The microtubule-associated kinase-like protein RUNKEL functions in somatic and syncytial cytokinesis

Authors


For correspondence (e-mail tamara.krupnova@zmbp.uni-tuebingen.de).

Summary

The microtubule (MT)-associated putative kinase RUNKEL (RUK) is an important component of the phragmoplast machinery involved in cell plate formation in Arabidopsis somatic cytokinesis. Since loss-of-function ruk mutants display seedling lethality, it was previously not known whether RUK functions in mature sporophytes or during gametophyte development. In this study we utilized RUK proteins that lack the N-terminal kinase domain to further examine biological processes related to RUK function. Truncated RUK proteins when expressed in wild-type Arabidopsis plants cause cellularization defects not only in seedlings and adult tissues but also during male meiocyte development, resulting in abnormal pollen and reduced fertility. Ultrastructural analysis of male tetrads revealed irregular and incomplete or absent intersporal cell walls, caused by disorganized radial MT arrays. Moreover, in ruk mutants endosperm cellularization defects were also caused by disorganized radial MT arrays. Intriguingly, in seedlings expressing truncated RUK proteins, the kinesin HINKEL, which is required for the activation of a mitogen-activated protein kinase signaling pathway regulating phragmoplast expansion, was mislocalized. Together, these observations support a common role for RUK in both phragmoplast-based cytokinesis in somatic cells and syncytial cytokinesis in reproductive cells.

Introduction

In plants, distinct mechanisms of cytokinesis achieve separation of daughter cells in different cell types. In cells of somatic tissues, the cytokinetic phragmoplast, consisting of microtubules (MTs), microfilaments and endoplasmic reticulum, supports the formation of a tubulo-vesicular structure, the cell plate (Jürgens, 2005). Phragmoplast MTs serve as tracks for Golgi-derived vesicles towards the phragmoplast midzone where they fuse. Microtubules rapidly depolymerize in areas of completed vesicle fusion at the lagging edge of the phragmoplast and polymerize at the leading edge, resulting in centrifugal phragmoplast and cell plate expansion towards the cortical division site (Müller, 2011).

In syncytial cytokinesis observed in male and female gametogenesis as well as in endosperm development, the radial MT arrays (RMA) define the nuclear-cytoplasmic domains, which are later confined by intersporal callose walls that are formed between so-called mini-phragmoplasts (Otegui and Staehelin, 2004).

Microtubule depolymerization in the phragmoplast is controlled by the well-defined mitogen-activated protein kinase (MAPK) signaling pathway NACK-PQR that targets MT associated proteins (MAPs) (Nishihama et al., 2002; Sasabe et al., 2011a; Sasabe and Machida, 2012). Experimental evidence suggests that a similar MAPK signaling pathway regulates syncytial cell plate formation in male meiocytes (Zeng et al., 2011). The MAPK signaling cascade is activated by the kinesin-like proteins HINKEL (HIK)/NtNACK1 in somatic cytokinesis and TETRASPORE (TES)/NtNACK2 in male meiocytes (Tanaka et al., 2004). While knock-out of HIK causes defects in somatic phragmoplast expansion (Strompen et al., 2002), knock-out mutants of TES display RMA defects in male meiocytes (Hülskamp et al., 1997; Spielman et al., 1997; Yang et al., 2003). Typical defects associated with mutations in MAPK signaling pathway components or their targets include arrested phragmoplast expansion and multinucleate cells, cell wall fragments and unfused vesicles in the plane of cell division (Hülskamp et al., 1997; Strompen et al., 2002; Yang et al., 2003; Müller et al., 2004; Tanaka et al., 2004; Kosetsu et al., 2010; Takahashi et al., 2010). Interestingly, defects in both male and female syncytial cytokinesis were observed in tes;hik double mutants, suggesting functional redundancy of TES and HIK (Tanaka et al., 2004; Oh et al., 2008).

Only a few proteins known to be required for phragmoplast expansion in somatic cells have also been shown to be involved in cytoskeletal rearrangements in germline cells, such as the evolutionarily conserved XMAP125 (Whittington et al., 2001; Hussey et al., 2002; Twell et al., 2002), the putative kinase TWO IN ONE (TIO) and PAKRP1/Kinesin-12A and PAKRP1L/Kinesin-12B (Lee et al., 2007; Ho et al., 2011). Moreover, TIO directly interacts with both kinesin-12 proteins (Oh et al., 2012).

