Plant Aurora kinases play a role in maintenance of primary meristems and control of endoreduplication

Authors


Author for correspondence:
Pavla Binarová
Tel: +420 241062130
Email: binarova@biomed.cas.cz

Summary

  • The conserved family of Aurora kinases has multiple functions during mitosis. The roles of plant Aurora kinases have been characterized using inhibitor treatments.
  • We down-regulated Aurora kinases in Arabidopsis thaliana using RNA interference (RNAi). We carried out a detailed phenotypic analysis of Aurora RNAi plants, biochemical and microscopic studies of AtAurora1 kinase together with AtTPX2 (targeting protein for Xklp2) and γ-tubulin.
  • Cell division defects were observed in plants with reduced expression of Aurora kinases. Furthermore, the maintenance of primary meristems was compromised and RNAi seedlings entered endoreduplication prematurely. AtAurora1, its activator AtTPX2, and γ-tubulin were associated with microtubules in vitro; they were attached to regrowing kinetochore microtubules and colocalized on spindle microtubules and with a subset of early phragmoplast microtubules. Only the AtAurora1 kinase was translocated to the area of the cell plate.
  • RNAi silencing of Aurora kinases showed that, in addition to their function in regulating mitosis, the kinases are required for maintaining meristematic activity and controlling the switch from meristematic cell proliferation to differentiation and endoreduplication. The colocalization and co-fractionation of AtAurora1 with AtTPX2, and γ-tubulin on microtubules in a cell cycle-specific manner suggests that AtAurora1 kinase may function to phosphorylate substrates that are critical to the spatiotemporal regulation of acentrosomal microtubule formation and organization.

Introduction

Aurora kinases are a family of conserved serine/threonine kinases that are major regulators of mitotic progression and cytokinesis. Aurora kinases that phosphorylate various mitotic regulators are overexpressed in many types of cancer cell (Carmena & Earnshaw, 2003). In metazoan cells, three members of the Aurora kinase family, Aurora A, B and C, are present and show different subcellular distributions. Aurora A is localized to the centrosome and along mitotic spindle microtubules in dividing cells, where it plays a major role in centrosome maturation and spindle pole organization and maintenance (Glover et al., 1995). Aurora A was also shown to be associated with centrosomin at a microtubular nucleation site in Drosophila and mammalian cells (Terada et al., 2003). Apart from its centrosomal functions, Aurora A regulates many events during mitotic entry. For example, it has been implicated in regulating cell cycle progression because the depletion of Aurora A delays chromosomal condensation and CDK1 (cyclin dependent kinase 1) activation (Liu & Ruderman, 2006). Aurora A, together with its activator TPX2 (targeting protein for Xklp2), plays an important role in spindle assembly, and an interaction between the two molecules is required to determine spindle length, primarily via the nucleation of microtubules (Kufer et al., 2002; Bird & Hyman, 2008). RanGTP (RAs-related Nuclear protein, GTP binding) releases TPX2 from a complex with importin, stimulates TPX2 to interact with Aurora A, and activated Aurora A consequently induces formation of bipolar spindles by phosphorylating numerous downstream targets (Tsai & Zheng, 2005).

Aurora B localized along the chromosomes during prophase, accumulates at the centromeres during metaphase, and localizes to the cleavage furrow in anaphase and to the midbody in telophase. Furthermore, as a chromosomal passenger, Aurora B is essential for normal chromosomal segregation and cytokinesis (Carmena & Earnshaw, 2003). A single site mutation within the Aurora A catalytic domain converts Aurora A to an Aurora B type of kinase that interacts with chromosome passenger proteins instead of with the TPX2 protein and shows subcellular localizations and functions characteristic of Aurora B (Fu et al., 2009). Moreover, Aurora B is responsible for mitotic arrest in the absence of functional Aurora A (Yang et al., 2005). Active Aurora B associated with spindle microtubules was also shown to promote spindle formation by transmitting spatial signals from the chromosomes (Tseng et al., 2010).

Only one type of Aurora kinase gene is found in the genome of Dictyostelium discoideum, and it was shown to have properties of both Aurora A and Aurora B kinases (Li et al., 2008). Similarly, in starfish, a single Aurora kinase performs the functions of Aurora A and B (Abe et al., 2010). In Chlamydomonas reinhardtii, the Aurora protein kinase CALK (Chlamydomonas aurora/Ipl1p-like protein kinase) was shown to be a key effector in the regulatory pathway of flagella disassembly (Pan et al., 2004). Later, an unexpected nonmitotic role of Aurora A was demonstrated in cilia disassembly in vertebrates (Pugacheva et al., 2007). Aurora kinases, which coordinate many events in cell division and differentiation, have therefore evolved to interact with a broad range of substrates in accordance with the needs of highly divergent organisms.

Three Aurora kinases, Aurora1, Aurora2 and Aurora3, are encoded in the Arabidopsis thaliana genome (Van Damme et al., 2004). All three plant Aurora kinases were shown to phosphorylate histone H3 (Demidov et al., 2005, 2009; Kawabe et al., 2005). Inhibition of Aurora kinase activities demonstrated their role in chromosomal segregation in tobacco (Nicotiana tabacum) BY2 cells (Kurihara et al., 2006), and highly homologous AtAurora1 and AtAurora2 localized to the nuclei of tobacco cells during interphase, and with mitotic microtubular arrays in dividing cells. AtAurora3 localization with chromosomes indicated a chromosomal passenger-like role similar to mammalian Aurora B kinase (Van Damme et al., 2004; Demidov et al., 2005; Kawabe et al., 2005). Homologs of chromosome passenger proteins have not been found in plant genomes; therefore, plant Aurora kinases may coordinate spatial and temporal events of mitosis/cytokinesis progression by targeting as yet uncharacterized substrates. Aurora kinases may also regulate acentrosomal nucleation of microtubules in plant cells at dispersed sites localized within nuclei, on membranes and on existing microtubules (Murata et al., 2005; Binarováet al., 2006).

The molecular mechanisms involved in the regulatory functions of Aurora kinases and their impact on plant growth and development are still not well understood. Our analysis of Arabidopsis thaliana plants with RNA interference (RNAi) phenotypes revealed an essential role of plant Aurora kinases in the control of cell division and differentiation. Furthermore, AtAurora1 and its activator AtTPX2 were implicated in γ-tubulin-driven microtubular regeneration and thus in processes of acentrosomal spindle and phragmoplast organization.

Materials and Methods

Plant material

In this study, Arabidopsis thaliana (L.) Heynh. plants, ecotypes Columbia, Landsberg erecta, and A. thaliana Columbia plants expressing Tua6-GFP (Abe & Hashimoto, 2005), Map4-GFP (Marc et al., 1998) and GFP-nuclear marker (Cutler et al., 2000), were used. Plants were grown under a regime of 8 h light : 16 h dark at 20°C for 4 wk, and thereafter under a regime of 16 h light : 8 h dark.

