The AtNACK1/HINKEL and STUD/TETRASPORE/AtNACK2 genes, which encode functionally redundant kinesins, are essential for cytokinesis in Arabidopsis


  • Hirokazu Tanaka,

    1. College of Bioscience and Biotechnology, Chubu University and CREST, Japan Science and Technology Corporation, 1200 Matsumoto-cho, Kasugai, Aichi 487-8501, Japan
    Search for more papers by this author
  • Masaki Ishikawa,

    1. Division of Biological Science, Graduate School of Science, Nagoya University, Chikusa-ku, Nagoya 464-8602, Japan
    Search for more papers by this author
    • a

      Present address: Laboratory of Plant Molecular Biology, The Rockefeller University, 1230 York Avenue, New York, NY 10021, USA

  • Saori Kitamura,

    1. Division of Biological Science, Graduate School of Science, Nagoya University, Chikusa-ku, Nagoya 464-8602, Japan
    Search for more papers by this author
  • Yuji Takahashi,

    1. Division of Biological Science, Graduate School of Science, Nagoya University, Chikusa-ku, Nagoya 464-8602, Japan
    Search for more papers by this author
  • Takashi Soyano,

    1. Division of Biological Science, Graduate School of Science, Nagoya University, Chikusa-ku, Nagoya 464-8602, Japan
    Search for more papers by this author
    • a

      Present address: Laboratory of Plant Molecular Biology, The Rockefeller University, 1230 York Avenue, New York, NY 10021, USA

  • Chiyoko Machida,

    1. College of Bioscience and Biotechnology, Chubu University and CREST, Japan Science and Technology Corporation, 1200 Matsumoto-cho, Kasugai, Aichi 487-8501, Japan
    Search for more papers by this author
  • Yasunori Machida

    Corresponding author
    1. Division of Biological Science, Graduate School of Science, Nagoya University, Chikusa-ku, Nagoya 464-8602, Japan
    Search for more papers by this author

  • Communicated by: Kiyotaka Okada

Correspondence: E-mail:


Cytokinesis is the critical step during which daughter cells are separated. We showed previously that a protein complex that consists of NACK1 (and NACK2) kinesin-like protein and NPK1 MAPKKK and its substrate NQK1 MAPKK are required for progression of cytokinesis in Nicotiana tabacum. The genome of Arabidopsis thaliana encodes homologues of NACK1 and NACK2, namely, AtNACK1/HINKEL and STUD/TETRASPORE/AtNACK2, respectively. Loss-of-function mutations in AtNACK1/HINKEL and STUD/TETRASPORE/AtNACK2 result in the occasional failure of somatic and male-meiotic cytokinesis, respectively. However, it is likely that these genes function redundantly to some extent in somatic tissues and female gametogenesis. We describe the phenotypes of Arabidopsis plants that have mutations in both the AtNACK1/HINKEL and STUD/TETRASPORE/AtNACK2 genes. These phenotypes suggest that the two genes are essential during both male and female gametogenesis. Female gametes with atnack1 atnack2 double mutations failed to cellularize and to generate a central cell, synergids and the egg cells. Male gametes with atnack1 atnack2 mutations were also not transmitted to the next generation. The AtNACK1/HINKEL and STUD/TETRASPORE/AtNACK2 genes for kinesin-like proteins have overlapping functions that are essential for gametogenetic cytokinesis. They appear to be essential components of a MAP kinase cascade that promotes cytokinesis of plant cells in both gametophytic (haploid) and sporophytic (diploid) proliferation.


Cytokinesis is the critical step in the cell division cycle during which the mother cell divides into two daughter cells. In plants, cytokinesis includes formation of specific arrays of microtubules (e.g. the phragmoplast in somatic cells) at the equatorial region between two daughter nuclei, fusion of Golgi-derived vesicles at the equatorial region, and expansion of the phragmoplast toward the parental cell walls, with eventual formation of the cross walls known as cell plates.

Several of the factors involved in vesicle fusion during cytokinesis in plants have been isolated (Nacry et al. 2000). The KNOLLE protein of Arabidopsis is a syntaxin-related protein that is involved in the fusion of Golgi-derived vesicles that results in formation of the cell plate during cytokinesis and it is concentrated at cell plates (Lukowitz et al. 1996; Waizenegger et al. 2000). Other factors, including KEULE, dynamin-related proteins (ADL1A and ADL1E) and KORRIGAN, appear also to be involved in vesicle fusion and the formation of cell plates (Waizenegger et al. 2000; Zuo et al. 2000; Kang et al. 2003). Homologues of some of these proteins are known to be involved in cytokinesis of animal cells (Bowerman & Severson 1999; Thompson et al. 2002).

The Golgi-derived vesicles are transported via microtubules in the phragmoplast, which is the cytokinetic apparatus of plant cells that expands centrifugally in the region between the two daughter nuclei. Both genetic and physiological evidence suggests that appropriate organization and/or dynamics of microtubules are critical for the progression of cytokinesis (Nishihama & Machida 2001; Mayer & Jürgens 2002). The gametophytic gemini pollen1 (gem1) mutation of Arabidopsis causes defects in cytokinesis during pollen mitosis I to varying extents (Park et al. 1998; Twell et al. 2002). The GEM1 gene, which is identical to the MICROTUBULE ORGANIZATION 1 gene (MOR1; Whittington et al. 2001), encodes a protein that is related to the MAP215 family of microtubule-associated proteins found in all eukaryotes. MOR1/GEM1 appears to associate with the microtubules of the phragmoplast, as well as with other types of microtubules, and to be involved in cytokinesis (Twell et al. 2002). Treatment of plant cells with reagents that affect the organization of microtubules interrupts the expansion of the phragmoplast (Nishihama & Machida 2001).

