ANGUSTIFOLIA contributes to the regulation of three-dimensional morphogenesis in the liverwort Marchantia polymorpha

ANGUSTIFOLIA (AN) of Arabidopsis thaliana (L.) Heynh. plays important roles in morphogenesis. Deficient mutants of AN express narrow and thicker leaves, twisted fruits and petals, and premature opening of flowers (Rédei, 1962; Tsukaya et al., 1994, 1996; Bai et al., 2010, 2013). At the cellular level, A. thaliana an leaves show reduced cell width expansion and enhanced cell thickness expansion in palisade tissue, a simpler epidermal cell shape and trichomes with two branches (Tsukaya et al., 1994; Hülskamp et al., 1994; Tsuge et al., 1996). It is thought that these phenotypes are caused by the abnormal arrangement of cortical MTs with altered orientation in palisade cells, which are more regularly aligned to be parallel to each other in the epidermal cells and have disrupted density gradient in the trichome cells (Kim et al., 2002; Folkers et al., 2002). A GFP-fused AN protein was distributed in a punctate manner in the cytosol, partially colocalizing with the trans-Golgi network (Minamisawa et al., 2011). In addition, AN protein is colocalized with stress granules under stress conditions, such as high temperature (Bhasin and Hülskamp, 2017). Moreover, an mutants show tolerance toward biotic and abiotic stresses, including pathogens, salt and osmotic stresses (Gachomo et al., 2013; Bhasin and Hülskamp, 2017). However, the molecular functions of AN in morphogenesis and stress responses are unknown. Use of a yeast two-hybrid system and pull-down assay resulted in the identification of several AN-interacting proteins; e.g. a receptor kinase STRUBBELIG/SCRAMBLED, a kinesin motor molecule ZWICHEL (ZWI), ANGUSTIFOLIA INTERACTING KINASE (AIK1) and ASYMMETRIC LEAF ENHANCER 3 (AE3), as well as several RNA-binding proteins (Folkers et al., 2002; Bai et al., 2013; Kwak et al., 2015; Bhasin and Hülskamp, 2017). Mutants of STRUBBELIG/SCRAMBLED and AN have similar phenotypes, such as twisted fruits, petals and roots, suggesting that these function cooperatively during morphogenesis [Fulton et al., 2009; Bai et al., 2013, who reported that DETORQUEO (DOQ) corresponds to AN; Kwak et al., 2015].

AN is widely conserved in Streptophyta, and plant ANs are homologs of animal C-terminal binding proteins (CtBPs), which function as transcriptional co-repressors (Kim et al., 2002; Folkers et al., 2002; Chinnadurai, 2007). Both CtBPs and ANs have high amino acid sequence homology to NAD(H)-dependent d-isomer-specific 2-hydroxy acid dehydrogenases (D2-HDH) (Kim et al., 2002). However, plant ANs lack the residues essential for activity of the D2-HDH motif (Kim et al., 2002; Tsukaya, 2006; Stern et al., 2007). Indeed, A. thaliana AN does not exhibit dehydrogenase activity and binding ability with adenovirus E1A protein, a well-known CtBP-binding protein in a dehydrogenase activity-dependent manner (Cho et al., 2008; Stern et al., 2007). Moreover, A. thaliana AN cannot rescue the zygotic lethal phenotype caused by loss-of-function of CtBP in the fruit fly Drosophila melanogaster (Stern et al., 2007). These findings suggest that plant ANs are functionally distinct from animal CtBPs. In contrast, plant ANs from the angiosperm Ipomoea nil (L.) Roth (Japanese morning glory), the gymnosperm Larix gmelinii (Dahurian larch) and the liverwort Marchantia polymorpha can rescue the phenotype of narrow leaves of the A. thaliana an-1 mutant, suggesting that the molecular functions of AN in plants are evolutionarily conserved (Cho et al., 2005; Lin et al., 2008; Minamisawa et al., 2011). Based on their sequence similarity, it is thought that animal CtBPs and plant ANs evolved from the NAD(H)-dependent D-hydroxyacid dehydrogenase (Stern et al., 2007). After divergence of animals and plants, the ancestral plant AN lost its dehydrogenase activity due to a deficient D2-HDH motif and the ancestral animal CtBP acquired the PLDLS-binding motif and RRT-binding motif required for co-repressor activity (Stern et al., 2007).

