Identification and characterization of the land-plant-specific microtubule nucleation factor MACET4

Plant cytokinesis is executed by a highly specialized cellular machine, the phragmoplast (Bajer, 1968; Becker, 1938; Hepler and Jackson, 1968). A typical phragmoplast consists of two nearly parallel microtubule arrays that sandwich the midzone, where assembly of the partition between daughter cells or cell plate takes place (Bajer, 1968; Becker, 1938; Samuels et al., 1995; Segui-Simarro et al., 2004; Wasteneys, 2002). Microtubules function as a scaffold for structural support of the cell plate and for transporting cell plate-building materials. The phragmoplast initially forms between the daughter nuclei at the end of anaphase (Cutler and Ehrhardt, 2002; Ueda et al., 2003). Completion of cell plate synthesis promotes microtubule disassembly in this region. The microtubule structural protein, tubulin, is released during depolymerization and is then recycled for assembling new microtubules at the outer edge of the cell plate (Yasuhara et al., 1993). As phragmoplast diameter is generally smaller that the diameter of the cell, continuous microtubule polymerization, de-polymerization and re-polymerization drives phragmoplast expansion towards the cortical division zone located at the plasma membrane of the cell. Attachment of the cell plate to a site within the cortical division zone leads to phragmoplast disassembly.

Phragmoplast microtubules are twice as dynamic as interphase microtubules (Hush et al., 1994; Smertenko et al., 2011). One of the underlying reasons for this difference is continuous depolymerization of microtubules in the region where cell plate synthesis is accomplished and polymerization of microtubules on the phragmoplast outer edge. What drives the depolymerization remains unknown, but new microtubules in the phragmoplast are nucleated by the γ-tubulin ring complex (γ-TuRC; Hashimoto, 2013). Furthermore, phragmoplast microtubules are shorter than those in interphase. A higher frequency of transition from polymerization to depolymerization (catastrophe) could be responsible for both restricting microtubule elongation and for increasing microtubule dynamicity.

On the tree of life, the phragmoplast appears in Charophyceae algae, which gives rise to land plants (embryophytes) (Buschmann and Zachgo, 2016; Leliaert et al., 2011). This fact suggests that higher tissue complexity, which accompanied adaptation to terrestrial life, was facilitated by cytokinesis with the phragmoplast. However, whether expansion is the intrinsic phragmoplast feature remains debatable considering findings that within the Charophyceae family expansion has been observed in Coleochaete orbicularis (Cook, 2004), but not in Chara zeylandica (Cook et al., 1998).

Overall phragmoplast morphology resembles that of the animal cell cytokinetic structure, the midbody (Otegui et al., 2005). In particular, both structures are composed of two anti-parallel microtubule arrays that overlap at the midzone. However, the midbody lacks expansion and possesses a structurally stable midzone, whereas the phragmoplast midzone undergoes rapid reorganization (Samuels et al., 1995; Smertenko et al., 2018). The majority of known phragmoplast microtubule regulators play a conserved role in midbody functions (McMichael and Bednarek, 2013). Considering the unique nature of phragmoplast expansion, we hypothesized that plants must have evolved specific proteins to facilitate this process. To date, the only plant-specific microtubule-binding protein localized in the phragmoplast known is Tan1 (Cleary and Smith, 1998; Smith et al., 2001). In angiosperms, Tan1 is described to function in determining the cell division plane orientation (Rasmussen et al., 2011; Walker et al., 2007). We hypothesize there are additional plant-specific proteins that determine features of phragmoplast organization which need to be studied in order to understand and model plant cytokinesis.

Here, we characterized embryophyte-specific microtubule-binding phragmoplast protein MACERATOR4 (MACET4) (Schmidt and Smertenko, 2016). MACET4 expresses during cell division and localizes to the anaphase spindle poles and phragmoplast distal zones where microtubule nucleation takes place. Biochemical analysis demonstrates that MACET4 oligomerizes, promotes microtubule nucleation and polymerization whilst inhibiting transition from depolymerization to polymerization (rescue). We propose that MACET4 oligomers promote microtubule nucleation and indirectly promote catastrophe by rapid depletion of the free tubulin pool in the cell.

The Arabidopsis protein MACET4 (At1g23790; Schmidt and Smertenko, 2016) localizes to microtubules when transiently expressed in tobacco (Nicotiana benthamiana) leaf pavement cells (Fig. S1A). MACET4 is part of a co-expression network with a key regulator of phragmoplast assembly, a member of the kinesin 7 family, NACK1/Hinkel (Fig. S1B), and with a key spindle assembly factor TPX2. The co-expression network of MACET4, NACK1 and TPX2 also includes known regulators of cell division including MAP65-3 (also known as PLE), the kinesins ATK5 and PAKRP1L, and cyclin B2. Conceivably, MACET4 is a cell-cycle-regulated gene. To test this hypothesis, we expressed a GUS reporter gene under control of the MACET4 promoter (Fig. 1). Peak GUS activity was detected in root and shoot meristems, lateral root primordia, young leaves, vasculature, developing stomata, pollen grains and young flowers. No discernible GUS activity was detected in differentiated leaves.

To test the localization of MACET4 during cell division, we expressed MACET4–GFP under the control of the ubiquitin promoter in tobacco BY-2 cells. However, transgenic lines could not be recovered, suggesting that overexpression of MACET4 was lethal. To overcome this problem, we expressed GFP–MACET4 under control of an estrogen-inducible promoter in BY-2 cells and found that MACET4 associates with cortical microtubules, the pre-prophase band and the phragmoplast, but remained cytoplasmic during metaphase (Fig. 2; Movie 1). MACET4 is recruited onto microtubules again during anaphase (Fig. 2). During cytokinesis, MACET4–GFP localized to the phragmoplast distal zones (Fig. 2).