The gene for NPK1 (nucleus- and phragmoplast-localized protein kinase 1) of tobacco encodes a member of the mitogen-activated protein kinase kinase kinase (MAPKKK) family (Banno et al. 1993; Nishihama et al. 1997; Machida et al. 1998; Nishihama & Machida 2001), which is preferentially expressed in tissues that contain proliferating cells (Nakashima et al. 1998). The NPK1 protein and its homologues appear to regulate the lateral expansion of the phragmoplast and the cell plate during cytokinesis (Nishihama et al. 2001; Jin et al. 2002; Krysan et al. 2002). The kinase activity of NPK1 increases during late M phase and NPK1 is localized at the leading edge of the equatorial zone of the phragmoplast during cytokinesis. Over-expression of a kinase-defective mutant form of NPK1 in tobacco cells results in inhibition of lateral expansion of the phragmoplast and formation of multinucleated cells with incomplete cell walls (Nishihama et al. 2001). Such over-expression of mutant NPK1 and silencing of the NPK1 gene in tobacco plants result in the formation of multinucleated cells with incomplete cell walls (Nishihama et al. 2001; Jin et al. 2002). Arabidopsis has three homologs of NPK1 (ANP1, ANP2 and ANP3; Nishihama et al. 1997). Loss of function of two of these three Arabidopsis homologues of NPK1 (e.g. ANP2 and ANP3) also results in formation of multinucleated cells with incomplete cell walls during vegetative growth (Krysan et al. 2002).

We have identified two M-phase-specific kinesin-like proteins in tobacco, designated NPK1-activating kinesin 1 (NACK1) and NACK2, which interact with NPK1 to increase its kinase activity (Ito et al. 1998; Nishihama et al. 2002). NACK1 is co-localized with NPK1 at the phragmoplast equator (Nishihama et al. 2002). Over-expression in tobacco cells of a mutant NACK1 protein that lacks the putative motor region results in failure of NPK1 to become concentrated at the phragmoplast equator and cytokinetic defects (Nishihama et al. 2002), suggesting that NACK1 might play an important role in both the subcellular localization and the activation of NPK1 MAPKKK for the positive regulation of cytokinesis.

Recently, we identified the NQK1/NtMEK1 and NRK1 proteins of tobacco as a MAPKK and MAPK, respectively, that act downstream of NPK1 (Soyano et al. 2003). NPK1 phosphorylates and activates NQK1/NtMEK1, which in turn activates NRK1 by phosphorylation. Analysis of the effects on cell division of over-expression of a kinase-defective mutant form of NQK1 and mutations in an Arabidopsis homologue of NQK1, ANQ1, showed that the NQK1 gene and related genes are required for the formation of the cell plate (Soyano et al. 2003). The activities of NPK1, NQK1 and NRK1 in tobacco cells increase at the late M phase of the cell cycle and decrease after the M phase (Nishihama et al. 2001; Soyano et al. 2003). These results led us to suggest the presence of a MAPK cascade, designated the NACK-PQR pathway (Soyano et al. 2003), which is activated by the interaction between NACK1 (NACK2) and NPK1 MAPKKK and positively regulates the formation of the cell plate.

The homologues of the NACK1 and NACK2 genes in Arabidopsis are designated AtNACK1 (At1g18370; Nishihama et al. 2002) and AtNACK2 (At3g43210; Nishihama et al. 2002), respectively, and they are identical to HINKEL (HIK) and STUD (STD)/TETRASPORE (TES), respectively (Strompen et al. 2002; Yang et al. 2003). Loss-of-function mutations in the AtNACK1/HIK gene result in malformed embryos and seedlings that contain abnormally large cells with cell wall stubs and multiple nuclei. These seedlings exhibit severe dwarfism and their growth ceases before bolting (Strompen et al. 2002; Nishihama et al. 2002). These results suggest that the AtNACK1/HIK gene might be involved in the progression of cytokinesis during embryogenesis and in the vegetative growth of Arabidopsis plants, even though it might not be essential during these growth stages. Plants with mutations in STD/TES/AtNACK2 exhibit normal features during vegetative and reproductive growth but cytokinetic defects become apparent during meiosis that normally results in the formation of microspores and abnormally large pollen grains are generated (Hülskamp et al. 1997; Spielman et al. 1997; Yang et al. 2003; our unpublished observations), suggesting that, while STD/TES/AtNACK2 might be essential for cytokinesis during the formation of microspores, it might not play a major role during vegetative and reproductive growth. However, since STD/TES/AtNACK2 transcripts can be detected at a relatively low level during these latter growth stages, as well as during meiosis (Yang et al. 2003; our unpublished observations) and since growth of atnack1/hik mutants is retarded, with abnormal morphology, STD/TES/AtNACK2 and AtNACK1/HIK might be functionally redundant during cytokinesis at certain stages of the growth of Arabidopsis plants. Indeed, since tobacco NACK2 and NACK1 activate NPK1 MAPKKK in a similar manner (Nishihama et al. 2002), it seems likely that NACK1-related kinesin-like proteins in both tobacco and Arabidopsis might be functionally redundant.

As described below, we attempted to make atnack1 atnack2 double mutants to examine the possible functional redundancy of these kinesin-like proteins in Arabidopsis. However, we failed to isolate plants that were homozygous for both mutations. Instead, we observed abnormal megagametogenesis in plants that were doubly heterozygous for atnack1 and atnack2 mutations. The atnack1 and atnack2 mutations affected cellularization (cytokinesis) synergistically during megagametogenesis, with the generation of abortive gametes. Considered together with the previously described functions of NACK1 and NACK2 in tobacco, these results indicate that the products of the AtNACK1/HIK and STD/TES/AtNACK2 genes are functionally redundant but are essential for promotion of the progression of cytokinesis in Arabidopsis.


