Evolutionary Linkage Between Eukaryotic Cytokinesis and Chloroplast Division by Dynamin Proteins

Evolutionary linkage between eukaryotic cytokinesis and chloroplast division by dynamin proteins

Shin-ya Miyagishima*†, Hidekazu Kuwayama‡, Hideko Urushihara‡, and Hiromitsu Nakanishi*

*Initiative Research Program, Advanced Science Institute, RIKEN, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan; and ‡Graduate School of Life and Environmental Sciences, University of Tsukuba, 1-1-1 Tennoudai, Tsukuba, Ibaraki 305-8572, Japan

Edited by Lynn Margulis, University of Massachusetts, Amherst, MA, and approved August 12, 2008 (received for review March 11, 2008) Chloroplasts have evolved from a cyanobacterial endosymbiont activity, such as transport vesicle budding, organelle division, and been retained for more than 1 billion years by coordinated cytokinesis, and pathogen resistance (10). In some cases, two or chloroplast division in multiplying eukaryotic cells. Chloroplast more functions have been assigned to the same protein (10). division is performed by ring structures at the division site, en- Among the dynamin family, chloroplast division proteins spe- compassing both the inside and the outside of the two envelopes. cifically localize at the chloroplast division site (7–9), and A part of the division machinery is derived from the cyanobacterial mutations specifically inhibit chloroplast division (8, 11), sug- cytokinetic activity based on the FtsZ protein. In contrast, other gesting that the proteins function exclusively in chloroplast parts of the division machinery involve proteins speciﬁc to eu- division in extant plants and algae. However, given that the karyotes, including a member of the dynamin family. Each member dynamin family already existed in eukaryotes before the emer- of the dynamin family is involved in the division or fusion of a gence of chloroplasts (4, 5, 10), the dynamin proteins involved in distinct eukaryotic membrane system. To gain insight into the kind chloroplast division probably are derived from those involved in of ancestral dynamin protein and eukaryotic membrane activity eukaryotic membrane systems other than that in the chloroplast. that evolved to regulate chloroplast division, we investigated the Understanding the eukaryotic membrane fission/fusion machin- functions of the dynamin proteins that are most closely related to ery, which has evolved into the division mechanism of organelles, chloroplast division proteins. These proteins in the amoeba Dic- should provide important insights into the question of how host tyostelium discoideum and Arabidopsis thaliana localize at the cells have regulated the division of endosymbionts to establish a sites of cell division, where they are involved in cytokinesis. Our permanent endosymbiotic relationship. It has been suggested results suggest that the dynamin for chloroplast division is derived that synchronization of the host–endosymbiont cell cycle and from that involved in eukaryotic cytokinesis. Therefore, the chlo- cosegregation are critical steps (12, 13), but it is not known how roplast division machinery is a mixture of bacterial and eukary- the synchronization was established in ancestral algae. otic cytokinesis components, with the latter a key factor in the In this study, we found that previously uncharacterized mem- synchronization of endosymbiotic cell division with host cell divi- bers of the dynamin family in plants and nonphotosynthetic sion, thus helping to establish the permanent endosymbiotic protists share a common ancestor with the plant-specific chlo- relationship. roplast division dynamin proteins. Our results show that these proteins of amoeba and plants are involved in eukaryotic cytokinesis. These results suggest that the dynamin used in chloro- t is widely believed that chloroplasts arose from a bacterial plast division is derived from that involved in eukaryotic cyto- endosymbiont related to extant cyanobacteria (1, 2). Although I kinesis. Application of cytokinetic dynamin to endosymbiont cell most of their genes have either been lost or transferred to the division may have enabled the synchronization of host– host nuclear genome, chloroplasts retain several features similar endosymbiont cell division such that each daughter cell can to cyanobacteria. Chloroplasts contain nucleoids and ribosomes, inherit an endosymbiont after cytokinesis. and they are not synthesized de novo (1, 2). Chloroplasts multiply