An evolutionarily conserved NIMA-related kinase directs rhizoid tip growth in the basal land plant Marchantia polymorpha

Directional cell growth is essential for the development of multicellular organisms. This is especially true for plants, owing to their sessile and autotrophic lifestyle. Plant cells are surrounded by rigid cell walls, which restrict motility; thus, directional cell growth is indispensable for morphogenesis and reproduction in land plants. Plant cell growth is driven by hydraulic turgor pressure, along with localized cell wall deposition and loosening of the cell wall. The turgor pressure causes the cell wall and plasma membrane to stretch, which increases the surface area of the cell. The direction of cell growth is primarily determined by the orientation of cellulose microfibrils in the cell wall, which, in turn, is directed by the cortical microtubule array (Green, 1962; Ledbetter and Porter, 1963). Plant cell growth is categorized as either diffuse or tip growth. Diffuse growth, which occurs in most plant cells, involves the even expansion of the entire cell surface, whereas tip growth, which occurs in specialized cells such as root hairs and rhizoids, results from spatially focused cell expansion, resulting in polarized growth and a filamentous cell morphology. Although tip growth is an evolutionarily conserved process in plants and has been extensively studied in root hairs, pollen tubes and protonemata (Rounds and Bezanilla, 2013), it is not known how a single growth point is established and maintained in this process.

Rhizoids are considered to be an early root system in land plant evolution (Jones and Dolan, 2012; Kenrick and Strullu-Derrien, 2014). Formed by the gametophytes of charophytes and seedless land plants, including bryophytes, lycophytes and monilophytes, rhizoids absorb nutrients and attach the gametophyte to a substrate (Jones and Dolan, 2012; Duckett et al., 2014). Although the mechanism of rhizoid development is not well understood, recent studies have revealed that the same regulatory mechanisms mediate the development of rhizoids in bryophytes and root hairs in angiosperms (Proust et al., 2016; Honkanen et al., 2016). The basic helix-loop-helix (bHLH) transcription factors ROOT HAIR DEFECTIVE 6 (RHD6) and RHD6-like (RSL) are conserved in land plants and positively regulate development of polar growing cells, such as caulonema cells, rhizoids and root hairs (Menand et al., 2007; Jang et al., 2011; Proust et al., 2016). Heterologous expression of PpRSL1 from Physcomitrella patens and MpRSL1 from Marchantia polymorpha rescued the loss of root hairs in the Arabidopsis thaliana rhd6 mutant (Menand et al., 2007; Proust et al., 2016). Another family of bHLH transcription factors, Lotus japonicus ROOTHAIRLESS LIKE (LRL), is conserved in streptophytes and participates in the development of rhizoids and root hairs (Tam et al., 2015; Breuninger et al., 2016). In addition, auxin commonly promotes the development of rhizoids and root hairs (Pitts et al., 1998; Sakakibara et al., 2003; Jang et al., 2011; Flores-Sandoval et al., 2015; Kato et al., 2015; Tam et al., 2015; Bowman, 2016). These results demonstrate that a well-conserved molecular mechanism mediates tip growth in land plants. Indeed, forward genetic screening has identified genes required for rhizoid growth and the Arabidopsis orthologs of some of these genes participate in tip growth in root hairs (Honkanen et al., 2016). However, the roles of these genes in rhizoid growth have just started to emerge and the mechanism governing rhizoid tip growth remains to be elucidated.

Never in mitosis A (NIMA)-related kinases (NEKs) constitute a family of mitotic kinases that have been shown to control various mitotic events in fungi and animals, such as the G2/M transition, centrosome separation and spindle formation (Osmani et al., 1988; Fry et al., 2012). By contrast, NEK members in plants regulate directional cell growth during interphase (Motose et al., 2008; Sakai et al., 2008; Motose et al., 2011). Consistent with this, NIMA kinase of the filamentous fungus Aspergillus nidulans, a representative member of the NEK family, has been shown to regulate directional growth of hyphae during interphase by microtubule organization (Govindaraghavan et al., 2014). The Arabidopsis genome has seven NEK members (AtNEK1-7), among which AtNEK6 plays a pivotal role in directional cell expansion (Motose et al., 2012; Takatani et al., 2015b, 2017). The epidermal cells of hypocotyls and petioles in Arabidopsis nek6 mutants exhibit ectopic outgrowths (Motose et al., 2008; Sakai et al., 2008). AtNEK6 interacts with other NEK family members and phosphorylates β-tubulin, resulting in depolymerization of cortical microtubules, during directional cell expansion (Motose et al., 2011; Takatani et al., 2017).

However, functional overlap between plant NEK family members hampers efforts to determine their precise roles in plant development. Here, we analyzed the role of MpNEK1, a single NEK gene in the genome of the early diverging land plant M. polymorpha, to elucidate the function and evolution of plant NEK family members; this species has the advantages of low genetic redundancy (Bowman et al., 2017) and various genetic tools (Ishizaki et al., 2008, 2013, 2015). Rhizoid cells of Mpnek1 knockout mutants exhibited abnormal growth with frequent changes in direction, and developed into cells with a distorted morphology. Our results demonstrate that MpNEK1 regulates directional growth of rhizoid cells through a mechanism that is evolutionarily conserved in land plants.