Isolation of T-DNA insertion alleles of the AtNACK2 gene

In an attempt to clarify the function of AtNACK2 in Arabidopsis, we performed PCR-based screening and a database search of populations of T-DNA insertion lines and isolated two mutant lines with insertions of T-DNA in the AtNACK2 gene. We isolated AtNACK2 cDNA by RT-PCR and determined the nucleotide sequence of the cDNA (Nishihama et al. 2002; GenBank accession no. AB088121). Comparison of the sequences of the AtNACK2 cDNA and the corresponding genomic DNA revealed that the AtNACK2-coding region consisted of 14 exons (Fig. 1A). T-DNA sequences in the atnack2-1 and atnack2-2 alleles were inserted in the ninth intron and the twelfth exon, respectively (Fig. 1A). The AtNACK2 gene was shown recently by Yang et al. (2003) to be identical to the STUD/TETRASPORE (STD/TES) gene.

Figure 1.

Morphological defects due to atnack2 alleles with T-DNA insertions. (A) The sites of insertion of the T-DNA in the two atnack2 mutants. The filled boxes indicate exons of the AtNACK2 gene, as deduced from a cDNA sequence (GenBank accession no. AB088121). Positions of T-DNA in atnack2 alleles are indicated by triangles. T-DNAs are not drawn to scale. Exons that encode a putative motor domain, a putative ATP-binding site and a putative NPK1-binding site, as predicted from a comparison with tobacco NACK1, are indicated. (B) Gross morphology of siliques 6 days after flowering (DAF). Siliques were shorter on plants homozygous for the atnack2-1 mutation than on wild-type plants and heterozygotes. (C) Developing seeds in a silique of a wild-type plant 6 DAF. (D) Developing seeds in siliques of an atnack2-1 plant with many unfertilized ovules. (E–J) Gross morphology of pollen (E, F, H, I) and the distribution of nuclei (G, J) of wild-type (E–G) and atnack2-1/atnack2-1 (H–J) plants. (E, H) Scanning electron micrographs of mature pollen. (F, I) Differential interference contrast (DIC) images. (G, J) Fluorescence images of the DAPI-stained pollen shown in (F) and (I), respectively. Arrowheads indicate positions of generative nuclei. (K–N) Sections of developing pollen obtained from wild-type (K, L) and atnack2-1 (M, N) plants. (L, N) Magnified views of boxed areas of (K) and (M), respectively. An analysis of serial sections revealed the presence of four nuclei in the large pollen grain in (N) even though only two nuclei are visible in this section (N). Scale bars: 1 mm for (B)–(D); 20 µm for (E)–(J); 50 µm for (K) and (M); and 10 µm for (L) and (N).

Production by atnack2 plants of enlarged pollen with incomplete cell walls during male gametogenesis

Unlike the morphology of atnack1/hik mutants, the morphology of atnack2 mutant plants was indistinguishable from that of wild-type plants during vegetative development (data not shown). The atnack2 plants produced flowers with normal morphology (data not shown). However, the siliques of atnack2 plants were shorter than those of wild-type plants (Fig. 1B) and their fertility was greatly reduced (Fig. 1D). Since atnack2-1 and atnack2-2 plants exhibited similar defects in terms of reduced fertility and deformation of pollen grains (see below), we focused our detailed phenotypic characterization on the atnack2-1 allele. Reciprocal crosses between wild-type and atnack2-1 plants revealed that the atnack2-1 plants were male-sterile (data not shown). Therefore, we examined the morphology of the pollen from atnack2-1 homozygous plants. Mature pollen grains from wild-type plants were ovoid (Fig. 1E) and contained two generative nuclei (Fig. 1G, arrowheads) and a vegetative nucleus (Fig. 1G, arrow). The pollen grains produced by atnack2 plants were larger and nearly spherical (Fig. 1H). Enlarged pollen grains from atnack2 plants often contained more than two generative nuclei (Fig. 1J, arrowheads), suggesting that cytokinesis and the subsequent separation of each pollen cell after pollen meiosis II had not been completed. In fact, atnack2-1 parents produced tetraspore pollen (multinucleated microspores; Fig. 1M,N). These defects were not observed in pollen produced by atnack2/+parents. Taken together, the results suggest that the AtNACK2 gene is required sporophytically for cytokinesis during male gametogenesis.

Patterns of expression of AtNACK1/HIK and STD/TES/AtNACK2 genes are similar but distinct

We next examined the patterns of accumulation of transcripts of the AtNACK1/HIK and STD/TES/AtNACK2 genes (Fig. 2). The results of semiquantitative RT-PCR indicated that, while both the AtNACK1/HIK and the STD/TES/AtNACK2 gene were expressed in tissues that contained proliferating cells, transcripts of the AtNACK1/HIK gene were more abundant in young plants than those of the STD/TES/AtNACK2 gene (Fig. 2A). As reported previously (Strompen et al. 2002), AtNACK1/HIK mRNA accumulated in a patchy pattern during embryogenesis (Fig. 2B). A similar patchy distribution of the AtNACK1/HIK transcript was observed in floral organs that contained dividing cells (Fig. 2D). STD/TES/AtNACK2 mRNA was detected in embryos with a similar distribution (Fig. 2C), but the signals were much weaker than those due to AtNACK1/HIK mRNA (Fig. 2B). A patchy pattern of signals is the hallmark of the accumulation of transcripts of genes that are expressed at specific stages of the cell cycle, such as genes for NACK1 and cyclins, in regions of shoot apical meristems (Fobert et al. 1994; Ito et al. 2001; our unpublished data). In flower buds, STD/TES/AtNACK2 transcripts were detected in the inner regions of developing anthers that contained pollen mother cells (Fig. 2E). Thus pattern of specific accumulation of STD/TES/AtNACK2 transcripts was consistent with the appearance of a major defect due to atnack2-1 during male gametogenesis.

Figure 2.

Accumulation of AtNACK1 and AtNACK2 RNAs in wild-type tissues. (A) Accumulation of AtNACK1 and AtNACK2 RNAs in various tissues. Semiquantitative RT-PCR was performed using mRNA from various tissues as indicated. Primers specific to the gene for α-tubulin (TUBA) were used as controls. Numbers of cycles of PCR reaction were indicated. Abbreviations are as follows: IF, inflorescences; FB, flower buds. (B, C) Accumulation of AtNACK1 (B) and AtNACK2 (C) RNAs in heart-stage embryos. (D, E) Accumulation of AtNACK1 (D) and AtNACK2 (E) RNAs in flowers. Arrows indicate the AtNACK2 transcripts that strongly accumulated at high levels in pollen mother cells. Scale bars: 20 µm for (B–E).