Liverworts, together with mosses and hornworts, make up the basal land-plant lineage bryophytes (Wickett et al., 2014). The liverwort M. polymorpha is used as a model plant because of its short life cycle, easy culture and crossing, lower level of genetic redundancy, and the availability of genetic information and gene manipulation systems (Ishizaki et al., 2016; Shimamura, 2016; Bowman et al., 2017). In bryophytes, the haploid gametophytic generation dominates the diploid sporophytic generation (Ishizaki, 2017). During gametophytic generation, M. polymorpha forms thalli as the main vegetative plant body. Thalli grow by repeated dichotomous branching (bifurcation) of apical notches, which is the site of cell proliferation (Shimamura, 2016; Solly et al., 2017). M. polymorpha can propagate not only by sexual reproduction via spore formation but also by asexual reproduction via gemma formation in the gemma cup on the thallus (Shimamura, 2016). As described above, the AN homolog from M. polymorpha (MpAN) can rescue the an-1 mutation of A. thaliana (Minamisawa et al., 2011). However, previous analyses of the roles and molecular functions of plant ANs were limited to the diploid sporophyte phase; no phenotype has been recognized in the haploid gametophyte phase. Therefore, analyses of the roles of MpAN in gametophytes of M. polymorpha will enhance our understanding of the roles of plant ANs in land plants and the evolution thereof.

Here, we report the morphology of Mpan knockout mutants of M. polymorpha. These mutants have abnormal twisted thalli and suppressed thallus growth along the growth axis due to irregular cell expansion. Moreover, cortical MTs were disordered in Mpan knockout mutants. In this article, we first report the roles of MpAN as a coordinator for arrangement of cortical MTs to form the well-ordered cell and tissue morphogenesis in gametophyte phase of M. polymorpha and discuss the comparison with other plant ANs.

MpAN is a broadly expressed gene in both the gametophyte phase, including gemmae, gemmalings, reproductive organs and sporelings, and the sporophyte phase, based on information available on the public database MarpolBase (marchantia.info) (Fig. S1A-C). To evaluate the roles of MpAN in M. polymorpha, we created Mpan knockout lines and Mpan genome-edited lines. The Mpan knockout lines were constructed by homologous recombination (Fig. S2A) (Ishizaki et al., 2013). Mpan-1^ko (male) and Mpan-1^ko (female) were isolated as described in the Materials and Methods section, and confirmed by genomic PCR (Fig. S2C). We also isolated Mpan genome-edited lines constructed using the CRISPR/Cas9 system: Mpan-4^ge (male) and Mpan-9^ge (female) (Fig. S2B,C).

In 12-day-old gemmalings, the knockout lines Mpan-1^ko (male) and Mpan-1^ko (female), and the genome-edited lines Mpan-4^ge and Mpan-9^ge showed less flattening of thalli than did the wild-type lines Takaragaike-1 (Tak-1) and Takaragaike-2 (Tak-2) (Fig. 1). To visualize thallus morphology in detail, we captured three-dimensional (3D) images by micro-computed tomography (micro-CT) (Fig. 2; Movies 1 and 2). Mpan-1^ko (male) showed an abnormal shape of the thallus, which was twisted along the growth axis. Moreover, the thallus morphology of the complementation lines pMpAN:MpAN-1/Mpan-1^ko (male) and pMpAN:MpAN-8/Mpan-1^ko (female) were restored to those of the wild-type lines, demonstrating that the abnormal thallus formation of the knockout lines was caused by loss-of-function of MpAN (Fig. 1). When using the deletion construct of the plant-specific C-terminal region in MpAN for the complementation test, the thallus phenotype of Mpan-1^ko was not rescued (Fig. S3). These results suggest that the plant-specific C-terminal region of MpAN has a role at least in thallus morphogenesis. Moreover, the same promoter region of MpAN for the complementation test was used to observe the expression pattern (Fig. S1C,D). MpAN was well expressed around apical notches known as active growth areas, in both pMpAN:Citrine-NLS and pMpAN:GUS transgenic plants.

M. polymorpha develops characteristic reproductive organs: antheridiophores and archegoniophores in males and females, respectively. The antheridiophores of Mpan-1^ko (male) showed an abnormal mushroom-like antheridial receptacle with a curled and waved margin, whereas the wild-type and complementation lines had flatter disc-like shapes (Figs S4 and S5). The archegoniophores of Mpan-1^ko (female) had thicker digitate rays compared with those of the wild-type and complementation lines (Fig. S4). Although MpAN expression was detected in both antheridiophores and archegoniophores of pMpAN:GUS transgenic plants, the GUS signal was not detected in the antheridium or archegonium, except for the younger antheridium (Fig. S1). Normal sporophytes and spores were produced following crossing of Mpan-1^ko (female) with Mpan-1^ko (male), indicating that the knockout lines have normal morphology and fertility (Fig. S6). In addition, MpAN expression was not detected in the early embryo stages following crossing of pMpAN:GUS transgenic plants (Fig. S1M).