To analyze the function of NEKs in plant development and evolution, we searched for NEK genes in the genomic sequence databases of the liverwort M. polymorpha and the moss P. patens. A single NEK gene was identified in each species and named MpNEK1 and PpNEK1, respectively. We cloned the corresponding cDNAs and confirmed that they encode NIMA-related kinases by sequencing [DDBJ/EMBL/GenBank accession numbers: LC274883 (MpNEK1) and LC274884 (PpNEK1)]. MpNEK1 and PpNEK1 encode proteins of 922 and 1018 amino acid residues, respectively (Fig. S1). Both proteins have a highly conserved N-terminal kinase domain and a variable C-terminal regulatory tail. Furthermore, of the seven Arabidopsis NEKs both MpNEK1 and PpNEK1 were most similar to AtNEK6 and their exon-intron structures were well-conserved (Fig. S1).

Our previous phylogenetic analysis showed that plant NEK genes can be classified into three clades: the AtNEK1/2/3/4, AtNEK5/7 and AtNEK6 clades (Takatani et al., 2015b). Both MpNEK1 and PpNEK1 belong to the AtNEK6 clade as do three NEK genes of the basal vascular plant Selaginella moellendorffii, suggesting that plant NEK members evolved from an ancestral NEK6-like gene (Takatani et al., 2015b). To examine this hypothesis, we expanded the phylogenetic analysis to include NEK genes of Klebsormidium flaccidum (a charophyte and sister lineage to all land plants), Sphagnum fallax (a moss) and Amborella trichopoda (a basal angiosperm) (Fig. S2). Klebsormidium has five NEK genes, one of which is present at the phylogenetic base of the land plant NEK family. Sphagnum has six NEK genes, all of which belong to the AtNEK6 clade (a subclade of bryophyte and lycophyte NEKs), implying recent divergence of the Sphagnum NEK genes. Amborella has three NEK genes, each of which belongs to a different clade (the AtNEK1/2/3/4, AtNEK5/7 and AtNEK6 clades), suggesting that three NEK classes are established during early angiosperm evolution. These findings support the notion that land plant NEK genes originated from one ancestral gene.

To determine whether MpNEK1 and AtNEK6 have conserved functions, we heterologously expressed an MpNEK1-GFP fusion gene driven by the AtNEK6 promoter (AtNEK6pro:MpNEK1-GFP) in the Arabidopsis nek6 mutant and obtained 14 independent transgenic lines. In all AtNEK6pro:MpNEK1-GFP lines, the ectopic outgrowth phenotype of nek6 was fully rescued (Fig. 1A). Furthermore, MpNEK1-GFP exhibited a localization pattern similar to that of AtNEK6-GFP (Fig. 1B). Both MpNEK1-GFP and AtNEK6-GFP localized to particles along the filaments. From these results, we conclude that land plant NEKs have an evolutionarily conserved role in directional cell growth.

We evaluated the expression pattern of MpNEK1 using real-time quantitative PCR (RT-qPCR). MpNEK1 transcripts were detected in all organs including the thallus, gemma cup and antheridiophores (Fig. S3). Next, we generated transgenic Marchantia lines expressing the β-glucuronidase reporter gene (GUS) under the control of the MpNEK1 promoter (MpNEK1pro:GUS). Strong GUS staining was observed in rhizoids (Fig. 2A) and in the meristematic apical notches of juvenile gemmalings (Fig. 2B). In older plants, GUS activity was observed in the broader region of the thallus and was more prominent in the meristem and midribs (Fig. 2C). MpNEK1pro:GUS was preferentially expressed in the rhizoids (Fig. 2C,D). We also generated transgenic lines expressing the fluorescent protein Citrine containing a nuclear localization signal under the control of the MpNEK1 promoter (MpNEK1pro:Citrine-NLS) and confirmed the preferential promoter activity of MpNEK1 in rhizoids (Fig. 2E,F) and meristematic regions (Fig. 2G).

To determine the function of MpNEK1, we generated knockout mutants of MpNEK1 through gene-targeted homologous recombination (Fig. S4, Ishizaki et al., 2013). The gene targeting construct was designed to disrupt MpNEK1 by replacing the first exon, including the start codon and an ATP-binding motif that is essential for kinase activity, with a hygromycin-resistant marker (Fig. S4). Approximately 500 independent transformants were screened using genomic PCR to detect homologous recombination events (Fig. S5). We isolated 13 independent knockout lines and examined their growth and morphology in detail. All of the Mpnek1 knockout lines exhibited similar morphological defects, especially in the rhizoids (Figs 3 and 4, Figs S6 and S7). In the wild type, rhizoids originated from the ventral epidermis and underwent tip growth to develop into straight filamentous structures (Fig. 3A,B, Fig. 4A,B). The Mpnek1 knockout rhizoids showed a distorted curly morphology (Fig. 3A,B, Fig. 4A, Figs S6 and S7). This phenotype was attributed to changes in growth direction rather than the deformation of already-formed rhizoids (Movie 1). In the wild type, most rhizoids penetrated the agar medium on which they were plated and had a transparent appearance when the agar plate was viewed from underneath (Fig. 3C). However, most rhizoids in Mpnek1 knockouts could not penetrate the agar and remained on the surface (Fig. 3C). The complex-thallus liverworts, including Marchantia, have unicellular dimorphic rhizoids: smooth rhizoids and pegged rhizoids (Duckett et al., 2014; Shimamura, 2016). Both rhizoid types exhibited a defect in growth directionality, but the pegged rhizoids had the least severe phenotype, which was characterized by a wavy and bumpy morphology (Fig. 4B). The number and length of rhizoids in the Mpnek1 knockouts appeared to be normal compared with the wild type (Fig. S8). Thus, MpNEK1 specifically regulates growth directionality in rhizoids, but is not required for the initiation and elongation of rhizoids.