The products of the AtNACK1/HIK and STD/TES/AtNACK2 genes are functionally redundant and essential for male and female gametogenesis

As described above, we postulated that the products of the AtNACK1/HIK and STD/TES/AtNACK2 genes might function redundantly to promote cytokinesis in certain organs. To investigate this possibility, we generated atnack1-1/+; atnack2-1/+ double heterozygotes and determined the genotypes of 196 self-pollinated progeny by examining the presence or the absence of the T-DNAs that caused the respective mutations. As shown in Table 1, we failed to obtain any atnack1-1 atnack2-1 double mutants or any plants that were homozygous for one mutation and heterozygous for the other. We obtained similar results when we examined the progeny of an atnack1-2/+; atnack2-1/+plant (Table 1). These results suggested that both AtNACK1/HIK and STD/TES/AtNACK2 might be essential for the viability of gametophytes and/or embryos. To determine whether atnack1 atnack2 gametes are viable, we performed reciprocal crosses between an atnack1-1/+; atnack2-1/+ double heterozygote and a wild-type parent (Table 2). When we used mutant pollen from the heterozygous atnack1-1 or atnack2-1 parent for crosses to a wild-type female parent, mutant alleles were transmitted to approximately 50% of the offspring, as expected (Table 2), indicating that male gametophytes with an atnack1-1 or atnack2-1 single mutation were viable. However, when we crossed the doubly heterozygous parents with wild-type plants, male and female gametes with both mutations did not participate in the development of the next generation (Table 2). These results suggested that 25% of ovules and pollen produced by atnack1/+; atnack2/+plants, with gametes that carried both the atnack1 and the atnack2 mutations, were not viable.

Table 1.  The progeny of atnack1/AtNACK1;atnack2/AtNACK2 parents
GenotypeNumber observedNumber expected*
  • *

    Expected values were calculated using the observed frequencies of alleles in control experiments with the self progeny of each heterozygous plant.

  • †,‡

    The genotypes of 196 (b) and 202 (c) progeny of plants with the respective indicated genotypes were determined by PCR.

  • Abbreviations are as follows: Wt, wild-type plants; Het, heterozygotes; Mut, plants homozygous for indicated mutant alleles.

Progeny of an atnack1-1/AtNACK1;atnack2-1/AtNACK2 parent
WtWt2613 (6.7%)
WtHet5126 (13.3%)
WtMut2313 (6.7%)
HetWt4425 (12.5%)
HetHet4849 (25.0%)
HetMut 025 (12.5%)
MutWt 412 (5.9%)
MutHet 023 (11.7%)
MutMut 012 (5.9%)
Progeny of an atnack1-2/AtNACK1;atnack2-1/AtNACK2 parent
WtWt2613 (6.6%)
WtHet4527 (13.2%)
WtMut2013 (6.6%)
HetWt5725 (12.5%)
HetHet4051 (25.0%)
HetMut 025 (12.5%)
MutWt1412 (5.9%)
MutHet 024 (11.8%)
MutMut 012 (5.9%)
Table 2.  Results of reciprocal crosses between atnack mutants and wild-type plants
CrossGamete frequency
AtNACK1 AtNACK2atnack1AtNACK2AtNACK1atnack2atnack1atnack2P-value
  • P values, as calculated by the chi-squared test, are indicated. P values indicating significant differences are indicated by an asterisk.

  • We determined the genotypes of 121 progeny from reciprocal crosses between atnack1/AtNACK1;atnack2/AtNACK2 and the wild type in order to infer the genotype of the gamete contributed by the mutant parent in each cross. The actual number of gametes observed for each genotype is shown. ‘Expected 1’ indicates the number of gametes expected when all four possible gametes are equally viable. ‘Expected 2’ indicates the number of gametes expected when the atnack1 atnack2 gametes are not viable.

  • §

    The progeny from crosses between the indicated heterozygous plants and the wild type were examined as controls.

Female mutant: atnack1-1/AtNACK1;atnack2-1/AtNACK2
 Observed21922 0 
 Expected 113131313<0.0001*
 Expected 2171717 0  0.1
Male mutant: atnack1-1/AtNACK1;atnack2-1/AtNACK2
 Observed272418 0 
 Expected 117171717<0.0001*
 Expected 2232323 0  0.6
Male mutant: atnack1-1/AtNACK1§
Male mutant: atnack2-1/AtNACK2§

The AtNACK1/HIK and STD/TES/AtNACK2 genes are essential for cellularization during megagametogenesis

We investigated the number and shape of nuclei in embryo sacs during female gametogenesis in atnack1/+; atnack2/+plants. A mature wild-type ovule contains an embryo sac, which usually contains four cells, namely, one central cell, one egg cell, and two synergid cells (Christensen et al. 1997; Drews et al. 1998). These cells are formed via simultaneous cellularization of a syncytium with eight nuclei. Two polar nuclei fuse to form a large central-cell nucleus and three anti-podal cells degenerate before maturation (Christensen et al. 1997; Drews et al. 1998).

Figure 3 and Table 3 show that approximately 25% of maturing ovules produced by atnack1-1/+; atnack2-1/+ plants contained multinucleated (3–7 nuclei) embryo sacs that contained no typical egg cells and synergid cells (Fig. 3D,H), although other ovules had those with four normal nuclei (Fig. 3A,E; Table 3). Such unusual embryo sacs often included five, six or seven nuclei in gynoecia that contained normal ovules just before maturation (Fig. 3C,G; Table 3, stage FG5). After wild-type siblings had matured, abnormal embryo sacs with three, four or five nuclei were frequently observed (Table 3, stages FG6 to FG7), indicating that the number of nuclei in the mutant embryo sacs decreased during the maturation of ovules. Unlike those in wild-type embryo sacs, the nuclei in the abnormal embryo sacs were often uniform in terms of size (Fig. 3G,H), which suggested that the nuclei in embryo sacs with both atnack1 and atnack2 mutations failed to follow the correct path to differentiation.

