb3r1-1. We selected T2 plants that were homozygous for myb3r1-1, myb3r4-1, and also for the transgene. Similarly, we prepared CDKA;1::MYB3R1 and CDKA;1::MYB3R4 vectors and obtained transgenic plants that were homozygous for the transgene in the myb3r1 myb3r4 background.

For constructing β-glucuronidase (GUS) reporter constructs, promoter regions of MYB3R1 (1.1 kb) and MYB3R4 (1.2 kb) were amplified by PCR and cloned upstream of the GUS gene in pPZP211 binary vector. Primers used for PCR were 5′-AACTGCAGTATTAGCCAATGAAGGTGGACTAGC-3′ and 5′-AAGTCGACAATTAAGACGCTGAGAATCCAGATG-3′ for MYB3R1, and 5′-AACTGCAGAGTCGGTAACATTCTGCCAGAGATG-3′ and 5′-AAGTCGACGAGCTTCAGAAATGGAAGTGGTTC-3′ for MYB3R4. The resulting constructs, MYB3R1::GUS and MYB3R4::GUS, were transformed into Col plants. Multiple lines were analyzed and had similar staining patterns.

Two KN promoter-deletion constructs were generated. KNΔMSA1 was amplified by PCR from the SacI/EcoRI KN expression construct(Völker et al., 2001)with the primer pair 5′-GTAATACGACTCACTATAGGGC-3′ and 5′-AAAGAATTCAAATATAGCCGTTGGGGCG-3′ and cloned in pGII BASTA. To generate the KNΔMSA2 construct, the two regions flanking the two proximal MSA elements in the KN promoter were amplified by PCR with the following primer pairs: (1)5′-TCTAGACCCGGGTTTCTCCTTTTTCTTATATTAGAAAGAAAGC-3′ and 5′-GAATTCCCTGCTCCCATATCCTTCATCG-3′ and (2)5′-GAATTCTCAAGAAGAGCTGAAACTGGTAATG-3′ and 5′-CACTGCGATTCTCTCTGATTC-3′. The PCR fragments were digested with XbaI and XbaI/EcoRI, respectively, and cloned in the KN genomic rescue fragment(Müller et al., 2003). Floral-dip transformation of kn-X37-2 heterozygous plants and selection for BASTA resistance were performed as described(Müller et al., 2003). Transgenic plants were analyzed for rescue of the kn seedling phenotype.

Western blots were performed as described(Lauber et al., 1997). The anti-KN serum (Lauber et al.,1997) was used at 1:6000. Myc-tagged KN protein expressed from the transgenes KNΔMSA1 and KNΔMSA2was detected with the mouse anti-c-myc monoclonal antibody 9E10 (Santa Cruz Biotechnology) at 1:250.