Previous reports established the role of the putative serine–threonine kinase gene RUNKEL (RUK) in somatic cytokinesis (Mayer et al., 1991; Krupnova et al., 2009). Five different ruk mutant alleles were defective in embryo and seedling development, displaying cytokinesis defects in somatic tissues such as cell wall stubs and multinucleate cells (Krupnova et al., 2009). Biochemical analysis demonstrated that RUK is a MAP associating with the preprophase band, spindle and phragmoplast during mitosis (Krupnova et al., 2009). In ruk mutants, phragmoplasts were not as broad as in the wild-type, while their midzone was expanded, and phragmoplast expansion was arrested, consistent with the requirement for the RUK gene product in phragmoplast-dependent cell plate formation (Krupnova et al., 2009).

In the present study we expressed truncated RUK proteins lacking the N-terminal kinase domain under control of the endogenous promoter in wild-type plants. We observed cellularization defects in microsporogenesis, interfering with intersporal callose wall formation in male tetrads, allowing us to characterize a previously unknown function for RUK in male meiosis. The defects in male meiocyte development were reminiscent of cellular abnormalities in weak tes alleles, while somatic cytokinesis defects in seedlings expressing the RUK truncated protein lacking the kinase domain resembled the ruk phenotype. We also documented abnormalities of endosperm cellularization in ruk mutants. Here we show that both defects in tetrads and endosperm cellularization result from disorganized radial RMAs. Furthermore, expression of truncated RUK proteins caused mistargeting of the kinesin HIK and the syntaxin KNOLLE.

Results

Expression of truncated RUK protein causes morphological defects

The RUK protein contains a putative N-terminal kinase domain, followed by a MT-binding domain (Figure S1a). Under the control of the RUK endogenous promoter, transgenic constructs comprising a series of deletions spanning the RUK MT-binding domain were expressed in Arabidopsis wild-type plants (Figure S1a). In T1 progenies, five of the seven gene fragments (FR2, -4, -5, -6, -7; Figure S1a) caused similar morphological defects in siliques (Figures 1a,b and S1a). Subsequent phenotypic and genetic analysis focused on transgenic lines expressing truncated protein FR5 which displayed the highest proportion (17%) of defects (Figure S1a). FR5 siliques were shorter than wild-type (Figure 1a) and contained unfertilized and aborted ovules (Figure 1b).

Figure 1.

Phenotype of FR5 plants. (a) Comparison of length differences of mature siliques from wild-type (wt) and transgenic plants expressing FR5. (b) Compared with wild-type, FR5-expressing plants display unfertilized ovules (arrowhead) and brown, aborted ovules (arrow), indicating partial sterility. Scale bar = 1 mm. (c), (d) Mature pollen grains (arrowheads) in anthers of wild-type (c) and FR5 plants (d). Scale bar = 20 μm.

To determine the origin of unfertilized ovules, we performed reciprocal crosses between FR5 T1 plants and wild-type. The pollination of wild-type flowers with FR5 pollen resulted in 30% unfertilized and 9% aborted ovules (Table S1). When wild-type was the pollen donor for FR5 plants, 39% unfertilized and 6% aborted ovules were counted. The frequency of sterile and aborted ovules was even higher in FR5 siliques after self-pollination (49% sterile and 6% aborted ovules; Table S1). Thus, transmission of both male and female gametophytes was reduced in FR5-expressing plants.

Fusion of the endogenous RUK promoter with the beta-glucuronidase (GUS) reporter gene was used to analyze gene expression of RUK in developing floral tissues including meiosis active stages (Figure S2a–c). Consistent with previously reported RUK function in mitosis (Krupnova et al., 2009), pRUK::GUS was detected in mitotically active tissues such as young rosette leaves (Figure S2d,e), lateral roots (Figure S2f) and embryos (Figure S2g).

In addition to shorter siliques and unfertilized ovules, up to 5% of FR5 T2 seedlings also exhibited moderate cytokinesis defects (Table S2) similar to ruk seedlings (Krupnova et al., 2009). FR5 seedlings displayed cell wall fragments in hypocotyls (Figure S3a–d) and roots (data not shown) and cell wall patterns in root meristems were disorganized (Figure S3e–g) further implying that FR5 expression interfered with endogenous RUK function.

We performed semi-quantitative RT-PCR analysis of RUK and FR5 mRNA in T2 and T3 progenies. Plants displaying strong mutant phenotypes had less endogenous RUK transcript in inflorescences compared with the wild-type, but FR5 transcript was stably expressed (Figure S3h,i), arguing for a dominant negative effect of FR5 expression.