Figure 3.

The atnack1 atnack2 embryo sacs are defective in positioning of nuclei. (A–D) Whole views of cleared ovules. (E–H) Magnified views of boxed regions in (A–D), respectively. (A, E) Terminal phenotype of the wild-type female gametophyte. The embryo sac contains a large central-cell nucleus and a small egg nucleus, as indicated by arrows in (E). (B, F) An ovule from an atnack1-1/AtNACK1 plant, showing a rare morphological abnormality. (C, G) An ovule before maturation from an atnack1-1/AtNACK1; atnack2-1/AtNACK2 parent. Note that nuclei of the relatively uniform size are concentrated in the micropylar region (indicated by arrows in G). (D, H) A mature ovule from an atnack1-1/AtNACK1; atnack2-1/AtNACK2 plant with four large nuclei (arrows in H). The micropylar region lacks visible nuclei (D). (I–N) Confocal laser scanning microscopy of developing (I–L) and mature (M–O) ovules obtained from atnack1-3/+; atnack2-2/+ parents. (I–K) Wild-type siblings at the indicated developmental stages. The inset in (K) shows three anti-podal nuclei in the same ovule in a different focal plane. (L) An ovule obtained from a silique that contained mainly wild-type siblings at early to mid stage FG5, exhibiting the atnack1 atnack2 phenotype. Mature ovules (stage FG7) containing an embryo sac with a wild-type phenotype (M) and with the typical atnack1 atnack2 phenotype (N, O). Developmental stages of ovules (FG4 through FG7) were determined as described elsewhere (Christensen et al. 1997). The stages of development of ovules with a mutant phenotype were deduced from those of wild-type siblings in the same gynoecia. Scale bars: 50 µm for (A–D); 10 µm for (E–H); and 20 µm for (I–O).

Table 3.  Frequency of ovules with defects in the number and the position of nuclei in the embryo sac
ParentsDevelopmental stageNumber of ovulesNumber of central-cell nuclei
  • Experiment 1: Gynoecia before pollination were collected from the indicated parents. Ovules were cleared and numbers of central-cell nuclei were determined. Numbers of mutant ovules with abnormally large numbers of central-cell nuclei are shown in bold type. Typical egg and synergid cells were not observed in these ovules.

  • Experiment 2: Mature ovules (stage FG7) were observed by confocal laser scanning microscopy (see Experimental procedures). Numbers of ovules with the typical mutant phenotype are shown in bold type.

  • *

    Gynoecia that contained mainly wild-type ovules with two unfused central-cell nuclei (late stage FG5) were counted.

  • Gynoecia that contained mainly mature wild-type ovules with a single central-cell nucleus (stage FG6 and stage FG7) were counted.

  • P values were calculated by the χ2-test when 25% of ovules (e.g. atnack1 atnack2 double-mutant ovules) were expected to exhibit the mutant phenotype.

  • §

    P values were calculated when 50% of ovules (e.g. atnack1 ovules) were expected to exhibit the mutant phenotype.

Experiment 1
 Ws FG6–7 (n= 111) 8 8518 0 00 00
 atnack1-1/AtNACK1;FG5 *(n= 80) 2 1346 1 (1.3%) 03 (3.8%)13 (16%)2 (2.5%)   0.95
 atnack2-1/AtNACK2FG6–7 (n= 92) 4 64 3 2 (2.2%)14 (15%)3 (3.3%) 2 (2.2%)0   0.6
 atnack1-1/AtNACK1FG6–7 (n= 249)2420121 0 1 (0.4%)0 2 (1%)0<0.0001§
 atnack2-1/atnack2-1FG6–7 (n= 190)1416510 0 00 01 (0.5%)
Experiment 2
 Col-0 FG7(n= 124)30 94 0 0 00 00
 atnack2-2/AtNACK2FG7(n= 107) 5 7312 (11%)16 (15%) 1 (0.9%)0 00   0.6
 atnack1-3/AtNACK1FG7(n= 120)23 95 2 (1.7%) 0 00 00<0.0001§
 atnack2-2/AtNACK2FG7(n= 97)18 79 0 0 00 00

To confirm that loss of AtNACK1 and AtNACK2 functions were responsible for the morphological defects, we generated double heterozygous plants that carried different alleles (atnack1-3/+; atnack2-2/+plants) and examined their phenotypes. We observed morphological defects similar to those described above (e.g. absence of typical egg and synergid cells) in approximately 25% of ovules (Table 3, experiment 2), confirming that the atnack1 and atnack2 mutations were responsible for the fate of megagametophytes. Ovules produced by atnack1/+ or atnack2/atnack2 plants exhibited similar defects but at a much lower frequency (Fig. 3B,F; Table 3).

To gain some insight into the developmental basis for the generation of the unusual embryo sacs, we followed the course of megagametogenesis in both wild-type and the doubly heterozygous plants. In wild-type ovules at stage FG4 (Fig. 3I), two nuclei were localized at the chalazal pole and the other two nuclei were located at the micropylar pole. Subsequently, at the early FG5 stage (Fig. 3J), nuclear division occurred without cytokinesis (cellularization) to generate a syncytium that contained eight nuclei, four of which were located in the chalazal region and four of which were located in the micropylar region. Prior to the mid FG5 stage, one nucleus migrated from the chalazal region to the micropylar region (Fig. 3K). After cellularization, three nuclei in the chalazal region (anti-podal-cell nuclei) degenerated, and a typical four-celled mature embryo sac remained (Fig. 3M). Development of atnack1 atnack2 embryo sacs deviated from that of their wild-type siblings as early as stage FG5 and not only at the mature stage, FG7 (Fig. 3L,N,O). There was a cluster of nuclei at the centre of each mutant embryo sac, suggesting that the positioning of nuclei was affected by the two mutations before (or just after) cellularization occurred in wild-type siblings.