by division, as do cyanobacteria (3). However, the chloroplast Results genome does not contain sufficient information for carrying out Phylogenetic Relationships in the Dynamin Family. To address the division, indicating that the host eukaryotic cell genome regu- questions of the type of dynamin family member that evolved lates the division of chloroplasts (3). into the chloroplast division protein and the membrane activity Chloroplast division is performed by the constriction of a in which the ancestor of the chloroplast division dynamin was division apparatus (ring) encircling the division site around the involved, we conducted phylogenetic analyses of the dynamin two envelope membranes (3–6). The division apparatus includes family. In these phylogenetic analyses, we included both previ- a plastid-dividing ring of unknown composition, FtsZ, and one ously characterized and uncharacterized members of the dy- of the dynamin family of proteins (4–6). FtsZ and its associated namin family of Plantae, Opisthokonta (fungi and animals), factors are descended from the cyanobacterial endosymbiont, Amoebozoa, Heterolobosea, Chromista (a stramenopile), and posttranslationally targeted into chloroplasts (4). In contrast, the Alveolata (ciliates). Because the dynamin family has a diverse dynamin family of GTPases is specific to eukaryotes, and the range of functions (10), it is expected that the evolutionary rates chloroplast division dynamin is recruited to the cytosolic side of vary with the specific functions. To avoid the generation of an the chloroplast division site (7–9). This suggests that the chlo- incorrect tree, which is commonly a problem with rapidly roplast division machinery is derived from both endosymbiotic evolving sequences (long-branch attraction artifacts), we con- (bacterial) and host (eukaryotic) cells. The cyanobacteria-descended components of the chloroplast division machinery evolved from the cell division machinery of Author contributions: S.-y.M., H.K., H.U., and H.N. designed research; S.-y.M., H.K., and H.N. the cyanobacterial endosymbiont (4, 5). In contrast, there is little performed research; S.-y.M. and H.K. analyzed data; and S.-y.M. wrote the paper. information about the origin of chloroplast division dynamin The authors declare no conﬂict of interest. proteins. The dynamin family of GTPase proteins self-assemble This article is a PNAS Direct Submission. into rings or spirals on the surface of eukaryotic membranes, †To whom correspondence should be addressed. E-mail: [email protected]. where they play roles in membrane fission or fusion (10). There This article contains supporting information online at www.pnas.org/cgi/content/full/ are divergencies in the dynamin family, and the function of each 0802412105/DCSupplemental. member has been assigned to a distinct eukaryotic membrane © 2008 by The National Academy of Sciences of the USA

15202–15207 ͉ PNAS ͉ September 30, 2008 ͉ vol. 105 ͉ no. 39 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0802412105 Downloaded by guest on October 2, 2021 Fig. 1. Phylogenetic relationships in the dynamin family of proteins. The tree shown is the maximum-likelihood tree constructed by the PhyML program based on alignment of 567 amino acid residues of 78 dynamin family proteins. The numbers at the nodes are local bootstrap values calculated using PhyML and ProtML analyses. Bootstrap values Ͼ80% are shown, and dashes indicate Ͻ80% support. Branch lengths are proportional to the number of amino acid substitutions, which are indicated by the scale bar below the tree. The tree includes the dynamin family of proteins of Plantae Arabidopsis thaliana (At), Oryza sativa (Os or locus ID start from Os), Chlamydomonas reinhardtii (Cr), Ostreococcus lucimarinus (Ot), Cyanidioschyzon merolae (Cm), Stramenopile Thalassiosira pseudonana (Tp), Amoebozoa Dictyostelium discoideum (Dd), Fig. 2. DlpA, DlpB, and DlpC are involved in cytokinesis in D. discoideum. (A) Heterolobosea Naegleria gruberi (Ng) Alveolata Tetrahymena thermophila The ⌬dlpA, ⌬dlpB, and ⌬dlpC mutant amoeba cells (observed using phase- (Tt), Paramecium tetraurelia (Pt), Opisthokonta Saccharomyces cerevisiae (Sc), contrast microscopy) are multinucleated (nuclei are shown with blue fluores- Schizosaccharomyces pombe (Sp), Homo sapiens (Hs), Drosophila melano- cence by DAPI staining) and larger than the wild-type (WT) cells. (B) Time gaster (Dm), and Caenorhabditis elegans (Ce). course of dlpA, dlpB, and