ESCRT Machinery Mediates Cytokinetic Abscission in the Unicellular Red Alga Cyanidioschyzon merolae.

In many eukaryotes, cytokinesis proceeds in two successive steps: first, ingression of the cleavage furrow and second, abscission of the intercellular bridge. In animal cells, the actomyosin contractile ring is involved in the first step, while the endosomal sorting complex required for transport (ESCRT), which participates in various membrane fusion/fission events, mediates the second step. Intriguingly, in archaea, ESCRT is involved in cytokinesis, raising the hypothesis that the function of ESCRT in eukaryotic cytokinesis descended from the archaeal ancestor. In eukaryotes other than in animals, the roles of ESCRT in cytokinesis are poorly understood. To explore the primordial core mechanisms for eukaryotic cytokinesis, we investigated ESCRT functions in the unicellular red alga Cyanidioschyzon merolae that diverged early in eukaryotic evolution. C. merolae provides an excellent experimental system. The cell has a simple organelle composition. The genome (16.5 Mb, 5335 genes) has been completely sequenced, transformation methods are established, and the cell cycle is synchronized by a light and dark cycle. Similar to animal and fungal cells, C. merolae cells divide by furrowing at the division site followed by abscission of the intercellular bridge. However, they lack an actomyosin contractile ring. The proteins that comprise ESCRT-I–IV, the four subcomplexes of ESCRT, are partially conserved in C. merolae. Immunofluorescence of native or tagged proteins localized the homologs of the five ESCRT-III components [charged multivesicular body protein (CHMP) 1, 2, and 4–6], apoptosis-linked gene-2-interacting protein X (ALIX), the ESCRT-III adapter, and the main ESCRT-IV player vacuolar protein sorting (VPS) 4, to the intercellular bridge. In addition, ALIX was enriched around the cleavage furrow early in cytokinesis. When the ESCRT function was perturbed by expressing dominant-negative VPS4, cells with an elongated intercellular bridge accumulated—a phenotype resulting from abscission failure. Our results show that ESCRT mediates cytokinetic abscission in C. merolae. The fact that ESCRT plays a role in cytokinesis in archaea, animals, and early diverged alga C. merolae supports the hypothesis that the function of ESCRT in cytokinesis descended from archaea to a common ancestor of eukaryotes.

In many eukaryotes, cytokinesis proceeds in two successive steps: first, ingression of the cleavage furrow and second, abscission of the intercellular bridge. In animal cells, the actomyosin contractile ring is involved in the first step, while the endosomal sorting complex required for transport (ESCRT), which participates in various membrane fusion/fission events, mediates the second step. Intriguingly, in archaea, ESCRT is involved in cytokinesis, raising the hypothesis that the function of ESCRT in eukaryotic cytokinesis descended from the archaeal ancestor. In eukaryotes other than in animals, the roles of ESCRT in cytokinesis are poorly understood. To explore the primordial core mechanisms for eukaryotic cytokinesis, we investigated ESCRT functions in the unicellular red alga Cyanidioschyzon merolae that diverged early in eukaryotic evolution. C. merolae provides an excellent experimental system. The cell has a simple organelle composition. The genome (16.5 Mb, 5335 genes) has been completely sequenced, transformation methods are established, and the cell cycle is synchronized by a light and dark cycle. Similar to animal and fungal cells, C. merolae cells divide by furrowing at the division site followed by abscission of the intercellular bridge. However, they lack an actomyosin contractile ring. The proteins that comprise ESCRT-I-IV, the four subcomplexes of ESCRT, are partially conserved in C. merolae. Immunofluorescence of native or tagged proteins localized the homologs of the five ESCRT-III components [charged multivesicular body protein (CHMP) 1, 2, and 4-6], apoptosis-linked gene-2-interacting protein X (ALIX), the ESCRT-III adapter, and the main ESCRT-IV player vacuolar protein sorting (VPS) 4, to the intercellular bridge. In addition, ALIX was enriched around the cleavage furrow early in cytokinesis. When the ESCRT function was

INTRODUCTION
Cytokinesis is a fundamental biological phenomenon in all organisms. However, in eukaryotes the mechanisms are diverse. A significant difference exists between a group of animals, fungi, and Amoebozoa (Amorphea; Burki, 2014) and the other groups (Excavates and Diaphoretickes) ( Figure 1A). Cells of Amorphea generally divide depending on constriction of the contractile ring (Pollard, 2017; Figure 1A), whereas those of other eukaryotic groups lack myosin-II, an essential ring component (Mishra et al., 2013; Figure 1A). The mechanism of cytokinesis has varied further in each group during evolution. One example is the cytokinesis of land plants whose cells divide by developing cell walls and membranes from the cell center toward the cell periphery (Muller and Jurgens, 2016). Our current knowledge of cytokinesis mostly depends on a limited number of model organisms. However, the mechanisms in different lineages warrant exploration to reveal the eukaryotic history and the core mechanisms of cytokinesis shared by eukaryotes.