To examine the cross walls in mature embryo sacs, we examined serial sections of mature ovules that were obtained from atnack1/+; atnack2/+ plants. A central cell, an egg cell, and synergid cells were apparent in embryo sacs with a wild-type phenotype (Fig. 4A,D,G). Typical cross walls between these cells were apparent in the wild-type (Fig. 4D, arrows). By contrast, there were no cross walls between the nuclei in mutant embryo sacs (Fig. 4B,C,E,F,H).

Figure 4.

The atnack1 atnack2 embryo sac failed to form typical cross walls between nuclei. (A–C) Sections of mature ovules obtained from atnack1-1/+; atnack2-1/+ parents, showing the wild-type phenotype (A) and two representative sections from a single ovule with the atnack1 atnack2 phenotype (B, C). Magnified views of the micropylar side of embryo sacs in (A–C) are shown in (D–F). Arrows in (D) indicate densely stained cross walls. Arrowheads in (D–F) indicate positions of nuclei. (G, H) Schematic representation of a normal embryo sac (G) with cross walls (red lines) and a mutant embryo sac with both the atnack1 and the atnack2 mutation. The mutant embryo sac contains two nuclei without any apparent cross walls between them (H). Bars: 50 µm for (A–C); and 20 µm for (D–F).


Our observations of cleared or sectioned ovules revealed that a typical egg cell and synergid cells failed to develop in approximately 25% of ovules generated by atnack1/+; atnack2/+plants carrying independently isolated mutant alleles, suggesting that embryo sacs with both mutations had reproducible morphological defects (Table 1; Fig. 3). These defects should, at least in part, explain the failure in the participation by female gametes with atnack1 atnack2 double mutations in the next generation, since egg and synergid cells are critical for the function of the female gametophyte. The synergid cells are essential for guidance of pollen tubes (Higashiyama et al. 2001) and the egg cell develops into the embryo after the double fertilization (Goldberg et al. 1994). Morphological abnormalities, such as the absence of obvious cross walls in the embryo sacs generated by atnack1/+; atnack2/+plants, suggest that both AtNACK1/HIK and STD/TES/AtNACK2 proteins are essential for cellularization during megagametogenesis.

The AtNACK1/HIK and STD/TES/AtNACK2 genes function redundantly during production of functional gametophytes

The atnack2 homozygous mutant exhibited defects only in cytokinesis during male gametogenesis (Fig. 1 and our unpublished observation). However, transcripts of the STD/TES/AtNACK2 gene accumulated in developing wild-type embryos and seedlings, as well as in the pollen mother cells (Fig. 2), implying that STD/TES/AtNACK2 functions in tissues other than pollen, a phenomenon that was not revealed by analysis of the single mutant.

The STD/TES/AtNACK2 protein is most closely related to the NACK2 kinesin-like protein of tobacco (Nishihama et al. 2002). AtNACK1/HIK and tobacco NACK1 are also similar, in terms of predicted amino acid sequences, to STD/TES/AtNACK2 and tobacco NACK2, respectively, although the relationship between NACK1-related proteins and NACK2-related proteins is relatively weak (Nishihama et al. 2002). Both NACK2 and NACK1 can activate NPK1 MAPKKK by binding to this latter protein. Mutations in AtNACK1/HIK result in partial defects in cytokinesis in embryos and cotyledons and lead to growth arrest of Arabidopsis plants before bolting (Strompen et al. 2002; Nishihama et al. 2002). These observations suggested that STD/TES/AtNACK2 and AtNACK1/HIK might be redundant in terms of cytokinetic functions during female gametogenesis, embryogenesis and the vegetative growth stage (Nishihama et al. 2002). Our genetic analysis showed that both male and female gametes that lacked the wild-type alleles of both of these genes (e.g. atnack1-1 atnack2-1) did not participate in development of the next generation (Table 2). However, each single-mutant gamete was basically functional (Table 2). Therefore, it seems likely that the AtNACK1/HIK and STD//TES AtNACK2 proteins have redundant functions in a process that is required for the viability of both male and female gametophytes.

AtNACK1/HIK and STD/TES/AtNACK2 are involved in the distribution of nuclei and cellularization during megagametogenesis

We found that, in the embryo sac of 25% of ovules from atnack1/+; atnack2/+ plants, the cells in the micropylar region occasionally lacked nuclei (Fig. 3). In addition, both the size and the positioning of the nuclei were abnormal. Given that cellularization occurs just prior to maturation stage (i.e. stage FG5) to generate the differentiated egg cell, synergid cells, and central cells in the wild-type (Christensen et al. 1997; Drews et al. 1998), we can reasonably speculate that the processes of development of these cells are tightly coupled and, indeed, our genetic analysis demonstrated a clear genetic link between cellularization and cell differentiation. There are at least two possible explanations, with respect to the roles of AtNACK1 and AtNACK2, for the molecular link between cellularization and the differentiation of cells in the embryo sac. One possibility is that the cellularization is a prerequisite for the subsequent differentiation of cells (Fig. 5, model 1). In this case, AtNACKs might act directly to promote cellularization and the nascent cell boundary-related structure (e.g. cell plate-related vesicles) might be generated in an AtNACK1- and AtNACK2-dependent manner during stage FG5, thereby ensuring the appropriate movement of nuclei and the subsequent differentiation of cells. This assumption is in agreement with the hypothesis, proposed by Christensen et al. (1997) that the cellularization begins at stage early FG5. In addition to a role in generating cross walls between cells, AtNACK1 and AtNACK2 (or their downstream factors) might act directly in the differentiation of cells and/or the appropriate positioning of nuclei, independently of the promotion of cytokinesis (Fig. 5, model 2). Supporting this model, a defect in nuclear positioning in atnack1 atnack2 embryo sac was observed as early as stage FG5, at which cellularization might have not been completed in wild-type siblings. Additional evidence also supports this latter explanation. Two gametophytic mutations in Arabidopsis, gfa3 and gfa7, affect cellularization and the fusion of polar nuclei during megagametogenesis (Christensen et al. 1998). Furthermore, the positioning of nuclei is relatively normal in mutant embryo sacs from gfa3/+ and gfa7/+ plants, in spite of the defect in cellularization (Christensen et al. 1998). Further detailed phenotypic analysis of these mutants and isolation of the corresponding genes that are involved in cellularization of the embryo sac might help us to clarify the relationships among the regulation of cellularization, the positioning of nuclei and the fusion of nuclei.