dlpC mRNA and DlpA protein levels during spore germination and cell culture mRNA were detected by semiquantitative RT-PCR cycB (encoding G2/M-specific cyclin B). The expression level indicates the structed phylogenetic trees by maximum-likelihood methods frequency of cells in M phase. TFIIS (encoding transcription elongation factor using amino acid sequences [ProtML (14) and PhyML (15)]. IIS) was used as an internal control. The DlpA protein was detected by The phylogenetic analyses showed a distant relationship be- anti-DlpA antibodies, which specifically detect DlpA, as shown at the Left. The tween chloroplast division DRP5B proteins (grouped in green in lowermost bands of Ponceau S staining show the same amount of total Fig. 1; originally named CmDnm2 in the red alga Cyanidioschy- protein was loaded in each lane. (C) Immunofluorescence images showing zon merolae and named ARC5 in Arabidopsis thaliana; later that DlpA localizes at the cleavage furrow during cytokinesis. Green fluores- renamed DRP5B, ref. 16) and other characterized members of cence corresponding to anti-DlpA antibodies and phase-contrast images of amoeba are overlaid. Orange fluorescence corresponding to the microtubules the dynamin family (Fig. 1). However, the phylogenetic tree stained by the anti-␣-tubulin antibody, and blue fluorescence showing nuclei revealed that the chloroplast division proteins are most closely are indicators of stages of the cell cycle: I1, interphase; M, metaphase indicated related to certain uncharacterized proteins (grouped in purple in by the short spindle; A, anaphase indicated by the elongated spindle; T1, T2, Fig. 1) of plants and protists that do not have chloroplasts. These and T3, telophase indicated by the cleavage furrow; and I2, interphase just ␮ ␮ uncharacterized members are three of the five Dictyostelium after cytokinesis. (Scale bars: A,20 m; C,10 m.) PLANT BIOLOGY discoideum (Amoebozoa) dynamin proteins, named DlpA, DlpB and DlpC, and plant and algal DRP5A proteins, in agreement with previous partial annotations (10, 16). In addition, a putative The uncharacterized members of the dynamin family (grouped protein, Dnm2 of Naegleria gruberi (Heterolobosea), was in- in purple in Fig. 1) are widespread in eukaryotes, including cluded in the uncharacterized group. Using BLAST searches of plants and algae, and three other eukaryotic groups which do not databases, we also found two putative proteins similar to DRP5A have chloroplasts (Amoebozoa, Heterolobosea, and Jakobidae). in the amoeba Entamoeba histolytica (Amoebozoa, GenBank Previous phylogenetic studies suggested that Amoebozoa, Het- accession numbers XP࿝653348 and XP࿝651307, omitted from the erolobosea, and Jakobidae had diverged from an ancestor of phylogenetic analyses because these made very long branches) plant cells before the acquisition of the chloroplast (2, 17). Taken and the expressed sequence tags from Histiona aroids (Jakobi- together, these results suggest that chloroplast division dynamin dae, GenBank accession numbers EC851519 and EC850393) proteins, which are specific to plant and algae, derived from an encoding a protein similar to DRP5A. ancestor of the uncharacterized members that is commonly The monophyly of plant and algal DRP5A (Fig. 1, branch d), shared by photosynthetic and nonphotosynthetic eukaryotes. that of plant and algal chloroplast division DRP5B (Fig. 1, branch e), and that of amoeba DlpA and DlpB (Fig. 1, branch DlpA, DlpB, and DlpC Are Involved in Cytokinesis in the Amoeba D. c) were strongly supported by the bootstrap values of 100/100 discoideum. To address the functions of the dynamin proteins (PhyML/ProtML). The monophyly of plant and algal DRP5A, from which the chloroplast division proteins were derived, we amoeba DlpA, and the heterolobosean Dnm2 was also sup- examined the functions of DlpA, DlpB, and DlpC in D. discoi- ported by the bootstrap values of 82/92 (Fig. 1, branch b). Finally, deum. When the genes were disrupted by homologous recom- our phylogenetic analyses suggest the monophyly of all of the bination, ⌬dlpA, ⌬dlpB, and ⌬dlpC mutants produced cells larger above-mentioned proteins (Fig. 1, branch a) by the bootstrap than those of the wild type, and a large population of the mutant values of 100/100. These results suggest that the chloroplast cells contained more than two nuclei (in contrast to wild-type division proteins DRP5A and the uncharacterized proteins share cells containing one or two duplicated nuclei) (Fig. 2A), similar a common ancestor. to the known cytokinetic mutants of D. discoideum (18). These