Genome analysis of plants from different taxa revealed that MACET4 homologs are only found in embryophytes (Fig. S2), suggesting that MACET4 evolved as plants began to colonize land. To determine the functions of MACET4, we analyzed T-DNA knockout mutants (Fig. S3). Seedlings of mce4-1 and Col-0 plants were grown at 8°C for 4 weeks. The root growth rate was not different (Fig. 3A); however, the root meristem zone was longer in mce4-1 than in Col-0 (Fig. 3B). Considering localization of MACET4 during cytokinesis, we examined morphology of the phragmoplast in mce4-1. The overall phragmoplast morphology showed no discernible phenotype in mce4-1; however, the phragmoplast length, measured as the distance between edges of the opposite distal zones (Smertenko et al., 2017) was greater in mce4-1 (Fig. 3C,D). This phenotype could be partially complemented by expressing proMACET4:MACET4-GFP in the mce4-1 background (Fig. 3C,D).

Analysis of mce4-1;proMACET4:MACET4-GFP lines revealed localization of MACET4 in the preprophase band and the phragmoplast, but not in the spindle (Fig. 3E–G; Movies 2 and 3). We also detected signal in the cortical division zone (Fig. 3G, frame 548). Overall, the localization pattern in the phragmoplast in A. thaliana roots was similar to that in the BY-2 cells, with exception that labeling of anaphase spindle was restricted to the spindle poles (Fig. 3G, frame 476). Labeling of anaphase spindle microtubules in BY-2 cells could be the consequence of protein overexpression driven by the inducible promoter. The expression level in the interphase cells was below the sensitivity of our approach. Thus, MACET4 is a cell-cycle-regulated embryophyte-specific protein that controls phragmoplast size.

The longer phragmoplasts in mce4-1 indicates that MACET4 contributes to regulating microtubule dynamics during cytokinesis. However, resolving the behavior of individual microtubules in the phragmoplast remains technically impossible due to the high concentration of free tubulin in this region (Vyplelová et al., 2018). Instead, we analyzed microtubule dynamics in leaf pavement cells of tobacco plants stably expressing Arabidopsis β6-tubulin–mCherry under the control of the CaMV35S promoter (Abe and Hashimoto, 2005). Expression of MACET4–GFP under control of the strong UBI10 promoter results in dense labeling of microtubules (Fig. 4A), which is different from the GFP puncta observed in the phragmoplast (Fig. 2; Fig. 3E–G). In an attempt to mimic the phragmoplast localization, we expressed MACET4–GFP under control of native promoter (proMACET4:MACET4-GFP). This approach was successful resulting in the GFP puncta localized along the microtubules and in the cytoplasm (Fig. 4A). We used both approaches to determine the impact of MACET4 on microtubule dynamics.

Detailed analysis of proMACET4:MACET4-GFP cells showed two patterns of GFP puncta association with the microtubule lattice: transient (<30 s) or persistent (>30 s). All persistent associations coincided with the microtubule crossover sites (n=50; Fig. 4B). Furthermore, all microtubule plus-end depolymerization events (n=43) were accompanied by the enrichment of GFP signal at the depolymerizing tip (Fig. 4B; Fig. S4A, Movie 4). Shrinking minus-ends of ‘treadmilling’ microtubules also accumulated MACET4–GFP (Fig. S4B,C). Enrichment of MACET4–GFP signal preceded cytoplasmic microtubule nucleation events (Fig. 4C; Movie 5) and coincided with the sites of branching microtubule nucleation (Fig. 4D; Movie 6). We found that all microtubule nucleation events under our experimental conditions occurred at sites enriched with MACET4 (n=12).

To measure the impact of MACET4 on microtubule dynamics, we artificially colored growing and shrinking microtubule ends using a published technique (Lindeboom et al., 2013). Growing and shrinking microtubules were apparent in both control and cells transfected with proMACET4:MACET4-GFP, but not in cells transfected with proUBN10:GFP-MACET4 (Fig. 4E; Movies 7 and 8). In the latter case, microtubules appeared to be stable and the only type of dynamics was wavering (Fig. 4E; Fig. S4D, Movie 9). Treatment of cells with an inhibitor of microtubule polymerization amiprophos methyl (APM) caused depolymerization of microtubules in the control, but microtubule fragments persisted in cells constitutively expressing MACET4 (Fig. S4E). Hence, overexpression of MACET4 stabilizes microtubule in vivo through the inhibition of catastrophe events.

Analysis of microtubule dynamics in cells transfected with proMACET4:MACET4-GFP showed that MACET4 reduces microtubule polymerization and depolymerization rates (Fig. 4F). Furthermore, MACET4 caused a greater frequency of depolymerizing microtubules (Fig. 4F; Fig. S4G) while not affecting the frequency of microtubule polymerization events (Fig. S4F). MACET4 also promoted the transition from polymerization to pause, from pause to catastrophe, and from polymerization to catastrophe. At the same time, transitions from catastrophe to polymerization and from catastrophe to pause were greater in these cells than in the control. Therefore, ectopic expression of MACET4 under control of native promoter, which results in lower levels of protein expression relative to that mediated by strong constitutive expression, causes a shift of bulk microtubule dynamics towards depolymerization.

Although MACET4 colocalizes with microtubules in cells (Fig. 4A,B), this interaction could be indirect through another microtubule-binding protein. To test whether MACET4 binds microtubules directly, we performed a co-sedimentation assay with recombinant MACET4 and Taxol-stabilized microtubules. MACET4 was recovered in the pellet with microtubules, whereas without microtubules it remained in the supernatant (Fig. 5A). Hence, MACET4 is a true microtubule-binding protein.