Sterile seeds were grown on half-strength Murashige and Skoog (MS) medium (Duchefa, the Netherlands), supplemented with 0.25 mM MES, 1% saccharose and 1% phytoagar. Selection of transformed seedlings was carried out according to Harrison et al. (2006). Arabidopsis thaliana suspension cultures of Columbia and Lansberg erecta were cultured as described in Drykováet al. (2003).

Molecular cloning of AtAurora1 and AtTPX2

AtAurora1 and AtTPX2 coding sequences were obtained by PCR amplification using a template of A. thaliana cDNA and platinum Pfx DNA polymerase (Invitrogen). The PCR primers were designed with attB sites (underlined) according to the manufacturer’s instructions. The following primers were used: AtAurora1, forward 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCATGGCGATCCCTACGGAGAC-3′ and reverse 5′-GGGGCACCACTTTGTACAAGAAAGCTGGGTCTTAAACTCTGTAGATTCCAGAAGGATC-3′; AtTPX2 for N-terminal fusion, forward1 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCATGGAAGCAACGGCGGAGGAATC-3′ and reverse1 5′-GGGGACCACTTTGTACAAGAAAGCTGGGTCTTATCTCATCTGACCAGCAGAGGC-3′, and for C-terminal fusion, forward1 and reverse2 5′-GGGGACCACTTTGTACAAGAAAGCTGGGTCTCTCATCTGACCAGCAGAGGC-3′. For N-terminal GFP fusion of a truncated version of AtTPX2 that lacked the N-terminal part of the molecule with AtAurora1 binding sites (AtTPX2ΔN-GFP), the primers forward2 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCATGCTTGAGCAAGAGGAACTGGAGAAG-3′ and reverse1 were used. PCR products were isolated and subsequently cloned by Gateway technology (Invitrogen). Gateway binary vectors pK7WGF2.0 for N-terminal GFP fusion, pH7WGR2.0 for N-terminal RFP fusion (Karimi et al., 2002), pMDC43 for C-terminal GFP fusion (Curtis & Grossniklaus, 2003) and pB7RWG2.0 for C-terminal RFP fusion (Karimi et al., 2002) were used. For RNAi silencing, pOpOff vectors (Wielopolska et al., 2005) were used.

The fusion vector expressing AtAurora1 under the control of its native promoter in-frame with the N-terminal GFP tag was constructed as follows. The complete coding sequence of AtAurora1 was PCR-amplified using a template of the sequenced AtAurora1 construct using the primer set AUR1MULTIgwFW01, 5′-GGGGACAGCTTTCTTGTACAAAGTGGAAATGGCGATCCCTACGGAG-3′, and AUR1MULTIgwRE01, 5′-GGGGACAACTTTGTATAATAAAGTTGGTTAAACTCTGTAGATTCCAGAAGG-3′. Similarly, the 5′ untranslated region adjacent to AtAurora1 was PCR-amplified using a template of freshly isolated A. thaliana chromosomal DNA using the primer set AUR15UTRgwFW01, 5′-GGGGACAACTTTGTATAGAAAAGTTGATGGAAATATCACTAAAAACACTTTTTTGC-3′, and AUR15UTRgwRE01, 5′-GGGGACTGCTTTTTTGTACAAACTTGGGAAGAAGAGATCAACCCAAC-3′. The resulting fragments corresponding to the AtAurora1 gene and the putative promoter sequence were cloned into the pGEMT-Easy P2R-P3 and pGEMT-Easy P4-P1R vectors, respectively, using Gateway technology (Invitrogen). Subsequently, a multisite LR reaction was performed with the above entry clones together with an E-GFP construct in the plasmid pGEMT-Easy 221 and the target binary vector pK7m34GW (purchased from Ghent University, Ghent, Belgium). All recombinations were confirmed by sequencing.

Stable transformation of cell suspension cultures and plants

Stably expressed AtAurora1-GFP, AtAurora1-RFP, AtTPX2-GFP and AtTPX2-RFP in A. thaliana suspension cultures of Columbia and Lansberg erecta were obtained according to the protocol of Mathur et al. (1998). AtAurora1-GFP, AtAurora1-RFP and AtAurora1-RNAi plants were obtained using the floral-dip method (Clough & Bent, 1998).

Quantitative real-time PCR (qPCR) analysis

qPCR experiments were performed following MIQE (The Minimum Information for Publication of Quantitative Real-Time PCR Experiments) recommendations (Bustin et al., 2009). Total RNA was isolated from A. thaliana control and AtAurora1-depleted plants using the Plant RNeasy Extraction Kit (Qiagen). Digestion of DNA during RNA purification was performed using the RNase-Free DNase Set (Qiagen). Then 100 ng of purified RNA was reverse-transcribed using the Transcriptor High Fidelity cDNA Synthesis Kit (Roche) with an anchored-oligo (dT)18 primer according to the manufacturer’s instructions. qPCR using EvaGreen Dye (Bio-Rad) was performed using the CFX96 Real-Time PCR Detection System (Bio-Rad). Three replicate PCR amplifications were performed for each sample. The PDF2 (regulatory subunit of protein phospatase 2A) gene (Czechowski et al., 2005) was used as a reference. Quantification of transcripts of each gene, normalized to the internal reference PDF2 gene, were determined using CFX Manager software (Bio-Rad). The transcript level of each target gene in control plants or the reference gene in control or RNAi plants was designated as 1.0. The primers used for real-time PCR were: PDF2, forward 5′-TAACGTGGCCAAAATGATGC-3′ and reverse 5′-GTTCTCCACAACCGCTTGGT-3′; AtAurora1, forward 5′-CCTACGGAGACACAACACCAG-3′ and reverse 5′-TTAAAGTCCATCTCTTTTGTGCAG-3′; AtAurora2, forward 5′-TGTGGCACTTTGGATTACCTT-3′ and reverse 5′-TCGACACTCGCATCATGTTC–3′; AtAurora3, forward 5′-CGAAGCCGAGAGTCAAAAAG-3′ and reverse 5′-GGGATCCTTAACAAGAAGCTGA-3′. To ensure specificity of primers, primer pairs were designed to span two neighbouring exons and were detected as a single peak in dissociation curve analysis.

Drug treatment

Taxol (Sigma-Aldrich) at a concentration of 5 μM, and amiprophos methyl (APM; Duchefa) at a concentration of 5 μM were diluted from dimethyl sulphoxide (DMSO) stock solutions.

