Arabidopsis Fused kinase and the Kinesin-12 subfamily constitute a signalling module required for phragmoplast expansion




The conserved Fused kinase plays vital but divergent roles in many organisms from Hedgehog signalling in Drosophila to polarization and chemotaxis in Dictyostelium. Previously we have shown that Arabidopsis Fused kinase termed TWO-IN-ONE (TIO) is essential for cytokinesis in both sporophytic and gametophytic cell types. Here using in vivo imaging of GFP-tagged microtubules in dividing microspores we show that TIO is required for expansion of the phragmoplast. We identify the phragmoplast-associated kinesins, PAKRP1/Kinesin-12A and PAKRP1L/Kinesin-12B, as TIO-interacting proteins and determine TIO-Kinesin-12 interaction domains and their requirement in male gametophytic cytokinesis. Our results support the role of TIO as a functional protein kinase that interacts with Kinesin-12 subfamily members mainly through the C-terminal ARM repeat domain, but with a contribution from the N-terminal kinase domain. The interaction of TIO with Kinesin proteins and the functional requirement of their interaction domains support the operation of a Fused kinase signalling module in phragmoplast expansion that depends upon conserved structural features in diverse Fused kinases.


Cytokinesis in plant cells depends upon the formation of a plant-specific cytokinesis apparatus, the phragmoplast, which guides the deposition of a membrane, bound the cell plate by expanding centrifugally to meet the plasma membrane (for review see Jürgens, 2005).

The phragmoplast arises initially from microtubules (MTs) of the spindle midzone at late anaphase and expands dynamically through the rapid exchange of tubulin dimers (Hush et al., 1994). Transport of Golgi-derived vesicles along the phragmoplast MTs and their fusion in the midzone enables the cell plate to form and expand at its margins (for review see Nacry et al., 2000; Jürgens, 2005), showing that the establishment and expansion of the phragmoplast are essential for cell-plate formation and outgrowth.

The phragmoplast, although dynamic, is stabilised temporarily by microtubule-associated proteins (MAPs) that include MAP65-3 and the phragmoplast-associated Kinesin-12 subfamily, which both play a role in keeping the MT plus ends anchored at the division site (Muller et al., 2004; Lee et al., 2007; Caillaud et al., 2008; Ho et al., 2011). Other MAPs such as the XMAP215 orthologue MOR1/GEM1 promote MT plus-end growth and disruption leads to short phragmoplast MTs and irregular phragmoplast and cell-plate growth (Whittington et al., 2001; Twell et al., 2002; Eleftheriou et al., 2005; Kawamura et al., 2006; Oh et al., 2010a). The initial stabilisation of the phragmoplast initial is short lived with depolymerisation of the MTs occurring in the centre of the phragmoplast and re-polymerisation occurring at the leading edge, and enables the phragmoplast to expand. Pioneering studies in tobacco have established that phragmoplast expansion is regulated via the turnover of MTs by the NACK-PQR pathway, which includes the Kinesin-like proteins NACK1 and NACK2 and a mitogen-activated protein kinase (MAPK) cascade (Takahashi et al., 2004). Phosphorylation of MAP65-1 through the MAPK cascade decreases its MT bundling activity, allowing the turnover of MTs for phragmoplast expansion (Sasabe et al., 2006a). Further studies of Arabidopsis homologues of various NACK-PQR components have revealed that this pathway operates in sporophytic and gametophytic cell types that include microspores (Tanaka et al., 2004; Oh et al., 2008; Takahashi et al., 2010). Moreover, recent data has shown that activation of the NACK1-stimulated MAPK cascade is also tightly regulated by cyclin-dependent kinase (CDK)-dependent inhibitory phosphorylation during early M phase and subsequent dephosphorylation during the onset or progression of cytokinesis (Sasabe et al., 2011a). However, a more comprehensive understanding of the signalling pathways that underlies phragmoplast expansion still awaits the functional integration of other signalling components.

Cytokinesis in sporophytic and gametophytic cells in plants exhibits differing modes to fulfill various cell type-specific requirements (Otegui and Staehelin, 2000). In this paper we focus primarily on cytokinesis of the haploid microspore, which is a determinative event in male germline development (Berger and Twell, 2011). Following the production of microspores by meiosis, the first step in pollen grain development is an unequal division at pollen mitosis I (PMI). This results in two distinct cells, a smaller generative or germ cell, which divides to form two sperm cells, and a larger vegetative cell that subsequently forms the pollen tube. Following nuclear division at PMI, formation of an asymmetrically placed phragmoplast is initiated to facilitate the separation of the germ cell from the vegetative cell and involves the canonical centrifugal expansion of the phragmoplast. However, this asymmetric cell division is exceptional in that the division plane is not determined by a preprophase band typical in somatic cells and instead of a planar wall, a hemispherical cell plate is synthesised around the nascent germ cell nucleus by the expanding phragmoplast ring (Terasaka and Niitsu, 1995; Oh et al., 2010b,c).

We have previously identified TWO-IN-ONE (TIO), the Arabidopsis orthologue of the Ser/Thr Fused kinase, as an essential component of the cytokinetic machinery that functions in sporophytic and gametophytic cell types (Oh et al., 2005). Plants carrying two-in-one (tio) mutant alleles show gametophytic failure of microspore cytokinesis resulting in the formation of short, incomplete cell plates that leads to the formation of binucleate (tio) pollen grains. Moreover, in cultured cells TIO is localized specifically to the phragmoplast midzone where the plus end of MTs are located, suggesting a specific role in phragmoplast structure or function (Oh et al., 2005).