Next, we determined the impact of MACET4 on microtubule polymerization by performing a turbidimetric assay. MACET4 increased the density of a tubulin solution in a concentration-dependent manner (Fig. S5A). Light-absorbance of a tubulin solution could increase due to microtubule polymerization, bundling or both. To distinguish between these possibilities, we measured the impact of MACET4 on bulk microtubule polymer production by performing a sedimentation assay. Different concentrations of MACET4 were added to a tubulin solution and incubated at 35°C for 20 min. The polymer was recovered by centrifugation and analyzed by SDS-PAGE (Fig. S5B). We found that the amount of microtubules in the pellet increased in a MACET4 concentration-dependent manner (Fig. S5C). This result demonstrates the ability of MACET4 to promote tubulin polymerization.

To test the bundling activity, we examined the effect of MACET4 on microtubules polymerized with ATTO488-labeled tubulin. Without MACET4 we observed formation of individual microtubules, whereas addition of MACET4 to the polymerization reaction caused formation of asters (Fig. 5B). Furthermore, the presence of MACET4 resulted in shorter individual microtubules relative to the control (Fig. 5C). The ability to induce formation of shorter microtubules in these assays prompted At1g23790 to be named MACERATOR4 (MACET4). However, we have no evidence that MACET4 reduces microtubule length through severing. Instead, microtubule elongation is likely to be inhibited by depleting free tubulin in the reaction. In this way, MACET4 increases the bulk of microtubule polymer while restricting elongation of individual microtubules.

The fact that MACET4 can induce asters, taken together with colocalization to the cytoplasmic microtubule nucleation sites in vivo (Fig. 5C), suggests that MACET4 possesses microtubule nucleation activity. To test this hypothesis, we created stable microtubule ‘seeds’ with ATTO488-labeled tubulin polymerized with the non-hydrolyzable GTP analog GMPCPP. We then mixed 5 µM Rhodamine-labeled tubulin with the ‘seeds’ and allowed the microtubules to polymerize. Tubulin concentration was optimized to prevent spontaneous microtubule nucleation but to allow polymerization on the seeds (Fig. 5D). In the presence of MACET4, 32% of microtubules formed de novo (defined as microtubule nucleation events that occur independently of the seed) relative to 0% in the control supplemented with MACET4 dialysis buffer (Fig. S5D). Furthermore, every seed in the presence of MACET4 had a microtubule extension, whereas as many as 28% of the seeds in the control reactions exhibited no microtubule extensions (Fig. S5E). Therefore, MACET4 promotes microtubule nucleation at subcritical concentrations of tubulin and facilitates microtubule elongation.

MACET4-induced asters resemble those generated from centrosomes. However, centrosomes promote nucleation of microtubules by recruiting multiple γ-tubulin ring complexes (γ-TuRCs) each consisting of many subunits. In gel-filtration chromatography experiments, MACET4 eluted just before the 443 kDa marker (Fig. 5E). Considering predicted molecular weight of 57 kDa, this approximately corresponds to the octamer, which is smaller than a γ-TuRC. To explain this observation we hypothesized that MACET4 facilitates aster formation by bundling early nucleation intermediates. To test this hypothesis we added MACET4 to a mixture of stable ‘seeds’ made with Rhodamine-labeled or ATTO488-labeled tubulin. Compared to the control reactions, addition of MACET4 caused three types of interactions between microtubules: parallel, branching and annealing (Fig. 5F,G). A known microtubule-bundling protein, MAP65, was used as a positive control, and induced formation of bundles distinct from those formed with MACET4. This outcome implies that MACET4 promotes aster formation through bundling of early microtubule nucleation intermediates, as MACET4-mediated bundling is reminiscent of asters rather than elongated bundles as seen with MAP65.

Although nucleation and branching activities provide a mechanism for the aster formation, understanding why microtubules formed in the presence of MACET4 are shorter required measurements of microtubule dynamics in vitro. In these experiments, we polymerized microtubules from Rhodamine-labeled tubulin on ATTO488-labeled seeds and followed microtubule dynamics. Kymographs showed that in the presence of MACET4, microtubules depolymerize down to the seed, whereas in the control reactions, rescues were common (Fig. 5H). Quantification of the kymographs demonstrates that MACET4 at the concentration 50 nM or 100 nM does not affect microtubule polymerization or depolymerization rates, but reduced the frequency of rescue by almost 3-fold (Fig. 5I). Therefore, MACET4 promotes microtubule nucleation and restricts microtubule elongation (Fig. 5J).

The primary structure of MACET4 lacks any known functional domains. The predicted secondary structure consists primarily of random coil with several α-helical regions (Fig. 6A). Alignment of MACET4 homologs from angiosperms, gymnosperms and mosses revealed three conserved regions (I, II and III) that coincide with α-helical regions (Fig. 6B). We tested whether one of these regions is a microtubule-binding domain by producing fragments of MACET4 that include either N-terminal conserved region I (amino acid residues 1–127) or C-terminal conserved regions II and III (amino acid residues 287–518).

Co-sedimentation assays with taxol-stabilized microtubules demonstrated that both regions are capable of binding microtubules (Fig. 6C). All MACET4^287-518 in the reaction mixture was recovered in the pellet with microtubules and no protein was detected in the supernatant, whereas some of MACET4^1-127 remained in the supernatant. Consistently with this outcome, transient expression experiments in N. benthamiana cells showed that whilst MACET4^287-518 associated with microtubules, GFP-MACET4^1-127 was cytoplasmic (Fig. 6D). This means the main microtubule-binding region of MACET4 encompasses the C-terminal conserved motifs.