In mammalian cells, cytokinesis proceeds by equatorial membrane furrowing followed by abscission of the intercellular bridge. The contractile ring containing actomyosin and septin filaments constricts to furrow the membrane (Green et al., 2012). As the ring closes, the midbody, the platform for the abscission machinery where the plus and minus ends of spindle microtubules overlap, is formed. The actin-capping protein that controls actin polymerization is required for the process (Terry et al., 2018). The intercellular bridge contains the spindle midzone microtubules and midbody. The septin filaments are reorganized into rings in the early intercellular bridge to assist bridge maturation (Renshaw et al., 2014;Karasmanis et al., 2019).
The endosomal sorting complex required for transport (ESCRT), a protein complex conserved among eukaryotes (Table 1), contributes to various membrane fusion/fission events such as multivesicular body formation at late endosomes and nuclear envelope fusion Schoneberg et al., 2017). In mammalian cells, ESCRT mediates scission of the intercellular bridge (Carlton and Martin-Serrano, 2007;Morita et al., 2007;Elia et al., 2011;Guizetti et al., 2011;Campsteijn et al., 2016;Schoneberg et al., 2017). The proteins comprising ESCRT-I-IV, the four subcomplexes of ESCRT, are sequentially targeted to the midbody. ESCRT-I recruits charged multivesicular body protein (CHMP) 4 in ESCRT-III by binding to CHMP6 in ESCRT-III by itself or through ESCRT-II (Christ et al., 2016). CHMP6 is a nucleation factor for ESCRT-III. A recent study showed that Septin (SEPT) 9, a constituent of the septin ring, associates with the ESCRT-I protein tumor susceptibility gene (TSG) 101 to assist the recruitment of ESCRT-II and demarcate the sites for ESCRT-III assembly (Karasmanis et al., 2019). The septin ring disassembles as ESCRT-III machinery develops (Karasmanis et al., 2019). In addition to ESCRT-I, the ESCRT-III adaptor protein apoptosis-linked gene-2-interacting protein X (ALIX) localizes at the midbody to separately recruit CHMP4 without binding to CHMP6 (Christ et al., 2016).
ESCRT possibly represents conserved machinery in eukaryotic cytokinesis inherited from the archaeal ancestor. In Sulfolobus, a thermophile archaeon, homologs of ESCRT-III proteins and VPS4, and the ESCRT-III scaffold cell division protein (Cdv) A are detected between daughter nucleoids of dividing cells, correlating with the site of membrane ingression (Table 1; Lindas et al., 2008;Samson et al., 2008Samson et al., , 2011Liu et al., 2017). They are necessary for cytokinesis from early to final stages. Whereas ESCRT-dependent cytokinesis is not universal in archaea (Makarova et al., 2010), recent studies support that eukaryotes have diverged from archaea encoding ESCRT (Zaremba-Niedzwiedzka et al., 2017; Table 1). However, in eukaryotes other than in animals, whether ESCRT mediates cytokinetic abscission is poorly understood. In the land plant Arabidopsis thaliana, elc mutation, a mutation of TSG101, results in the production of multinucleated cells (Spitzer et al., 2006). Although the mechanism underlying induction of the phenotype is unclear, it may reflect conserved functions of ESCRT in eukaryotic cytokinetic abscission. Some similarities between the animal midbody and plant phragmoplasts, arrays of microtubules on the division plane, have been indicated in a previous study (Otegui et al., 2005).