For construction of luciferase (LUC) reporter plasmids, promoter regions of CYCB1;1 (1.1 kb), CYCB1;2 (0.7 kb), CDC20.1 (0.9 kb), AtNACK1/HIK (1.4 kb) and KN (2.2 kb) were amplified by PCR. Primers used for PCR were 5′-AACTGCAGAAGCTTACAATTGTGTGGGAACCATAGC-3′ and 5′-AAGTCGACTCTCTCAGACTAAAATCTCAGG-3′ for CYCB1;1,5′-CACCTTCAGATGATAGTGTACTCAC-3′ and 5′-TTCTCTTTCGTAAAGAGTCTCTGCG-3′ for CYCB1;2,5′-AACTGCAGTGAAGAACATGCTTATCACACGTC-3′ and 5′-AAGTCGACAAGCTAGCGAAGAGGGAATCGTTC-3′ for CDC20.1,5′-CACCACAGACTGAAAGCGACTTGATAGT-3′ and 5′-TCAGGCAGCTAAGAATGTAGAATC-3′ for AtNACK1/HIK, and 5′-CACCAGGAAAAATTAGCTTCACGAG-3′ and 5′-CATAAACGATTTCGTCATCAAGTC-3′ for KN. The amplified fragments were inserted upstream of the LUC reporter gene in pUC18. For mutagenesis of the MSA elements, a PCR fragment of the KNpromoter was cloned in pENTR-/D-TOPO (Invitrogen, Carlsbad, CA). Using this plasmid as a template, PCR was performed with the primer pair 5′-CCCTGCAGCTATATTTTGGCGATGTATCCAATGTCGTC-3′ and 5′-AGCTGCAGGGGCGAAGAAATCACAGTATTGGCTAATTTC-3′. The amplified fragment, containing the entire vector sequence, was cut by PstI and then circularized by end-end ligation. The resulting plasmid contained KN promoter fragments with all three MSA elements being mutated(KNΔMSA), which was then subcloned upstream of the LUC reporter gene in pUC18. Expression plasmids, 35S::MYB3R1 and 35S::MYB3R4, were constructed by inserting full-length cDNA sequences downstream of the double cauliflower mosaic virus (CAMV) 35S promoter in pJIT60 (Guerineau and Mullineaux,1993). The 35S::CYCB1 expression plasmid, isolation of tobacco BY-2 protoplasts, transfections and LUC activity assays were described previously (Araki et al.,2004).

Cells were observed with differential interference contrast (DIC) and fluorescence optics (Olympus BX54 microscope). Seedlings and ovules were fixed in an ethanol:acetic acid (9:1) solution, cleared overnight in Hoyer's solution (a mixture of 20 g of chloral hydrate, 1 ml of glycerol, 6 ml of water and 1.5 g of gum arabic) and viewed with DIC. Epidermal peels were obtained from silique valves using tweezers, mounted in Hoyer's solution and imaged by DIC. For Toluidine Blue staining, epidermal peels were immersed in a solution containing 0.1% w/v Toluidine Blue O and 50 mM sodium phosphate (pH 7.0), rinsed with water and mounted in Hoyer's solution. For DAPI staining,epidermal peels were fixed in 2% glutaraldehyde, mounted in a solution containing 1 μg/ml DAPI, 1% Triton X-100 and 2% glutaraldehyde, and viewed with fluorescence optics.

RNA was extracted from inflorescences containing young flower buds or MM2d cells using Trizol reagent according to the manufacturer's instructions(Invitrogen). Poly(dT) cDNA synthesis was performed using a SuperScript First-Strand Synthesis System for RT-PCR according to the manufacturer's instructions (Invitrogen). Quantification was performed on a Light Cycler 1.5 system (Roche, Mannheim, Germany) using the SYBR Premix EX Taq (Takara Biochemicals, Japan). Thermal cycling conditions were as follows: 10 seconds at 95°C, and then 40 cycles of 5 seconds at 95°C, 20 seconds at 60°C. For gene expression analysis in wild type and mutants, three replicate assays were performed with RNA isolated from distinct individuals. Results were normalized to the expression of ACT2 mRNA. Information on the primer sets used for real-time RT-PCR is available upon request.

Three tobacco genes, NtmybA1, NtmybA2 and NtmybB, encode R1R2R3-Myb transcription factors that bind to MSA elements(Ito et al., 2001). The Arabidopsis genome contains five related but functionally uncharacterized genes, MYB3R1-MYB3R5(Stracke et al., 2001), and there are four genes encoding R1R2R3-Myb proteins in the rice genome (The TIGR rice database, http://rice.tigr.org/). By phylogenetic analysis of the conserved DNA-binding domains, Arabidopsis MYB3R1 and MYB3R4 form a clade together with NtmybA1,NtmybA2 and two rice proteins (Fig. 1A). That these Myb proteins might perform an identical biological function is also suggested by two stretches of conserved amino acid sequences in their C-terminal half (Fig. 1B). Another clade comprised MYB3R3, MYB3R5 and two rice proteins but no previously characterized Myb protein, whereas Arabidopsis and rice appear to lack orthologs of the tobacco NtmybB transcriptional repressor(Fig. 1A).