To determine the functions of MACET4^1-127 and MACET4^287-518 domains, we tested their activities in vitro. Turbidimetric assays revealed that MACET4^287-518 increases the optical density of the tubulin solution, whereas MACET4^1-127 lacked any discernible activity relatively to the control (Fig. 6E). Furthermore, the lag phase of tubulin as measured by assessing the optical density at 350 nm (OD₃₅₀) was shorter in the presence of MACET4^287-518 relative to full-length MACET4 or control (Fig. 6F). A shorter lag phase indicates the ability of the C-terminal region to promote microtubule nucleation and that the C-terminal region has enhanced function in the absence of the N-terminal region.

Measurements from the bulk microtubule polymer assay revealed that MACET4^287-518 promoted tubulin polymerization in a concentration-dependent manner (Fig. 6G,H). MACET4^287-518, but not MACET4^1-127, caused microtubule aster assembly (Fig. 6J). Microtubules in the asters originate from a bright central core, which seemingly functions similar to a centrosome in animal cells. We attempted to produce the cores by mixing 1 µM MACET4^287-518 with 5 µM ATTO488–tubulin. Supplementing this reaction with 10 µM Rhodamine–tubulin produced red asters on the green cores (Fig. 6K). No cores formed in the reactions containing ATTO488–tubulin alone. The proportion of short microtubules was greater in the reactions containing MACET^287-518 relative to the control (Fig. 6I). Taken together, these data indicate that: (1) the conserved C-terminal region of MACET4 possesses all activities of full-length MACET4, (2) the N-terminal region dampens activity of the C-terminal region, and (3) MACET4 and tubulin form microtubule-nucleating centers.

All thus far known microtubule-associated proteins in the phragmoplast are also found in animals. The only exception is Tan1, which directly binds microtubules (Smith et al., 2001) and cooperates with another microtubule-binding protein AIR9 in maintaining cell division plane orientation (Martinez et al., 2017). The role of Tan1 in the regulation of phragmoplast microtubule dynamics and its contribution to the evolution of phragmoplast remains to be determined. Here, we identify an embryophyte-specific microtubule-binding protein in the phragmoplast, called MACET4 (Schmidt and Smertenko, 2016). This finding, taken together with our knowledge about Tan1, supports the hypothesis that phragmoplast evolution required unique plant microtubule regulators.

Functional characterization demonstrates that MACET4 promotes microtubule polymerization at a concentration of tubulin that is incapable of self-nucleation in vitro and decreases the lag phase of microtubule polymerization in the turbidimetric assays. We also found colocalization of MACET4 with all free and lattice-associated microtubule nucleation sites in vivo. Taken together, MACET4 appears to be a microtubule-nucleation factor.

Microtubule nucleation in the phragmoplast was shown to be facilitated by the evolutionarily conserved γ-TuRC, which comprises six main subunits, GCP1–GCP6 (γ-tubulin is GCP1) and several associated proteins (Hashimoto, 2013). γ-TuRC nucleates microtubules preferentially on the lattice of existing microtubules, resulting in the formation of a branched microtubule network (Hashimoto, 2013). The knockout of both genes encoding γ-tubulin in Arabidopsis results in gametophytic lethality (Pastuglia et al., 2006) demonstrating functional non-redundancy of this nucleation pathway. By contrast, knockout of MACET4 results in a mild phenotype, which suggests either a non-essential role of MACET4 in microtubule nucleation or functional redundancy of MACET4 with its close homolog MACET5. Detailed characterization of MACET4 and MACET5 double-knockout mutants would provide more insights on the role of MACET4 in vivo.

Animals have an alternative to the γ-TuRC microtubule nucleation mechanisms. One of them relies on ‘targeting protein for Xenopus laevis kinesin-like protein 2’ (TPX2; Gruss et al., 2001; Wittmann et al., 1998). TPX2 promotes microtubule nucleation during metaphase spindle assembly by reducing the rate of dissociation for tubulin subunits, thus stabilizing early nucleation intermediates (Reid et al., 2016; Roostalu et al., 2015; Wieczorek et al., 2015). Despite a different primary structure, the secondary structure of TPX2 and MACET4 consists of several predominantly α-helical domains interspersed by unstructured regions. Both TPX2 and MACET4 can nucleate microtubules and induce aster formation in a solution of purified tubulin (Brunet et al., 2004; Schatz et al., 2003). Plants have a TPX2 that binds microtubules of the mitotic spindle and plays a key role in spindle assembly (Vos et al., 2008). However, TPX2 does not bind microtubules during cytokinesis. By contrast, MACET4 does not bind microtubules in the mitotic spindle. This means plants could exploit distinct microtubule-nucleation mechanisms for assembling the mitotic spindle or the phragmoplast.

Another evolutionarily conserved microtubule nucleation mechanism relies on XMAP215 (also known as Stu2, chTOG and Alp14 and CKAP5 in various systems). This protein promotes microtubule polymerization and nucleation both in vitro and in vivo, and localizes to the microtubule nucleation sites and centrosomes in yeast and animal cells (Flor-Parra et al., 2018; Gunzelmann et al., 2018; Roostalu et al., 2015; Thawani et al., 2018; Wieczorek et al., 2015). XMAP215, like MACET4, also combines microtubule nucleating and polymerization activities, but the plant XMAP215 homolog MOR1 (also known as GEM1) binds along phragmoplast microtubules (as opposed to the puncta seen for MACET4) and localizes to the midzone (Kawamura et al., 2006; Twell et al., 2002). Distinct localization patterns indicate that MOR1 and MACET4 perform different functions during cytokinesis.