Because ESCRT is a conserved multifunctional complex, the presence of ESCRT genes in the genome does not necessarily suggest its involvement in cytokinesis. To determine whether ESCRT is primordial core machinery FIGURE 1 | Distribution of representative proteins in the contractile ring, EF1α, and ESCRT in eukaryotes that divide centripetally and a scheme for C. merolae cytokinesis. (A) The phylogenetic tree is based on Burki (2014). Branch length does not represent evolutionary distance. Amorphea generally divide depending on the actomyosin contractile ring. Other eukaryotes lack myosin-II, an essential component of the contractile ring. Septins are part of the contractile ring in animal cells, whereas in fungal cells, they form separate ring structures. Some contractile ring components are found in the intercellular bridge (ICB), which is not depicted in the figure. Protein localization (to the cleavage furrow or intercellular bridge) or the presence of genes in species were investigated by literature and/or BLAST searching.
The list of proteins in mammals and S. cerevisiae excluding CdvA is based on Schuh and Audhya (2014). In the C. merolae genome, ESCRT and ALIX homologs were searched by BLAST using S. cerevisiae sequences as queries. The results for ESCRT homologs were consistent with those of Leung et al. (2008). CdvA homologs in eukaryotes were searched by BLAST using the S. acidocaldarius sequence as a query. "−" no detectable homologs. The information of Sulfolobus is based on Lindas et al. (2008); Samson et al. (2008), and Liu et al. (2017). Both Sulfolobus species have four CHMP family proteins. It is unclear which eukaryotic CHMP protein is the closest. "+" indicates that the homologous sequence is present in the genomes of "Asgard" archaea, the group proposed to be the closest to eukaryotes (Zaremba-Niedzwiedzka et al., 2017). Blank indicates that the protein or gene is not mentioned in the above results. Underlined proteins localize at the midbody or intercellular bridge (mammals) or between daughter nucleoids (Sulfolobus). The references are in this legend or the text. * VPS28 and EAP30 are required for midbody localization of TSG101 and EAP20, respectively (Christ et al., 2016). Proteins in bold font were examined in this study.
for eukaryotic cytokinesis, we explored ESCRT functions in the acidothermophilic unicellular red alga Cyanidioschyzon merolae that branched early in eukaryotic evolution (Yoon et al., 2004(Yoon et al., , 2006. In addition to the phylogenetical position, C. merolae provides an excellent experimental system. The cell (∼2 µm in diameter) has a simple structure (Kuroiwa, 1998). The genome (16.5 Mb, 5335 genes) has been completely sequenced (Matsuzaki et al., 2004;Nozaki et al., 2007). Genetic transformation is feasible (Ohnuma et al., 2008;Fujiwara et al., 2013), and a light and dark cycle highly synchronizes cell cycle progression and thus the timing of cytokinesis in a population (Suzuki et al., 1994; Supplementary Figure S1A). Unlike other algae and plants, C. merolae does not have a rigid cell wall. It divides through membrane furrowing at the equator (constriction stage) that takes several minutes, followed by scission of the intercellular bridge (abscission stage), a stage that completes within a minute ( Figure 1B; Supplementary Figures  S1A,B). C. merolae lacks the actomyosin contractile ring and septins ( Figure 1A). The actin gene does not seem to be expressed in C. merolae, and staining with phalloidin, which detects F-actin, is negative (Suzuki et al., 1995;Takahashi et al., 1995;Matsuzaki et al., 2004). Moreover, no myosin heavy chain gene or septin genes are present in the C. merolae genome (Matsuzaki et al., 2004). The only protein that has been linked to C. merolae cytokinesis is elongation factor (EF) 1α, which accumulates at the cleavage furrow ( Figure 1B, Supplementary Figure S2; Imoto et al., 2011), as observed in Tetrahymena (Numata et al., 2000) and sea urchin eggs (Fujimoto and Mabuchi, 2010). Sea urchin EF1α bundles actin filaments and maintains the contractile ring structure (Fujimoto and Mabuchi, 2010). However, in C. merolae, actin filaments are probably absent and thus the function of EF1α in cytokinesis is unclear.
In this study, we investigated localization of ESCRT proteins in C. merolae by immunofluorescence and examined the effects of a dominant-negative mutant of VPS4 on cytokinesis. Five homologs of ESCRT-III proteins (CHMP1, CHMP2, and CHMP4-6), ALIX, and VPS4 localized at the intercellular bridge before cytokinetic abscission. ALIX also located close to the cleavage furrow early in the constriction stage. The expression of mutant VPS4 caused abscission failure, indicating that ESCRT mediates cytokinetic abscission in C. merolae.