Flow cytometry analysis

Suspensions of nuclei were prepared according to Galbraith et al. (1983) in Otto buffer (Doležel & Göhde, 1995). After staining with 4′,6-diamidino-2-phenylindole (DAPI) (2 μg ml−1), fluorescence intensity was measured using a FACSVantage SE flow cytometer (Becton Dickinson, San José, California, USA). Data were collected using CellQuest software (Becton Dickinson, USA). Mean endopolyploidy was analysed according to Barow & Jovtchew (2007).

Chromocentre labelling

Chromocentre labelling and counting were performed as described in Fransz et al. (2002). First leaves of A. thaliana control and RNAi seedlings were fixed in a 3 : 1 (v/v) ethanol/acetic acid mixture and, after hydrolysis in 0.1 N HCl, were stained for 10 min with 0.1 mg ml−1 of DAPI.

Cell extracts, differential centrifugation and immunopurification

Suspension cells were homogenized in liquid nitrogen and extracted in buffer (50 mM Na-Hepes, pH 7.5, 150 mM NaCl, 1mM MgCl26H2O and 1 mM EGTA with a protease and phosphatase inhibitor cocktail), as described in Drykováet al. (2003). Differential centrifugation was performed at 10 000 g for 10 min (S10), at 20 000 g for 30 min (S20) and at 100 000 g for 1 h (S100). The pellets (P10, P20 and P100) were dissolved in extraction buffer in a volume equal to the volume of the corresponding supernatants (S10, S20 and S100), and analysed by SDS-PAGE and by immunoblotting with anti-GFP antibody (Abcam, Cambridge, UK).

Co-immunoprecipitation was performed using GFP-Trap A and RFP-Trap A (ChromoTek, Planegg-Martinsried, Germany) according to the manufacturer’s instructions. In a standard procedure, cell extract S20 from the A. thaliana cell culture expressing AtAurora-RFP and AtTPX2-GFP (protein concentration 3–4 mg ml−1) was incubated with GFP-Trap or RFP-Trap beads for 1.5 h at 4°C. The immunoprecipitated proteins were released by elution with glycine, pH 2.5. Extracts from the A. thaliana cell culture expressing α-SNAP-RFP (soluble N-ethylmaleimide sensitive factor NSF attachment protein; At3g56190) that is predominantly present in cytoplasm (our unpublished data; A. thaliana eFP Browser; http://bar.utoronto.ca) were used as a negative control. Proteins in the eluates were resolved by 8% SDS-PAGE and analysed for AtAurora1-RFP and AtTPX2 by western blotting with anti-GFP and anti-RFP (Abcam and ChromoTek) antibodies.

Electrophoresis and immunoblotting

Proteins separated by SDS-PAGE were transferred onto a nitrocellulose membrane by wet electroblotting and immunoreactive bands were detected with primary and secondary antibodies and visualized using a Supersignal ECL kit (Thermo Scientific Pierce Protein Research Products, Rockford, Illinois, USA) according to the manufacturer’s instructions.

Microtubule sedimentation assay

Microtubule sedimentation experiments were performed as described previously by Drykováet al. (2003). Microtubules were polymerized from the S70 and S100 extracts that were supplemented with GTP and taxol to final concentrations of 1 mM and 20 mM, respectively. Taxol polymerized brain microtubules (Cytoskeleton Inc., Denver, Colorado, USA) were added to the polymerization mixture at a concentration of up to 10 μg ml−1 (plus brain microtubules variant). In controls, taxol and GTP were omitted. Microtubules were pelleted through a 40% sucrose cushion and after washing were resuspended in SDS sample buffer and loaded, together with the input and supernatants, on SDS-PAGE gels. Proteins were visualized by silver staining and analysed by western blotting.

Immunofluorescence

Arabidopsis thaliana seedlings or suspension cultures were processed for immunofluorescence as described in Binarováet al. (2006). Primary antibodies were: anti-α-tubulin monoclonal antibody DM1A (Sigma), monoclonal anti-γ-tubulin TU-32 (kindly provided by Pavel Dráber, Institute of Molecular Genetics, Prague, Czech Republic), affinity-purified rabbit polyclonal antibody AthTU (Drykováet al., 2003), anti-GFP antibody (Abcam), anti-AtKNOLLE antibody (Rose Biotech, Winchendon, Massachusetts, USA) and anti-kinesin KN-03 (ExBio, Prague, Czech Republic). The secondary antibodies used were anti-mouse fluorescein isothiocyanate (FITC)-conjugated antibody and anti-rabbit Cy3-conjugated antibody (Jackson Laboratories, West Grove, PA, USA). FM4-64 (Molecular Probes; Invitrogen) was applied as described by Völker et al. (2001).

Microscopy

Microscopy was performed on an IX81 motorized inverted research microscope with CellR (Olympus, Tokyo, Japan) equipped with a disk scanning unit (DSU) and digital monochrome CCD camera (CCD-ORCA/ER). To avoid filter crosstalk, fluorescence was detected using HQ 480/40 exciter and HQ 510/560 emitter filter cubes for FITC and HQ 545/30 exciter and HQ 610/75 emitter filter cubes for Cy3 (both AHF Analysen Technique, Tubingen, Germany). Excitation and emission wavelengths were 488 and 505–532 nm for FITC and 543 and 566–600 nm for Cy3. Green fluorescent protein was excited using 488 nm and imaged using an emission filter at 500/550 nm. Images were processed and analysed using CellR software (Olympus) and Quick Photo Camera software version 2.3 (Olympus). Figures were prepared in Corel Draw 9 and Adobe Photoshop 7.0. The quantitative colocalization analyses were performed using NIH Image J software with the Colocalization Finder plugin (http://rsbweb.nih.gov/ij/pluginsNakamura et al., 2007) and Pearson’s correlation coefficients (Rr) were calculated.

Database search

To identify genes with a similar pattern of gene expression to AtAurora1, we searched the publicly available expression databases ATTED II (Obayashi et al., 2011) and SuperCluster Editor 2.7 (Menges et al., 2008). To identify predicted domains and putative interaction sites, the ELM (Puntervoll et al., 2003), SUMOsp 2.0 (Ren et al., 2009), SMART (Letunic et al., 2006), and Pfam (Finn et al., 2006) databases were used. ClustalW2 was used for identity/similarity evaluation (Larkin et al., 2007). T-Coffee (Notredame et al., 2000) was used to perform multiple sequence alignment and to obtain phylograms.