Both XMAP215 (including homologs from other systems) and TPX2 cooperate with γ-TuRC in promoting microtubule nucleation: TPX2 binds γ-tubulin through a conserved C-terminal domain whereas the XMAP215 of fission and budding yeasts is targeted to the γ-TuRC by the TACC ortholog Alp7 or Spc72 (Alfaro-Aco et al., 2017; Flor-Parra et al., 2018; Gunzelmann et al., 2018). Cooperation of MACET4 with γ-TuRC will have to be established by analyzing their colocalization at the nucleation sites. Considering colocalization of MACET4 with all microtubule nucleation sites in the interphase cells, both pathways are likely to cooperate. Overexpression of MACET4 in interphase cells does not have an impact on the frequency of microtubules nucleation events. It is possible that MACET4 nucleation activity is suppressed during interphase or masked by TPX2-, MOR1- or γ-TuRC-mediated nucleation mechanisms.

Despite promoting microtubule nucleation, TPX2 and MACET4 appear to have a different impact on microtubule dynamics: TPX2 reduces the frequency of catastrophe and increases the frequency of rescue leading to longer microtubules in the spindle (Reid et al., 2016; Roostalu et al., 2015; Wieczorek et al., 2015), whereas MACET4 promotes catastrophe in vivo and suppresses the rescue in vitro. Suppression of rescues in vitro could be indirect through depletion of free tubulin as the consequence of more-efficient microtubule nucleation. Partial depletion of the free tubulin pool could also be the reason for the overall suppression of microtubule polymerization in cells expressing MACET4 under control of its native promoter. Release of free tubulin during catastrophe would eventually enable nucleation of new or rescue of the existing microtubules (Fig. 3H). In this way MACET4 would ultimately cause formation of many short microtubules, which are typical for the phragmoplast (Segui-Simarro et al., 2007). Consistent with this hypothesis, knockout of MACET4 leads to longer phragmoplasts. Our data also suggests that activity of MACET4 is modulated by the availability of free tubulin and by the cell cycle stage-dependent signaling processes.

Determining how MACET4 increases phragmoplast length would require careful analysis of phragmoplast microtubule dynamics in the mutant background. However, considering its strong nucleation activity, MACET4 could function in concentrating tubulin from all parts of the cytoplasm in the phragmoplast region. If this hypothesis is correct, then phragmoplast microtubules would be less dynamic in mce4-1 strains. Furthermore, root apical meristem size was increased in mce4-1 at 8°C through an as-yet-unknown mechanism.

In addition to promoting microtubule nucleation, MACET4 induces formation of asters from the non-polymerized tubulin in vitro in the manner that is similar to that mediated by centrosomes and TPX2 proteins (Gruss et al., 2002; Mitchison and Kirschner, 1984; Schatz et al., 2003). Asters formed in the reactions containing free tubulin, whereas mixing MACET4 with polymerized microtubules resulted in bundling and annealing. TPX2 has been proposed to form asters by bundling microtubules during nucleation (Schatz et al., 2003). Given the available evidence, induction of aster formation by MACET4 and by TPX2 is coupled with the microtubule nucleation activity. Tobacco histone H1 was also shown to induce aster formation (Nakayama et al., 2008); however, whether this activity is coupled to the nucleation is yet to be determined.

One important question is how does MACET4 promote microtubule nucleation? Our experiments demonstrate that, under control of the ubiquitin promoter, MACET4 localizes along the microtubule lattice and suppresses microtubule depolymerization in a TPX2-like manner. This means that a high concentration of MACET4 inhibits catastrophe. Under physiological protein levels, a high local concentration of MACET4 could be achieved through oligomerization, as MACET4 forms octamers in vitro and foci in vivo. These oligomers could function as the microtubule nucleation sites by inhibiting catastrophe.

A member of the MACET gene family, MACET1 (also known as CORD1), has been recently reported to play a role in the destabilization of cortical microtubules during formation of pits in the wall of xylem cells (Sasaki et al., 2017). The mechanism for microtubule destabilization by CORD1/MACET1 was proposed to be through inhibition of interaction between microtubules and the plasma membrane. Overexpression of CORD1/MACET1 resulted in microtubule wavering (Sasaki et al., 2017). We found that overexpression of MACET4 also induced wavering, but not the destabilization of microtubules. Instead microtubule depolymerization was suppressed. According to these findings, MACET4 and CORD1/MACET1 have distinct activities. Although we cannot exclude that decoration of microtubules by MACET4 in the overexpression situation causes wavering by out-competing the plasma-membrane linkers, the physiological meaning of this mechanism needs careful examination in cells expressing MACET4 under control of the native promoter. The density of microtubule labeling in this situation may not be sufficient to out-compete the interaction between microtubules and the membrane linkers unless these linker proteins bind to the same region on the microtubule lattice.

The direct binding of CORD1/MACET1 to microtubules has not been yet confirmed, but considering it shows a high conservation in the MACET4 microtubule-binding region, CORD1/MACET1 is likely a microtubule-binding protein. Experiments in vivo have demonstrated that CORD1/MACET1^200-505 decorates microtubules, whereas CORD1/MACET1^301-505 is cytoplasmic (Sasaki et al., 2017). By contrast, MACET4^287-518 could bind microtubules both in vitro and in vivo and perform all major activities of the full-length MACET4 in vitro. Thus, there appears to be a functional diversification amongst the MACET/CORD family members for the regulation of microtubule dynamics during cell division and vascular cell development. Detailed characterization of biochemical properties of MACET1 in the future would unravel the mechanisms underlying different activities of these proteins.

In conclusion, we identified and functionally characterized the first embryophyte-specific microtubule regulator in the phragmoplast, MACET4. MACET4 is a microtubule nucleation factor that accumulates at the distal zones of the phragmoplast, where microtubule nucleation takes place. Increase of the phragmoplast length in mce4-1 strains indicates that MACET4 is essential for maintaining microtubule length in the phragmoplast. Our data together indicate that MACET4 facilitates microtubule turnover in the phragmoplast by maintaining the balance between nucleation and depolymerization. This work advances our knowledge on the mechanisms underlying phragmoplast assembly and paves the way for understanding phragmoplast evolution.