TIO has homology only to Fused kinase orthologues in various eukaryotes that include Drosophila, human and Leishmania. The kinase domain is most highly conserved with further limited homology in the C-terminal region containing four predicted ARM/HEAT repeats (Oh et al., 2005). ARM and HEAT repeats are degenerate sequences that can form conserved potential super-helical structures that may facilitate protein–protein interactions (Andrade et al., 2001; Coates, 2003). Drosophila Fused (dFu), the founding member of the Fused kinase family, acts as a positive regulator of Hedgehog (Hh) signalling, which is mediated by a cytoplasmic multiprotein complex composed of three major proteins: the kinase dFu, the Kinesin-like protein Costal 2 (Cos2), and the transcription factor Cubitus Interruptus (Ci). It is known that the dFu–Cos2–Ci complex is associated with MTs via Cos2, sequestering full-length Ci-155 in the cytoplasm for further processing to a shorter repressive form Ci-75. In the presence of Hh, dFu-induced phosphorylation of Cos2, together with other events, dissociates the complex from MTs and accumulates full-length Ci-155, thus stimulating transcription of Hh target genes (Ruel et al., 2007). More recently, the Dictyostelium orthologue of the Fused kinase, tsunami, was reported to be essential for polarization and chemotaxis (Tang et al., 2008) and mouse Fu was shown to be required for motile ciliogenesis (Wilson et al., 2009), illustrating the divergent cellular roles of the conserved Fused kinase amongst eukaryotes.

To further elucidate the plant-specific role of the Arabidopsis Fused kinase (TIO) in cytokinesis we have visualized GFP-labelled phragmoplasts in vivo and confirmed that cytokinesis defects in tio-3 microspores arise from failure of phragmoplast expansion. We present evidence that an active kinase domain and the C-terminal ARM/HEAT repeat domain are both required for TIO function in cytokinesis. We also identify protein interactions between TIO and the phragmoplast-associated Kinesin-12 subfamily and show that the function of TIO in planta and its interaction with Kinesin-12 depend largely on the C-terminal domain consisting of four ARM/HEAT repeats and two highly conserved sequence motifs. Our results suggest that TIO and Kinesin-12 constitute a signalling module that is required to support phragmoplast expansion and cell-plate growth in plant cells.


TIO is required for phragmoplast expansion

We reported previously that in tio mutants the callosic cell plate between the vegetative and generative nuclei in dividing microspores fails to expand completely leading to the formation of callosic cell-plate fragments or stubs adjacent to the parental wall (Oh et al., 2005; Figures 1h and S1c). As cell-plate expansion is facilitated by the expansion of the phragmoplast we investigated whether failure of cell-plate expansion in tio mutants results from altered phragmoplast organization or behaviour. To visualize phragmoplast MTs we crossed +/tio-3, a null T-DNA insertion allele, with a GFP-tubulin marker line (ProUBQ14-GFP-TUA6), which provides an effective tool to visualize MTs throughout male gametogenesis (Oh et al., 2010a; Movie S1). tio-3 mutant plants are fully penetrant heterozygotes and produce an equal proportion of wild type (wt) and mutant microspores in anthers. Although mutant spores fail to transmit the mutation, tio-3 is maintained in heterozygotes through limited (1–3%) female transmission and +/tio-3 plants can be selected based on their BASTA resistance (Oh et al., 2005). Progeny of +/tio plants homozygous for ProUBQ14-GFP-TUA6 were grown with or without the addition of BASTA to obtain GFP-tubulin marked +/tio-3 and control (wt) plants from the same population. GFP-labelled phragmoplasts in microspores of wt and +/tio-3 plants were scored for the presence of either unexpanded/short or expanded/ring structures (Figure 1a,b). Over 1000 phragmoplasts were counted from multiple individuals (n > 20).

Figure 1.

 Structure and position of phragmoplasts and cell plates in wt and +/tio-3 microspores.
(a, b) Fluorescence images illustrating GFP-TUA6 MT-labelling of phragmoplast structure and progression in wt microspores.
(c) Frequency of ring compared with short phragmoplasts observed in microspores of wt and +/tio-3 plants.
(d) Frequency of mislocalized phragmoplasts in wt and +/tio-3.
(e, f) Short phragmoplasts observed in +/tio-3, either in the centre between vegetative and generative nuclei or off-centre.
(g, h) Pollen at early bicellular stage was viewed for callose walls after double staining with DAPI (left) and aniline blue (right) in wt and +/tio-3. VN, vegetative nucleus; GN, generative nucleus.

Scoring of phragmoplasts in microspores from wild-type plants revealed an overall ratio of 3:1 ring:short structures, whereas microspores from +/tio-3 plants showed a significantly reduced ratio of 1:1 ring:short phragmoplasts (Figure 1c and Table S1). Heterozygous +/tio-3 plants therefore show a much greater proportion of short phragmoplasts, accounting for 47% of the structures observed (Figure 1c), compared with only 25% in wild type. Considering that tio-3 is heterozygous, any fully expanded ring structures are likely to represent the wild-type population of spores. The increased proportion of short phragmoplasts in +/tio-3 plants indicates that tio-3 mutant microspores establish the phragmoplast but these fail to expand and do not form a typical ring structure as in the wild type. Additionally, when the position of short phragmoplasts in wt and +/tio-3 was compared, the majority (98%) of short phragmoplasts in wt microspores were located centrally between the two daughter nuclei (Figure 1d,e), but those observed in tio-3 were often located off-centre, outside the inter-nuclear zone (61%), although still within the vicinity of the daughter nuclei (Figures 1d,f and S1a; Table S2). This situation probably accounts for the incomplete callosic walls, which are observed in mutant tio-3 spores (Oh et al., 2005), illustrated with aniline blue staining in Figures 1(h) and S1(c). Our observations clearly show that TIO is not required for the initial establishment of the phragmoplast, but has an essential role in the regulated expansion of the phragmoplast and cell plate.