Thallus growth and morphology in the Mpnek1 knockouts also appeared to be largely normal compared with the wild type. However, the thalli of Mpnek1 knockouts exhibited some minor morphological defects. When the knockouts were grown on a soil surface, they developed an unusual wavy thallus morphology (Fig. S9). Thus, MpNEK1 appears to function in thallus development. No obvious phenotypes were observed in the morphology of gemma cups, air pores, thallus margin and assimilatory filaments (Fig. S10).

Next, we transgenically expressed an MpNEK1-Citrine fusion driven by the endogenous MpNEK1 promoter (MpNEK1pro:MpNEK1-Citrine) in the Mpnek1 knockouts. The rhizoid growth defect caused by the Mpnek1 knockout was rescued in all 14 independent transgenic lines (Fig. 5A). The rhizoids of the complementation lines were straight (Fig. 5A, Movie 2), whereas those of the Mpnek1 knockouts were helical and/or twisted (Fig. 5A, Movie 1). To determine the subcellular localization of MpNEK1 in the rhizoids, we observed the complemented lines expressing MpNEK1-Citrine by confocal microscopy. MpNEK1-Citrine localized to the apical dome of rhizoids, which is the region where tip growth occurs. MpNEK1-Citrine accumulated to high levels in small particles and along the cytoskeletal filamentous structure (Fig. 5B, Movie 3). The MpNEK1 particles moved along the filamentous structures (Movie 4, Fig. S11). In the tip region of the apical dome, MpNEK1 particles accumulated in intersections of cytoskeletal filaments and changed the position of the intersections (Movie 4, Fig. S11). In the lateral region of the apical dome, MpNEK1 particles tended to move away from the apical region (Movie 4, Fig. S11). The localization pattern of MpNEK1-Citrine is similar to that of AtNEK6-GFP, which associates with and moves along cortical microtubules (Motose et al., 2011; Takatani et al., 2017).

To examine whether localization of MpNEK1-Citrine is dependent on microtubules, we analyzed the effects of microtubule-depolymerizing drugs, oryzalin and propyzamide. Treatment with oryzalin and propyzamide disrupted the MpNEK1-Citrine localization pattern, causing fluorescence signal to be dispersed throughout the rhizoid (Fig. 6A). By contrast, treatment with the actin filament-depolymerizing drug latrunculin B only slightly affected the localization of MpNEK1-Citrine, which still accumulated in the apical dome of rhizoids (Fig. 6A).

To investigate colocalization of MpNEK1-Citrine with microtubules, we stably expressed β-tubulin 2 (MpTUB2; Buschmann et al., 2016) fused with TagRFP at the N terminus under the control of the cauliflower mosaic virus 35S RNA (CaMV35S) promoter (CaMV35Spro:TagRFP-MpTUB2) in the Mpnek1 knockouts complemented with MpNEK1-Citrine. MpNEK1-Citrine localized to microtubule foci at the tip of the apical dome in rhizoids (six independent transgenic lines; Fig. 6B, Fig. S12A). The colocalization of MpNEK1-Citrine and microtubules was confirmed by double immunolabeling using antibodies against GFP and α-tubulin (Fig. S13). Next, we stably expressed the actin marker Lifeact-TagRFP under the CaMV35S promoter (CaMV35Spro:Lifeact-TagRFP) in the Mpnek1 knockouts complemented with MpNEK1-Citrine. MpNEK1-Citrine did not show colocalization with actin filaments in rhizoids (five independent transgenic lines; Fig. 6C, Fig. S12B). These results indicate that MpNEK1 localizes to microtubules in the apical dome of rhizoids.

Because AtNEK6 has been shown to regulate microtubule organization (Motose et al., 2011; Takatani et al., 2017), we hypothesized that MpNEK1 also regulates microtubule organization. To test this possibility, we introduced CaMV35Spro:TagRFP-MpTUB2 into the wild type and Mpnek1 knockouts to visualize microtubules in rhizoids (Fig. 7). In the apical region of wild-type rhizoids, fine microtubules were observed and cytoplasmic fluorescence derived from free tubulin dimers was clearly detected (Fig. 7A,C,G). However, thick microtubules became more prominent and the number of microtubules decreased further away from the apical dome (Fig. 7A). By contrast, thick microtubules accumulated in the apical domes of rhizoids in the Mpnek1 knockouts and fine microtubules were not observed in this region (Fig. 7B,D). We quantified microtubule density in the apical dome of rhizoids in the wild type and Mpnek1 knockouts (Fig. 7E). Microtubule density was significantly increased in the Mpnek1 knockouts. Next, we measured skewness of the microtubule fluorescence intensity distribution (an index for the bundled filaments; Higaki et al., 2010). The skewness was significantly higher in the Mpnek1 knockouts, which showed increased microtubule bundling (Fig. 7F). Furthermore, the fluorescence derived from free cytoplasmic tubulin dimers was decreased in the Mpnek1 knockouts (Fig. 7G). These results indicate increased microtubule density and bundling in the Mpnek1 knockouts.