RESULTS
The C. merolae genome encodes homologs for 11 ESCRT proteins and ALIX (Table 1). We refer to these homologs according to the names of mammalian proteins except for the homolog of mammalian CHMP5, CHMP5/VIG1 (Vacuolar inheritance gene 1), which was previously characterized in C. merolae (Fujiwara et al., 2010;Yagisawa et al., 2018). To understand ESCRT functions in cytokinesis, we first examined the localization of ESCRT-III, the structure most directly involved in membrane deformation. We labeled CHMP2 using specific antibodies (Supplementary Figure S3A). In a synchronized culture under a light-dark cycle, the protein was expressed throughout the cell cycle with an increased level during the dark period (Supplementary Figure S3B). Immunofluorescence showed that CHMP2 localized on the punctate cytoplasmic structures and intercellular bridge of cytokinesis (Figure 2A). Next, we examined whether other ESCRT-III components localize with CHMP2 at the intercellular bridge using strains that ectopically expressed proteins fused to hemagglutinin (HA)-tags. C. merolae encodes two CHMP1 homologs (CMR340C and CMQ376C; Table 1). CHMP1-HA (CMR340C) localized at the intercellular bridge with CHMP2 ( Figure 2B). CHMP1-HA (CMQ376C) was not expressed consistently with the lack of the expressed sequence tag (EST) of the native gene (data not shown; Matsuzaki et al., 2004). CHMP4-HA, CHMP5/VIG1-HA, and CHMP6-HA localized at the intercellular bridge with CHMP2 ( Figure 2B).
To further examine the involvement of ESCRT in C. merolae cytokinesis, we detected the localization TSG101, a major component of ESCRT-I, and ALIX. TSG101-HA was detected on the cytoplasmic puncta, but not on the intercellular bridge ( Figure 3A). Although we also tested N-terminally tagged HA-TSG101, it was not expressed (data not shown). In contrast to TSG101, FLAG-tagged ALIX localized to the intercellular bridge ( Figure 3B and Supplementary Figures S4A,B). During early constriction, ALIX-FLAG also located around the cleavage furrow (Figures 3C,D and Supplementary Figures S4A,B). The signals partially overlapped with those of EF1α (Figures 3C,D). In the other stages (G1, M, and late constriction), ALIX-FLAG was mainly localized close to the cell membrane and on some cytoplasmic structures (Supplementary Figures S4A,B).
Our attempts to knock out some ESCRT genes were unsuccessful, suggesting that ESCRT disruption is lethal in C. merolae. An ATPase-inactive dominant-negative mutant of VPS4 blocks cytokinetic abscission in mammalian cells (Carlton and Martin-Serrano, 2007;Morita et al., 2007). To further clarify the role of ESCRT in cytokinesis, we expressed the corresponding mutant VPS4 (E292Q) in C. merolae cells.
When expressed under control of the native promoter sequence, VPS4-HA localized on the intercellular bridge ( Figure 4A). To assess the effect of the mutation on cytokinesis, wild-type ( WT ) or the mutant ( E292Q ) VPS4-HA were expressed under the control of a heat-inducible promoter in the synchronized culture. The cells were subjected to heat treatments at the beginning of the dark period (G2/M phase, as shown in Figures 1B, 4B,C). VPS4 WT -HA cells completed cell division in 12 h after the onset of heat shock, which was similar to untreated cells (Figures 4D,E). In contrast, induction of VPS4 E292Q -HA accumulated cells with notably elongated intercellular bridges (Figures 4D,F-H). Most of these long intercellular bridges were spanned by the spindle (Figure 5A) and positive for VPS4 E292Q -HA and CHMP2 ( Figure 5B).

DISCUSSION
ESCRT potentially represents a component of the most ancient conserved machinery for cytokinetic abscission in eukaryotes. However, little is known about such ESCRT functions in eukaryotes other than in animals. In this study, we revealed that ESCRT is an essential component for cytokinetic abscission in C. merolae, an early diverged eukaryote.