To examine the role of the Arabidopsis R1R2R3-Myb genes MYB3R1 and MYB3R4 in cell cycle and development, we characterized T-DNA insertion mutants from the SALK collection(Alonso et al., 2003). The T-DNA insertions of myb3r4-1 and myb3r4-2 were close to each other in the second exon (Fig. 1C). The myb3r4-1 insertion occurs within the open reading frame N-terminal to the Myb domain and causes an 18-bp deletion,impairing protein function (see below). The myb3r4-2 insertion is located 35 bp upstream of the initiation codon. However, this mutation reduced the level of MYB3R4 transcript to less than 3% of the wild-type level(Fig. 1D). The T-DNA of myb3r1-1 is inserted in the eighth intron of MYB3R1, which is flanked by exons encoding conserved amino acid sequences(Fig. 1C). Quantitative RT-PCR analyses with two different primer pairs indicated a normal level of MYB3R1 mRNA 5′ to the insertion site[Fig. 1C, primer pair (1)] but no mRNA with proper splicing around the insertion site[Fig. 1C, primer pair (2)],suggesting that myb3r1-1 transcript encodes a truncated protein that lacks conserved sequences (Fig. 1D). The myb3r1-1 myb3r4-1 and myb3r1-1 myb3r4-2double mutants, however, expressed low levels (∼20% of wild-type level) of properly spliced MYB3R1 transcript, which may encode wild-type protein (Fig. 1D). In summary,the T-DNA insertions appear to represent loss-of-function alleles of MYB3R1 and MYB3R4 genes, although they may still allow for some residual activity, especially of MYB3R1 in the double mutant.

myb3r1 and myb3r4 single mutants displayed no macroscopic defect, although myb3r4 plants had very weak cellular defects of cytokinesis, to be described later. The myb3r1-1 myb3r4-1 and myb3r1-1 myb3r4-2 double mutants showed essentially identical phenotypes. Most of these plants were viable and developed into mature,fertile plants, although some macroscopic abnormalities were found in a subpopulation, which include abnormal seedling morphology and impaired shoot elongation. In addition, we found that all the double mutant individuals had cellular defects in cytokinesis. The defective cytokinesis was fully or partially complemented by transgenes that expressed MYB3R1 or MYB3R4 from the CDKA;1 promoter (data not shown, see Table 1), supporting that the T-DNA insertions inactivate the MYB3R1 and MYB3R4 genes at least partially. All experiments reported below were done with the myb3r1-1 myb3r4-1 double mutant, whose phenotype was slightly stronger than that of the myb3r1-1 myb3r4-2 double mutant.

Defective cytokinesis was found in the myb3r1 myb3r4 embryo, most frequently during the first or second division after fertilization. These embryos were rod- or club-shaped and contained multiple nuclei that were not separated by cell walls (Fig. 2B, compare with wild type shown in Fig. 2A). This phenotype may have resulted from multiple rounds of nuclear division without cytokinesis. In support of this interpretation, we frequently observed embryos in which all nuclei progressed synchronously through metaphase(Fig. 2C). A similar defective embryo phenotype has been reported for the kn keule double mutant in which somatic cytokinesis is completely blocked(Waizenegger et al., 2000). The most notable defects of cytokinesis were found in stomata, as reported for other cytokinesis-defective mutants (Falbel et al., 2003; Söllner et al., 2002; Yang et al.,1999). In normal stoma development, a guard mother cell divides symmetrically to produce two guard cells whose closely apposed common cell walls then partially detach from one another to form the pore(Fig. 2D). Stomata were irregularly shaped in the double mutant and thus similar to those of other cytokinesis-defective mutants, lacking the ventral wall or containing cell wall stubs (Fig. 2E-G). The surrounding guard cell was binucleate, which suggests that these abnormal stomata resulted from a failure of the guard mother cell to undergo cytokinesis (Fig. 2I, compare with the wild-type stoma shown in Fig. 2H). Cytokinesis-defective stomata were detected in most organs at similar frequencies of 20-30%, including cotyledons, cauline leaves,inflorescence stems, silique valves and sepals, but were almost absent from rosette leaves (Fig. 2N). Gapped cell walls and cell wall stubs were also observed in epidermal cells of most organs, except rosette leaves (Fig. 2J-M). Thus, the requirement of the MYB3R1 and MYB3R4 genes for cytokinesis may vary between organs and cell types.