TIO interacts with the phragmoplast-associated Kinesin-12 subfamily proteins

To identify TIO-interacting proteins involved in gametophytic cytokinesis, we constructed a yeast-two-hybrid cDNA library from spore preparations enriched for haploid microspores and early bicellular pollen. To take advantage of an unamplified library the screen was performed by co-transforming cDNA prey and the TIO bait vector into yeast strain AH109. A total of approximately 3 × 105 co-transformants were screened with an approximately 3 Kb C-terminal region of TIO (termed BD-CL) encoding the C-terminal 1021 aa region but excluding the N-terminal kinase domain (Figure 2a). Several rounds of selection to isolate strong interactors led to the identification of a partial cDNA clone, AD-12A, encoding the C-terminal 101 aa of At4g14150. Interestingly this gene encodes Kinesin-12A, which is known to play a critical role in phragmoplast organization and cytokinesis at pollen mitosis I (Lee et al., 2007). Positive interaction was confirmed by retransforming yeast with the isolated AD-12A plasmid and TIO bait vectors encoding different regions of TIO (Figure 2a). BD-TF (TIO full length, 1322 aa), BD-CL (C-terminal 1021 aa) and BD-CS (C-terminal 323 aa) showed strong interactions with AD-12A, whereas two N-terminal kinase domain baits, BD-KIN (296 aa) and BD-N (656 aa), failed to interact, restricting the interacting site to the C-terminal 323 aa of TIO (Figure 2b).

Figure 2.

 Identification of TIO-interacting proteins.
(a) Diagram of various bait and prey vectors co-transformed into yeast strain AH109.
(b) Direct Y2H tests between various TIO baits and the AD-12A prey.
(c) Direct Y2H tests between various TIO baits and either AD-12BS or AD-12BL prey.
(d–i) Representative examples of the results of BiFC assays testing interactions between fragments of TIO and Kinesin-12A or Kinesin-12B.

As Kinesin-12A and Kinesin-12B are highly homologous and play redundant roles in cytokinesis (Lee et al., 2007), we also tested interaction between TIO and Kinesin-12B. Two prey vectors were tested, AD-12BS or AD-12BL, encoding either the C-terminal 107 or 217 aa region of Kinesin-12B respectively. Both AD-12BS and AD-12BL interacted with TIO in the same manner as Kinesin-12A with a slight difference in interaction strength (Figure 2c).

We further verified interactions in planta using bimolecular fluorescence complementation (BiFC) assays. Full-length fragments encoding Kinesin-12A and Kinesin-12B, under the control of the 35S promoter, were fused to either N-terminal (YFPN) or C-terminal (YFPC) fragments of YFP and co-infiltrated into tobacco leaves with CL or CS fragments of TIO (Table S3). Strong YFP fluorescence that reported positive interaction was observed when CL-YFPC was co-infiltrated with either YFPN-Kinesin-12A or YFPN-Kinesin-12B (Figure 2d,e); CS was also capable of interaction with both Kinesin-12A and Kinesin-12B (Figure 2f,g and Table S3). Collectively, these results demonstrate that the C-terminal region of TIO is capable of interaction with the C terminus of both Kinesin-12 subfamily members.

The ARM repeat-containing C-terminal region of TIO is required for interaction with Kinesin-12

The results of the yeast-two-hybrid assays and BiFC tests described above demonstrate that the CS region of TIO is sufficient for interaction with Kinesin-12. This region consists of approximately 100 aa before the ARM repeat region, approximately 180 aa comprising the four ARM repeats and approximately 40 aa beyond. To further dissect the interacting region we generated a new series of constructs (Figure 3). The BD-CS construct was truncated from the C-terminal end to create a series of progressively longer fragments, these were comprised of the region before the ARM repeat region (95 aa, BD-CSARM0), up to the end of the first (141 aa, BD-CSARM1), the second (185 aa, BD-CSARM2), the third (224 aa, BD-CSARM3), and the last (279 aa, BD-CSARM4) of the ARM repeats, and the next 20 aa close to the C terminus (299 aa, BD-CSΔC). None of these constructs was capable of interacting with AD-12A and AD-12B in yeast. Even BD-CSΔC, which lacks only 23 aa from the C terminus compared with BD-CS, completely abolished the strong interaction (Figure 3). This result indicates that interaction of TIO with the Kinesin-12 subfamily is dependent on the presence of four ARM repeats and the extreme C-terminal sequence beyond the repeats.

Figure 3.

 Yeast-two-hybrid assays.
(a) A diagram of various baits and preys used to co-transform yeast strain AH109.
(b) Yeast growth on selective media including −LT, −HLT, −HLT + 1 mm 3AT, −AHLT. Freshly grown colonies on the −LT master plate were streaked on the plate containing each drop out medium and incubated for 3–5 days at 30°C. The positions of the four ARM repeats are indicated. The sizes of proteins encoded are shown. Asterisks in black or white represent modifications of the Fax and Nag motifs within the second and third ARM repeats respectively.