To examine in detail the differences between the wild-type and Mpan knockout lines, we traced the development of gemmalings for 10 days (Fig. 3). After day 5, the Mpan knockout [Mpan-1^ko (male) and Mpan-1^ko (female)] lines showed enhanced vertical twisting growth. On the other hand, the fresh weight of 10-day-old gemmalings in Tak-1 and Mpan-1^ko were 9.98±2.76 mg (n=22) and 9.39±1.34 mg (n=19), respectively, indicating no significant difference using Student's t-test (P=0.41). Next, we measured the shape of Tak-1 and Mpan-1^ko gemmalings visualized with the modified pseudo-Schiff propidum iodide (mPS-PI) staining. The ‘length’ was defined as distance between the apical notches, and the ‘width’ as distance from the trace of the stalk to the opposite side across the length at 90° (Fig. 4A). In addition, the gemma/thallus shape index was calculated as the ratio of the width to length. Mpan-1^ko from the gemma stage exhibited a shorter length on day 3 compared with Tak-1, resulting in a more square-shaped thallus in Mpan-1^ko than in Tak-1 (Fig. 4, Fig. S7). For longer-term observations, we cultured Tak-1 and Mpan-1^ko (male) on rockwool (Fig. S8). By day 35, Tak-1 had formed radially expanded thalli, whereas Mpan-1^ko showed an overlapped and waved form around the basal part of the thalli, which developed earlier or at the early notch plastochron stage (Solly et al., 2017), and were slightly wider and rounded around the distal growing part. Although the gross shape of the thalli of the Mpan knockout lines was abnormal, they formed air pores after day 5 and gemma cups at a later stage (Fig. 3, Fig. S8).

In general, the extent and location of cell proliferation are important for morphogenesis. Gemma of M. polymorpha have two apical notches, where cell proliferation occurs, and grow by bifurcation of the apical notches (Fig. S8A; Solly et al., 2017). However, observation of the apical notches in knockout lines was hampered by the wavy margin of their thalli. Thus, the cell proliferation region was visualized by 5-ethynyl-2′-deoxyuridine (EdU) staining (Nishihama et al., 2015; Yin and Tsukaya, 2016) (Fig. 5). The apical notches on Tak-1 and Mpan-1^ko were similar in number, location and timing of bifurcation (Fig. 5), indicating that the knockout line had the normal growth pattern of thallus.

Next, to identify the factor(s) that influence the thallus shapes of the knock-out lines, we enumerated S-phase cells visualized by EdU staining in the apical notches of 3-day-old gemmalings. The number of EdU signals in the apical notches of Mpan-1^ko was identical to that of Tak-1 (Fig. S9). Thus, cell proliferation is not altered in the Mpan knockout lines. In contrast, the apical notches of Mpan-1^ko were more open than those of Tak-1 (Fig. S9), suggesting an altered growth orientation.

Cell expansion is important in tissue morphogenesis. We cultured thalli under weak blue light to stimulate and evaluate cell elongation. Thalli of Tak-1 and Mpan-1^ko elongated in the vertical direction against the plate surface (Fig. 6A-D). Mpan-1^ko had a more wavy and jagged thallus margin than that of Tak-1. To compare thallus elongation between Tak-1 and Mpan-1^ko, we measured the thallus length of samples mounted on a glass slide (Fig. 6E). Elongation along the growth axis of Mpan-1^ko was suppressed compared with that of Tak-1. This tendency was also seen under standard white light (Fig. 4, Fig. S7). On the other hand, the thickness of thalli did not show any significant difference between Tak-1 and Mpan-1^ko (Fig. S10). In addition, in this growth condition, cell proliferation levels were the same (Fig. S9C). Next, we observed cell morphology by scanning electron microscopy (Fig. 7). In Tak-1, most epidermal cells elongated along the growth direction (Fig. 7A-D), but the elongation of epidermal cells in Mpan-1^ko was suppressed (Fig. 7E-H), particularly around the air pore, resulting in epidermal cells of heterogeneous morphology. Statistical analysis confirmed these differences in epidermal cell elongation between Tak-1 and Mpan-1^ko (Fig. 7I-K). Mpan-1^ko showed greater circularity and a smaller surface area in epidermal cells and a broader cell elongation angle. Moreover, thalli of Tak-1 had a smooth flat surface, whereas those of Mpan-1^ko had a rough uneven surface (Fig. 7B,F). Epidermal cell morphologies were also observed in longitudinal sections (Fig. S11). Epidermal cells of Mpan-1^ko had smaller, longer and slightly thicker cells with greater circularity. These results suggest that MpAN controls both directional and non-directional cell expansion.