The gene dosage of MYB3R1 had a quantitative effect on the severity of the cytokinesis defect in myb3r4 homozygous plants. Cytokinesis-defective stomata were much less frequent in the myb3r4single mutant than in the myb3r1 myb3r4 double mutant (1.1% versus 22.1% in the outer epidermis of silique valve), whereas myb3r1/MYB3R1 myb3r4/myb3r4 plants, with just one copy of MYB3R1, showed an intermediate frequency of 8.7% defective stomata. The observed severity of the cytokinesis defects in each genotype was inversely correlated with the abundance of properly spliced MYB3R1 transcript (data not shown). By contrast, neither the myb3r1 single mutant nor myb3r1/myb3r1 myb3r4/MYB3R4 plants displayed abnormal stomata, suggesting that MYB3R4 makes a larger contribution to cytokinesis than does MYB3R1, although cytokinesis is promoted by the combined level of expression of the two functionally redundant Myb genes, MYB3R1 and MYB3R4.

MYB3R1 and MYB3R4 are homologous to transcriptional activators NtmybA1 and NtmybA2 in tobacco, which activate G2/M phase-specific genes CYCB1and NACK1 through binding to their MSA elements(Ito et al., 2001). To examine whether MYB3R1 and MYB3R4 have similar function, we compared mRNA levels of several cell cycle-related genes between myb3r1 myb3r4 and wild-type plants, analyzing RNA from inflorescences and young flower buds by quantitative RT-PCR (Fig. 3A). No changes in abundance were detected for CDKA;1, histone H4, CYCD3;1,CYCA2;2 and CYCA3;1 genes, which lack MSA motifs and are expressed constitutively or before G2/M in synchronized ArabidopsisMM2d cells (Menges et al.,2003). Potential target genes of MYB3R1 and MYB3R4 should contain MSA motifs and show G2/M phase-specific expression in synchronized cells. The transcript levels of five such genes, CYCA1;1, CYCB2;1, CYCB1;2,CYCB1;4 and CDC20.1, were differentially reduced, whereas the mRNA level of CYCB1;1 was slightly increased(Fig. 3A).

The severe cytokinesis defects in myb3r1 myb3r4 double mutants suggested some genes that have essential functions in cytokinesis and are upregulated during G2/M phase, such as AtNACK1/HIK, PLEIADE(PLE)/MAP65-3, ANQ1 and KN, might be downregulated(Menges et al., 2003). Although all these genes contain MSA motifs in their promoter regions,transcript abundance for AtNACK1/HIK, PLE/MAP65-3 and ANQ1 was unchanged or even increased in the double mutant(Fig. 3A). By contrast, the level of KN mRNA was decreased dramatically to approximately 30% of the wild-type level.

Since the gene dosage of MYB3R1 in the myb3r4 mutant background influenced the severity of cytokinesis defects, we examined whether it also affected the expression of potential target genes KN and CYCB2;1 (Fig. 3B). No changes were detected in the myb3r1 single mutant, whereas transcript levels of both KN and CYCB2;1 were significantly reduced in the myb3r4 single mutant, even more strongly in the myb3r1 myb3r4 double mutant, and to intermediate levels in MYB3R1/myb3r1 myb3r4/myb3r4 plants. Thus, the two Myb proteins MYB3R1 and MYB3R4 may quantitatively activate transcription of G2/M phase-specific genes, with MYB3R4 making the larger contribution.