Miyagishima et al. PNAS ͉ September 30, 2008 ͉ vol. 105 ͉ no. 39 ͉ 15203 Downloaded by guest on October 2, 2021 results indicate that nuclear division occurs normally but cytokinesis is defective in the mutant cells. Since single-gene muta- A D tions have an affect on cytokinesis, it is suggested that the functions of DlpA, DlpB, and DlpC are not redundant and are distinct in the activity of cytokinesis. To examine the relationship between the expression of dlpA, dlpB, and dlpC and cell division, we germinated spores of D. discoideum and compared the mRNA levels of the dlpA, dlpB, B C and dlpC and DlpA proteins during cell cycle progression after germination (Fig. 2B). After synchronous germination of the wild-type spores was induced by heat shock treatment (19), spores germinated within 8 h and amoebas in the G1 phase (19) appeared. Cells started to divide after 24 h and reached a stationary phase (G2 phase) (19) between 72 and 96 h. Like the E case of the G2/M cyclin gene (cycB), dlpA, dlpB, and dlpC gene expression begins between 12 and 24 h. Immunoblot analysis using anti-DlpA antibodies showed that the DlpA protein appeared at the onset of log-phase (24 h), reached a maximum level around mid–log-phase, and then disappeared during the stationary phase (Fig. 2B). These results suggest that the DlpA protein expresses specifically during the M phase and that the protein level is regulated by transcription and protein degradation according to the phase of the cell cycle. To examine whether DlpA is directly involved in cytokinesis, we examined the localization of DlpA throughout the cell cycle. Immunofluorescence microscopy using anti-DlpA antibodies revealed that DlpA localizes at the cleavage furrow during cytokinesis (Fig. 2C). DlpA localization was detected at the F G cleavage furrow in cells during telophase and at sites where cells had divided (Fig. 2C). The above results indicate that DlpA, and probably DlpB and DlpC, are directly involved in cytokinesis in D. discoideum.

DRP5A Is Involved in Cytokinesis in A. thaliana. DRP5A proteins in H I plants and algae were predicted to be involved in chloroplast division based on the similarity between DRP5A and chloroplast division DRP5B proteins (16). However, the above results showing the involvement of DlpA in cytokinesis in D. discoideum and the monophyly of DlpA, plant and algal DRP5A, and chloroplast division DRP5B proteins (Fig. 1, branch a) raised the possibility that DRP5A proteins are involved in cytokinesis rather than chloroplast division. To examine the function of plant DRP5A proteins, we expressed a DRP5A-GFP fusion protein using the DRP5A pro- Fig. 3. DRP5A is involved in cytokinesis in A. thaliana. (A) DRP5A-GFP moter in A. thaliana. DRP5A-GFP signals were detected in expression by the DRP5A promoter (Right) showing the presence of DRP5A in some cells of a root tip (Left, differential interference contrast image). (B)A meristematic tissues (root tips are shown in Fig. 3 A and B) and magnified image of the root tip showing patchy localization and bar-shaped meristemoid cells in the leaf epidermis (Fig. 3C), but no signal localization of DRP5A-GFP in two distinct cells and no signal in other cells. (C) was detected in other somatic cells or chloroplast division sites. DRP5A-GFP exhibits patchy signals in the meristemoid of leaf epidermis but These observations suggest that DRP5A expression is limited to exhibits no signal in other epidermal cells, including guard cells. (D) Colch- dividing cells and that DRP5A is involved in activities other than icine-mediated (ϩcol) arrest of root cells in the M phase reveals DRP5A-GFP chloroplast division. In the root tip, the fluorescent signal was expression in all cells, but S-phase arrest by aphidicolin (ϩaph) abolishes limited to a certain number of cells (Fig. 3 A and B; the two cells DRP5A-GFP expression. (E) Immunofluorescence images showing DRP5A lo- at the lower left are fluorescing, but the others are not in Fig. calization at cell plates during cytokinesis. DRP5A localization in root tip cells 3B), with this mosaic pattern being similar to the expression (Left) and a series of images showing the transition of DRP5A localization during the cell cycle (Right). Green fluorescence indicates DlpA localization by patterns of cell cycle–regulated proteins, such as cyclin B (20). the anti-DRP5A