Results

Phylogenetic analyses revealed the specificities of plant Aurora kinases AtAurora1, AtAurora2 and AtAurora3 (Kawabe et al., 2005) and their distances from mammalian A, B, and C Aurora kinases (Supporting Information Fig. S1). AtAurora1 and AtAurora2 were present as a segmental genome duplication between chromosomes 2 (AtAurora2) and 4 (AtAurora1); they have 95% protein sequence similarity (Fig. S2), which is higher than the similarity between human Aurora kinases A and B (57% similarity). Replacement of the Gly-198 residue within the catalytic domain of mammalian Aurora A with the equivalent residue Asn-142 of Aurora B converts the substrate specificity, regulator interactions, and function of Aurora A into those of Aurora B (Fu et al., 2009). As shown in Fig. S1, the equivalent residue within the catalytic domain of plant AtAurora1, AtAurora2 and AtAurora3 is a Gly, which suggests that the plant Aurora kinases lack the functional divergence of Auroras A and B. A database search revealed predicted sites for interactions of plant AtAurora1, AtAurora2 and AtAurora3 with cyclins, a conserved degradation box, putative sumoylation sites, MAPK (Mitogen-activated protein kinase) and protein phosphatase-1 docking motifs, and a retinoblastoma interaction site (Fig. S1b).

RNAi-silenced plants demonstrate cell division defects, arrested apical meristematic development, and ectopic meristem formation

We generated RNAi lines in which AtAurora1 was silenced with an RNAi construct spanning the entire mRNA. The T1 generation of A. thaliana transformants showed a high proportion of seedlings with pronounced growth retardation and severe developmental defects (Fig. 1a). Quantitative RT-PCR analysis of AtAurora gene expression levels (comparative Cq method; Pfaffl, 2001) showed a major reduction in transcript levels for AtAurora1 as well as AtAurora2 and AtAurora3 (Fig. 1b). Development of primary root meristems in RNAi seedlings was severely impaired, as shown in Fig. 1(c,d). Reinitiation of cell division occurred in some of the most severe phenotypes; an example of an ectopic meristematic region formed from a cotyledon is shown in Fig. 1(e). Primary root meristems were arrested in c. 85% of 14-d-old RNAi seedlings, and later, adventitious roots were initiated and differentiated from the hypocotyl boundary (Fig. 1f) or even from the central part of the hypocotyl (Fig. 1h).

Figure 1.

Phenotype of Arabidopsis thaliana plants with RNA interference (RNAi) down-regulated AtAurora kinases. (a) Representative images of 6-d-old RNAi seedlings that, compared with control plants, show defective development of primary root and shoot meristems. (b) Relative expression of AtAurora1, AtAurora2 and AtAurora3 kinases in a representative sample of 14-d-old RNAi plants (plant numbers 16, 17, 18 and 19) showed a significant reduction in transcript levels for all three kinases compared with the control wild-type plants. Error bars represent ± SE. (c, d) A 6-d-old RNAi seedling with primary root development arrested at (c) the triangular embryo stage or (d) the bent cotyledon embryo stage. (e) A 6-d-old RNAi seedling with (*) arrested primary root (R) and cotyledon leaf (L) development; (**) light and fluorescence microscopy showing an ectopic meristem formed from the base of a quiescent cotyledon leaf (arrowheads). (f) The lateral root meristem (arrow) of a 6-d-old RNAi plant was initiated and differentiated in close vicinity to the arrested primary meristem (arrowhead). (g) Representative images of 14-d-old RNAi seedlings with retarded development of primary root and shoot meristems. (h) Forty-five-day-old RNAi plants with arrested primary root meristem (arrowhead) and adventitious roots formed from the central part of the hypocotyl (arrow). (i, j) Swollen root hairs with loss of polarity and root hairs with two growth axes in 20-d-old RNAi seedlings. Bars: (a) 1 cm; (c–f) 1 mm; (g, h) 1 cm; (i, j) 10 μm.

In c. 11% of seedlings, primary roots developed further. However, regular cell files above the arrested meristematic zone were disturbed (Fig. 1i). Immunolabelling with the cytokinesis-specific syntaxin KNOLLE showed defects of cell plate formation, as demonstrated for 7-d-old RNAi roots in Fig. S3(a). Cells of irregular shape observed above a shortened division zone indicated that anisotropic growth was affected (Fig. 1i); cortical microtubules visualized by immunoflorescence were bundled and randomized compared with regular parallel cortical arrays in the wild-type roots (Fig. S3b). Ectopic root hair formation was observed in roots with shortened elongation zones; root hairs were often bulged and showed two axes, suggesting defective polar tip growth (Fig. 1j).

True leaves of RNAi plants were fragile, with signs of abaxiality. Young leaves developed under reduced levels of AtAurora kinases showed a needle-like linear shape and in severe cases lacked leaf lamina, suggesting that leaf axial patterning may be affected (Fig. 2a). Sporadically, trumpet-shaped leaves occurred with abaxial characters on the outside of the cup and adaxial characters inside the cup (Fig. 2b).

Figure 2.

Aerial parts of Arabidopsis thaliana plants with RNA interference (RNAi) down-regulated AtAurora kinases. (a) Curled fragile leaves and needle-like leaves (arrow). (b) A trumpet-shaped leaf. (c) Long extension in the apical division zone of a leaf and a trichome with an increased branch point number (arrow) and aberrant branch points. (d) A trichome with an increased branch point number and with a swollen base. (e) Clustered stomata, single guard cells (inset), incomplete guard cell division (arrow) and disturbed lobe formation in the leaf of an RNAi plant. (f) Frequencies of clustered stomata in stomata units (SUs) in AtAurora RNAi-expressing plants (grey, = 255) and control plants (black, = 250). (g) Frequencies of stomata with cytokinetic defects (grey) vs those with correct division (black); = 352 for control and = 329 for RNAi plants. Bars: (a, b) 1 mm; (c, d) 0.1 mm; (e) 10 μm.

Trichomes that developed under reduced levels of Aurora kinases exhibited supernumery branches (Fig. 2c) with the aberrant location of primary and secondary branch points (Fig. 2d). Severely affected leaves without signs of dorsoventral polarity were often trichomless (Fig. 2a). AtAurora RNAi plants showed defective stomatal development with guard cells that varied in size, spatial arrangement and developmental stage (Fig. 2e). We found that 48% of stomata were in clusters of two to five. Incompletely divided stomata or single guard cells (22%) indicated that, besides stomatal patterning, cell division was also affected (Fig. 2e).

Plants with down-regulated Aurora kinases showed higher levels of endoreduplication

Microscopic analysis of RNAi seedlings with a nuclear GFP marker showed that, compared with the wild-type plants, the first true leaves of RNAi plants showed a mixture of large and small nuclei (Fig. 3a). Large spindle-like strangulated nuclei were observed in hypocotyl cells of RNAi seedlings (Fig. 3b). Similarly, spindle-like or strangulated filamenous nuclei were typically found in root hairs of RNAi plants (Fig. 3c,d).

Figure 3.