To analyze the genetic relationship between KN and the two regulators, we crossed the myb3r1 myb3r4 double mutant with the knheterozygote to generate F2 progeny with different genotypes at those three loci. An initial survey of 48 F2 plants with cytokinesis defects in the epidermis of silique valves revealed that all these plants were homozygous for myb3r4, indicating that a single copy each of the wild-type alleles of MYB3R4 and KN are sufficient for normal cytokinesis. For a detailed analysis, we genotyped 96 F3 individuals derived from self-fertilized F2 plants that were homozygous for myb3r4, but heterozygous for myb3r1 and kn. The severity of cytokinesis defects as represented by the frequency of abnormal stomata in the outer epidermis of silique valve was dramatically increased in both myb3r4single mutant and MYB3R1/myb3r1 myb3r4/myb3r4 background by the heterozygous mutation of kn (Fig. 4A). There was also a comparable increase in the frequency of epidermal cells of silique valves that had gapped cell walls or cell wall stubs (Fig. 4A), suggesting that KN becomes limiting when the level of activating Myb transcription factor is reduced.

Genotyping of 96 F3 plants suggested that the heterozygous mutation of kn may dramatically increase the frequency of embryonic lethality of myb3r1 myb3r4. One surviving plant of this genotype was tiny and produced a few small curly leaves, and subsequently stopped growing entirely. We identified a further eight such plants among ∼500 F3 plants, and examined their phenotype. In the rosette leaves of these plants, the frequency of cytokinesis-defective stomata was much higher than that observed in the myb3r1 myb3r4 plants (45.2% versus 1.5%, n=5). In addition,gapped cell walls were frequently found in epidermal pavement cells and palisade cells of rosette leaves when plants had the heterozygous mutation of kn in the myb3r1 myb3r4 background(Fig. 4B,C). Enlarged cells with multiple nuclei were also observed in roots of this genotype but not in the absence of the heterozygous mutation of kn(Fig. 4D).

kn heterozygotes on their own did not show any phenotype of defective cytokinesis (Fig. 4A). Thus, the observed genetic interaction between KNand the two Myb genes was not simply because of additive effects of each mutation. Rather, the reduced copy number of functional KN gene might have further decreased the already reduced level of KN gene expression in the myb3r1 myb3r4 double mutant, leading to enhanced defects of cytokinesis. To test this idea, we analyzed transcript abundance of KN in the F3 individuals (Fig. 4E). The relative decrease in KN mRNA level in each genotype was roughly correlated with the frequency of cytokinesis-defective stomata and epidermal cells (compare Fig. 4E with Fig. 4A),supporting the notion that inactivation of MYB3R1 and MYB3R4genes causes reduced KN gene expression, which, in turn, leads to defective cytokinesis.

To test whether enhanced KN gene expression could rescue the cytokinesis defects of myb3r1 myb3r4 plants, we generated transgenic plants that expressed KN protein from heterologous promoters. Expression of KN protein from the CaMV 35S promoter failed to rescue the cytokinesis-defective kn mutant embryo, which was probably because of weak activity of the CaMV 35S promoter in mitotically dividing cells(Völker et al., 2001). We tested the promoters of the CDKA;1 and RPS5A genes, which are active in proliferating tissues(Hemerly et al., 1993; Weijers et al., 2001). We confirmed that kn homozygous mutant seeds were partially rescued by the CDKA;1::KN and PRS5A::KN. CDKA;1::KN transgenic plants that were homozygous for kn developed past the seedling stage and produced a few small leaves, before growth ceased at the vegetative stage. The RPS5A::KN transgene rescued kn mutant plants more effectively, resulting in the formation of fertile flowers. Introduction of these transgenes into the myb3r1 myb3r4 double mutant also partially rescued their cytokinesis defects, significantly reducing the frequency of cytokinesis-defective cells, which include abnormal stomata, epidermal cells with gapped cell walls and the multinucleate single-celled embryos without cross walls (Table 1).