Full-length MACET4 cDNA was amplified by PCR from Arabidopsis thaliana Col-O leaf cDNA using primers MACET4fwdattB1 and MACET4revattB2 (Table S1) and cloned into Gateway binary vectors pUBN or pUBC (Grefen et al., 2010) harboring green fluorescent protein (GFP) for N- and C-terminal tagging, respectively. Plasmids were transformed into Agrobacterium tumefaciens strain GV3101. Transfected A. tumefaciens were grown overnight in 3.0 ml of YEB at 30°C, shaking at 200 rpm until optical density at 600 nm (OD₆₀₀) was between 0.5 and 1.0 for MACET4 clones, and OD₆₀₀≈1.0 for p19 clone. Cells were collected by centrifugation at 3000 g for 5 min and washed two times with infiltration medium composed of 10 mM MES pH 5.6, 10 mM MgCl₂, and 200 µM Acetosyringone. After the second wash, bacteria were resuspended in 1.0 ml infiltration medium and incubated at room temperature for 2–5 h. Cells harboring MACET4 constructs were mixed with cells harboring p19 at the ratio 1:1. N. benthamiana plants were grown at 21°C under a light cycle of 16 h light and 8 h dark for 3–4 weeks. Infiltration constructs were injected into the abaxial leaf side of either wild-type plants or CaMV35S:Tub6-mCherry stable transformants (Abe and Hashimoto, 2005). Leaves were imaged at 2–5 days after infiltration using a Leica SP8 or Leica SP8X confocal microscope. GFP was excited at 488 nm and mCherry was excited at 561 nm. To analyze colocalization of MACET4–GFP with the microtubule nucleation sites, the nucleation events were identified first on the TuB6–mCherry channel, and then MACET4–GFP channel was examined for the evidence of MACET4 accumulation at the nucleation site. Kymographs were constructed using ImageJ (Schindelin et al., 2012).

Full-length MACET4 cDNA was amplified using primers MACET4 fwd attB2r and MACET4 rev attB3 (Table S1). An estradiol-inducible promoter was amplified from the pMDC7 vector (Curtis and Grossniklaus, 2003) using primers pMDC7fwdattB4 and pMDC7revattB1r. Fragments were recombined into the gateway vector pH7m34 together with GFP sequence so that the GFP was located on MACET4 N-terminus. Plasmids were transformed into A. tumefaciens strain LBA4404. BY-2 tissue culture cells were transformed with agrobacterium. Induction was performed with 5 µM estradiol for 48 h. Cells were imaged with a Leica SP8 confocal microscope.

A T-DNA insertional mutant from an Arabidopsis Biological Resource Center (ABRC) seed stock (SALK_070182.53.50) was genotyped for left border insertions using primers Salk070182fwd. RNA was extracted from seedlings and flowers of T-DNA lines homozygous for the left border insertion. cDNA was made with the MaximaH reverse transcriptase kit. cDNA flanking the insertion was amplified using primers complimentary to exon sequences: Salk079547fwd and Salk079547rev for mce4-1. cDNA upstream of the insertion was amplified using primers MACET4_fwd and MACET4 T_DNA_rev. cDNA downstream of the insertion was amplified using primers MACET4 T-DNA_fwd and MACET4_rev. cDNA was also amplified using the positive control primers, Salk073077_fwd and Salk073077_rev, which amplify a region on chromosome 3 within the MACET1 gene (Table S1). The lack of amplification upstream, downstream and flanking the SALK_070182.53.50 insertion in mce4-1 was interpreted as a knockout.

Arabidopsis seeds were surface-sterilized with 75% ethanol with 0.05% (v/v) Triton X-100 for 15 min, 95% (v/v) ethanol with 0.05% Triton X-100 for 8 min. Seeds were then rinsed with 95% (v/v) ethanol, dried for 15 min, spread on half-strength MS-medium supplemented with 0.5% (w/v) sucrose, and incubated in 24 h light at 22°C. Two-day-old seedlings were transferred to half-strength MS plates and grown at 8°C for 4 weeks. Roots were imaged and root length was measured with ImageJ (Schindelin et al., 2012). Col-0 seedlings were used as a control. One-way ANOVA with Kruskal–Wallis test and Dunn's multiple comparison post-test were performed in GraphPad Prism version 5.01.

Arabidopsis seeds were germinated on agar plates as above. Three-day-old seedlings of Col-0, mce4-1 or proMACET4:MACET4-GFP were harvested and fixed for 1 h at room temperature in 3.7% paraformaldehyde, 50 mM PIPES, pH 6.8, 5 mM EGTA, 2 mM MgSO₄, 0.4% (v/v) Triton X-100 and 0.02% (w/v) glutaraldehyde. Roots were then washed three times for 5 min each time in PBST [PBS with 0.1% (v/v) Tween 20]. Cell walls were digested for 7 min at room temperature using an enzyme mixture containing 2% (w/v) Dricelase, 0.4 M Mannitol, 5 mM EGTA, 15 mM MES pH 5.0, 1 mM PMSF, and 10 µg/ml each of leupeptin and pepstatin A. Seedlings were rinsed twice for 5 min each time in PBST followed by in PBST supplemented with 2% (w/v) bovine serum albumin and 0.05% (w/v) NaN₃ for 30 min. Seedlings were incubated overnight at 4°C in rat primary anti-tubulin antibody YOL 1/34 (Fisher Scientific, 50174608) diluted 1:300 in the latter buffer. Following three washes for 2 h each in PBST, seedlings were incubated overnight at 4°C in secondary anti-rat-IgG Alexa Fluor 488-conjugated secondary antibody diluted 1:400 as above. After two washes for 30 min each in PBST, DNA was stained with a solution of DAPI (10 mg/ml) in PBST for 1 h at room temperature. Root tips were excised from the seedlings and mounted in Vectashield (Vector Labs). The coverslips were sealed using nail varnish. Phragmoplast microtubules in root tips were imaged with an SP8 confocal microscope. Phragmoplasts were measured with ImageJ and statistical analysis was performed with Graphpad Prism 5.1.