FR5 plants are pollen defective

The results of the reciprocal crosses suggested that FR5 expression caused a reduction of gamete transmission. Thus, we investigated the development of pollen and found that anthers of FR5 plants contained abnormal pollen grains (Figures 1c,d and 2). In the wild-type, mature tricellular pollen contained two compact generative nuclei and one diffuse vegetative nucleus (Figure 2a). In contrast, FR5 pollen grains varied in size, shape and number of nuclei. Only 24% of pollen grains produced by FR5 plants appeared normal, whereas 48% enclosed de-condensed or elongated nuclei (Figure 2b,c) or contained enlarged nuclei and variable numbers of vegetative and generative nuclei (Figure 2d,e). The remaining 28% of FR5 pollen grains were aborted (Figure 2f,i).

Figure 2.

Pollen grain phenotypes of FR5 plants. (a)–(e) Epifluorescence and bright-field light microscopy of 4′,6-diamidino-2-phenylindole (DAPI)-stained pollen grains. (a) Tri-nuclear wild-type pollen displaying one vegetative and two generative nuclei. (b), (c) Enlarged FR5 pollen grains exhibit diffuse (b) or elongated nuclei. (c) Arrowheads indicate nuclei. (d), (e) FR5 pollen grains contain enlarged nuclei (d) and inconsistent numbers of vegetative and generative nuclei (e). (f) FR5 pollen grains abort before maturation. (g) qrt tetrads contain four wild-type-looking pollen grains of equivalent size and shape. (h), (i) FR5 pollen grains in the qrt mutant background are asymmetrically placed within the tetrad, variable in size and occasionally abort (asterisks). Scale bar = 27 μm.

In FR5, a considerable proportion of pollen grains were enlarged. While the ellipsoid, diploid wild-type pollen grains were about 25.5 μm long and about 20.3 μm wide on average, pollen of randomly chosen FR5 T3 plants ranged from 31.7 to 58.9 μm in length and from 26.3 to 32.2 μm in width (Table S3). The sizes of FR5 pollen could be categorized into two approximate length classes of approximately 30–40 μm (class I, Table S3, light gray) and approximately 40–60 μm (class II, Table S3, dark gray). Notably, diploid wild-type pollen was smaller than either class I or class II of FR5 pollen while tetraploid wild-type pollen was within the class I size range (Yu et al., 2010) (Table S3).

Similar defects were observed when FR5 plants were crossed with the quartet (qrt) mutant, which fails to separate the four microspores after meiosis due to the persistence of pectin polysaccharides (Figure 2g) (Preuss et al., 1994). In contrast to the qrt mutant, tetrads of qrt-expressing FR5 contained two, three or four microspores of variable size and number of nuclei (Figure 2h,i). In summary, the abnormalities observed in FR5 pointed towards defects occurring during early gametogenesis, but not during pollen mitosis.

FR5 microspores display cellularization defects

Upon meiosis the microspore mother cell (MMC) gives rise to four microspores, collectively termed the tetrad. A callose wall surrounds the tetrad, and microspores are separated by centripetal formation of the intersporal callose wall during meiocyte cellularization (Enns et al., 2005). Eventually, dissolution of the callose walls releases mature microspores from the tetrad. To investigate the impact of FR5 expression on microsporogenesis, we visualized callose of meiocytes at the tetrad stage with methyl blue staining (Enns et al., 2005). In wild-type tetrads, intersporal callose walls were clearly present around and between the post-meiotic nuclei (Figure 3a). In contrast, intersporal callose walls were entirely absent in approximately 15% and only partially formed in approximately 80% of FR5 tetrads (= 125; Figure 3b–d), indicative of incomplete division and similar to the weak tes-2 phenotypes (Figure 3e,f). Thus, FR5 expression resulted in defects in meiocyte cellularization.

Figure 3.

Cellularization defects in FR5 and tes-2 male meiocytes. (a) In wild-type (wt) tetrads intersporal callose walls separate four newly formed microspore nuclei after cellularization. (b)–(d) Intersporal callose cell walls in FR5 male meiocytes are completely absent (b) or incomplete (c, d). (e), (f) tes-2 male meiocytes display incomplete intersporal callose cell walls. The external callose wall is mechanically removed in (f). Scale bar = 10 μm.