The KN gene contains three MSA motifs within 59 bp of its promoter region (Fig. 5A). To evaluate the function of these MSA motifs in KN gene expression, we analysed different promoter deletion constructs of a genomic rescue clone of KN (Müller et al.,2003). The promoter deletion construct KNΔMSA1 (Fig. 5A) retaining a fragment up to -286 bp from the KN start codon was sufficient for normal expression(Fig. 5B) and rescuing activity of Myc-tagged KN protein (data not shown). Because this promoter fragment still contained two MSA motifs with high homology to the consensus sequence TC(T/C)AACGG(T/C)(T/C)A (Ito,2000), we deleted these two MSA motifs from the Myc-KN rescue construct by replacing them with two additional restriction sites(Fig. 5A). This mutant construct, KNΔMSA2, did not rescue kn mutant plants (data not shown). The KNΔMSA2 construct encoded a Myc-tagged KN protein (Fig. 5A), which can be distinguished from protein encoded by endogenous KN gene. In transgenic plants carrying the KNΔMSA2, Myc-KN expression was reduced below detection level, indicating that MSA-dependent activation of KN expression is essential for KN activity in cytokinesis(Fig. 5C).

To directly demonstrate that MYB3R1 and MYB3R4 activate the KNgene promoter, we performed co-transfection assays in which the Myb transcription factor genes and a fusion between the KN promoter and the LUC reporter gene were introduced into tobacco BY-2 protoplasts. We have used this system previously to demonstrate that transfection of an NtmybA2 expression plasmid resulted in increased activity of reporter constructs such as CYCB1::LUC and NACK1::LUC fusions(Ito et al., 2001). We have further shown that this transactivation activity of NtmybA2 was dramatically increased by co-transfection of 35S::CYCB1, possibly because the activity of NtmybA2 was enhanced upon phosphorylation by the cyclin/CDK complex(Araki et al., 2004). Expression of the KN::LUC reporter construct was stimulated approximately twofold by co-transfection of 35S::MYB3R4, and this activation was further increased by the additional transfection of a 35S::CYCB1 expression plasmid(Fig. 6A). In comparison,KN::LUC reporter activity was unchanged when 35S::MYB3R1 was transfected, but was activated approximately twofold when 35S::CYCB1 was additionally co-transfected (Fig. 6A). Thus,MYB3R1 and MYB3R4 act as transcriptional activators of the KN gene promoter, with MYB3R4 being more effective than MYB3R1. Activation of the KN promoter by MYB3R4 is dependent on the presence of MSA elements,because a mutant KN promoter that lacks all three MSA elements(KNΔMSA) was no longer activated by 35S::MYB3R4 alone or in combination with 35S::CYCB1 (Fig. 6B).

We also tested LUC reporter constructs fused to the promoters of other genes expressed during G2/M transition(Fig. 6C). Transfection of 35S::MYB3R4, either alone or in combination with 35S::CYCB1, resulted in the activation of CYCB1;2 and CDC20.1 promoters, consistent with the decreased expression of these genes in the myb3r1 myb3r4 double mutant. By contrast, the promoters of CYCB1;1 and AtNACK1/HIK, whose transcript levels were unchanged in the double mutant, were also nearly unchanged by co-transfection of either 35S::MYB3R4 alone or 35S::MYB3R4 plus 35S::CYCB1 (Fig. 6C). These results again suggest that MYB3R4 acts as a transcriptional activator for various G2/M phase-specific genes, and that its transactivation varies between potential target genes.