We found that five ESCRT-III proteins, CHMP1 (CMR340C), CHMP2, CHMP4, CHMP5/VIG1, and CHMP6, localized at the intercellular bridge of C. merolae (Figures 6A,B). In mammalian cells, CHMP1-6, including its isoforms (Carlton and Martin-Serrano, 2007;Morita et al., 2007Morita et al., , 2010Dukes et al., 2008;Yang et al., 2008;Bajorek et al., 2009;Elia et al., 2011;Guizetti et al., 2011;Carlton et al., 2012;Goliand et al., 2014;Christ et al., 2016), and IST1 (Agromayor et al., 2009;Goliand et al., 2018) localize at the midbody. C. merolae is devoid of genes encoding CHMP3 and IST1. In addition, ESCRT-III genes, except for CHMP1, exist as a single copy. Thus, ESCRT-III machinery in C. merolae is simpler in terms of protein composition. Electron microscopy has shown that mammalian ESCRT-III proteins either form or assist in forming a spiral of 17 nm-diameter filaments underlying the intercellular bridge membrane (Guizetti et al., 2011;Mierzwa et al., 2017;Schoneberg et al., 2017). The identification of such a structure is challenging in C. merolae because of the short duration of the abscission stage and small size of the intercellular bridge. Thus, further extensive studies are required to elucidate the structure involved in cytokinetic abscission. C. merolae TSG101 appeared to be absent from the intercellular bridge. Mammalian ESCRT-I and ALIX localize at the midbody to separately target ESCRT-III (Christ et al., 2016). ESCRT-I depends on CHMP6 to recruit other ESCRT-III proteins, whereas the ALIX route does not (Christ et al., 2016). C. merolae CHMP6 resided at the intercellular bridge (Figures 6A,B). Thus, it potentially has a role unrelated to ESCRT-I. ESCRT-I is found in all major eukaryotic taxa but was secondarily lost in some species (Williams and Urbe, 2007;Leung et al., 2008). Although we cannot completely rule out the possibility that the addition of epitope-tags altered the localization of the protein or that the antibody could not react with the protein because of poor accessibility, the absence of TSG101 from the intercellular bridge may suggest a major role of ALIX in recruiting ESCRT-III. In mammalian cells, both ESCRT-I and ALIX are recruited by centrosome protein 55 kDa (CEP55), a midbody protein. However, CEP55 is absent in C. elegans and Drosophila melanogaster, although they depend on ESCRT for cytokinetic abscission (Green et al., 2013;Lie-Jensen et al., 2019). In Drosophila, ALIX is recruited to the midbody by Pavarotti, a homolog of human mitotic kinesin-like protein (MKLP) 1 (Lie-Jensen et al., 2019). In C. merolae, CEP55 or MKLP1 homologs have not been found. Thus, upstream mechanisms to recruit ESCRT appear to vary among organisms.
In contrast to TSG101, C. merolae ALIX was enriched at the intercellular bridge (Figures 6A,B). We also detected ALIX around the cleavage furrow early in the constriction stage ( Figure 6A). Of related interest is CdvA, a scaffold protein for ESCRT-III in the archaea Sulfolobus (Lindas et al., 2008;Samson et al., 2011). It localizes to the mid-region of the cell, corresponding to membrane ingression sites from the beginning to final stages of cell division. The localization precedes that of ESCRT-III (Samson et al., 2011). Because C. merolae and eukaryotes other than Amorphea lack the actomyosin contractile ring, understanding the role of ALIX or ESCRT in the constriction stage would be of interest for future study.