The importance of an intact C terminus was also confirmed with full-length TIO constructs that were C-terminally truncated. BD-TF was modified to include a region up to the end of the last ARM repeat (1279 aa, BD-TFARM4) or the next 20 aa before the C terminus (1299 aa, BD-TFΔC) (Figure 4). These constructs were tested with AD-12A and AD-12B preys and showed no or highly reduced interaction compared with BD-TF. As the last 23 aa, absent in BD-TFΔC, did not contain any significantly conserved motifs, we tested whether it can be substituted with an unrelated sequence. In BD-TFΔC3M (1299 aa + 3 myc), we added 42 aa of triple myc sequence that was also used in microspore-targeted complementation analysis in planta (see below). Interestingly, this construct allowed some yeast growth at lower stringencies when tested with AD-12A and AD-12B (Figure 4), which suggested that protein length after the repeats may be important as well as the amino acid composition. In parallel, an N-terminally truncated TIO bait, BD-Δ50CS (C-terminal 271 aa) lacking approximately 50 aa before the first ARM repeat, showed highly reduced or no interaction with AD-12A or AD-12BS (Figure 4). Taken together, these results show that the entire CS region of TIO is necessary for interaction with Kinesin-12 proteins.

Figure 4.

 Yeast-two-hybrid assays.
(a) A diagram of various baits and preys used to co-transform yeast strain AH109.
(b) Yeast growth on selective media including −LT, −HLT, −HLT + 1 mm 3AT, −AHLT. Three microlitres of serially diluted cultures were spotted on the plate containing each drop out medium and incubated for 3–5 days at 30°C. The positions of the four ARM repeats are indicated. The sizes of proteins encoded are shown. The 42 aa triple myc sequence is differently shaded.

Structurally conserved motifs in the C-terminal ARM domain of TIO are required for interaction with Kinesin-12

It is possible that the structure formed by the putative ARM domains is the key to this interaction and we previously reported homology between the C-terminal domain of TIO and that of Fused kinases from other organisms (Oh et al., 2005). Recent advances in the availability of genome sequence data allowed us to investigate this homology further. The C-terminal amino acid sequences of 20 Fused kinases, including 18 plant and two protozoan sequences were aligned and submitted to WEBLOGO (Schneider and Stephens, 1990; Crooks et al., 2004). This analysis identified two potentially important highly conserved motifs, the Fax motif (FAIGNAAY in Arabidopsis) and the Nag motif (NAAGSALN in Arabidopsis), that occur in the second and the third ARM repeats respectively. These motifs were also found to be conserved in fungal and vertebrate Fused kinase proteins (Figure S2). We disrupted these motifs within the BD-CS construct; for each motif we made two types of disruption, one with similar chemical composition [Fax target (BD-CSFT), YGGAQGGW; Nag target (BD-CSNT), QGGAGGDQ], and the other with different functional properties [Fax random (BD-CSFR), LTFRDTTH; Nag random (BD-CSNR), DTTRTMCD]. All four modified constructs abolished the interaction with AD-12A and AD-12B in yeast (Figure 3). We also performed BiFC assays in tobacco leaves using CLFT or CLNT co-infiltrated with either KINESIN12A or KINESIN12B and did not observe the strong YFP signals observed with intact CL constructs, further suggesting that these motifs are required for interaction of TIO with Kinesin-12A or Kinesin-12B (Figure 2h,i and Table S3).

To explore why the integrity of the C-terminal domain of TIO is so critical for interaction, a structural 3D protein model was predicted using the I-TASSER on-line server (Roy et al., 2010). This prediction revealed a possible structure for TIO (C value −2.16) that consists of a super-helical structure in which the C-terminal domain forms a pocket (Figure S3a,e and Movie S2). The ARM repeats are integral to this structure, and the last 40 aa (yellow and white) form the completion of this pocket, a finding that suggested a reason for the significance of their conservation, and their requirement for function and interaction with Kinesin-12. In this model the Fax and Nag motifs (green), located in the second (pink) and the third (turquoise) ARM repeats, are positioned next to each other at the base of the pocket (Figure S3) and led us to speculate that together they may form a binding domain for interaction with Kinesin-12 proteins. Similar structures are predicted for rice (C value −1.86), human (C value −1.70) and Tetrahymena (C value −1.66) Fused orthologues (Figure S3b–d).

Highly conserved active kinase residues in TIO positively contribute to protein interactions with Kinesin-12

The kinase domain, the most conserved region in all Fused kinase orthologues, includes an invariant putative ATP-binding site (Lys, K35) and a kinase active site (Asp and Asn, D127 and N132). To explore whether TIO kinase activity might also contribute to Kinesin-12 protein interactions we substituted each of these residues with Ala (A35, A127, and A132). As the bait protein that contained a kinase domain alone, BD-KIN, does not interact with the Kinesin-12 preys (Figure 2a–c), substitutions were introduced into the TIO full-length bait BD-TF termed BD-KA and BD-DANA respectively and tested for interaction with AD-12A and AD-12BS (Figure S4). As a result, both BD-KA and BD-DANA showed a significantly reduced interaction compared with BD-TF, a finding that indicated that the invariant active kinase residues also contributes to TIO protein interactions with Kinesin-12A and Kinesin-12B in the context of the native TIO protein.

Microspore-targeted complementation of tio-3 provides an in vivo function assay for TIO interaction domains

To assess the role of identified TIO domains functionally, we generated a series of constructs corresponding with those used in yeast and BiFC assays and performed a phenotypic complementation assay, scoring the efficiency of rescue of gametophytic cytokinesis in the null allele, +/tio-3 (Oh et al., 2005). As TIO plays an important role in cytokinesis in both sporophytic and gametophytic cells, we targeted the expression of TIO variants to microspores to avoid potential sporophytic lethal effects. Thus, we used the MSP1 (microspore-promoter 1) sequence to drive transgene expression, enabling the role of different TIO domains in post-meiotic cytokinesis to be tested. Full-length TIO tagged with a 3× myc sequence, ProMSP1-TF-3MYC, fully complemented the +/tio-3 phenotype in T1 lines (Figure 5a), showing a reduction of the mean percentage mutant phenotype to 24% compared with 52% in untransformed +/tio-3 plants. All lines screened (n = 12) showed restoration to the wild-type phenotype, confirming that this assay provides a robust system to assess the in vivo role of TIO domains (Figure 5a).