To investigate microsporogenesis in more detail we performed ultrastructural analysis. In wild-type MMCs (Figure 4b,d) the intersporal callose wall is initiated from the periphery and wall formation proceeds centripetally to separate microspores. In contrast, in FR5 MMCs, inward-directed formation of intersporal callose walls was often initiated but not completed, and callose wall maturation was arrested or delayed (Figures 4a,c,g and S4c). In post-meiotic FR5 MMCs and similar to wild-type, mitochondria and plastids assembled at the so-called organelle band in the plane of cell division (Otegui and Staehelin, 2004). Also in FR5, individual mini-phragmoplast MTs were present in close proximity to vesicles in the plane of cell division (Figure S4a).

Figure 4.

Ultrastructural analysis of FR5 male meiocytes.(a) Mature FR5 male meiocyte features misplaced wavy and incomplete intersporal callose walls. (b) In wild-type male meiocytes intersporal callose walls form by centripetally directed furrowing. (c) Magnification of boxed area in (a). (d) Magnification of boxed area in (b). (e)–(g) FR5 microspores directly after meiocyte release, before mitosis I. (e) Internal cell wall (arrowheads) without exine layer separates FR5 microspores. (f) Incomplete cell wall (boxed area) with exine layer (Ex) separates mutant FR5 microspores. (g) Cell wall stub (CWS) in FR5 microspore. C, callose wall; P, plastid; M, mitochondrion; N, nucleus; T, tapetum. Scale bars = 1 μm.

Furthermore, FR5 microspores contained furrowing or partially internalized exines in addition to the outer exine (Figures 4f and S4e) and displayed persisting cell wall stubs (Figures 4g and S4d), which were generally absent from wild-type microspores (Otegui and Staehelin, 2004). As a consequence, microspores released from FR5 tetrads were typically multinucleate (Figure S4b) and incompletely separated (Figure S4e,f).

Radial MT systems in FR5 microspore mother cells are disorganized

Before cellularization in the MMC, RMAs radiate from four post-meiotic nuclei and overlap at the future plane of cell division (Otegui and Staehelin, 2004). Upon cytokinesis, mini-phragmoplasts assemble in the region of former RMA overlap and facilitate the formation of the intersporal callose wall. We analyzed the MT cytoskeleton in FR5 and wild-type meiocytes using immunostaining to visualize α-tubulin (Figures 5a–f and S5). The organization of the meiotic single (= 30; Figure S5a,b) and double bipolar spindles (= 25; Figure S5e,f) and the RMAs at the end of telophase I (= 25; Figure S5c,d) was normal.

Figure 5.

Immunolocalization of tubulin in tetrads and microspores of wild-type (wt) and FR5 plants and in ruk endosperm. (a), (b) Radial microtubule arrays (RMA) in developing tetrads emanate from nuclei (N) in (a) wild-type microspore mother cell (MMC) and in (b) FR5 MMC. (c), (d) Organization of RMA tetrad during intersporal callose wall formation (arrowheads). (c) The RMAs are well-organized in the wild-type tetrad and (d) partially disorganized in the FR5 tetrad indicated by uneven distribution of nuclei. (e) Four microspores of wild-type plants released from tetrad show the interphase radial microtubules (MTs) around each nucleus. (f) FR5 microspores are not discretely released from the tetrad. Interphase radial MTs are in close proximity of sister nuclei (arrowheads). (g)–(k) Radial microtubule array organization in wild-type endosperm (g, h) and in ruk−/− endosperm (i, k). (h), (k) Magnification of MT arrays in endosperm. In wild-type endosperm, nuclei are evenly spaced and separated by radial MTs (arrowheads in g). (i), (k) The ruk−/− endosperm exhibits irregular spacing between nuclei and displays disorganized radial MT organization (arrowheads in i). Scale bars = 5 μm.

In telophase II, bundled MTs assembled in RMAs, radiating from each of the four evenly spaced post-meiotic nuclei in wild-type MMCs (Figure 5a,c). Similarly, RMAs in FR5 MMCs were formed but often showed abnormal organization (Figure 5b,d). For example, MT bundles radiated asymmetrically from nuclei (Figures 5b,d and S6a), occasionally large gaps were present between MT bundles (Figure S6b), and individual nuclei were unevenly spaced (Figure S6a). Insufficient separation of nuclei and asymmetry of RMAs emanating from nuclei were also observed in tes-2 mutants (Figure S6c,d). After cellularization, FR5 microspores often did not completely separate (Figure 5f) compared with wild-type microspores (Figure 5e). Related defects were observed in tes-2 microspores (Figure S6e,f). Therefore, our observations suggest that FR5 interfered with the organization of RMAs. The resemblance of FR5 and tes-2 microsporogenesis supports a shared pathway.