We found that VPS4 localized to the intercellular bridge (Figures 6A,B). The phenotype of cells expressing VPS4 E292Q -HA was strikingly similar to that observed after overexpression of VPS4, either WT or dominant-negative forms (Carlton and Martin-Serrano, 2007;Morita et al., 2007), or disruption of spastin (Connell et al., 2009) in mammalian cells. This phenotype is also reminiscent of that in the archaea Sulfolobus overexpressing truncated ESCRT-III proteins, which exhibit long intercellular bridges (Liu et al., 2017). Therefore, C. merolae . The results showed that 94.5 ± 3.7% of the long intercellular bridges were positive for spindles, and 94.1 ± 5.3% were positive for VPS4 E292Q -HA that colocalized with CHMP2. Arrowheads, locations of the VPS4E292Q-HA signal on the long intercellular bridge. BF, bright field. Scale bars, 2 µm. VPS4 plays a pivotal role in scission of the intercellular bridge, as seen in these organisms. C. merolae VPS4 E292Q -HA resided with CHMP2 on the long intercellular bridge, suggesting that the dynamics of ESCRT-III regulated by the AAA-ATPase activity of VPS4 are critical for cytokinetic abscission in this organism. Finally, our data indicate that ESCRT mediates cytokinetic abscission in eukaryotic cells that lack the contractile ring and septins, and in the eukaryotic intercellular bridge that is considerably smaller than that of mammalian cells. In mammalian cells, the contractile ring is required for midbody formation (Hu et al., 2012). The polymerization state of actin controls midbody maturation which is essential for the appropriate assembly of ESCRT-III (Terry et al., 2018). The clearance of F-actin from the intercellular bridge after the furrow closure is also a limiting step in ESCRT-III recruitment (Fremont et al., 2016). Septins function in both the contractile ring and intercellular bridge. They are essential for maturation and stabilization of the intercellular bridge as well as proper ESCRT-III assembly (Renshaw et al., 2014;Addi et al., 2018;Karasmanis et al., 2019). The inhibitory effects of F-actin on ESCRT-III recruitment, and the role of septins in ESCRT-III assembly may be confined to animals or eukaryotic groups in which cell division is dependent on the contractile ring. The mammalian midbody is >1 µm in diameter (Mullins and Biesele, 1977;Green et al., 2012), and ESCRT-III is targeted for >40 min before abscission (Stoten and Carlton, 2018). However, in C. merolae, the intercellular bridge is ∼200 nm in diameter and requires less than 1 min to be cleaved (Supplementary Figure S1B). Thus, C. merolae appears to control cytokinetic abscission more simply than in mammalian cells. Importantly, regardless of these differences and the phylogenetic distance, ESCRT components mediate cytokinetic abscission in these organisms.
In summary, we demonstrate that five ESCRT-III proteins, ALIX, and VPS4 localize at abscission sites to mediate cytokinetic abscission in C. merolae. We also show that ESCRT functions in cytokinesis of an organism that lacks the contractile ring and septins. The fact that ESCRT mediates cytokinesis in archaea, animals, and the early diverged red alga C. merolae supports the idea that ESCRT is the primordial machinery for cytokinetic abscission in eukaryotes. We expect that exploring other lineages of eukaryotes that undergo ESCRT-mediated cytokinetic abscission and characterization of their mechanisms should further advance our understanding of the conserved mechanisms and evolution of eukaryotic cytokinesis.

Strain Generation
The primers and plasmids used to generate strains are listed in Supplementary Table S1. All strains except for the ALIX-FLAG strain were generated by integration of DNA fragments into the upstream region of the URA5.3 gene (Fujiwara et al., 2015). To add tags to ESCRT proteins, plasmids containing transformation cassettes, which included the upstream (-2300 to -898 bp) of the URA5.3 gene (CMK046C), genes encoding ESCRT proteins with their promoter region, 3 × HA-tag, the 3 UTR of β-tubulin, and the URA5.3 gene with the promoter region, were generated using an In-Fusion HD Cloning kit (Clontech). For CHMP2-, CHMP4-, and CHMP5/VIG1-HA, PCR products #1, #5 and one of #2-#4 (Supplementary Table S1) were used. For other ESCRT proteins, PCR products #6 and one of #7-#11 were used. Plasmids containing a heat shock promoter (Sumiya et al., 2014) and VPS4-HA were generated by fusing PCR products #12 and #13 using the In-Fusion HD cloning kit. The plasmid for dominantnegative VPS4-HA (E292Q; a mutation in conserved Walker B motif; Hanson and Whiteheart, 2005) was prepared by In-Fusion cloning of PCR product #14. Transformation of C. merolae was conducted as described previously (Ohnuma et al., 2008;Fujiwara et al., 2015). Briefly, the M4 strain (a point mutant of URA5.3; Minoda et al., 2004) was transformed with PCRamplified cassettes from each plasmid (#15) using a polyethylene glycol-mediated method. The transformants were selected for uracil independence in starch placed on solidified MA2 medium (Fujiwara et al., 2013). Establishment of the strain ALIX-FLAG was performed following the procedures of Takemura et al., 2019a. PCR products #16 and #17, corresponding to the 3portion (from + 997 to + 2496, where +1 is the first base position of the initiation codon) and the 3 -downstream region (from +2497 to +3996) of the CMC051C (ALIX) ORF, respectively, were inserted into StuI-digested pMKTf (Takemura et al., 2018) to construct the plasmid pMKTf-ALIX-Tagging. Subsequently, transformation cassette #18 was amplified from pMKTf-ALIX-Tagging by PCR and used to transform the uracil-auxotroph T1 strain (Taki et al., 2015) as described previously (Takemura et al., 2019a). Transformants were selected on uracil-free MA2 plates using the top starch method as described previously (Takemura et al., 2019b).