Next, we investigated the effects of various microtubule-targeting drugs on rhizoid growth (Fig. 8, Fig. S14). Treatment with the microtubule-stabilizing drug taxol altered the direction of rhizoid growth at concentrations of 0.3 and 1 µM (Fig 8A,B, Fig. S14B). The direction of rhizoid growth became random and the rhizoids showed distorted morphology (Fig. 8B, Fig. S14B). Thus, taxol treatment phenocopied the rhizoid growth defect of the Mpnek1 knockouts. Treatment with the microtubule-depolymerizing drug propyzamide induced wavy growth and strongly suppressed elongation of rhizoids (Fig. 8A-C, Fig. S14). In the presence of 0.3 µM propyzamide, rhizoids had a wavy morphology, whereas mock-treated rhizoids did not (Fig. 8B, Fig. S14B). At a concentration of 1 µM propyzamide, rhizoid growth was severely suppressed (Fig. 8A,C, Fig. S14A) and some rhizoids branched (Fig. S14B). These results suggest that microtubule depolymerization and polymerization are required for the directional growth of rhizoids.

We further investigated the effects of taxol and propyzamide on microtubule organization. Transgenic lines expressing Citrine-MpTUB2 under the CaMV35S promoter were grown on agar medium containing taxol or propyzamide and observed by confocal microscopy (Fig. 9). Taxol treatment resulted in increased microtubule density and bundling (Fig. 9B-D,G,H). Propyzamide treatment caused depolymerization of microtubules (Fig. 9E-G). Thus, taxol stabilizes microtubules, whereas propyzamide depolymerizes microtubules in rhizoids. These results were confirmed by using CaMV35Spro:TagRFP-MpTUB2 (Fig. S15). Furthermore, microtubule density was increased to a similar extent in Mpnek1 knockouts and taxol-treated wild type (Fig. S15).

Although we used low concentrations of taxol and propyzamide, there is a possibility of side effects in any pharmacological approach. Thus, we next used a transgenic approach, in which the microtubule-binding domain (MBD) of mammalian microtubule-associated protein 4 (MAP4-MBD) fused with GFP at the N terminus was driven under the control of CaMV35S promoter or elongation factor 1α (EF1α) promoter (Althoff et al., 2014). GFP-MAP4-MBD has been shown to stabilize cortical microtubules and affects organ growth in Arabidopsis (Hashimoto, 2002). We isolated more than 20 independent lines for each construct. Some rhizoids of GFP-MAP4-MBD lines showed distorted morphology (Fig. S16A), suggesting that MAP4-MBD affects directional growth of rhizoids. However, GFP-MAP4-MBD was not expressed efficiently in either construct. In CaMV35Spro:GFP-MAP4-MBD lines, a strong microtubule-like signal was observed in mucilage hair cells (Fig. S16B). The microtubule-like signal was also weakly observed in some thallus cells. However, we could not conclude whether MAP4-MBD affects directional growth of rhizoids by stabilizing microtubules.

Because AtNEK6 phosphorylates tubulin in vitro (Motose et al., 2011; Takatani et al., 2017), we investigated whether MpNEK1 also phosphorylates tubulin. Recombinant MpNEK1 or AtNEK6 fused with glutathione S-transferase (GST) at the N terminus were expressed in Escherichia coli and purified using glutathione beads. The GST-tagged MpNEK1 or AtNEK6 was incubated with or without purified Arabidopsis tubulin in the presence of [γ-³²P] ATP. GST-MpNEK1 showed autophosphorylation activity as AtNEK6 (Fig. 10A). In addition, GST-MpNEK1 strongly phosphorylated tubulin (Fig. 10A). To determine whether MpNEK1 phosphorylates β-tubulin or α-tubulin, we incubated GST-MpNEK1 with the recombinant Arabidopsis β-tubulin 4 (TUB4) or α-tubulin 6 (TUA6). GST-MpNEK1 preferentially phosphorylated TUB4 but weakly phosphorylated TUA6 (Fig. 10B). AtNEK6 has been shown to preferentially phosphorylate β-tubulin in vitro (Motose et al., 2011; Takatani et al., 2017). These results indicate that plant NEKs have conserved molecular functions.