Levels of endoreduplication were higher in RNA interference (RNAi) seedlings compared with wild-type seedlings. (a) The first true leaf of 15-d-old Arabidopsis thaliana wild-type and RNAi plants expressing the GFP nuclear marker. Enlarged nuclei of irregular shape were observed in the RNAi plants. (b) Nuclei of hypocotyl cells of RNAi plants with spindle-like strangulated nuclei (arrows). (c, d) Aberrant short primary root and root hairs with filamentous (arrow) and strangulated nuclei (arrowheads). (e) Nuclear DNA distribution of 14-d-old RNAi seedlings and control plants. Bars: (a–c) 10 μm; (d) 0.1 mm.

Flow cytometric analysis (FCM) of the DNA content of 2-wk-old RNAi plants showed an increase in the DNA content of a high proportion of cells with 8C and 16C DNA, while 2C and 4C cells remained similar to those of wild-type seedlings (Figs 3e, S4). We determined whether the observed ploidy increase is attributable to increased chromosomal number as a result of aberrant mitosis (polyploidy) or to the process of repeated S-phases, known as endoreduplication, which is developmentally switched on in A. thaliana plants when cells exit the proliferative stage and start elongating. To do this, we stained cells with DAPI and counted the number of chromocentres (heterochromatin aggregates in interphase nuclei that correspond to mitotic centromeres) in leaf epidermal cells (Fransz et al., 2002). We found that the number of chromocentres in RNAi plants (9.8 ± 1.1) did not differ from the control chromocentre number (9.6 ± 0.9). Therefore, chromocentre counting confirmed that the increased amount of DNA was not attributable to polyploidy but was caused by extra replication cycles.

Collectively, these data revealed that, in plants with silenced AtAurora kinases, defects in primary meristematic proliferation and cell division defects were accompanied by entry into an endoreduplication programme.

AtAurora1 kinase, AtTPX2 protein and γ-tubulin were associated with plant microtubules in vitro and were colocalized on microtubular arrays in a cell cycle-specific manner

To link cell division defects observed in RNAi plants specifically to AtAurora1, we studied the kinase and its activator AtTPX2 protein. We expressed AtAurora1-GFP using its own promoter in cultured cells and A. thaliana seedlings. A signal for AtAurora1-GFP kinase was observed with the mitotic microtubules from preprophase to anaphase, with a subset of phragmoplast microtubules in the anaphase/telophase transition and in the cell plate area (Fig. 4a,b). When expression of the AtAurora1-GFP kinase was driven from the 35S promoter, the localization pattern of AtAurora1 kinase with microtubules was similar to localization observed with the native promoter (Video S1). The AtAurora1-GFP signal with microtubules was stronger and a small portion of the constitutively expressed protein was present in nuclei and in the cytoplasm. We did not observe a gain of function phenotype in seedlings or cells expressing 35S AtAurora1-GFP from Gateway vectors. Double immunofluorescence of AtAurora1-GFP expressed from the native promoter and α-tubulin revealed a weak signal for the AtAurora1 kinase localized with the prospindle around nuclei and with the preprophase band of cortical microtubules (Fig. 4c). Double immunolocalization also confirmed that the AtAurora1 kinase was associated with a specific subset of early phragmoplast microtubules adjacent to chromatin and distal to the cell plate (Fig. 4c). A portion of the AtAurora1 kinase was translocated to the area of cell plate formation (Fig. 4a,b; Video S1). When AtAurora1-GFP or -RFP was transiently overexpressed from the 35S promoter, a weak signal was detected with cortical microtubules (Fig. S5).

Figure 4.

AtAurora1 kinase localization with microtubules in dividing cells of Arabidopsis thaliana. (a) AtAurora1-GFP expressed from the native promoter in A. thaliana cultured cells. AtAurora1-GFP localizes with the preprophase nucleus and with the mitotic spindle from metaphase to late anaphase. During early telophase, the kinase accumulates in the phragmoplast area (arrowheads) and at the newly formed cell plate (arrows). (b) Time-lapse images of AtAurora1-GFP in a dividing root meristematic cell of A. thaliana. The AtAurora1-GFP signal localized with the metaphase (10 min) and anaphase (14 min) spindle, with the early telophase phragmoplast facing towards chromatin (16 min and 19 min; arrowheads), and with the cell plate (16–36 min; arrows). (c) Double immunofluorescence labelling of AtAurora1-GFP expressed from the native promoter and microtubules. AtAurora1 localized with perinuclear microtubules of the prospindle (arrows) and decorated microtubules of the preprophase band (arrowheads) during preprophase. While AtAurora1 localized along the entire length of microtubules from metaphase to anaphase, only part of the phragmoplast distal to the cell plate was decorated with AtAurora1 kinase in late anaphase and in early telophase (arrowheads). Bars, 10 μm.

We investigated whether AtAurora1 kinase was present on microtubules together with its activator protein AtTPX2, similar to the Aurora A-TPX2 complex reported in vertebrates (Kufer et al., 2002). The A. thaliana homologue of the TPX2 protein with microtubular localization was described by Vos et al. (2008). A similar localization pattern for AtTPX2-GFP was found in our experiments; the signal was present in nuclei, in the vicinity of pre-mitotic nuclei, and with microtubular kinetochore fibres in mitosis. Furthermore, we observed translocation of AtTPX2 from anaphase poles to a subset of early phragmoplast microtubules (Fig. 5a). Analysis of the cell lines with stable co-expression of AtTPX2-GFP/AtAurora1-RFP revealed that both proteins colocalized on microtubules from preprophase to early telophase (Fig. 5b, Video S2). AtAurora1 and AtTPX2 were exclusively localized on the specific subset of phragmoplast microtubules in early telophase and only the AtAurora1 kinase was observed at the midzone and accumulated at the cell division plane (Fig. 5b). Colocalization of AtAurora1-RFP and AtTPX2-GFP was quantified and its significance determined with Pearson’s correlation coefficient (Rr) using colocalization plugins of ImageJ (http://rsbweb.nih.gov/ij/plugins/). As shown in Fig. 5(b), a high degree of colocalization for both proteins in cells was confirmed from prophase to early anaphase, while a lower correlation coefficient (Rr = 0.7) in early telophase reflected translocation of a portion of the kinase signal to the cell plate area. Colocalization analysis of AtTPX2-RFP and the MBD-GFP domain of the MAP4 protein and/or analysis of AtAurora1-GFP colocalization with mitotic spindle-associated kinesin recognized by the KN-03 antibody were performed as negative controls (Fig. S6). Compared with the high correlation coefficient (Rr = 0.9) generated for AtTPX2-GFP and AtAurora1-RFP colocalization in the metaphase and early anaphase spindle (Fig. 5b), the correlation coefficient for AtTPX2-RFP and MBD-GFP ranged between 0.5 and 0.7 in metaphase and was even lower in early anaphase, when the AtTPX2-RFP signal was enriched on spindle microtubules closer to the poles while the MBD-GFP signal localized along the entire length of the microtubules. Kinesin recognized by the KN-03 antibody and the AtAurora1-GFP protein both localized on the mitotic spindle but showed only a weak positive association (Rr ranged between 0.4 and 0.6) compared with that found for AtTPX2-GFP and AtAurora1-RFP (Figs 5b, S6).