antibodies. Orange fluorescence indicates microtubules by Data from the Genevestigator site (http://www.genevestigator- the anti-␣-tubulin antibody, blue fluorescence by DAPI staining indicates .ethz.ch) show M phase–specific accumulation of the DRP5A nuclei, and differential interference contrast images are indicators of the mRNA detected by cell cycle synchronization in an A. thaliana stages of cell cycle. I, interphase; P, prophase indicated by the preprophase cultured cell line (21). When DRP5A-GFP–expressing plants band of microtubules, M, metaphase indicated by the spindle; T1, early were treated with colchicine to arrest the cell cycle at the M telophase indicated by the expanding phragmoplast; T2, midtelophase indi- phase, most of the root tip cells displayed fluorescence (Fig. 3D). cated by the expanded phragmoplast; T3, late telophase. (F) Positions of In contrast, aphidicolin treatment to arrest the cell cycle at the T-DNA insertion of drp5A-1 (No-0 background) and drp5A-2 (Col-0 back- S phase resulted in little fluorescent signal (Fig. 3D). These ground). Exons are depicted as rectangles. (G–I) Phenotypes of drp5A-2 seedlings compared with those of the wild type (Col-0) at 16°C. The same results results suggest that DRP5A protein expression is also specific to were obtained in drp5A-1 seedlings compared with No-0 wild-type plants. (G) the M phase, and the protein level is controlled, at least in part, Ten-day-old wild-type (WT) and drp5A-2 seedlings. (H and I) Shown are by the transcription level and the degradation of the proteins disarrangement of root tip cells in drp5A-2 (H) and incomplete (Lower Left, after the M phase. arrow) or disordered (Lower Right, arrows) formation of cell plates in drp5A-2 In the root tip, speckles of fluorescence were observed in some (I). (Scale bars: A, C, D, and H,20␮m; B, E, and I,10␮m.)

15204 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0802412105 Miyagishima et al. Downloaded by guest on October 2, 2021 cells, but a bar-shaped localization pattern was observed in other division of endosymbiotic organelles, mitochondria, and chlo- cells (Fig. 3B), suggesting that the localization of DRP5A roplasts (4, 5, 10). The question addressed by this study is the changes during the cell cycle. To verify whether DRP5A-GFP kind of dynamin that evolved into the dynamin proteins specific reflects the true localization of endogenous DRP5A, and to to chloroplast division, and the eukaryotic membrane activity examine the transition of DRP5A localization during the cell that evolved into the chloroplast division machinery after the cycle, we performed immunofluorescence microscopy using endosymbiosis of a cyanobacterium. anti-DRP5A antibodies. In keeping with the DRP5A-GFP fu- We have shown that chloroplast division DRP5B proteins sion protein results, the antibody results showed speckles and share a common ancestor with previously uncharacterized mem- bar-shaped structures (Fig. 3E). Simultaneous labeling of tubulin bers of the dynamin family, the function of which has been and DNA showed the transition of DRP5A localization during assigned by our analyses to cytokinesis both in the plant A. the cell cycle (Fig. 3E). Interphase cells showed no specific thaliana and the amoeba D. discoideum. The phylogenetic studies signal, whereas in the prophase, speckles of DRP5A were revealed a monophyly of the dynamin proteins involved in the detected around the nucleus. The speckles were dispersed in the chloroplast division and eukaryotic cytokinesis, and the exis- cytosol of cells in the metaphase or early telophase. In late- tence of the latter in plants and algae, in addition to protists, telophase cells, most of the speckles were detected at the cell which do not have chloroplasts. These results suggest that (i)a plate, whereas at the end of cell division, DRP5A was detected common ancestor of the chloroplast division and the cytokinetic uniformly at the cell plate. These results suggest that DRP5A dynamin proteins was involved in cytokinesis in ancestral non- plays a role in cytokinesis. photosynthetic eukaryotes, and (ii) the chloroplast division To confirm that DRP5A is required for cytokinesis rather than DRP5B proteins were derived from the cytokinetic dynamin chloroplast division, we examined two independent insertional proteins. mutant lines of DRP5A (Fig. 3F). Although the mutant lines Some members of the dynamin family possess multiple func- grew normally when seeds were