The MACET4 promoter region, defined as the 2.3 kb region upstream of the MACET4 transcriptional start site, was amplified using primers MACET4ProFWDattB1 and MACET4ProrevattB2 (Table S1). The promoter region was cloned into Gateway vector pBI101G upstream of the glucouronidase (GUS) reporter gene. Wild-type Arabidopsis plants were transformed using the floral dip method with A. tumificiens harboring proMACET4:GUS (Clough and Bent, 1998). GUS activity was detected in T2 generation by using a previously published procedure (Weigel and Glazebrook, 2002). To generate C-terminal GFP fusions, proMACET4 and the open reading frame of MACET4 were cloned together into the gateway vector pMDC107 (Curtis and Grossniklaus, 2003). mce4-1 T-DNA null mutant plants were transformed using floral dip method, and plants from the T2 generation were imaged. Membranes in roots were stained in 1 µM FM464 for 5 min and rinsed in water prior to imaging.

MACET4 full-length cDNA or fragments were amplified using primers listed in Table S1, and cloned into the pGAT4 vector. A majority of the experiments were performed with non-labeled MACET4 in order to exclude potential artifacts associated with addition of the fluorescent protein. Protein expression was conducted in E. coli strain BL21(DE3)Rosetta-2. Bacterial cultures were grown at 30°C until the OD₆₀₀ reached ∼0.45 and induced with 1 mM IPTG for 3 h. Cells were collected by centrifugation at 3000 g for 5 min at room temperature. The pellet was resuspended in extraction buffer containing 50 mM HEPES pH 7.0, 300 mM NaCl, 20 mM imidazole, 30 mM β-mercaptoethanol, and 8 M urea. Cells were extracted overnight at room temperature. Debris was removed by centrifugation at 150,000 g for 2 h. Supernatant was poured into a nickel resin column and mixed for 10 min. Non-specifically bound proteins were removed with two washes of 40 mM imidazole with 4 M urea, and then three washes with 60 mM imidazole and 3 M urea. Bound proteins were eluted with 250 mM imidazole with 2 M urea. Eluate was concentrated and further purified using a Superdex 200 Increase column. Purified protein was dialyzed overnight at 4°C against buffer containing 20% (v/w) glycerol, 50 mM PIPES-KOH pH 6.8, 2 mM EGTA, 2 mM MgSO₄, 10 mM DTT and 50 mM KCl.

Tubulin was purified from bovine brain using a published procedure (Castoldi and Popov, 2003). Microtubules were polymerized by adding 1 mM GTP, 10 µM Taxol and MTSB (50 mM PIPES-KOH, pH 6.8, 2 mM EGTA and 2 mM MgSO₄) to tubulin at final concentration of 40 µM. The mixture was incubated at 37°C for 30 min. Microtubules at 2 mg/ml microtubules in MTSB supplemented with 200 mM KCL were mixed with 2 µM MACET4, MACET4^1-127 or MACET4^287-518 and centrifuged at 100,000 g at 25°C for 10 min. 50 µl of supernatant was aspirated and mixed with 50 µl of 2× SDS-PAGE sample buffer. The remaining supernatant was discarded. The pellet was washed with 150 µl of warm MTSB supplemented with 0.5% (w/v) Tween-20 and resuspended in 200 µl of SDS-PAGE sample buffer. Protein samples were heated at 95°C for 5 min, then 20.0 µl of the samples were separated on a 12.5% (w/v) polyacrylamide resolving gel, 4% (w/v) polyacrylamide stacking gel at 30 mA. The gels were rinsed in deionized water, stained with colloidal Coomassie Brilliant Blue G-250 overnight and then distained with deionized water.

To measure microtubule polymerization through a turbidimetric analysis, we prepared a reaction mixture containing 1 mM GTP, 20 µM tubulin and MACET4 in 120 µl of MTSB buffer. Reactions were mixed on ice in quartz cuvettes. Cuvettes were transferred from ice to a pre-warmed spectrophotometer chamber. The OD₃₅₀ was measured at 350 nm every minute for 30 min or every 10 s for 15 min.

For measuring microtubule polymerization through sedimentation analysis, the reaction contained 1 mM GTP, 8 µM Tubulin, 0.05% (w/v) Tween-20 and MACET4 protein in MTSB, at a final volume of 100 µl. Reactions were incubated at 37°C for 20 min then centrifuged at 85,000 g at 37°C for 10 min. Supernatant was aspirated and the pellet was carefully washed with a 0.05% (w/v) solution of Tween 20 in deionized water at room temperature. The pellet was resuspended in 200 µl SDS-PAGE sample buffer and heated at 95°C for 5 min. 20 µl of protein was separated on a 12.5% (w/v) polyacrylamide resolving gel with a 4% (w/v) polyacrylamide stacking gel at 30 mA as described above. Band intensity was measured using ImageJ and charts were created with Graphpad Prism.