We examined the expression patterns of MYB3R1 and MYB3R4during the cell cycle in cultures of Arabidopsis MM2d cells synchronized by aphidicolin block and release. Progression of the cell cycle was monitored by mitotic index and transcript abundance of histone H4, an S phase-specific gene. Histone H4 mRNA level and mitotic index peaked at 2 and 12 hours, respectively, after aphidicolin release(Fig. 7A,B). Transcript levels of MYB3R4 and KN changed in a similar pattern during the cell cycle, and their peak was recorded 2 hours before the peak of mitotic index (Fig. 7A,B). By contrast,the level of MYB3R1 mRNA was nearly unchanged(Fig. 7A).

To examine expression domains of MYB3R1 and MYB3R4 in vivo, we generated transgenic plants carrying 1.1 kb of the MYB3R1promoter or 1.2 kb of the MYB3R4 promoter upstream of the GUS gene and examined GUS expression patterns by X-gluc staining. In both MYB3R1::GUS and MYB3R4::GUS lines, X-gluc staining was observed throughout the cotyledons and rosette leaves in 12-day-old seedlings, where vascular tissues were strongly stained(Fig. 7C,D). In the root, the vascular cylinder was darkly stained in both lines(Fig. 7E,F). The division zone of primary root tips and emerging lateral roots were stained in MYB3R4::GUS lines, whereas such staining was absent in MYB3R1::GUS lines (Fig. 7E-H). Instead, MYB3R1::GUS expression was observed in the columella root cap (Fig. 7G). Floral organs in young flower buds were stained in both MYB3R1::GUS and MYB3R4::GUS lines(Fig. 7I,J). In MYB3R1::GUS flowers, strong expression was observed in vascular tissues of filaments and anthers. In ovules of both MYB3R1::GUS and MYB3R4::GUS lines, the developing embryo and maternal tissues were weakly and uniformly stained (Fig. 7K,L).

In summary, MYB3R4 is expressed in a cell cycle-dependent manner in synchronized cells, and its promoter is active in proliferating tissues,such as root tips and emerging lateral roots. However, the expression domains of MYB3R1::GUS and MYB3R4::GUS are not restricted to the meristematic tissues, and levels of expression do not generally correlate with cell division activity. This may suggest that MYB3R1 and MYB3R4 are post-transcriptionally regulated such that their ability for transcriptional activation is enhanced only in tissues with high cell-division activities. One such mechanism might be CDK-dependent phosphorylation, which activates the transactivation potential of NtmybA2 in tobacco cells(Araki et al., 2004). Consistently, both MYB3R1 and MYB3R4 contain multiple consensus phosphorylation sites by CDK (S/T-P-X-K/R), and their activities are enhanced by CYCB1 in tobacco cells (see Fig. 6B).

R1R2R3-Myb proteins NtmybA1 and NtmybA2 are transcriptional activators that bind to MSA elements, which are necessary and sufficient for transcription of the G2/M phase-specific genes CYCB1 and NACK1 in tobacco cells (Ito et al., 1998; Ito et al., 2001). In Arabidopsis, MSA elements are present in most G2/M phase-specific genes (Menges et al., 2005). MSA element-binding R1R2R3-Myb proteins might thus act as general positive regulators of mitotic events by inducing genes with essential mitotic functions. In this study, we have characterized T-DNA insertion mutants of Arabidopsis MYB3R1 and MYB3R4 genes, which are homologous to tobacco NtmybA1 and NtmybA2, and observed downregulation of several G2/M phase-specific genes including CDC20.1, CYCB2;1 and KN in the myb3r1 myb3r4 double mutant. The primary defect was no or incomplete cytokinesis of several cell types in different organs,which correlated with a reduced transcript level of KN, a gene required for cytokinesis. The KN gene contains three MSA motifs in its promoter and is expressed in a cell cycle-dependent manner preferentially during G2/M phase in embryos and cultured cells(Lukowitz et al., 1996) (this study). As shown here, the two promoter-proximal MSA elements are essential for a functionally sufficient level of KN transcription, and transient expression of MYB3R4 activated the KN gene promoter in an MSA-dependent manner in tobacco cells. These findings strongly suggest that MYB3R4 activates KN transcription in vivo by binding to its MSA elements.