In A. thaliana an mutants, it is thought that morphological phenotypes such as altered cell expansion direction and irregular trichomes are caused by abnormal arrangement of cortical MTs (Kim et al., 2002; Folkers et al., 2002). Thus, we examined whether MpAN also affects the orientation of cortical MTs in the epidermal cells of elongated thalli by immunostaining using an anti-α-tubulin antibody (Fig. 8). Cortical MTs in Tak-1 were orientated mainly perpendicular to the growth direction and were aligned in parallel, whereas in Mpan-1^ko, they slightly tended to orientate along the growth direction. These MT orientations are coincident with epidermal cell morphology (Fig. 7). Thus, regulation of cortical MT arrangement by MpAN contributes to well-ordered thallus morphogenesis by controlling cell expansion in the gametophytes of liverworts. Moreover, we observed the subcellular localization of MpAN-Citrine driven by the MpAN promoter (Fig. S12). MpAN-Citrine was observed as widely spread signal and diffuse punctate structures in the cytosol, similar to the GFP-fused AN localization in A. thaliana (Minamisawa et al., 2011; Bhasin and Hülskamp, 2017). The punctate structures of citrine signals were not colocalized with the TagRFP-TUB2 signal that visualizes MTs.

Although the molecular functions of AN were already known to be evolutionarily conserved among land plants in terms of whether homologs can complement the an-1 mutation phenotypes of A. thaliana (Cho et al., 2005; Lin et al., 2008; Minamisawa et al., 2011), the roles of AN in the basal land plant M. polymorpha are unclear. Moreover, the roles of AN in the haploid gametophyte phase are unknown, because previous analyses have been restricted to diploid sporophytes of A. thaliana. In this study, to assess the roles of plant ANs in the haploid phase of basal land plants, we analyzed thallus development in Mpan knockout mutants of M. polymorpha. Mpan knockout mutants showed various morphological phenotypes, such as abnormal thalli that were twisted along the growth axis, suppressed thallus growth and an antheridial receptacle with a curled and wavy margin (Figs 1–4, Figs S4, S5, S7 and S8). This antheridial receptacle phenotype (Figs S4 and S5) and abnormal thalli may have similar causes, because the antheridial receptacle is similar in anatomy to the thallus (Shimamura, 2016). When the Mpan knockout lines were cultured under thallus-elongation conditions, suppressed elongation and an abnormal direction of epidermal cell expansion were apparent (Fig. 7, Fig. S11). Moreover, the knockout lines showed disordered cortical MT orientation (Fig. 8). Cortical MTs control the direction of cellulose microfibrils in the cell wall to determine the direction of cell expansion (Ledbetter and Porter, 1963; Paredez et al., 2006; Chen et al., 2016). These results suggest that MpAN coordinates the formation of well-ordered cell and tissue morphologies by regulating cortical MTs in gametophytes, similar to the situation in A. thaliana sporophytes.