Figure 5.

 Percentage of abnormal pollen phenotypically resembling tio-3.
(a) Full complementation of the tio-3 mutant is observed in lines expressing the ProMSP1-TF-3MYC (full-length) construct (TF); constructs in which the ATP-binding site [ProMSP1-KA-3MYC (KA)], or kinase active site [ProMSP1-DANA-3MYC (DANA)] within the kinase domain of TIO are mutated fail to complement the tio-3 mutant.
(b) Truncation of TIO after ARM4 [ProMSP1-TFARM4-3MYC (ARM4)] prevents this construct from rescuing the tio-3 mutant; however, with an additional 20 aa beyond ARM4 ProMSP1-TFΔC-3MYC (ΔC) is capable of rescue. ProMSP1-TFΔC [ΔC (no myc)], which lacks the 3-myc tag fails to rescue the tio-3 mutant phenotype.
(c) Mutation of the conserved Fax [ProMSP1-TFFT-3MYC (TFFT) and ProMSP1-TFFR-3MYC (TFFR)] or Nag [ProMSP1-TFNT-3MYC (TFNT) and ProMSP1-TFNR-3MYC (TFNR)] motifs in full-length TIO constructs also prevents the rescue of tio-3. N = 12 [except lines: ProMSP1-TFΔC (n = 11), ProMSP1-TFFT-3MYC (n = 6), ProMSP1-DANA-3MYC (n = 11)]. Error bars show standard error of the mean. *Statistically significant difference from tio-3 using unpaired two-tailed t-test (< 0.0001).

To investigate whether kinase activity may be required for TIO function two constructs harbouring amino acid substitutions at the active kinase sites, equivalent to BD-KA and BD-DANA that were previously tested in yeast (Figure S4), were introduced into +/tio-3 plants. Plants that expressed ProMSP1-KA-3MYC or ProMSP1-DANA-3MYC failed to complement the +/tio-3 mutant phenotype (Figure 5a), which suggested that TIO kinase activity is required for its role in phragmoplast expansion.

When C-terminally truncated TIO constructs were assayed, the ProMSP1-TFARM4-3MYC construct was unable to rescue the tio-3 mutant phenotype, showing a mean frequency of 46% aberrant pollen (Figure 5b). This finding suggests that the C-terminal sequence beyond the ARM repeat region is required for TIO function. Consistent with the yeast-two-hybrid results, ProMSP1-TFΔC-3MYC significantly rescued the tio-3 mutant phenotype, but to a lesser extent (32%) than ProMSP1-TF-3MYC (24%), while ProMSP1-TFΔC without the 3-myc tag failed to complement the tio-3 mutant phenotype (Figure 5b), which indicated that the last 23 aa may be substituted, in part, by an unrelated sequence.

Substitution mutants in the conserved Fax and Nag motifs shown to be important for TIO interaction with Kinesin-12 subfamily proteins in yeast and in BiFC assays were also tested. None of the four constructs tested was capable of rescuing the tio-3 mutant phenotype (Figure 5c), which suggested that these motifs are critical for TIO function in cytokinesis. Verification of the expression of these constructs in young buds was demonstrated by reverse transcription polymerase chain reaction (RT-PCR) (Figure S5). This result highlights the importance of the ARM repeat-containing C-terminal region of TIO and suggests that the failure of these constructs to rescue the tio-3 mutant is likely to be due to the failure of the interaction between TIO and Kinesin-12 proteins.


Our results reveal an essential role for the Arabidopsis Fused kinase, TIO, in phragmoplast expansion and identify important domains and conserved amino acid motifs involved in TIO function. We uncover a protein interaction between TIO and the two phragmoplast-associated Kinesin-12 subfamily members that is reminiscent of the conserved interaction of metazoan Fused homologues with kinesin-related proteins (Chen et al., 2005; Tay et al., 2005). The requirement for kinase active sites and Kinesin-12 interaction domains in TIO function therefore supports the operation of a signalling module in phragmoplast expansion that depends upon a conserved molecular interaction.

TIO has a specific role in phragmoplast expansion

We previously reported TIO to be an essential protein required for cell-plate expansion during cytokinesis at PMI (Oh et al., 2005). Here, using a GFP-tubulin marker we show that, in the absence of TIO, the phragmoplast persists as a short structure that is not sufficient to deposit a complete cell plate between the vegetative and germ cell nuclei (Figures 1h and S1c). In wild-type microspores, the phragmoplast, which starts as a short barrel-shaped structure, expands centrifugally as a curved ring structure. We observed that microspores from +/tio-3 plants exhibited a much higher ratio of short phragmoplasts to ring structures than microspores from wild-type plants. Failure of phragmoplast expansion in tio-3 microspores is, therefore, the major cause of failure of expansion of the cell plate.