Our genetic analysis suggests that Myb proteins MYB3R1 and MYB3R4 may have redundant function, but MYB3R1 clearly contributes much less to KNand CYCB2;1 gene activation and promotion of cytokinesis, and its contribution is only detected in the absence of MYB3R4. This difference cannot be attributed to differences between their promoters or remaining activities of T-DNA insertion alleles because (1) CDKA;1::MYB3R1 was unable to rescue reduced KN expression in myb3r4 plants (data not shown), and (2) 35S::MYB3R4, but not 35S::MYB3R1, on its own was able to activate KN and other MSA-containing promoters in tobacco cells. Transactivation activity of MYB3R1 was evident only when CYCB1 was co-expressed, which may enhance its transactivation potential(Araki et al., 2004). MYB3R4 was expressed in proliferating tissues and preferentially during G2/M phase in synchronous cell cultures, whereas expression of MYB3R1 appeared to be unrelated to the cell cycle. Taken together,MYB3R4 may be a major positive regulator for transcription of G2/M phase-specific genes, whereas MYB3R1 may assist MYB3R4 and may have other unrelated biological functions.

The seedling lethality of the cytokinesis-defective kn mutant was not rescued by a KN genomic clone lacking the two promoter-proximal MSA elements, indicating that MSA-binding activator(s) are essential for KN expression. However, the myb3r1 myb3r4 double mutant is viable and fertile, although it shows cytokinesis defects, and KNexpression is not completely abolished. The apparent discrepancy might be explained by residual activity of MYB3R1 in the double mutant. However, an alternative explanation is suggested by the observation that the myb3r1 myb3r4 double mutation did not decrease, but even slightly increased, the expression of the CYCB1;1 gene (this study), although the presence of MSA elements is essential for its transcription(Planchais et al., 2002; Li et al., 2005). This may not be due solely to remaining activity of MYB3R1 in the double mutant, because the myb3r1-1 single mutation, which eliminated the accumulation of the normal MYB3R1 transcript, did not affect the CYCB1;1transcript level (data not shown). In addition, we showed that the CYCB1;1 promoter was not activated by 35S::MYB3R1 (data not shown)nor 35S::MYB3R4 (see Fig. 6C)in tobacco cells. Thus, we speculate that Arabidopsis might express additional transcriptional activator(s) that act(s) through MSA elements redundantly with MYB3R1 and MYB3R4. This hypothesis may also explain the differential effects of the myb3r1 myb3r4 double mutation on their potential target genes, by assuming differential contributions made by the redundant factor(s) on each target gene. Arabidopsis contains three other R1R2R3-Myb genes, MYB3R2, MYB3R3 and MYB3R5,which, however, may not be simply functionally redundant with MYB3R1and MYB3R4, because their mutations do not enhance the cytokinesis defect of myb3r4 single mutant and myb3r1 myb3r4 double mutant (M.I., unpublished). Candidates for such redundant factor(s) might be a TCP family transcription factor (Li et al., 2005) and a Myb domain-containing protein(Planchais et al., 2002), both of which bind to the promoter of the CYCB1;1 gene. Identification and functional characterization of the redundant activator(s), and possibly genetic studies on MYB3R2, MYB3R3 and MYB3R5 genes as well,would unravel the mechanisms that regulate gene expression during the G2/M phase in Arabidopsis.

We thank Miki Yosioka and Hiro Iguchi for their technical assistance;Masami Sekine and Hirokazu Tanaka for technical advice; the SALK Institute and the ABRC for T-DNA insertion lines; and Yasunori Machida for helpful discussion. This work was supported in part by a Grant-in-Aid for Scientific Research onPriority Areas (grant No. 17027015; Molecular Basis of Axis and Signals in Plant Development) from the Ministry of Education, Culture, Sports and Technology, Japan to M.I. and by the Nissan Science Foundation to M.I.