In this study, we identified NEK genes in the bryophytes Marchantia and Physcomitrella. Similar to various fungi, both species have only one NEK gene, whereas most organisms, including animals and plants, have multiple NEK genes (Parker et al., 2007; Takatani et al., 2015b). Arabidopsis has seven NEK genes, which have been shown to function redundantly in directional cell growth and microtubule organization (Motose et al., 2011, 2012; Takatani et al., 2015b). Therefore, analysis of the bryophyte NEK genes will provide valuable insight into their fundamental role in land plant evolution and morphogenesis. Our functional analysis of MpNEK1 demonstrates its crucial role in the directional growth of rhizoids. Mpnek1 knockouts exhibited a defect in rhizoid growth directionality, but not in elongation, implying that MpNEK1 has a specific function in establishing and maintaining growth polarity. This is consistent with the role of AtNEK6 in the directional expansion of epidermal cells. Our finding that heterologous expression of MpNEK1 rescued the outgrowth phenotype of an Arabidopsis nek6 mutant indicates that an evolutionarily conserved NEK-dependent mechanism underlies cellular growth polarity. NEKs in land plants fundamentally regulate directional cell expansion and this primary function might have been acquired early in land plant evolution. Considering that animal and fungal NEK members mainly regulate mitotic events, land plant NEKs might be specialized to regulate cell expansion and growth polarity in interphase cells. Because rhizoids are thought to be an ancestral rooting system inherited from Charophycean algae (Jones and Dolan, 2012; Kenrick and Strullu-Derrien, 2014), it is likely that NIMA-related kinases served as growth regulators during the evolution of the early rooting system of land plants, which supported the colonization of terrestrial environments.

We aimed to identify the molecular function of MpNEK1 in directional tip growth. MpNEK1-Citrine localized to microtubules in the growing apical dome of rhizoids. In addition, MpNEK1-Citrine particles localized to the crossover sites and fused with each other. This is reminiscent of the microtubule localization and dynamics of AtNEK6. We also found that the thick microtubule bundles, which are normally absent from the apical dome, extended into the apical domes of Mpnek1 knockout rhizoids, and that the fine microtubules observed in the tips of wild-type rhizoids were absent. Thus, MpNEK1 might reorganize and destabilize microtubules in the apical dome during rhizoid tip growth (a working hypothesis is shown in Fig. S17). Treatment with the microtubule-stabilizing drug taxol led to a defect in rhizoid growth directionality, phenocopying the Mpnek1 knockouts and further supporting the notion that MpNEK1 promotes microtubule destabilization and turnover. As AtNEK6 has been shown to direct cell expansion by depolymerizing microtubules (Motose et al., 2011; Takatani et al., 2017), all members of the land plant NEK family might share an evolutionarily conserved molecular function in directional cell growth, possibly through regulation of microtubule stability.

Previous studies revealed that actin microfilaments are essential for tip growth in plants, whereas the function of microtubules in this process is controversial (Rounds and Bezanilla, 2013). However, several lines of evidence do suggest that microtubules regulate the directionality of tip growth. For example, it has been shown that cortical microtubules are depolymerized during root hair initiation and are then reorganized into parallel and helical arrays longitudinal to the growing axis (Baluška et al., 2000; Sieberer et al., 2002; Takahashi et al., 2003; Van Bruaene et al., 2004). Treatment of root hairs with microtubule inhibitors has been shown to lead to wavy growth and the formation of multiple growth tips (Bibikova et al., 1999). In addition, genetic analyses have demonstrated that kinesin motor proteins regulate the directionality of tip growth through microtubule organization (Eng and Wasteneys, 2014; Hiwatashi et al., 2014; Eng et al., 2017). Although longitudinal microtubules similar to those present in root hairs were observed in rhizoids (Murata et al., 1987; Corellou et al., 2005; Duckett et al., 2014), microtubule function in rhizoid growth remains to be demonstrated. Nevertheless, the dynamic reorganization of microtubules is a common feature of tip growth and is indispensable for the polarity of cell growth. Thus, MpNEK1 might promote microtubule reorganization to establish growth polarity in rhizoids. However, it remains unclear how microtubule reorganization/destabilization regulates growth directionality. In fungi and animals, microtubules establish and maintain growth polarity through a positive-feedback loop in which microtubules deliver polarity proteins to a specific site in the cell cortex and these polarity proteins subsequently anchor microtubules at the cortex to enhance cellular asymmetry (Siegrist and Doe, 2007). In root hairs of Arabidopsis, positive feedback involving the NADPH oxidase RHD2, reactive oxygen species and Ca²⁺ influx regulates determination and maintenance of growth sites (Takeda et al., 2008). However, microtubules are not required for the localization of RHD2 to the growing tips. Although it is not clear whether a microtubule-based positive-feedback mechanism functions in rhizoid tip growth, NEK-mediated microtubule reorganization might be involved in such a feedback loop (see detail in Fig. S17). Further analysis of NEK-mediated growth polarity in rhizoids of the basal land plant Marchantia should reveal much about the fundamental mechanism of microtubule-dependent tip growth in plants and other systems.