Interestingly, the cortical MTs in leaf epidermal cells of A. thaliana an mutants were more regularly aligned in parallel compared with the wild-type lines (Kim et al., 2002; Folkers et al., 2002), whereas the thallus epidermal cells of the Mpan knockout mutants had more diverse orientations than those of the wild-type lines (Fig. 8). Although these phenomena appear to be opposing, an mutants in the two species share similar defects, i.e. the final cell morphogenesis are perturbed, because in the wild-type lines, the cell morphology differs markedly between leaf epidermis of A. thaliana and thallus epidermis of M. polymorpha. In fact, the organ-level morphological phenotypes were similar between A. thaliana and M. polymorpha; i.e., Mpan knockout mutants formed twisted thalli (Figs 1-3), and the fruits, petals and roots of A. thaliana an mutants are twisted (Fulton et al., 2009; Bai et al., 2010, 2013; Kwak et al., 2015). A. thaliana with mutations in genes encoding tubulin and MT-associated proteins also show twisted (helical) growth in some organs (e.g. Smyth, 2016). In those mutants, the twisting orientations are fixed. Indeed, the twisting of thalli in Mpan knockout mutants was uniform and anti-clockwise. Therefore, MpAN and A. thaliana AN likely have similar functions in the regulation of cortical MT arrangement. How AN regulates the arrangement of cortical MTs is still unknown in A. thaliana. Further research on M. polymorpha will supply us with clues to discover its mechanism. A putative leucine-rich repeat receptor-like kinase, STRUBBELIG/SCRAMBLED (SUB/SCM), physically interacts with AN, and the sub/scm mutants share twisted phenotypes with an in A. thaliana (Fulton et al., 2009; Bai et al., 2013; Kwak et al., 2015). SUB belongs to the STRUBBELIG-RECEPTOR FAMILY (SRF) along with eight other members (SRF1-8) (Eyüboglu et al., 2007). We found one homolog, MpSUB (Mapoly0090s0074.1.p), in M. polymorpha using a blast search and phylogenic analysis. Further comparative research between MpAN and MpSUB will provide new clues to understanding the molecular function of plant ANs.

Recently, A. thaliana AN has been suggested to contribute to abiotic and biotic stress responses (Gachomo et al., 2013; Bhasin and Hülskamp, 2017). However, in this work we could not detect the clear differences in the response to various culture conditions (abiotic stresses and phytohormones) between the WT and the Mpan knockout mutant (Fig. S13).

During land-plant evolution, regulatory genes were acquired to facilitate terrestrialization and development of 3D tissues, resulting in a more complex body plan (Graham et al., 2000; Ishizaki, 2017; Bowman et al., 2017). The molecular functions of AN in land plants are likely evolutionarily conserved, because all plant AN homologs can complement the an-1 mutation phenotypes of A. thaliana (Cho et al., 2005; Lin et al., 2008; Minamisawa et al., 2011). In this study, we show that MpAN regulates the orientation of cortical MTs to ensure correct gametophyte morphogenesis. To our knowledge, this is the first report of the role of AN homologs in the haploid gametophyte phase. On the other hand, knockout of MpAN did not affect sporophyte morphogenesis (Fig. S6). Thus, these findings support the evolutionary conservation of the role of ANs as coordinators of tissue morphology at least in the dominant generations: the gametophyte phase in M. polymorpha and the sporophyte phase in A. thaliana, respectively. Genomic analysis of the filamentous charophyte Klebsormidium nitens suggests that an ancestor of K. nitens acquired genes related to organization of complex multicellular systems despite its simple body plan (Hori et al., 2014; Ohtaka et al., 2017). Indeed, the K. nitens genome harbors one AN homolog, but that of the single-celled Chlorophyta Chlamydomonas reinhardtii does not (Fig. S14). Although animal CtBPs and plant ANs evolved from NAD(H)-dependent D-hydroxyacid dehydrogenase (Stern et al., 2007), the molecular functions of animal CtBPs are distinct from those of plant ANs (Tsukaya, 2006; Stern et al., 2007). In brief, ancestral plant AN was derived from a D-hydroxyacid dehydrogenase in Charophyta; it co-opted to regulate cortical MT orientations (possibly in charophytes) and then contributed to 3D tissue morphology in the gametophytes of basal land plants. According to the evolution of land plants, specifically the transition in dominance from the gametophyte to sporophyte phases, the roles of AN in gametophyte morphogenesis in bryophytes may have been used for sporophyte morphogenesis in vascular plants.

The M. polymorpha female accession Tak-2 and male accession Tak-1 were used as the wild-type lines (Ishizaki et al., 2008). Wild-type and transgenic plants were cultured on half-strength Gamborg's B5 medium (Gamborg et al., 1968) containing 1% (w/v) agar under 50-60 µmol m⁻² s⁻¹ continuous white light at 22°C. To induce the reproductive phase, 2-week-old thalli from gemmae cultured on half-strength Gamborg's B5 medium were transferred to rockwool and cultured under continuous 50-60 µmol m⁻² s⁻¹ white light supplemented with 20-40 µmol m⁻² s⁻¹ far-red light (730 nm) irradiation (VBL-T600-1, Valore). To induce elongation of thalli, wild-type and transgenic plants were cultured on half-strength Gamborg's B5 medium containing 1% (w/v) sucrose and 1% (w/v) agar under continuous 15 µmol m⁻² s⁻¹ blue light (470 nm) irradiation (LED-B, EYELA) at 22°C.