RUNKEL is involved in endosperm cellularization

Previous analysis of RUK function during endosperm development revealed mild cellularization defects including nuclei of different size, multinucleate cells and incomplete cell walls (Sorensen et al., 2002). In the endosperm, RMAs and mini-phragmoplasts are also critical for cellularization. To further investigate the potential role of RUK in these cellularization events, we investigated the MT cytoskeleton in ruk endosperm. We took advantage of the ruk/− mutants and studied the MT organization of endosperm in siliques of ruk/+ heterozygous plants (Figure 5g,h). Therefore, seeds containing the readily identifiable ruk/− embryos were selected and the associated endosperm was examined. The endosperm surrounding phenotypically wild-type embryos displayed dense MT bundles, radiating from evenly spaced nuclei and overlapping in the future plane of cell division (Figure 5g). In contrast, the endosperms that nurtured ruk−/− mutant embryos showed unsymmetrical distribution of individual nuclei accompanied by less dense RMAs with fewer bundles (Figure 5h). Thus, the disorganization of RMAs is likely to be the cause of the cellularization defects observed in ruk.

FR5 truncated protein does not co-localize with MT arrays

In root meristematic tissue, hemagglutinin (HA)-tagged RUK co-localized with preprophase bands, spindle and phragmoplast MTs during mitosis and the cytokinesis-specific syntaxin KNOLLE associated with the cell plate in late telophase (Krupnova et al., 2009; Figure S7b). The FR5 deletion fragment comprised a large portion of the protein domain that associated with MTs in a MT co-sedimentation assay (Krupnova et al., 2009), but lacked the kinase domain entirely (Figure S1a). The HA-tagged FR5 protein did not associate with phragmoplast MT arrays (Figure S7a) or any earlier stage of mitosis (data not shown) but accumulated in the cytoplasm. Furthermore, the KNOLLE protein distributed diffusely in the plane of cell division in late telophase, implying inefficient KNOLLE targeting to the cell plate in FR5 mitotic cells. Repeated attempts to localize HA–RUK and HA–FR5 in meiocytes were not successful, probably due to the inability of anti-HA antibody to enter the callose cell wall.

Disruption of RUK function by FR5 expression results in mistargeting of HIK

Since defects in FR5 mutants and ruk were similar to defects observed in tes-2 and hik, we suspected a genetic interaction between RUK, HIK and its homolog TES. To test this hypothesis, we investigated HIK localization in FR5 mutant seedlings, which showed modest cytokinesis defects (Figure S3c,d,f,g).

We established lines expressing both Myc-tagged HIK (Myc-HIK) and FR5 proteins by crossing Myc-HIK and FR5 lines. In Arabidopsis wild-type cells, Myc-HIK mainly accumulated at the phragmoplast midzone and co-localized with the cell plate-specific syntaxin KNOLLE during early and late cytokinesis (Figure 6a) in agreement with previous data on GFP-tagged NACK1/HIK in tobacco BY2 cells (Sasabe et al., 2011a). In contrast, in cytokinetic FR5 cells, confinement of Myc-HIK to the midzone was less pronounced. Instead Myc-HIK was dispersed around the cell plate and between daughter nuclei in early and late cytokinesis and only partially co-localized with KNOLLE syntaxin (Figure 6b). Intriguingly, Myc-HIK signal in FR5 cells resembled the localization of phragmoplast microtubules.

Figure 6.

The kinesin HINKEL (HIK) is mislocalized during FR5 somatic cytokinesis. Immunolocalization and confocal microscopy of Myc-HIK in FR5 cytokinesis cell. (a) In wild-type cells Myc-HIK co-localizes with KNOLLE syntaxin and is almost exclusively restricted to the midzone in early and late cytokinesis. (b) In FR5 cells, Myc-HIK partially co-localizes with KNOLLE at the midzone and is diffusely distributed (arrowheads) between re-formed daughter nuclei near the cell plate. Scale bar = 10 μm. Red, MycHIK; green, KNOLLE; blue, 4′,6-diamidino-2-phenylindole (DAPI).