M. polymorpha accession Takaragaike-1 (Tak-1, male) and Takaragaike-2 (Tak-2, female) were grown on the half-strength Gamborg's B5 medium solidified with 1% agar at 22°C under continuous white light. Gametangiophore formation was induced under continuous white light with far-red irradiation as described previously (Chiyoda et al., 2008). For pharmacological analysis, taxol, oryzalin, propyzamide and latrunculin B (Wako Pure Chemical Industries) were dissolved in dimethyl sulfoxide (DMSO) and added to B5 media such that final concentrations of DMSO were less than 0.2%. Dormant gemmae were collected and plated on B5 media containing drugs. A. thaliana accession Colombia and nek6-1 (ibo1-4) (Motose et al., 2008) were grown on half-strength Murashige and Skoog (MS) agar medium as described previously (Takatani et al., 2015a).

To generate a targeting vector of MpNEK1, 5′- and 3′-homologous arms (approximately 4.5-kb length) were amplified from Tak-1 genomic DNA by PCR using KOD FX Neo (Toyobo) with the primer pairs shown in Table S1 (MpNEK-PacI-F1-Gib and MpNEK-PacI-R1-Gib, MpNEK-AscI-F1-Gib and MpNEK-AscI-R1-Gib). The amplified 5′- and 3′-fragments were cloned into the PacI and AscI sites of pJHY-TMp1 (Ishizaki et al., 2013), respectively, with the Gibson assembly system (New England BioLabs). The MpNEK1 targeting vector was transformed into F1 sporelings derived from sexual crosses between Tak-1 and Tak-2 as previously described (Ishizaki et al., 2008). Transformants were selected with 10 µg/ml hygromycin B and 100 µg/ml cefotaxime (Sanofi). The gene-targeted lines were screened by genomic PCR as previously described (Ishizaki et al., 2013). The primers used in this screening are shown in Table S1 and Fig. S5.

Other vectors were mainly constructed based on Gateway technology (Life Technologies). The full-length MpNEK1 was amplified from Tak-1 cDNA by PCR using KOD plus (Toyobo) with the primers MpNEK1-F-cacc and MpNEK1-R-nonstop and cloned into a pENTR/D-TOPO cloning vector (Life Technologies). The full-length PpNEK1 was amplified from cDNA of a NIBB-David strain by PCR using KOD plus with the primers PpNEK1-F-cacc and PpNEK1-Rrev1-nonstop and cloned into pENTR/D-TOPO. These vectors were confirmed by sequencing and designated as pENTR/D-TOPO-MpNEK1 and pENTR/D-TOPO-PpNEK1, respectively. To generate the AtNEK6pro:MpNEK1-GFP construct, the AtNEK6 promoter region including a 2.1-kb region upstream of the initiation codon, was amplified from genomic DNA of Arabidopsis Columbia accession by PCR using KOD plus with the primers NEK6-F(-2104)-NotI and NEK6(-1)-R-NotI. The PCR fragment was digested by NotI and ligated into the unique NotI site of pENTR/D-TOPO-MpNEK1 to generate pENTR/D-TOPO-AtNEK6pro:MpNEK1. This entry vector was used in the Gateway LR reaction (Life Technologies) with the Gateway binary vector pGWB550 (Nakagawa et al., 2007) to generate AtNEK6pro:MpNEK1-GFP construct. This vector was introduced into the Arabidopsis nek6-1 mutant by the floral dip method (Clough and Bent, 1998). Transformants were selected with 30 µg/ml hygromycin B and 200 µg/ml cefotaxime and T3 lines were analyzed in further experiments.

To construct MpNEK1pro:GUS, the MpNEK1 genomic region, including a 4.5-kb region upstream of the initiation codon and 84-b coding region corresponding to 28-amino-acid length, was amplified from Tak-1 genomic DNA by PCR using KOD plus with the primers MpNEK1-F-4.7k-cacc and MpNEK1-R-Ex1 and cloned into pENTR/D-TOPO. This entry vector was used in the LR reaction with the Gateway binary vector pMpGWB104 (Ishizaki et al., 2015) to generate the MpNEK1pro:GUS construct, in which the GUS reporter gene was translationally fused with the first 28-amino-acid sequence of MpNEK1. In addition, the entry vector was used in the LR reaction with pMpGWB115 (Ishizaki et al., 2015) to generate MpNEK1pro:Citrine-NLS construct, in which Citrine-NLS was translationally fused with the first 28-amino-acid sequence of MpNEK1. MpNEK1pro:GUS and MpNEK1pro:Citrine-NLS vectors were introduced into regenerating thalli of Tak-1 as previously described (Kubota et al., 2013). Transformants were selected with 10 µg/ml hygromycin B and 100 µg/ml cefotaxime.

To construct MpNEK1pro:MpNEK1-Citrine, the MpNEK1 promoter region, including a 4.5-kb region upstream of the initiation codon, was amplified from Tak-1 genomic DNA by PCR using KOD plus with the primers MpNEK1-F-4.7k-cacc and MpNEK1pro-R1 and cloned into pENTR/D-TOPO. This vector was confirmed by sequencing and designated as pENTR/D-TOPO-MpNEK1pro. The full-length MpNEK1 was amplified from pENTR/D-TOPO-MpNEK1 by PCR using KOD plus with the primers MpNEK1-F-AscI-Gibson and MpNEK1-R-AscI-Gibson. The amplified fragment was cloned into the AscI site of pENTR/D-TOPO-MpNEK1pro by the Gibson assembly system to generate pENTR/D-TOPO-MpNEK1pro:MpNEK1. This entry vector was used in the LR reaction with the Gateway binary vector pMpGWB307 (Ishizaki et al., 2015) to generate the MpNEK1pro:MpNEK1-Citrine construct. This plasmid was introduced into regenerating thalli of the Mpnek1 knockout line #2-4 by the method described by Kubota et al. (2013). Transformants were selected with 0.5 μM chlorsulfuron (Sigma-Aldrich) and 100 µg/ml cefotaxime.