An interesting phenotype observed in dividing +/tio-3 microspores was that 61% of the short phragmoplasts were located off-centre outside the inter-nuclear zone (Figures 1f and S1a). This situation raises several possibilities for the origin of these mislocalized short phragmoplasts. First, the phragmoplast initial may be established off-centre and persists at this position. However, given that the axis of division in tio-3 microspores appears regular (Oh et al., 2005), and the phragmoplast is assembled from the late anaphase spindle midzone, this may be least likely. Moreover, TIO is located specifically to the phragmoplast midzone but not to other cortical or mitotic MT arrays that could influence the initial positioning of the phragmoplast (Oh et al., 2005). Second, the phragmoplast initial in tio-3 microspores may be established normally in the inter-nuclear zone but persists in the absence of expansion, allowing it to drift out of position. This situation may be more likely as short phragmoplast structures at single off-centre locations were observed, rather than more complex incomplete ring structures. Third, some mislocalized short phragmoplasts could arise from the asymmetric expansion of a central phragmoplast initial. The phragmoplast does have limited capacity to expand as some tio-3 microspores form cell plates that extend from the centre to the spore wall. In such cases GFP-labelled phragmoplast MTs, that are detected only at the leading edge of the expanding cell plate, may appear as short off-centre MT arrays. Nevertheless, the increased proportion of short phragmoplasts in tio-3 microspores clearly accounts for the incomplete cell plates observed in tio-3 that appear as unattached islands or as fragments at single asymmetric sites close to the microspore wall (Oh et al., 2005; Figure S1).

TIO and Kinesin-12 – a signalling module required for phragmoplast expansion in dividing microspores

Yeast-two-hybrid assays revealed that TIO was also capable of interacting with Kinesin-12B and BiFC assays further verified TIO-Kinesin-12 protein interactions in planta (Figure 2). Kinesin-12A and Kinesin-12B localize specifically to the phragmoplast midzone where the plus end of the MTs are juxtaposed (Lee and Liu, 2000). The specific location of Kinesin-12 proteins and the defective cytokinesis phenotypes in kinesin-12a/b double-mutant pollen are similar to those described for TIO and tio mutants, supporting the significance of the protein interactions identified in this study. On the other hand, while the heterozygous null allele, tio-3, is fully penetrant and results in approximately 50% of pollen with cytokinetic defects, kinesin-12a/b double homozygous mutant plants still produce 42% of wild-type pollen (Lee et al., 2007). Moreover, while down-regulation of TIO by inducible RNAi in somatic cells results in multinucleate cells with cell wall stubs (Oh et al., 2005), the sporophytic growth of kinesin-12a/b double-mutant plants was not affected (Lee et al., 2007). These observations suggest that other kinesins could have redundant roles with Kinesin-12 in male gametophytic and sporophytic cells that could involve interactions between TIO and unknown kinesins.

Kinesin-12 members are proposed to serve as dynamic integrators required for organization and/or maintenance of the anti-parallel phragmoplast MT array based on the loss of phragmoplast MTs as a mirrored set and the lack of cell-plate formation in kinesin-12a/b double-mutant microspores (Lee et al., 2007). In contrast, all tio mutants, including the null allele tio-3, produce short but significant cell plates stained with aniline blue (Oh et al., 2005; Figure S1). These observations suggest that Kinesin-12 members function to organize and maintain the anti-parallel phragmoplast MT array while TIO functions specifically during phragmoplast expansion. From this point of view, in the absence of both Kinesin-12 members the disorganized anti-parallel MTs fail to establish a functional phragmoplast midzone causing the mass of callose aggregates observed at the cell cortex (Lee et al., 2007). In contrast, we propose that in tio mutants the anti-parallel MT arrays of the early phragmoplast are established through the activity of both Kinesin-12 members, thereby the delivery of Golgi-derived vesicles to the midzone and establishment of the cell plate appears to be normal at early stages. Consequently, however, the lack of TIO signalling is proposed to deregulate Kinesin-12 proteins or prevent MT destabilization, which in turn inhibits the re-organization of MTs into anti-parallel arrays at the leading edge of the cell plate, resulting in incomplete phragmoplast and cell-plate expansion.

Signalling pathways underlying phragmoplast expansion

The major signalling pathway known to regulate phragmoplast expansion is the NACK-PQR pathway that promotes the destabilization of phragmoplast MTs through MAP65 activity (Sasabe et al., 2006b). In somatic cells of tobacco and Arabidopsis, this pathway is triggered by the kinesins NACK1 and NACK2 [HINKEL (HIK) and TETRASPORE (TES) in Arabidopsis] that interact with the MAPKKK, NPK1 (ANP1, ANP2, and ANP3 in Arabidopsis). NPK1 in turn phosphorylates NQK1 (ANQ1/AtMKK6), which subsequently phosphorylates NRK1 (multiple members of MAPK including AtMPK4 in Arabidopsis). Phosphorylation of MAP65-1 by NRK1 decreases the MT bundling activity of MAP65-1 and so enables MT turnover required for phragmoplast expansion (Sasabe et al., 2006a).