Discussion

In this study we report a newly uncovered function for the kinase-like protein RUK in syncytial cytokinesis during male microsporogenesis and further elucidate its function during endosperm development. Previously, analysis of seedling-lethal ruk mutants revealed that the RUK gene product is a component of the phragmoplast expansion machinery during embryogenesis and early seedling development (Krupnova et al., 2009). Yet the seedling-lethal phenotypes of the available ruk mutant alleles obstructed analysis of a potential function in mature plants and gametophytes. Our gene expression analysis of pRUK::GUS transgenic plants confirmed RUK gene expression in floral organs, which was recently corroborated by the presence of RUK in the transcriptome of Arabidopsis male meiocytes (Yang et al., 2011). In the present study, the expression of different truncated RUK fragments under control of the RUK promoter in wild-type plants caused severe cellularization defects in meiocytes displaying incomplete callose wall formation, internalization of the exine layer and multiple nuclei in microspores in addition to cytokinesis defects in somatic tissues. Deficiencies in syncytial cytokinesis appear to be caused by abnormally organized RMAs in telophase II of meiocytes, most likely due to a dominant negative effect on RUK function. Similarly, the abnormal RMA organization in the ruk/− endosperm led to cellularization defects. Furthermore, in FR5 transgenic lines, Myc-HIK was no longer restricted to the midzone, but dispersed in a pattern resembling the phragmoplast.

An increase in number and size of nuclei as observed in FR5-expressing lines is indicative for failure in cellularization after meiosis (Zeng et al., 2011). Similar to the formation and expansion of the phragmoplast in somatic cytokinesis, the formation of the RMAs and mini-phragmoplasts during sporogenesis and endosperm development most likely relies on a functional MAPK signaling cascade which coordinates MT turnover and MT bundling by phosphoregulation-dependent activity of MAP65. The kinesin-like protein TES/AtNACK2 is specifically required for MMC development by activation of a MAPK signaling cascade in microspores (Tanaka et al., 2004). Severe tes-1 mutant phenotypes exhibited enlarged pollen with an increased number of nuclei and a complete loss of callose walls due to the failure to organize normal RMAs (Yang et al., 2003). Members of the MAP signaling cascade are well characterized in tobacco, and conservation of the pathway in Arabidopsis microsporogenesis and in somatic tissues is well supported by genetic and biochemical experimental evidence (Sasabe and Machida, 2012). The absence of intersporal callose walls in the mpk4 tetrads, reminiscent of the tes-1 mutant, implicated MPK4 kinase in microsporogenesis (Zeng et al., 2011) in addition to its role in somatic cytokinesis (Kosetsu et al., 2010). MPK4 phosphorylated the MT cross-linking root-specific protein AtMAP65-3/PLEIADE (PLE), AtMAP65-1 and AtMAP65-2 in vitro (Sasabe et al., 2011a,b), indicating that MPK4 is a shared component of MAPK signaling of both somatic and syncytial cytokinesis.

In contrast to strong tes-1 alleles (Yang et al., 2003), weaker tes-2 alleles exhibit partially formed intersporal callose walls (this study) and internalized exines (Spielman et al., 1997), similar to the defects observed in FR5. Our cytoskeletal analysis of tes-2 microspores provides evidence for abnormally organized RMAs in tes-2, consistent with recently reported abnormal RMAs in the weak tes-4 alleles (De Storme et al., 2012) and with RMA defects observed in FR5. Relevant to this, induction of abnormal RMAs by cold treatment during telophase II also resulted in incompletely separated microspores with multiple or fused multiploid nuclei and gave rise to enlarged pollen grains (De Storme et al., 2012). Thus both genetic intervention and inhibition of RMA biosynthesis by cold treatment result in incompletely formed intersporal callose walls. The similarity of RMA defects in FR5, weak tes and mpk4 mutants supports the role of RUK in a common pathway underlying microspore development.

Abnormal RMA organization and subsequent defects in pollen development were also observed as a result of abnormal meiotic spindle alignment in FORMIN mutants of Arabidopsis (Li et al., 2010). In wild-type metaphase II meiocytes, the two spindles were oriented perpendicular to each other, whereas in afh14 mutants the spindles were oriented in parallel (Li et al., 2010). However, in FR5 metaphase II, spindle orientations never deviated from the perpendicular wild-type arrangement, indicating that RMA defects in FR5 did not originate from spindle defects.