The MpAN targeting vector used for homologous recombination was constructed according to Ishizaki et al., (2013). The 5′ and 3′ ends of the 4.5 and 4.3 kb homologous arms of MpAN, respectively, were amplified from Tak-1 genomic DNA by PCR using KOD FX Neo (Toyobo). To amplify the 5′ end, the forward primer 5′-CTA AGG TAG CGA TTA ATG TTT CAC AAA AGG CAA TAT TCA-3′ and reverse primer 5′-CCG GGC AAG CTT TTA ATG ACC GCC TTA TCA GCA CAA C-3′ were used. To amplify the 3′ end, the forward primer 5′-AAC ACT AGT GGC GCG CAC TCA AGC TCC GTC TAG GG-3′ and reverse primer 5′-TTA TCC CTA GGC GCG GAG GTG CGA TCT GGA AAA AC-3′ were used. The amplified 5′- and 3′-end fragments were cloned into the PacI and AscI sites of pJHY-TMp1 (Ishizaki et al., 2013), respectively, using the In-Fusion HD cloning kit (Clontech). MpAN-oligoA 5′-GCA CCC AGC CTC TCG CCT CCC GAG GCA TCG TCA GTT TTA GAG CTA GAA-3′ and MpAN-oligoB 5′-TTC TAG CTC TAA AAC TGA CGA TGC CTC GGG AGG CGA GAG GCT GGG TGC-3′ were incubated together at 95°C for 5 min to enable annealing. The annealed oligo was cloned into pMpGE_En01 digested with SacI and PstI using the In-Fusion HD cloning kit. The sequence between attL1 and attL2 within pMpGE_En01 was inserted into the binary vector pMpGE010 by LR reaction using LR clonase II (Life Technologies) (Sugano et al., 2018). The vector used for complementation analysis was constructed based on that of Ishizaki et al. (2015). The genomic sequence from the putative promoter region (5 kb upstream of the start codon) to the terminator region (1 kb downstream of the stop codon) of MpAN was amplified from Tak-1 genomic DNA by PCR using KOD FX Neo (Toyobo) with the following primer pairs: proMpAN-Fw 5′-CAC CAG AGT TTG TAT GTA CAG CGG-3′ and tarMpAN-Rev 5′-ACA GGA TCC TAG TTG AGG AC-3′. The PCR product, proMpAN:MpAN:tarMpAN, was cloned into pENTR/D-TOPO (Life Technologies). The plant-specific C-terminal region deleted entry vector was constructed by mutagenesis on pENTR/D-TOPO-proMpAN:MpAN:tarMpAN using the following primer pair: 5′-AAA AGC CTA GTT TAC GTT ATG GAT GAT-3′ and 5′-GTA AAC TAG GCT TTT CGT CGC CCT GCT CTT-3′. The region between attL1 and attL2 within the entry vectors were inserted into the destination vector pMpGWB301 (Ishizaki et al., 2015) by LR reaction using LR clonase II (Life Technologies). The promoter:reporter fusion constructs proMpAN:GUS and proMpAN:Citrine-NLS were also constructed based on those of Ishizaki et al. (2015). The 5 kb promoter region of MpAN was amplified using proMpAN-Fw as described above and proMpAN-Rev 5′-TGC CAG TTC AGT AAT GCC TAC C-3′, and cloned into pENTR/D-TOPO. The constructed entry vector was used for LR cloning with the destination vectors pMpGWB304 and pMpGWB315 (Ishizaki et al., 2015). pENTR/D-TOPO-proMpAN:MpAN was constructed by mutagenesis on pENTR/D-TOPO-proMpAN:MpAN:tarMpAN using the following primer pair: 5′-GCT TAG CAA GGG TGG GCG CGC CGA C-3′ and 5′-CCA CCC TTG CTA AGC CAG CGT ACT A-3′. The constructed entry vector was used for LR cloning with the destination vector pMpGWB307 for the proMpAN:MpAN-Citrine construct (Ishizaki et al., 2015).