Figure 5.

AtTPX2 (targeting protein for Xklp2) colocalized with AtAurora1 on microtubules in Arabidopsis thaliana cells. (a) AtTPX2-GFP signal was present with microtubules from preprophase to early telophase, where it localized on the specific subset of phragmoplast microtubules (arrows). (b) AtTPX2-GFP colocalized with AtAurora1-RFP on mitotic microtubules (Rr > 0.9) and with early phragmoplast microtubules (arrowheads) when the AtAurora1-RFP signal became localized at the cell division plane (Rr > 0.7). Individual frames presented in (b) are part of the sequence of frames in Supporting Information Video S2. (c) Distribution of AtAurora1 in cellular fractions after differential centrifugation. The majority of the AtAurora1-GFP protein was present as a soluble pool in the cytoplasm (S100), and a small portion was present in the pellets (P10, P20 and P100). (d) AtAurora1 and AtTPX2 proteins were pulled down with in vitro polymerized plant microtubules. Microtubules were prepared from extracts from the cell line expressing AtAurora1-RFP and AtTPX2-GFP by taxol-driven polymerization and analysed by immunoblotting for α-tubulin, GFP, RFP and γ-tubulin. The signal for AtTPX2 is weaker because of degradation (AtTPX2-GFP* indicates the proteolysis degradation product). Inputs: S70, 70 000 g supernatant; S, supernatant after the taxol polymerization; MTs+TAX, microtubular pellet; MTs+BMT, microtubular pellet from the assay supplemented with brain microtubules (BMTs); MTs-TAX, the negative control. (e) Graphical presentation of the AtTPX2 protein molecule with Aurora binding sites and targeting sites for microtubules. The 442–758 aa C-terminal part of the molecule lacking Aurora binding sites (AtTPX2ΔN) is marked in red. (f) Localization of AtTPX2ΔN-GFP in dividing cells of A. thaliana. The AtTPX2ΔN-GFP signal was present with microtubules from preprophase to late anaphase, while the whole molecule AtTPX2 and AtAurora1 persisted with microtubules until early telophase, as shown in (a). Bars: (a, b, f) 10 μm. MTs, microtubules.

Fractionation of cell extracts by differential centrifugation showed only a minor portion of AtAurora1-GFP in low- and high-speed pellets, and the majority of the AtAurora1-GFP protein was present in the soluble cytoplasmic S100 pool (Fig. 5c). To confirm biochemically the association of AtAurora1 and AtTPX2 with microtubules, we performed in vitro taxol-driven polymerization of plant microtubules (Drykováet al., 2003). High-speed S100 or S70 supernatants were used (Fig. S7). AtAurora1 kinase-GFP pelleted together with taxol-polymerized plant microtubules (Fig. 5d). As shown by Vos et al. (2008), the AtTPX2 protein is highly unstable in plant cell extracts, and this makes detection by western blotting difficult. However, despite a reduction in signal, we could detect AtTPX2 together with AtAurora1 in microtubular pellets (Fig. 5d). AtAurora1 kinase-GFP expressed both from the 35S promoter and from the native promoter was pelleted with taxol-polymerized microtubules. The kinase pool with microtubules was not further enriched when the assay was supplemented with taxol-stabilized brain microtubules. The kinase was also not pelleted in minus-taxol controls. Results from microtubule spin-down experiments are thus consistent with microscopic observations of AtAurora1 and AtTPX2 localizing with microtubules (Fig. 5b, Video S2).

To confirm the interaction of AtAurora1 and the AtTPX2 protein revealed by colocalization analysis at the biochemical level, we performed co-purification experiments. As shown in Fig. S8, AtAurora1-RFP copurified with AtTPX2-GFP isolated on a GFP Trap. Several bands for the AtTPX2-GFP protein were immunodetected using the anti-GFP antibody in a sample of purified AtTPX2-GFP as a result of the above-mentioned instability of the AtTPX2 protein in plant cell extracts. When extracts of cells expressing α-SNAP-RFP were used as a negative control, we did not detect α-SNAP-RFP nonspecifically bound to the GFP Trap bead (Fig. S8).

Consistent with our previous data (Drykováet al., 2003), γ-tubulin pelleted with the taxol-polymerized fraction of microtubules (Fig. 5d). Double immunofluorescence analysis confirmed that AtAurora1 and AtTPX2 colocalized with γ-tubulin in the prospindle during preprophase, along the entire length of spindle microtubules during metaphase and anaphase, and with a subset of early phragmoplast microtubules (Fig. 6a,b). While AtAurora1 and AtTPX2 labelling disappeared from phragmoplast microtubules during early telophase, γ-tubulin remained associated with phragmoplast microtubules until the end of cytokinesis (Fig. 6a,b). As reported previously, γ-tubulin is present on kinetochore microtubule stubs during recovery from depolymerization by APM (Binarováet al., 2000). Our data suggest that AtAurora1 and AtTPX2 were loaded to microtubules regrowing from the kinetochore area, together with γ-tubulin (Fig. 6c,d).

Figure 6.

AtAurora1 and AtTPX2 (targeting protein for Xklp2) localized on microtubules together with γ-tubulin in a cell cycle-specific manner. (a, b) Immunolabelling of (a) AtAurora1 and (b) AtTPX2 with γ-tubulin and 4′,6-diamidino-2-phenylindole (DAPI)-stained chromatin in AtAurora1-GFP- and/or AtTPX2-GFP-expressing Arabidopsis thaliana cells. AtAurora1 and AtTPX2 localized together with γ-tubulin on microtubules from preprophase until early telophase (arrowheads). Later in telophase, γ-tubulin persisted on phragmoplast microtubules while (a) AtAurora1 was observed in the cell plate area (arrows) and (b) the signal for AtTPX2 diminished. (c, d) Immunolocalization of (c) AtAurora1-RFP and (d) AtTPX2-GFP together with α-tubulin and γ-tubulin in cells after 3 h of amiprophos methyl (APM) treatment. Both AtAurora1 and AtTPX2 colocalized with γ-tubulin on persistent kinetochore microtubular stubs where kinetochore microtubules were regrowing. Bars, 10 μm.