CaMV35Spro:TagRFP-MpTUB2 and CaMV35Spro:Citrine-MpTUB2 (R.N., unpublished) were constructed in the Gateway binary vectors pMpGWB302 and pMpGWB105 (Ishizaki et al., 2015), respectively. CaMV35Spro:TagRFP-MpTUB2 was introduced into F1 wild-type sporelings derived from crosses between Tak-1 and Tak-2 or into regenerating thalli of MpNEK1 knockout line #2-4 as described above. CaMV35Spro:Citrine-MpTUB2 was introduced into F1 wild-type sporelings derived from crosses between Tak-1 and Tak-2.

TagRFP-MpTUB2 cloned into pENTR-D/TOPO (R.N., unpublished) was used in the LR reaction with the Gateway binary vector pMpGWB202 (Ishizaki et al., 2015) to generate the CaMV35Spro:TagRFP-MpTUB2 construct. This construct was introduced into regenerating thalli of MpNEK1 knockout complemented with MpNEK1-Citrine (line #2) to generate double-labeled lines expressing both MpNEK1-Citrine and TagRFP-MpTUB2. Transformants were selected with 100 µg/ml gentamycin (Wako Pure Chemical Industries) and 100 µg/ml cefotaxime.

The actin marker Lifeact cloned into pENTR-D/TOPO was generated from Lifeact-Venus constructed in pENTR-D/TOPO (Era et al., 2009) by removing Venus sequence with the KOD-plus mutagenesis kit and the primers shown in Table S1. This construct (pENTR-D/TOPO-Lifeact) was used in the LR reaction with the Gateway binary vector pMpGWB228 to generate the CaMV35Spro:Lifeact-TagRFP construct. This construct was introduced into regenerating thalli of MpNEK1 knockout complemented with MpNEK1-Citrine (line #2) to generate double-labeled lines expressing both MpNEK1-Citrine and Lifeact-TagRFP. Transformants were selected with 100 µg/ml gentamycin and 100 µg/ml cefotaxime.

To construct CaMV35Spro:GFP-MAP4-MBD and EF1α pro:GFP-MAP4-MBD, the GFP-MAP4-MBD was amplified from the genomic DNA of the Arabidopsis CaMV35Spro:GFP-MAP4-MBD line (Hamant et al., 2008) by PCR using KOD plus with the primers shown in Table S1 and cloned into pENTR/D-TOPO. This vector was confirmed by sequencing and designated as pENTR/D-TOPO-GFP-MAP4-MBD. This entry vector was used in the LR reaction with the Gateway binary vectors pMpGWB102 and pMpGWB103 (Ishizaki et al., 2015) to generate CaMV35Spro:GFP-MAP4-MBD and EF1α pro:GFP-MAP4-MBD constructs, respectively. These plasmids were introduced into F1 wild-type sporelings derived from crosses between Tak-1 and Tak-2. Transformants were selected with 10 µg/ml hygromycin and 100 µg/ml cefotaxime. All transformants of Marchantia described above were subjected to at least two rounds of transplantation of gemmae and used for further experiments.

For phylogenetic analysis of plant NEKs, we used a conserved region in N-terminal kinase domain (corresponding to amino acids 8-262 of AtNEK6). Sequence information was collected in Phytozome (ver. 12, http://phytozome.jgi.doe.gov/pz/portal.html) and the Klebsormidium database (http://www.plantmorphogenesis.bio.titech.ac.jp/~algae_genome_project/klebsormidium/). To generate a rooted maximum-likelihood tree, sequences were aligned using CLUSTAL Omega (http://www.clustal.org/omega/) and a phylogenetic tree was visualized using FigTree (http://tree.bio.ed.ac.uk/software/figtree/). To generate a rooted maximum-likelihood bootstrap consensus tree, sequences were aligned using MAFFT (http://mafft.cbrc.jp/alignment/software/) and phylogenetic analyses were conducted in MEGA7 (Kumar et al., 2016). The bootstrap consensus tree inferred from 500 replicates is taken to represent the evolutionary history of the taxa analyzed. Branches corresponding to partitions reproduced in less than 30% bootstrap replicates are collapsed. The percentages of replicate trees in which the associated taxa clustered together are shown next to the branches.