Generation of an Antibody Against C. merolae CHMP2 DNA fragments encoding the CMB008C and pQE80 expression vector (Qiagen) were amplified by PCR with the primers listed in Supplementary Table S2. The fragments were fused and circularized using the In-Fusion HD cloning kit, resulting in a construct containing the six-histidine tag at the N-terminus of CMB008C. The recombinant proteins were purified with His-Trap columns (GE Healthcare Life Sciences) and used to raise antibodies in rats (T. K. craft, Ltd.).

Time-Lapse Imaging
A synchronized culture at M phase was mounted on coverslips, which had pieces of surgical tape at the corners, and was then incubated for 30 min at room temperature. After removing excess medium, the coverslips were inverted and placed in glassbottom dishes. The dishes were transferred into a chamber for live imaging (BZ-H3XD; Keyence) at 40 • C. Images were obtained under a microscope (BZ-X700; Keyence) using a ×100 objective.

Microscopy
For immunofluorescence, cells were fixed with methanol containing 1% formaldehyde and 10% DMSO at −20 • C overnight. The fixed cells were centrifuged at 1500 × g at 4 • C, washed once with cold methanol (−20 • C), and then twice with PBS. For blocking, the cells were treated with either Blocking One (Nakarai Tesque) for 15 min at 4 • C or 5% BSA for 30 min at 37 • C. The antibody reaction was performed for 1 h at 4 • C. Primary antibodies were diluted in PBS and used at the following dilutions: 1:500 for rat anti-CHMP2, 1:100 for rabbit anti-αtubulin (Fujiwara et al., 2009), 1:500 for guinea pig anti-EF1α (Imoto et al., 2011), 1:1000 for mouse anti-HA (Clone 16B12; BioLegend), and 1:1000 for mouse anti-DYKDDDDK tag (to detect FLAG-tag; Clone 1E6; Wako). Fluorescent secondary antibodies (Thermo Fisher Scientific) were diluted in PBS and applied at 1:1000 for Alexa Fluor 488 and 1:100 for Alexa Fluor 555. DNA was stained with 1 µg/ml 4 ,6-diamidino-2-phenylindole (DAPI). Images were acquired under the BZ-X700 fluorescence microscope using the × 100 objective. For Alexa Fluor 488, the GFP filter was used. The emission filter of the TRITC filter was changed to XF3022 (580DF30; Omega Optical) for Alexa Fluor 555 to avoid signals of chloroplast autofluorescence. To analyze the length of intercellular bridges, cells were fixed with 1% glutaraldehyde and imaged under the BZ-X700 microscope using the × 100 objective. The length was measured using ImageJ software (Schneider et al., 2012). For Figure 4D, cells were fixed with 1% glutaraldehyde and stained with 1 µg/ml DAPI. Images were obtained under a fluorescence microscope (BX51; Olympus) with a × 40 objective and CCD camera (C7780, Hamamatsu Photonics). The following filter sets were used: U-MWU2 (Olympus) for DAPI and U-MWIG2 (Olympus) for chloroplast autofluorescence. Heat maps of the signal intensities were generated in Image Lab software (Bio-Rad). All images were adjusted for contrast using Photoshop software (Adobe Systems).

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on reasonable request to the corresponding author.

AUTHOR CONTRIBUTIONS
FY and TF formulated the concept, designed the study, performed the experiments, analyzed and interpreted the data, and drafted the manuscript. TT, YK, and NS performed the experiments, analyzed and interpreted the data, and drafted the manuscript. NS performed the experiments and interpreted the data. SN, YI, OM, and KT designed the study and interpreted the data. SM, HK, and TK contributed to the concept, designed the study, interpreted the data, and drafted the manuscript.