Figure 5.

A diagram showing the proposed roles of the AtNACK1/HIK and STD/TES/AtNACK2 kinesin-like proteins in cellularization and the localization of nuclei during megagametogenesis. The diagram shows developing embryo sacs (large ellipses) and their nuclei (small circles). The AtNACK1 and AtNACK2 proteins might act to regulate, positively, the formation of a cell boundary-related structure (shown in red) that restricts localization of nuclei during stage FG5 (model 1). These kinesin-like proteins might also act to inhibit potential mislocalization of the nuclei (blue) that will become the nuclei of the anti-podal cells, egg, and synergids (model 2).

Possible targets of AtNACK1/HIK and STD/TES/AtNACK2 during gametogenesis

Our analysis indicated that AtNACK1/HIK and STD/TES/AtNACK2 function redundantly to promote cytokinesis during megagametogenesis. The results of reciprocal crosses between an atnack1/+; atnack2/+ parent and a wild-type parent suggested that AtNACK proteins might play a redundant roles in male gametogenesis although it remains unclear whether they are essential for cytokinesis during pollen mitosis I and/or pollen mitosis II. Since atnack1/hik mutants exhibit defects in cytokinesis in the diploid generation (Strompen et al. 2002; Nishihama et al. 2002), both gametophytic and sporophytic generations seem to involve a common molecular mechanism for cytokinesis. It remains to be determined whether common factors downstream of NACK-related kinesin-like proteins (NACK1, NACK2, AtNACK1/HIK and STD/TES/AtNACK2) are involved in the two generations. In tobacco, a signalling pathway involving the NACK1 protein (the NACK-PQR pathway), which positively regulates cytokinesis, has been identified (Nishihama et al. 2001, 2002; Soyano et al. 2003). This NACK-PQR pathway involves the localization of NACK1 and NPK1 MAPKKK to the phragmoplast equator, activation of NPK1 MAPKKK via interaction with NACK1 (Nishihama et al. 2002), activation by NPK1 of NQK1/NtMEK1 MAPKK via phosphorylation, and activation by active NQK1 of NRK1/NTF6 MAPK (Soyano et al. 2003). The NACK-PQR pathway appears to be conserved in tobacco and Arabidopsis because loss of function of the Arabidopsis homologues of the NPK1 and NQK1 genes (e.g. anp2 anp3 double mutant and the anq1 mutant), as well as atnack1 mutations, results in defects in cytokinesis in various tissues at the sporophytic generation (Krysan et al. 2002; Nishihama et al. 2002; Soyano et al. 2003).

The triple mutant combination of loss-of-function mutations in three ANP genes (anp1 anp2 anp3) that encode Arabidopsis homologues of NPK1 was not transmitted through either male or female gametes (Krysan et al. 2002). This result also indicates that the three ANP genes of Arabidopsis are functionally redundant and essential for formation of functional gametophytes. It will be of interest to determine whether the embryo sac of the anp1 anp2 anp3 triple mutant has a phenotype identical to that of the atnack1 atnack2 double mutant (e.g. defective cellularization and nuclear positioning during megagametogenesis).

Our data provide novel evidence for involvement of AtNACK1/HIK and STD/TES/AtNACK2 in cytokinesis during the gametophytic generation. They also suggest that these proteins might function mainly in somatic and meiotic cell division, respectively, even though they appear to function somewhat redundantly. Further genetic analysis of components of the NACK-PQR pathway might allows us to elucidate the mechanism that positively regulates cytokinesis in both the gametophytic and sporophytic generations.

Experimental procedures

Isolation of T-DNA insertion mutants

We screened 133 440 T-DNA-tagged Arabidopsis plants, generated at the Arabidopsis Knockout Facility of the University of Wisconsin (Krysan et al. 1999), for atnack2 mutations by PCR according to the instructions at and we identified an atnack2-1 allele. We used primers specific for the T-DNA left-border (JL-202, 5′-CATTTTATAATAACGCTGCGGACATCTAC-3′) and AtNACK2 (AtNA2-2F, 5′-AGAGTAGATTTGTAGCGTTTAGGTCTTAT-3′; and AtNA2-1R, 5′-CAGAGATTAGATGATCATATGGTTAGGAT-3′). The atnack2-2 (SALK_143396) and atnack1-3 (SALK_092154) mutants were identified in a population of T-DNA tagged Arabidopsis plants generated at the Salk Institute (Alonso et al. 2003). Insertion of the T-DNA in the atnack2-2 allele was verified by PCR using primers specific to AtNACK2 (AtNACK2-1249F, CAGTCTGAACTTGATCTAGAAAGA; and AtNACK2-2500R, GCTTCAAAGCTCTTATACTCGA) and a primer specific to the left border of T-DNA (LBb1, GTGGACCGCTTGCTGCAACT). A T-DNA sequence was inserted in the ninth exon in the atnack1-3 allele, as verified by PCR using LBb1 and a primer specific to the AtNACK1 sequence (ANK1-092154F, GCGGCCATATACCTTACAGA). The atnack1-3 mutant had a growth defect that was essentially identical to but slightly weaker than that of atnack1-1 and atnack1-2 plants.