We have reported previously that loss of both HIK and TES in dividing microspores leads to incomplete cell-plate expansion similar to tio mutants, implicating the NACK-PQR pathway in cytokinesis at PMI. Here we propose that a signalling module that involves the TIO Fused kinase and the phragmoplast MT plus-end associated Kinesin-12 are also critically required for phragmoplast expansion. Whether these two signalling modules are actively engaged via protein interactions or may be linked indirectly to regulate, for example the phosphorylation of different MAP65 family members, remains to be determined. According to a recent report (Sasabe et al., 2011b) GFP-MAP65-1 and GFP-MAP65-2 decorate the entire phragmoplast, whereas GFP-MAP65-3 specifically decorates the phragmoplast midzone. On the other hand, Ho et al. (2011) reported that the location of Kinesin-12 at the plus ends of phragmoplast MTs is dependent on MAP65-3, which selectively cross-links the interdigitating MTs in the midzone. Based on these results, one possible scenario is that MAP65-3 cross-links interdigitating phragmoplast MTs and subsequently Kinesin-12 and TIO are guided to and located in the vicinity. Later, while the NACK-PQR pathway destabilizes both the interdigitating and non-interdigitating MTs, a signalling module of TIO and Kinesin-12 destabilizes preferentially interdigitating MTs by phosphorylating substrates such as MAP65-3 in the phragmoplast midzone (Figure 6a). A second scenario is that complete turnover of the interdigitating MTs, which are more stable than non-interdigitating MTs, could require both inactivation of MAP65-3 through the NACK-PQR pathway and release of Kinesin-12 proteins from MTs, possibly that involves phosphorylation-induced conformational change by TIO (Figure 6b). In Drosophila Hh signalling, a model has been accepted that dFu-induced phosphorylation of kinesin Cos2 could cause the Hh signalling complex to move along or release from MTs, which in turn prevents the Ci processing, allowing the intact Ci to act as a transcriptional activator (Nybakken et al., 2002; Ruel et al., 2007). Based on published results in Arabidopsis it has been demonstrated that MAP65-3 has a critical role in the interaction between interdigitating MTs (Muller et al., 2004; Caillaud et al., 2008; Ho et al., 2011). More importantly however, Kinesin-12 members are required to keep the phragmoplast midzone discrete as the lack of Kinesin-12 members in the presence of MAP65-3 causes anti-parallel MTs to overlap to form continuous bundles (Lee et al., 2007). This emphasizes that in order to maintain interdigitating anti-parallel MTs in the division site, without the cellular defects seen in the MAP65-3 null mutant, pleide, and kinesin12a/b mutants, two cellular devices operate, the ‘bundling activity’ of MAP65-3 and the ‘anchoring activity’ of Kinesin-12. Therefore, it is plausible to expect that the complete turnover of the interdigitating MTs could involve a twofold destabilization process by NACK-PQR and TIO-Kinesin-12 signalling modules. A third scenario would be that after delivery of TIO to the midzone by Kinesin-12, TIO could act to trigger the NACK-PQR pathway by phosphorylating AtNACK1/2 kinesins or ANP1/2/3 MAPKKKs to promote the dynamic turnover of MTs and phragmoplast expansion (Figure 6c).

Figure 6.

 Three possible scenarios by which the TIO-Kinesin-12 signalling module could contribute to phragmoplast expansion along with the known NACK-PQR pathway.
(a) The TIO-Kinesin-12 module may preferentially destabilize interdigitating MTs through MAP65-3, while the NACK-PQR pathway destabilizes phragmoplast MTs through MAP65 members on both the interdigitating and non-interdigitating MTs.
(b) TIO could complete the turnover of more stable interdigitating MTs by promoting the release of Kinesin-12 proteins from midzone MTs in addition to the inactivation of MAP65-3 through the NACK-PQR pathway.
(c) The delivery of TIO to the phragmoplast midzone by Kinesin-12 could trigger the NACK-PQR pathway to promote the turnover of MTs for the phragmoplast expansion. MAP65s denotes multiple MAP65 family members on non-interdigitating MTs of the phragmoplast midzone.

Divergent functions of Fused kinases involve conserved structure and mode of action

A conserved feature of invertebrate and vertebrate Fused kinase orthologues as positive regulators of Hh signalling is their interaction with Cos2-related kinesins within a cytoplasmic multiprotein complex that links their activity to MTs (Osterlund and Kogerman, 2006). Moreover, Fused kinase orthologues with Hh-independent divergent functions, such as mouse Fu, were shown to interact with kinesins (Wilson et al., 2009) and the association of the Dictyostelium Fused kinase tsunami (tsuA) with MTs (Tang et al., 2008) could also involve unknown kinesins. Importantly, the interaction of Arabidopsis Fused kinase with Kinesin-12 proteins indicates that the mode of action of Fused kinase proteins and their association with MT-dependent signalling is conserved in plants. Yeast-two-hybrid and BiFC assays indicate that interaction of TIO with Kinesin-12A and Kinesin-12B occurs via the C-terminal domain, which is a conserved feature of other Fused kinases. In Drosophila the C-terminal domain of dFu was reported to be necessary for the direct binding of Fu to Cos2 (Ascano et al., 2002), and tsuA is dependent on the C-terminal ARM repeat or TH region for MT localization (Tang et al., 2008).

The conservation of secondary structure of Fused kinase within eukaryotes indicates the ancient lineage of this protein, and although Fused kinases have evolved diverse cellular roles, their mode of action through interaction with kinesins appears to have been maintained. As illustrated, Fused kinases from different organisms are predicted to have similar structures (Figure S3), with conserved positioning of the Fax and Nag domains in the ARM repeat region. These domains are integral for the interaction of TIO and Kinesin-12, and their positioning within this structure may provide the key to the interaction of Fused with divergent kinesins in other organisms.

The failure of TIO constructs that contain mutations in either the ATP-binding site or the kinase active site to rescue the tio-3 mutant (Figure 5) further suggests that the function of TIO is dependent upon its kinase activity. In Drosophila, mutation of the conserved Lys residue also leads to loss of dFu function, (Therond et al., 1996) and KA and DANA mutations were sufficient to prevent dFu activation of a kinesin-dependent reporter (Fukumoto et al., 2001). Likewise, kinase-dead versions of tsuA were unable to complement the aggregation defects of Dictyostelium tsunami mutants (Tang et al., 2008). In Drosophila, dFu phosphorylates multiple substrates including dFu itself and Cos2, and an important avenue of future research to explore how TIO kinase signalling promotes phragmoplast expansion will be to identify TIO substrates and their phosphorylation sites. Interestingly, the integrity of the TIO kinase active sites was shown to contribute to the interaction of TIO with Kinesin-12 in yeast (Figure S4) indicating that Kinesin-12 may be a substrate, although we acknowledge that the lack of interaction of KA and DANA could arise from overall conformational changes. In addition to TIO itself, other potential substrates located in the phragmoplast midzone could include MAP65 family members or even regulators of the NACK-PQR pathway.