We prepared a GFP fusion with AtTPX2 that lacked the N-terminal part of the molecule (AtTPX2ΔN-GFP) where two conserved Aurora kinase binding sites are located (Fig. 5e). As expected, the truncated version of AtTPX2, with a preserved microtubule binding motif, localized with microtubules. However, translocation from the poles of the anaphase spindle to early phragmoplast microtubules in the vicinity of chromatin, as observed for the intact AtTPX2 protein and AtAurora1 kinase, was not observed for AtTPX2ΔN-GFP in cultured cells or in dividing cells of A. thaliana seedlings (Figs 5f, S9). Colocalization analysis (Fig. S10) confirmed that signal from AtTPX2ΔN-GFP lacking the Aurora binding sites disappeared from anaphase kinetochore fibres earlier than was observed for the intact AtTPX2-binding AtAurora1.

These experiments showed that, similar to vertebrate Aurora A, plant AtAurora1 was colocalized with its activator AtTPX2 on mitotic microtubules. The active kinase plus AtTPX2 complex localized to microtubules from preprophase to early telophase together with γ-tubulin. Interaction with AtAurora1 kinase is likely to be required for stability of AtTPX2 and its cell cycle-specific localization with microtubules in late mitosis.

AtAurora1, but not AtTPX2 or γ-tubulin, was translocated to the cell plate area

While AtTPX2/AtAurora1 formed a gradient emanating from the poles to a specific subset of phragmoplast microtubules during the late mitosis/telophase transition, only AtAurora1 kinase relocalized to the cell plate area during cytokinesis (Fig. 7a,b). A distinct midzone signal for AtAurora1 kinase became detectable in early telophase, and the spatial and temporal localization pattern of the kinase with the cell plate was similar to that demonstrated by staining with the endocytic tracker FM4-64 (Fig. 7c). Double immunofluorescence showed colocalization of AtAurora1 kinase with cytokinesis-specific syntaxin KNOLLE. Both proteins were present on the expanding cell plate until it reached the mother cell wall (Fig. S11). AtAurora1, AtTPX2, and γ-tubulin localized on microtubules from preprophase until early telophase, but only AtAurora1 was found in the cell plate area; this suggests a cytokinesis-specific function.

Figure 7.

AtAurora1-GFP colocalization with cell plate-specific markers during telophase in cultured Arabidopsis thaliana cells. Both AtAurora1-GFP and FM4-64 were localized in the cell plate area (arrowheads) from early until late telophase. Bars, 10 μm.

Discussion

The presence of conserved sites that mediate interactions with cell cycle regulatory molecules, as well as data from publicly available expression databases (Menges et al., 2008; Obayashi et al., 2011), suggests that the plant Aurora kinases may regulate multiple steps throughout the cell cycle, similarly to mammalian Aurora A and Aurora B. However, functional data on plant Aurora kinases are limited. A specific inhibitor of Aurora B kinase, hesperadin, was shown to reduce AtAurora3 kinase activity in vitro and to affect chromatid separation in tobacco BY2 cells. Hesperadin inhibition of AtAurora1 kinase activity was also reported (Kurihara et al., 2006; Demidov et al., 2009). However, treatments with hesperadin were carried out at doses approx. 40 times higher than the doses used for specific inhibition of Aurora B kinase in mammalian cells; nonspecific inhibitory effects on cyclin-dependent kinases or MAP kinases, as reported for the drug by Hauf et al. (2003), cannot be excluded. Therefore, we used an RNAi approach to analyse post-embryonic development of A. thaliana plants containing reduced levels of AtAurora kinases.

In addition to their role in cell division, AtAurora kinases are required for the maintenance of meristematic activity and for entering endocycles

The presence of single guard cells, multinuclear cells, and irregular cell files indicates defective cell division, as expected for plants with impaired functions of Aurora kinases. However, we also found that, depending on the strength of the phenotype, development of primary meristems was arrested or delayed. As postembryonic development in plants is mediated by meristems which provide a source of newly generated undifferentiated cells, a failure of cells within primary meristems to progress through the mitotic cycle in Aurora RNAi plants had a pronounced impact on growth and development. There are many types of mutants with defective shoot or root meristematic activity, including mutants with specific inhibition of CDK complexes (Andersen et al., 2008), transcription factor mutants (Casson et al., 2009), mutants in the auxin response pathway (Himanen et al., 2002) and D-type cyclin mutants (Nieuwland et al., 2009). Our data showed that the plant Aurora kinases are implicated in maintaining meristematic cells in the cell cycle.

The reduction or absence of root meristematic activity observed in Aurora RNAi seedlings was accompanied by premature differentiation of cells in the elongation zones. Cells with large nuclei or with several nuclei of irregular shape and size indicated defective cell divisions in roots as well as in the aerial part of the seedlings. We found that endoreduplication was higher in AtAurora RNAi plants than in wild-type plants. The ability to initiate endoreduplication is particularly relevant during development; in A. thaliana, endoreduplication occurs in 12–13-d-old seedlings (Galbraith et al., 1991; Beemster et al., 2005). Our finding of increased levels of endoreduplication in 14-d-old Aurora RNAi plants indicates premature entry into endocycles under reduced levels of Aurora kinases. Similarly, our observation of trichomes with increased numbers of irregularly located branch points in RNAi plants implicates the kinases in the control of endocycles, as the level of endoreduplication is known to be critical for regular trichome branching (Sugimoto-Shirasu & Roberts, 2003).

The transition from mitotic cycles to endocycles is interlinked with the developmental progression of cell proliferation, cell expansion, and differentiation. The timing of endocycle onset needs to be precisely controlled by the cell cycle regulatory machinery; for example, modified expression of Cdk1 inhibitory Wee1 kinase, or SUMO E3 (Small Ubiquitin-like Modifier E3) ligase, resulted in premature entry into endoreduplication (Sun et al., 1999; Ishida et al., 2009). By contrast, overexpression of the cyclin D gene CycD3;1 resulted in a failure of differentiating cells to exit the mitotic cell cycle and undergo endoreduplication (Dewitte et al., 2003).

We found that, in seedlings with down-regulated Aurora kinases, continuing cell division or endoreduplication occurred, depending on the cell type; the ectopic cell divisions were observed in stomata-forming cell files, whereas enlarged cells with large nuclei that are hallmarks of endoreduplication were observed in the epidermal cell files of hypocotyls. Similar developmental defects were described in seedlings overproducing E2Fa-DPa (transcription factor with domain for dimerization), which is a component of the retinoblastoma-related (RBR)-E2F pathway (De Veylder et al., 2002). In mammalian cells it has been shown that Aurora B phosphorylates the retinoblastoma (RB) protein in vitro and in vivo at serine 780 (Nair et al., 2009). Inhibition of Aurora B led to RB hypophosphorylation accompanied by endoreduplication which occurred independently upon cyclin-dependent kinase inhibition. Aurora B kinase has been suggested to directly regulate the RB protein and the postmitotic checkpoint by preventing endoreduplication in cells with aberrant mitosis. However, further experiments are needed to establish whether phosphorylation of an RBR protein by plant Aurora kinases might occur and thereby prevent premature entry into endoreduplication.