Total RNA was isolated from the thallus, gemma cup and antheridiopore of one-month-old Tak-1 by phenol/chloroform extraction and subsequent lithium chloride precipitation (Takahashi et al., 1992). Gemma cups were separated from the thalli using a scalpel and then gemma cups and remaining thalli without gemma cups were immediately frozen in liquid nitrogen for subsequent RNA extraction. For each sample, 0.5 µg of total RNA was reverse transcribed to cDNA using ReverTra Ace reverse transcriptase (Toyobo) according to the manufacturer's protocol. Real-time PCR was performed on a thermal cycler Dice Real Time System (Takara) using the KAPA SYBR FAST qPCR Kit (Kapa Biosystems) according to the manufacturer's protocol. Transcript levels of MpEF1α or MpACT1 were used as a reference for normalization (Kubota et al., 2014; Saint-Marcoux et al., 2015). Primers used in RT-qPCR are listed in Table S1. RT-qPCR experiments were performed using four biological replicates.

For histochemical GUS staining, MpNEK1pro:GUS-expressing Marchantia plants were grown on half-strength B5 medium for 7 and 14 days under continuous white light. GUS staining was performed as described previously (Ishizaki et al., 2012) and at least six independent lines were observed for GUS staining patterns using a stereoscopic microscope S8APO0 (Leica Microsystems) equipped with a CCD camera DFC500 (Leica) or a light microscope DM5000B (Leica) equipped with DFC500. The morphology of gemmalings and rhizoids was also observed using the microscopes described above.

To analyze the localization of MpNEK1-Citrine and microtubule organization, MpNEK1:MpNEK1-Citrine and CaMV35Spro:TagRFP-MpTUB2 plants were grown on half-strength B5 medium for 5 and 7 days under continuous white light, respectively. The gemmaling was placed on a hole slide glass (www.t-sgmt.co.jp/products/?ca=1) (26×76 mm, Toshinriko) with small aliquots of water, covered with a glass strip [22×40 mm (or 24×36 mm), Thickness No. 1, Matsunami], and observed under a FV1200 confocal laser-scanning microscope (Olympus) equipped with a high-sensitivity GaAsP detector and silicone oil objective lenses (30× and 60×, Olympus). Silicone oil (SIL300CS, Olympus) was used as immersion media for the objective lenses. The images were analyzed using ImageJ (National Institutes of Health, USA) and LPixel ImageJ plugins (LPixel, Japan). Microtubule density and the skewness of the microtubule fluorescence intensity distribution were measured according to Takatani et al. (2017) and Higaki et al. (2010), respectively.

Immunostaining was conducted to detect microtubules and MpNEK1-Citrine in 7-day-old gemmalings of Mpnek1 knockout complemented with MpNEK1-Citrine according to Motose et al. (2011). Rat monoclonal anti-α-tubulin antibody YL1/2 (Abcam, ab6160) and mouse monoclonal anti-GFP antibody (clones 7.1 and 13.1) (Roche, 11814460001) were used as primary antibodies at 1/100. Alexa-568-labeled anti-rat-IgG antibody (Thermo Fisher Scientific, A-11077) and Alexa-488-labeled anti-mouse-IgG antibody (Thermo Fisher Scientific, A-11001) were used as secondary antibodies at 1/250. Rhizoids were observed with a FV1200 confocal laser-scanning microscope (Olympus).

The cDNA encoding full-length MpNEK1 was cloned into pENTR/D-TOPO-MpNEK1 and transferred to pDEST15 by LR reaction to generate GST-MpNEK1 construct. GST-MpNEK1, GST-AtNEK6, His-TUB4 and His-TUA6 were expressed in E. coli and purified as previously described (Motose et al., 2008, 2011). The native tubulin was purified from Arabidopsis T87 cell suspension culture using a GST-Stu2 TOG1/2 domain affinity column (Widlund et al., 2012) as previously described (Hotta et al., 2016). One microgram of substrates (purified tubulin, His-TUB4 or His-TUA6) were incubated with or without 1 µg GST-MpNEK1 or GST-AtNEK6 in 20 µl of kinase buffer (50 mM Tris-HCl pH 7.2, 10 mM MnCl₂, 1 mM dithiothreitol, 50 µM ATP and 10 µCi [γ-³²P]ATP) at 25°C for 15 min. Reaction products were separated by SDS-PAGE and stained with a Coomassie Brilliant Blue (CBB) staining kit (Apro Science). Phosphorylated proteins were detected with an image analyzer (FLA-7000 IP, Fujifilm) and imaging plates (BAS-MS2040, Fujifilm).

We thank Sakiko Ishida (Kyoto University) for assistance with gene targeting; Shigeyuki Tsukamoto (Kobe University) for technical assistance with transformation; Professor Tsuyoshi Nakagawa (Shimane University) for pGWB550; Professor Takashi Ueda, Dr Atsuko Era, Dr Kazuo Ebine and Dr Takehiko Kanazawa (NIBB) for pENTR-Lifeact-Venus and advice on plasmid construction; Dr Olivier Hamant (ENS-Lyon) for the Arabidopsis CaMV35Spro:GFP-MBD line; Dr David N. Drechsel (Max Planck Institute) for pGEX-6P-1-Stu2; Dr Takashi Hotta and Professor Takashi Hashimoto (NAIST) for technical assistance with tubulin purification; Professor Yuichiro Takahashi (Okayama University) for use of FLA-7000; and Professor Mitsuyasu Hasebe (NIBB) for the Physcomitrella NIBB-David strain.