Tubulin was labeled with NHS-Biotin, NHS-ATTO-488, or NHS-Rhodamine by following a previously published procedure (Peloquin et al., 2005). The mixtures of labeled and unlabeled tubulin were supplemented with 1 mM GMPCPP and 1 mM DTT in BRB80 buffer (80 mM PIPES-KOH pH 6.9, 1 mM MgCl₂ and 1 mM EGTA) to reach a final tubulin concentration of 40 µM. The mixture was incubated on ice for 5 min, clarified by centrifuging at 85,000 g for 5 min at 2°C, snap frozen in 10 µl aliquots and stored at −80°C. To polymerize tubulin, an aliquot of the seed mix was thawed in a 37°C water bath and supplemented with 5.0 µl of pre-warmed buffered glycerol. The reaction was incubated at 37°C for 20 min, and then gently resuspended in 150 µl of warm BRB80 plus 1 mM DTT, layered over a 60% glycerol cushion, pH 6.8, containing 10 µM Taxol, and centrifuged at 85,000 g for 5 min at 35°C. The glycerol cushion and supernatant were aspirated and the pellet was gently resuspended in 30 µl of warm BRB80 with 1 mM DTT. Seeds were further diluted in BRB80 plus 1 mM DTT as necessary.

Microtubule seeds were made from the mixture of non-labeled and Rhodamine-labeled tubulin, or non-labeled and Atto488-labeled tubulin at the stoichiometric ratio of 3:1, as described above. Each reaction contained a mixture of ‘red’ and ‘green’ seeds in 20 µl BRB80. Reactions were incubated for 5 min at 37°C, then 8.0 µl of the reaction was placed on a 22×22 mm coverslip and imaged with a Leica TIRF DMI6000 microscope equipped with iLas 2 TIRF and FRAP system (Biovision Technologies). A minimum of 15 images were taken for each reaction. Seed encounters were scored manually and calculated as the percentage of total events.

A reaction mixture contained non-labeled and ATTO488-labeled tubulin at a ratio of 6:1 at a final concentration of 20 µM, and 1 mM GTP in 20 µl of MTSB. Reactions were mixed on ice then incubated for 20 min at 35°C. Microtubules were fixed by adding 100 µl of warm (35°C) MTSB supplemented with 0.5% (v/v) glutaraldehyde for 15 min. The reaction was further diluted in 600 µl of room temperature MTSB. 5 µl of reaction mix was added to a 22×22 mm coverslip and placed on a microscopy slide. Imaging was performed with a Leica TIRF microscope as above. A minimum of 15 images were taken for each reaction. Single microtubule lengths were measured using ImageJ and statistical analysis was performed with Graphpad Prism.

Microtubule dynamics reconstitution experiments were performed using previously published procedures (Bieling et al., 2010). PEG-coated coverslips (MicroSurfaces, Inc.) were functionalized with 45 mg/ml solution NHS-biotin in dimethylformamide by sandwiching 80 µl of the biotin solution between two 22×22 mm coverslips and incubating at 75°C for 1 h. The coverslips were then rinsed twice in dimethylformamide and six times in RNase-free water. For functionalizing with NeutrAvidin, biotinylated coverslips were incubated for 10 min at room temperature with solution containing 50 µg/ml NeutrAvidin, 0.05% (w/v) K-casein, 80 mM PIPES (pH 6.8, KOH), 85 mM KCl, 85 mM KoAC, 0.25% (w/v) Brij-35, 4 mM MgCl₂, 1 mM EGTA, 1 mM GTP, 2 mM ATP and 10 mM β-mercaptoethanol. Coverslips were then rinsed in the same buffer, but without NeutrAvidin and K-casein, three times for 5 min each time, followed by in RNase-free water. Washed coverslips were blotted and air-dried.

To construct the reaction chamber, the functionalized coverslips and microscopy slides were cut in half along the long axis. Then, two pieces of double-sided tape were placed on the slide and the functionalized surface of the coverslip was placed perpendicular and face down on the tape so that edges of the coverslip extended over the edge of the slide, creating a ‘landing stage’ and chamber for buffer delivery. Nail varnish was used on the edge of the coverslip to secure the coverslip to the slide at the tape-attachment point. The chamber was rinsed twice with warm BRB80 plus DTT and then incubated for 5 min at room temperature with seeds prepared from ATTO488-labeled, biotin-labeled and non-labeled tubulin at the ratio 1:1:2. To flush the non-attached seeds, 80 µl of warm BRB80 plus DTT was added on one side of an extended coverslip and wicked with a paper towel from the other side. The chamber was then rinsed twice with 80 µl of ensemble buffer [80 mM PIPES-KOH, pH6.8, 85 mM KCl, 85 mM KoAC, 0.25% (w/v) Brij-35, 4 mM MgCl₂, 1 mM EGTA, 1 mM GTP, 2 mM ATP, 10 mM β-mercaptoethanol, 0.05% (w/v) β-casein, 0.1% (w/v) methylcellulose and Antifade system (250 nM glucose oxidase, 64 nM catalase, 40 mM D-glucose)].

Microtubule polymerization assays were performed on a microscope-stage pre-heated to 35°C. 5 µM tubulin and MACET4 were mixed on ice in ensemble buffer. 80 µl of ice-cold reaction solution was wicked through the chamber and imaging began immediately. Kymographs were made using the ImageJ kymograph plugin. Polymerization rates were calculated using Microsoft Excel. Statistical analysis was performed using Graphpad Prism.

Gene co-expression analysis was performed using ATTED-II (http://atted.jp/). Sequences were aligned using ClustalX (Thompson et al., 1997). The phylodendrogram was constructed using the Bootstrap/Jackknife method with PAUP 4.0 (Sinauer Associates). Bootstrap values were calculated from 1000 replicates; only groups with bootstrap scores of 60 or above were retained in the phylodendrogram. Selaginella MACET was used as an outgroup.

We thank ABRC for providing T-DNA seed stocks, Deirdre Fahy for technical assistance and Takashi Hashimoto for mCherry–TuB6 constructs.