Defects in male gametogenesis also result in defective pollen formation. The diploid MMCs undergo two nuclear meiotic divisions followed by cellularization, giving rise to equally sized haploid microspores in telophase II (Berger and Twell, 2011). In gametogenesis the microspores pass through two rounds of pollen mitosis and produce mature tricellular pollen containing two generative sperm cells and one vegetative cell (Berger and Twell, 2011). The TIO kinase regulates phragmoplast expansion and cell plate formation in microspores (Oh et al., 2005) and interacts with PAKRP1/Kinesin-12A and PAKRP1L/Kinesin-12B via its C-terminal ARM domain (Oh et al., 2012). These kinesins act redundantly in post-meiotic microspore development (Lee et al., 2007) and localize at the phragmoplast midzone to anchor MT plus ends (Lee et al., 2007). Furthermore, PAKRP1/Kinesin-12A and PAKRP1L/Kinesin-12B mislocalize in the MAP65-3/PLE allele dyc283 (Ho et al., 2011). MAP65-3/PLE maintains interdigitation of antiparallel MT plus ends in the phragmoplast midzone (Li et al., 2010; Ho et al., 2011) by phosphorylation-dependent cross-linking activity (Sasabe et al., 2011b). There is no evidence for a role of RUK in post-meiotic mitosis I and mitosis II since cytokinesis defects were not observed in ruk microspores (= 2000; data not shown).

In support of the notion that various proteins are shared between tissue-specific MAPK signaling pathways in both somatic and syncytial cytokinesis, endosperm cellularization was found to be altered in knolle, pleiade, hinkel and runkel mutants which also display severe cytokinesis defects during embryogenesis (Sorensen et al., 2002). In addition, gene expression of HIK, PAKRP1/Kinesin-12A and PAKRP1L/Kinesin-12B and MAP65-1 was up-regulated during endosperm development (Day et al., 2009).

During embryo development the TES homolog HIK/AtNACK1 is required for phragmoplast expansion (Strompen et al., 2002). Cytokinesis defects observed in ruk and FR5 embryos and seedlings are reminiscent of hik mutants (Strompen et al., 2002). Thus, ruk and FR5 mutants combine features of hik and tes mutants suggesting that RUK, similar to MPK4, might act in a HIK-dependent pathway during somatic cytokinesis and together with TES during syncytial cytokinesis. Interestingly, in ruk mutants phragmoplasts are arrested and showed structural defects with a wider, KNOLLE-positive midzone than wild-type, similar to defects in map65-3/ple (Müller et al., 2004; Krupnova et al., 2009). Also in FR5 root cells, accumulation of KNOLLE at the phragmoplast midzone was dramatically disturbed, suggesting that loss of RUK function either by point mutation (Krupnova et al., 2009) or by FR5 expression interfered with KNOLLE localization at the cell plate as a consequence of abnormal MT organization in somatic phragmoplasts. In addition, the disturbance of RUK function in FR5 mutant seedlings results in inefficient targeting of HIK, suggesting that RUK is required for specific MT plus end localization of HIK. Previous analysis provided evidence for a regulatory function of RUK in phragmoplast expansion (Krupnova et al., 2009). One possible mechanism is that RUK controls phragmoplast expansion by regulating HIK-mediated MT plus end targeting, which initiates MAPK-mediated MAP65 deactivation and release of cross-linked antiparallel MTs in the midzone. In summary we propose that RUK is a shared component of somatic and syncytial cytokinesis to promote coordination of MT turnover and bundling activity via MAPK-mediated signaling. Future genetic analyses, live cell imaging of MT dynamics and localization studies will clarify the mechanism of RUK function and its precise biochemical relationship with HIK, TES and MAPK components.

Experimental Procedures

Cloning, plant transformation and selection of transgenic plants

The 4.1 kb full-length RUK coding sequence (CDS) (Krupnova et al., 2009) was used as a template for the cloning of the truncated RUK fragments FR1FR7 into the binary vector pGreen0229 (http://www.pgreen.ac.uk/JIT/pG0229.htm). A 6 × HA-tag sequence was fused to the N-terminal of RUK, directly after the start codon. The 5′ RUK regulatory sequence was inserted upstream of the HA coding sequence and the 3′ untranslated region (UTR) of RUK downstream of the RUK stop codon. The primer combinations used for cloning of each construct FR1FR7 are summarized in Figure S1(b). The resulting plasmids were transformed into wild-type Arabidopsis thaliana plants (Col-0). The progeny of three independent FR5 T1 lines (#50, #59, #79) were selected for subsequent characterization. The 5.1-kb genomic DNA containing the HIK coding and the 5′, 3′ HIK regulatory sequences were cloned into pCB302 binary vector (Xiang et al., 1999) using the NotI restriction site. The Myc-tag sequence was fused to the N-terminal part of HIK directly after the start codon using Kpn2I digestion. The resulting plasmid pCB302MycHIK was transformed into heterozygous hik-Con8 allele lines to confirm the complementation of the mutant phenotype. The floral-dip method (Clough and Bent, 1998) was used to obtain all transgenic lines via Agrobacterium tumefaciens transformation (GV3101 (pMP90) strains replicated experimental constructs). Transformants in T1 and subsequent generations were identified by spraying BASTA® (200 gl−1 glufosinate-ammonium, Bayer CropScience, Germany, http://www.bayercropscience.de; 1:1000) or by selection of seedlings on media containing phosphinotricin (Duchefa, http://www.duchefa.com/).