Collectively, our results support an essential role for Fused kinase signalling in phragmoplast expansion that involves the interaction of TIO with Kinesin-12 in the phragmoplast midzone. In plants, this molecular module is essential not only for the establishment of the anti-parallel array but also for the dynamic expansion of the phragmoplast. Our results further highlight the link between Fused signalling and MTs through conserved interaction with kinesins as an ancestral feature of Fused kinases (Ruel et al., 2007; Tang et al., 2008; Wilson et al., 2009). The structural basis of the interaction of Fused kinases with different kinesin partners therefore deserves further investigation to more fully understand how this molecular module has been co-opted into different signalling pathways in distinct evolutionary lineages.

Experimental Procedures

Plant growth and vector construction

Arabidopsis wild type (Col-0) and +/tio-3 plants were grown under continuous fluorescent lighting at 20–22°C. For complementation analysis all constructs were made based on a vector (proMSP1-NN-3MYC) containing the TIO full-length cDNA clone and the MSP1 promoter as described (Honys et al., 2006). For detailed information see Appendix S1.

For plant transformation, +/tio-3 plants were selected on soil supplemented with 30 μg ml−1 BASTA (glufosinate ammonium, DHAI PROCIDA), and transformed by floral dipping (Clough and Bent, 1998). Transformants were double selected with phosphinothricin and kanamycin as described (Honys et al., 2006) and confirmed by transgene-specific PCR using primers that match TIO and the 3-myc tag (Table S4). Additional information in Appendix S1.

Yeast-two-hybrid experiments

Developing pollen was isolated from young buds (approximately 1 mm) containing uninucleate and bicellular stage pollen of Arabidopsis wild type Columbia-0 as described by Honys and Twell (2004) and total RNA extracted using the RNeasy Plant Mini Kit (Qiagen, The yeast-two-hybrid cDNA library was constructed using the Matchmaker 3 library construction and screening kit (Clontech, with 2 μg of total RNA and pGADT7-rec prey vector. For various bait and prey vector constructions, PCR-amplified fragments were cloned between either NcoI and XmaI or NdeI and XmaI sites in pGBKT7 and pGADT7 (Clontech) using primers listed in Table S4. All constructs were verified by enzyme digestion and sequencing.

Yeast strain AH109 was transformed according to the manufacturer’s manual (Clontech) and grown on minimal media plates for 3–5 days at 30°C. For more detailed information see Appendix S1.

Bimolecular fluorescence complementation

BiFC constructs were created using Multisite Gateway® Technology (Invitrogen,, which enabled the fusion of the CaMV35S promoter, YFP fragment and cDNA of interest in different combinations. cDNA was amplified using Phusion DNA polymerase (Finnzymes, with primers containing suitable attachment site (att) adapters and full-length attB sites incorporated into each fragment in a second high fidelity PCR. Fragments were cloned into pDONRP4P1R for the promoter, pDONR207 or pDONR221 for the central fragment, and pDONRP2R-P3 for the C-terminal fragment – all vectors were verified by sequencing. The three fragments were combined in a multipart LR reaction into pK7m34GW, and were used to transform Agrobacterium tumefaciens (GV3101). Infiltration of young tobacco leaves was achieved using methods described previously (Borg et al., 2011). The two vector strains were combined in infiltration medium, each with an OD of 0.3. After 36–48 h 1 cm2 leaf sections were mounted in water and screened immediately for YFP fluorescence.


For scoring of pollen phenotypes in transformed lines, pollen was isolated and stained with DAPI as described in Oh et al. (2005). Fluorescence and confocal laser scanning microscopy (CLSM) was performed using the methods and equipment described in Brownfield et al. (2009). For in vivo imaging of ProUBQ14-GFP-TUA6 lines, buds were dissected in 0.3 m mannitol and microspores imaged immediately. Further details of equipment used for counting of microspore phragmoplasts can be found in Appendix S1.

Motif identification and protein modelling

TIO orthologous sequences, including both verified and predicted proteins, were aligned by ClustalW using MacVector and submitted to weblogo (Schneider and Stephens, 1990; Crooks et al., 2004) to identify conserved motifs. For alignment of the C-terminal domain of Fused orthologues (Figure S2), the C terminus of TIO (1100–1322 aa) was submitted to NCBI blast (Altschul et al., 1990) and corresponding regions of selected hits were compiled and submitted to ClustalW2 (Larkin et al., 2007; Goujon et al., 2010). The resulting document was shaded according to those amino acids that matched TIO.

The TIO protein sequence and those of other Fused orthologues Oryza sativa Os12g0433500, Homo sapiens serine/threonine 36 and Tetrahymena thermophila (XP_001029696) were submitted to the I-TASSER on-line server for protein structure predictions (Zhang, 2008; Roy et al., 2010; Roy et al., 2011). The resulting models were colour coded in Swiss Pdbviewer (Guex and Peitsch, 1997) and a movie generated using PyMOL 1.4 (


We thank Anthony Wardle and June Saddington for assistance with plant growth and maintenance and Ralf Schmid for generating the movie file of the TIO model. This research was supported by the UK Biotechnology and Biological Sciences Research Council (BBSRC) (BB/E001017/1 and BB/I011269/1 to DT), by the Basic Science Research Program, National Research Foundation of Korea (Grant 2011-0011141 to SKP) and by the Next-Generation BioGreen21 Program, Rural Development Administration, Republic of Korea (Plant Molecular Breeding Center No. PJ008137 to SKP).