The transformation procedures for homologous recombination, the CRISPR/Cas9 system, and the promoter:reporter fusion constructs were performed as described previously (Ishizaki et al., 2008) using spores obtained by crossing Tak-2 with Tak-1. Transformants were selected on half-strength B5 agar medium containing 1% (w/v) agar, 10 mg l⁻¹ hygromycin (Wako Pure Chemical Industries) and 100 mg l⁻¹ cefotaxime (Claforan; Sanofi-Aventis). Two Mpan knockout lines, a female (#49) and a male (#121), constructed by homologous recombination were isolated. These Mpan knockout lines were crossed to generate F1 spores. After spore germination, a female and male line were again isolated. In this study, Mpan-1^ko (female) and Mpan-1^ko (male) were used. Thallus transformation was performed as described previously (Kubota et al., 2013). Thalli from Mpan-1^ko (female) and/or Mpan-1^ko (male), and Agrobacterium containing the pMpGWB301-proMpAN:MpAN:tarMpAN vector, the pMpGWB301-proMpAN:MpAN ΔCT:tarMpAN vector or the pMpGWB307-proMpAN:MpAN vector were used for transformation. pMpGWB304-proMpAN and pMpGWB315-proMpAN were transformed to Tak-1 and Tak-2. Additionally, the pMpGWB307-proMpAN:MpAN vector and the pMpGWB202-TagRFP-MpTUB2 vector for the CaMV35Spro:TagRFP-MpTUB2 construct (Otani et al., 2018) were co-transfomed to Mpan-1^ko (male). Transformants were selected on half-strength B5 agar medium containing 1% (w/v) agar, 0.5 µM chlorsulfuron and 100 mg l⁻¹ cefotaxime (Claforan; Sanofi-Aventis).

Genomic DNA extraction from thalli and genomic PCR were performed as described previously (Ishizaki et al., 2013). To verify the Mpan knockout and Mpan genome-edited lines, the following primer pairs were used: MpAN-a 5′-GGT AGG CAT TAC TGA ACT GGC A-3′ and MpAN-b 5′-TTT TCA CCA AAG GTG GGA AG-3′. The primer sets for the sex chromosome-specific DNA markers rbm27, rhf73 and rbf62 were as described previously (Fujisawa et al., 2001).

Total RNAs were extracted using an RNeasy Plant Mini Kit (Qiagen). cDNAs were synthesized from the extracted RNAs using a PrimeScript RT reagent kit (TAKARA). MpAN transcripts were amplified using the forward primer 5′-AAG GGT TCT GGA AAA AAG GG-3′ and the reverse primer 5′-GCT AAG CCA GCG TAC TCC A-3′. MpEF1α transcripts were amplified using the forward primer 5′-TCA CTC TGG GTG TGA AGC AG -3′ and the reverse primer 5′-GCC TCG ACT AAA GCT TCG TG-3′. The PCR was performed using Quick Taq HS DyeMix (TOYOBO).

GUS staining of thalli proMpAN:GUS transgenic plants was performed as described previously (Althoff et al., 2014).

To visualize S-phase cells, the Click-iT EdU Imaging Kit (Life Technologies) was used. Gemmalings cultured for several days were soaked in half-strength Gamborg's B5 medium with 10 μM EdU at 22°C for 2 h under continuous white light. Thalli were fixed in FAA [50% (v/v) ethyl alcohol, 2.5% (w/v) glacial acetic acid, and 2.5% (w/v) formalin] and were degassed by centrifugation at 10,000 g for 10 s. The fixed samples were stained for EdU with Alexa Fluor 488-azide as described previously (Yin and Tsukaya, 2016). Samples were mounted on glass slides with chloral hydrate solution (200 g chloral hydrate, 20 g glycerol and 50 ml dH₂O). Alexa Fluor 488 fluorescence was visualized using the M16F fluorescence microscope (Leica) at excitation and detection wavelengths of 488 and 519 nm, respectively.

Three-day-old gemmalings were soaked in half-strength Gamborg's B5 medium with 10 μM EdU at 22°C for 1 h under continuous white light. The samples were fixed in 50% (v/v) methyl alcohol and 10% (v/v) acetic acid and were degassed at 10,000 g for 10 s. Samples in fixative solution were incubated at 4°C for at least 12 h. After fixation, the samples were incubated in 80%, 90% and 100% (v/v) ethanol and 90% (v/v) acetone at 80°C for 5 min each, followed by fixative solution at 22°C for 30 min. After four washes with 0.5% (v/v) Triton X-100 in PBS, samples were incubated with an EdU detection cocktail for 30 min in the dark. The samples were rinsed with 0.5% (v/v) Triton X-100 in PBS and incubated in 1% (w/v) periodic acid at 22°C for 45 min. The samples were rinsed again with 0.5% (v/v) Triton X-100 in PBS and incubated in 0.1 mg ml⁻¹ propidium iodide solution for 1 h. After four washes with 0.5% (v/v) Triton X-100 in PBS, the samples were transferred onto slides and incubated with chloral hydrate solution for 30 min. Before observation, a coverslip was placed on top of the samples. Alexa Fluor 488 fluorescence and propidium iodide were visualized using the FV1000 confocal laser scanning microscope (Olympus). Z-series fluorescence images in 2 μm steps were obtained at excitation and detection wavelengths of 473 and 490-540 nm for Alexa Fluor 488, and at 559 and 575-675 nm for propidium iodide, respectively. Z-projection images were created using ImageJ software (Schneider et al., 2012).