AtAurora1 and AtTPX2 function with microtubules from preprophase to the late mitosis/telophase transition

As previously shown by Demidov et al. (2005) and Kawabe et al. (2005), AtAurora1 kinase, expressed from a constitutive 35S promoter in tobacco BY2, was associated with microtubules. Using AtAurora1 expressed from its own promoter in A. thaliana, we have demonstrated that association of the kinase with microtubules and with the forming cell plate was not the result of overproduction of the protein in heterologous tobacco cells. As shown recently by Kettenbach et al. (2011), interaction with activating subunits and spatial distribution are crucial factors in the regulation of functional modules of Aurora kinases. The TPX2 protein is the activating subunit for Aurora A (Bayliss et al., 2003), and activation of mammalian Aurora A kinase in vitro was observed also with the plant homologue AtTPX2 (Vos et al., 2008). Our data on colocalization of AtTPX2 with AtAurora1 kinase on microtubules suggest that AtTPX2 may guide Aurora kinase to microtubules and, in analogy with other systems, TPX2 may spatially and temporally modulate the function of plant AtAurora1.

Aurora A kinase activity is required for nucleation of microtubules with or without centrosomes (Tsai & Zheng, 2005). A growing body of evidence suggests that γ-tubulin is directly involved in the coordination of signals at the microtubule organizing centre with the cortex or with kinetochores (Cuschieri et al., 2007; Lin et al., 2011). The presence of AtAurora1 and its activating subunit AtTPX2 with γ-tubulin on microtubules suggests a link between this kinase and γ-tubulin-mediated processes such as acentrosomal plant microtubule nucleation from existing microtubules (Murata et al., 2005). Furthermore, Aurora kinases may be among multiple kinases that affect proteins of the MAP65 family, as Aurora kinase phosphorylation sites were predicted in synthetic peptides generated from the AtMAP65 sequence (Smertenko et al., 2006). We found in the publicly available Affymetrix expression databases that MAP65-4, a gene encoding a protein that cross-links newly formed spindle microtubules (Fache et al., 2010), is among genes with correlation to the AtAurora1 gene (Menges et al., 2008; Obayashi et al., 2011). We suggest that the active AtAurora1 and AtTPX2 regulatory module may act with the γ-tubulin nucleation complex and other substrates such as the microtubular crosslinker MAP65-4, or perhaps with molecular motors, in spatiotemporal regulation of acentrosomal plant spindle formation. Recently, a kinase-independent function of Aurora A was reported to exist along with Aurora A kinase function in Caenorhabditis elegans, and both forms closely cooperate in the assembly of spindle microtubules (Toya et al., 2011). Addressing the question of whether similar nonkinase and kinase multifunctionality also exists for AtAurora1 or other plant Aurora kinases presents a challenge for future research.

We found that the association of AtTPX2 and AtAurora1 may protect AtTPX2 from degradation, as has been reported for mammalian TPX2/AuroraA counterparts (Steward & Fang, 2005). However, while in animal cells the Aurora A/TPX2 complex is degraded at mitosis exit, a polar gradient of active AtAurora1/AtTPX2 persisted on microtubules until early telophase, when it localized with a specific subset of early phragmoplast microtubules in the vicinity of separating chromatin. After AtTPX2 degradation, the AtAurora1 kinase was translocated from that site to the cell plate area at the midzone. AtAurora1 thus behaves during the mitosis–cytokinesis transition similarly to Aurora B, which forms a gradient centred on the midzone during anaphase that is dependent on activation of the kinase on the subpopulation of microtubules (Fuller et al., 2008).

The presence of AtAurora1 kinase at the active cortex zone during prophase, and at the cell division site throughout cytokinesis, as shown by Demidov et al. (2005) and Kawabe et al. (2005) and demonstrated in our experiments, may transmit cortical information to the cell division plane. In animal cells, the correct location of a cell division site is dictated mainly by the proper localization of chromosomal passenger protein complexes that are substrates for Aurora B kinase at the cleavage furrow. No chromosomal passengers, such as INCENP (Inner centromere protein), survivin and borealin, are found in higher plant genomes. Expression of the kinesin-like protein NACK1 (NPK1-activating kinesin 1), an activator of NPK1 (tobacco MAPK kinase kinase – MAPKKK) that regulates lateral expansion in cytokinesis, and expression of the cytokinesis-specific syntaxin KNOLLE, correspond to expression of AtAurora1 in the Affymetrix expression databases (Menges et al., 2008; Obayashi et al., 2011). Consistent with that finding, AtAurora1 colocalized with KNOLLE to the nascent cell plate and persisted in the plane of cell division until the cell plate fused with the mother cell wall. Furthermore, as we have shown previously (Weingartner et al., 2004), correct localization of the kinesin-like protein NACK1 in the midzone and with the cell plate that resembles cytokinetic AtAurora1 localization depends on regular function of cell cycle machinery. However, a role of AtAurora1 kinase in the MAP kinase–KNOLLE pathway which provides spatial information for plant cytokinesis is only hypothetical and needs to be validated experimentaly.

Down-regulation revealed that, despite the absence of centrosomes as well as homologues of chromosomal passengers in plant cells, the Aurora kinases play a conserved role in the regulation of cell division. The presence of AtAurora1 and AtTPX2 on microtubules and the translocation of AtAurora1 to the area of cell plate formation may serve as an effective mechanism for defining critical microtubule-bound and cytokinetic substrates. Furthermore, our data suggest that plant Aurora kinases are an integral component of regulatory signalling for meristem maintenance, differentiation, and entry into the endoreduplication programme.

Acknowledgements

We thank CSIRO, MGP and Dr Ian Moore of the University of Oxford for providing pOpOff Gateway binary vectors, Dr Richard Cyr (Pennsylvania State University, USA) for providing the Map4-GFP vector, and Dr Roger Y. Tsien (Howard Hughes Medical Institute and Dept of Pharmacology, San Diego, CA, USA) for mRFP. We thank Dr Pavla Suchánková (Centre of the Region Haná for Biotechnological and Agricultural Research, IEB ASCR, Olomouc, Czech Republic) for the FCM analysis. This work was supported by grants 204/07/1169, 204/09/P155 and GD 204/09/H084 from the Grant Agency of the Czech Republic, grants LC06034 and LC545 from the Ministry of Education, Youth and Sports of the Czech Republic, grant IAA500200719 from the Grant Agency of the Czech Academy of the Sciences, grant No. CZ.1.05/2.1.00/01.0007 to B.P. from the Centre of the Region Haná for Biotechnological and Agricultural Research, and Institutional Research Concepts grants AV0Z50200510 and AV0Z50380511.

Ancillary