Semi-quantitative RT-PCR analysis

For RNA extraction (RNeasy Mini Kit, Qiagen, http://www.qiagen.com/) flower buds of stages 1–5 and 6–18 (according to Smyth et al., 1990) were collected from wild-type and FR5 transgenic plants. One microgram of total RNA was reverse transcribed using random hexamer primers according to the manufacturer's instructions (Fermentas, http://www.thermoscientificbio.com/fermentas/). For subsequent PCR reactions, four microliters of RT reactions were used as templates in each 25-μl PCR reaction. The primer combinations used to amplify endogenous and transgenic RUK and ACTIN are listed in Figure S1(b).

Microscopy

For nuclear staining, mature anthers were crushed on a microscope slide to release the pollen and mounted in nuclear staining solution [1 μg/ml 4′,6-diamidino-2-phenylindole (DAPI); Sigma, http://www.sigmaaldrich.com/] in phosphate buffered saline (PBS) pH 7.4, 5% DMSO, 0.01% Tween). For callose staining, the flower buds were fixed overnight in a methanol/acetone (3:1) solution and subsequently washed three times in PBS. Tetrads were prepared from young anthers and mounted in 0.01% methyl blue (Sigma) and 1 μg/ml DAPI solution. Specimens were viewed with a Zeiss Axiophot microscope using a 100 W mercury lamp. Whole-mounts of plant tissues were prepared and sections were analyzed as previously described (Friml et al., 2003). For electron microscopy analysis, FR5 and wild-type anthers were high-pressure frozen, freeze-substituted (acetone supplemented with 2.5% osmium tetroxide) and embedded in epoxy resin as previously described (Stierhof and El Kasmi, 2010).

Whole-mount immunolocalization

Immunolocalization in MMCs and male tetrads was done with mouse anti-α-tubulin (Sigma; 1:600) and goat anti-mouse-Cy3 (Dianova, http://www.dianova.com/; 1:600). The MMCs were prepared on microscope slides as described for callose staining. Anthers were gently flattened on coated slides (Polysine; Thermo Scientific, http://www.thermoscientific.com/), resulting in the release of MMCs. Subsequently, slides were dried for 10 min, covered with a thin layer of gelatin/agarose (0.75% gelatin, 0.75% agarose and 0.3% sucrose) and dipped in liquid nitrogen.

The ovules were fixed in 4% formaldehyde on ice for 1 h and the endosperm was isolated from ovules and transferred to microscope slides coated with 2% gelatin.

Immunolocalization in seedling root tips and in endosperm was carried out as previously described (Völker et al., 2001) with mouse anti-HA (Covance, http://www.covance.com/; 1:500), mouse anti-Myc monoclonal antibody 9E10 (Santa Cruz Biotechnology, http://www.scbt.com/; 1:600), tubulin with mouse anti-α-tubulin (see above) and rabbit anti-knolle (1:3000) for KNOLLE localization. Secondary antibodies were goat anti-rabbit-FITC (Dianova; 1:600) and goat anti-mouse-Cy3 (see above). Images were processed using Adobe Photoshop software (http://www.adobe.com/).

Acknowledgements

We thank Ramon Torres-Ruiz for the tetraploid A. thaliana Col-0 seeds, Florian Faessler, Arvid Herrmann and Jacob Boerold for assistance in immunostaining analysis, Hugh Dickinson for tes-1 and tes-2 seeds, Kathrin Schrick and Farid El-Kasmi for critical reading and comments on the manuscript. This work was funded by the Deutsche Forschungsgemeinschaft (Ju 179/12-1) and the University of Tübingen, and we thank Gerd Jürgens for his generous support.

Ancillary