Micro-CT images of 10-day-old gemmalings cultured on half-strength Gamborg's B5 medium were obtained using the R_mCT2 micro-CT system (Rigaku) at 90 kV (180 µA) for 4.5 min with a 13 mm field of view. 3D images were reconstructed using OsiriX software (Pixmeo).

Samples fixed in FAA were dehydrated through a graded ethanol series and isoamyl acetate, and dried using the JCPD-5 critical point dryer (JEOL). The samples were mounted using carbon tape and coated with platinum using the JEC-3000FC sputter-coater (JEOL). Images of epidermal cell surfaces were captured using the JSM-6510LV scanning electron microscope (JEOL). Epidermal cell margins were traced using Photoshop software (Adobe) and analyzed using ImageJ software.

Gemmalings were fixed overnight in FAA [50% (v/v) ethyl alcohol, 2.5% (w/v) glacial acetic acid and 2.5% (w/v) formalin] and were degassed. The fixed samples were incubated in 70%, 90% and 100% (v/v) ethanol for 10 min each, followed by 100% (v/v) ethanol for 20 min. Dehydrated samples were then preincubated in a solution of 50% (v/v) Technovit 7100 (Kulzer) for 30 min and in 100% Technovit resin for 30 min. The samples were incubated in 100% resin with 0.05% (w/v) Neutral Red overnight. The samples were embedded in Technovit resin following the manufacturer's protocol. Sections of 7 μm were cut with the HM360 microtome (MICROM), stained with Toluidine Blue and photographed using a DM4500 light microscope (Leica).

Microtubules were visualized based on methods used in previous publication (Pasternak et al., 2015). Gemmalings were transferred into 2% (w/v) paraformaldehyde in PEMT buffer [50 mM PIPES, 2 mM EGTA, 2 mM MgSO₄, and 0.05% (v/v) Triton X-100 (pH 7.2)]. After degassing, the samples were incubated for 40 min at 37°C with gentle shaking. After three washes with water, the water was replaced with 100% methanol and the samples were incubated at 60°C for 10 min. Water was gradually added until a methanol concentration of 20% was reached, and the samples were rinsed again with water. The samples were then incubated in cell-wall digestion solution [0.2% (w/v) driselase and 0.15% (w/v) macerozyme in 2 mM MES (pH 5.0)] at 37°C for 40 min. After three washes with PBS with 0.1% (v/v) Triton X-100 (PBSTX), the samples were treated with membrane-permeabilization solution [3% (v/v) IGEPAL C630, 10% (v/v) dimethylsulfoxide in PBS] at 37°C for 20 min and rinsed four times with PBSTX. The samples were blocked in blocking buffer [2% (w/v) bovine serum albumin in PBSTX] for 30 min and incubated at 37°C for 1 h in blocking buffer containing 1000-fold diluted anti-α-tubulin antibody (Sigma T5168). After three washes with PBSTX, the samples were incubated at 37°C for 1 h with 500-fold-diluted anti-mouse IgG (H+L) secondary antibody conjugated to Alexa Fluor 488 (Thermo Fisher A-11017). Z-series images of Alexa Fluor 488 fluorescence were captured in 0.1 μm steps using the FV1000 confocal laser scanning microscope (Olympus). Cortical MT orientation was quantified using a mathematical morphology method described previously (Kimori et al., 2016).

We thank Dr Akihiro Fujikawa for assistance with the micro-CT scanning (National Institute for Basic Biology); Dr Yuki Kondo (The University of Tokyo) for assistance with the confocal microscopy, and his helpful and kind support; Dr Hiroyuki Koga (The University of Tokyo) for assistance with the immunostaining; and Emma Sarath (The University of Tokyo) for support with manuscript preparation.