Genetic analysis

Arabidopsis thaliana ecotypes Wassilewskija (Ws) and Columbia (Col-0) were used as the wild-type. Plants were grown as previously described (Tanaka et al. 2001). To determine genotypes of AtNACK1 and AtNACK2 loci, genomic DNA was isolated from individual plants and subjected to analysis by PCR with the following primers: for the atnack1-1 mutation, AtNA1-B34R2, ANK1-B34F2 and JL-270; for the atnack1-2 mutation, AtNA1-B5F2, AtNA1-B5R2 and JL-270; and for the atnack2-1 mutation, AtNA2-6F2, AtNA2-7R2 and JL-270. The JL-270 sequence is complementary to the sequence of the T-DNA left border and was used for detection of T-DNA insertion alleles. Nucleotide sequences of primers were as follows: JL-270, 5′-TTTCTCCATATTGACCATCATACTCATTG-3′; ANK1-B34F2, 5′-CGTGCCCTTAAACTCCTTGA-3′; AtNA1-B34R2, 5′-TTCCAGACCCCTGAAAGAGA-3′; AtNA1-B5F2, 5′-GAAAGACTACGATGCGCCAA-3′; AtNA1-B5R2, 5′-CTCCTCTGCTGCGTTTTGAA-3′; AtNA2-6F2, 5′-CGTAGTTTATTGACACTTACTACAGT-3′; and AtNA2-7R2, 5′-GCTACTTTTTGCTGCAGATGTTTCA-3′


For examinations by scanning electron microscopy (SEM) of the morphology of pollen grains, anthers were dissected from open flowers, frozen in liquid nitrogen and observed as previously described (Tanaka et al. 2001). For visualization of nuclei in pollen grains, pollen was processed essentially as described by Park et al. (1998). Open flowers were collected into a staining solution [0.1 m sodium phosphate (pH 7), 1 mm EDTA, 0.1% Triton X-100, 0.4 µg/mL 4′,6-diamidino-2-phenylindole (DAPI)]. After brief vortex mixing and centrifugation, pellets of pollen grains were transferred on to glass slides. Fluorescence from DAPI was observed under excitation by ultraviolet light using a fluorescence microscope (Axioplan 2; Carl Zeiss, Oberkochen, Germany) that was equipped with a cooled CCD camera system (Photometrics, Tucson, AZ, USA). Ovules were cleared essentially as described elsewhere (Aida et al. 1997). Gynoecia containing ovules at appropriate stages were cut from open flowers and carpels were slit to facilitate fixation. The dissected gynoecia were fixed overnight in a mixture of ethanol and acetic acid (9 : 1, v/v) at room temperature. After rehydration in a graded ethanol series (90%, 70%, 50% and 30%) for 20 min in each solution, gynoecia were cleared in the 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). Cleared ovules were dissected from gynoecia and observed under a microscope equipped with differential interference contrast (DIC) optics (Carl Zeiss). Photographs were taken with a CCD camera (AxioCam; Carl Zeiss). Nuclei were counted on a series of focal planes within each embryo sac. Ovules for observation under a confocal laser scanning microscope (CLSM) were prepared as described elsewhere (Christensen et al. 1997). Specimens were observed under a CLSM (LSM 510; Carl Zeiss) with excitation from an argon laser (488 nm) and an emission filter (505–550 nm). Developmental stages of ovules and gynoecia were determined as described elsewhere (Christensen et al. 1997). Samples for semithin sections were prepared essentially as previously described (Nishimura et al. 1993). Gynoecia were dissected from flowers and carpels were partially slit and vacuum-infiltrated with fixative [4% paraformaldehyde, 1% glutaraldehyde and 0.06 m sucrose in 0.05 m cacodylate buffer (pH 7.4)]. They were fixed for 2 h at room temperature and then postfixed in 1.5% osmium tetroxide in cacodylate buffer (0.15 m) for 4 h at 4 °C. The samples were dehydrated in a graded ethanol series, treated with propylene oxide for 1 h and infiltrated with a mixture of propylene oxide and Epon resin (1 : 1, v/v) for 1 h at room temperature. The samples were then embedded in Epon resin, which was allowed to polymerize at 60 °C for 72 h. Semi-thin sections of 350–500 nm thickness were cut with a diamond knife (Sumi knife XAC; Sumitomo Electric Industries, Ltd, Osaka, Japan) on an ultramicrotome (Ultracut UCT, Leica Microsystems, Inc., Tokyo, Japan), transferred to glass slides and stained with a 0.05% aqueous solution of toluidine blue for 1 min. Sections were observed and photographed under a microscope (Axioplan 2; Carl Zeiss) equipped with a CCD camera system (AxioCam; Carl Zeiss).

Analyses of gene expression

Amplification by RT-PCR was performed as described elsewhere (Hamada et al. 2000), using primers specific to AtNACK1 (ANK1-B34F2; and ANK1-092154R, 5′-GCTCTCCGATTTCCATTTCCA-3′) and AtNACK2 (ANK2-1249F; and ANK2-800R, 5′-CCTGAAATCTGGAGTCTGTTCA-3′). In situ hybridization was performed as described elsewhere (Tanaka et al. 2001). A 551-bp and a 522-bp products of PCR [nt 1134 to nt 1684 of AtNACK1 cDNA (AB081599) and nt 1163 to nt 1684 of AtNACK2 cDNA (AB088121)] were cloned into pBluescript (SK-) (Stratagene, La Jolla, CA, USA) to generate pNU815 and pNU816, respectively. An anti-sense RNA probe for AtNACK1 was generated by linearizing pNU815 with ClaI, with subsequent synthesis of RNA by T3 RNA polymerase. The anti-sense probe for AtNACK2 mRNA was generated using pNU816, HindIII and T3 RNA polymerase.


The authors thank Dr Ryuichi Nishihama for helpful discussions. The authors also thank the Arabidopsis Biological Resource Center for seeds of the Wassilewskija ecotype and of atnack1 and atnack2 mutant lines. This work was supported in part by a grant from the Bio-orientated Technology Research Advancement Institution and by a Grant-in-Aid for Scientific Research on Priority Areas (no. 14036216) from the Japanese Ministry of Education, Culture, Sports, Science and Technology. H.T. was supported by a grant for Core Research for Evolutional Science and Technology from the Japan Science and Technology Agency.