germinated under conventional tions. For example, dynamin proteins in animals are involved in conditions (21°C), mutant seedlings grew more slowly than those both endocytosis and cytokinesis (10, 24). The mitochondrial of the wild type at a lower temperature (16°C) (Fig. 3G). In both division dynamin proteins in certain organisms are also involved conditions, the chloroplast size and number per cell in the dlp5A in peroxisomal division (10). Although cytokinetic DRP5A and mutants were normal, whereas the chloroplast division dlp5B chloroplast division DRP5B share a common ancestor, their mutant cells contained chloroplasts that were fewer in number completely distinct subcellular localization and nonoverlapping and larger than in the wild type (8, 9, 11). Microscopic exami- mutant phenotypes suggest that the functions of DRP5A and nation of the mutant root tips revealed perturbation of the cell DRP5B are completely distinct in the extant plants, at least in A. H array (Fig. 3 ) and formation of incomplete or twisted cell plates thaliana. in mutant cells (Fig. 3I), similar to other A. thaliana mutants of It has been suggested that the synchronization of host– cytokinetic proteins (22). The same phenotypes were observed endosymbiont cell division is a critical step in the permanent in two independent drp5A alleles, suggesting that the observed fusion of the two partners (12, 13). There are eukaryotic species effects are caused by mutations in the gene. These results suggest that contain transient photosynthetic endosymbionts called klep- that loss of DRP5A affected cell plate formation and that, rather toplasts in the cells. For example, Hatena arenicola has a than chloroplast division, DRP5A is involved in cytokinesis, at transient green algal endosymbiont, and this photosynthetic least under lower temperature conditions, which often occur in the wild. endosymbiont is inherited by only one daughter cell during cell The lack of detectable phenotypes in the drp5A mutants at division (12). Hatena arenicola (Katablepharidophyta) bears a normal temperatures may be caused by redundant or overlap- close evolutionary relationship with cryptophytes, in which ping functions provided by other proteins. The genome of A. permanent chloroplasts are established (25). A similar situation thaliana has only a single copy of the DRP5A gene, and the most has been observed in Dinoflagellata (26) and Cercozoa (27), in closely related protein of DRP5A is DRP5B. We reexamined which certain species have established the division synchroniza- GFP-DRP5B localization in A. thaliana (8, 9), and the localization, whereas others still have a transient endosymbiotic rela- PLANT BIOLOGY tion was specific to the chloroplast division site, as previously tionship with photosynthetic eukaryotes or cyanobacteria. The observed (8, 9), whereas no localization was observed around the above observations suggest that division synchronization was cell plate. Also, we reexamined the phenotypes of the drp5B necessary for the establishment of permanent chloroplasts (13). mutants (8, 9, 11), but the pattern of cell division was normal. However, how this synchronization was established has remained These results suggest that the function of DRP5B is specific to largely unknown. Our results suggest that the host cells have chloroplast division. made use of the M phase–specific cytokinetic dynamin proteins Arabidopsis thaliana contains several members of the DRP1 for the division of the endosymbiont. As a result, the division of and DRP2 cytokinetic dynamin proteins (16, 23), although they the endosymbiont is limited to the M phase, so that each are evolutionarily distantly related to DRP5A (Fig. 1). To daughter cell inherits a daughter endosymbiont. examine the potential existence of functional redundancy, we Notably, DRP5B chloroplast division dynamin is expressed overexpressed DRP5A-GFP or DRP5A (K75A), which corre- specifically during the M phase in the primitive red alga Cyan- sponds to K44A dominant-negative mutant of human dynamin idioschyzon merolae, which contains only a single chloroplast per 1. However, the overexpressers grew normally, and the cell cell, and therefore chloroplast division and cell cycle are syn- division patterns were normal (data not shown). Although at chronized in this alga (7). This is consistent with M phase– present it is unclear why the cytokinetic defects of the drp5A specific expression of DRP5A and Dlp proteins for cytokinesis mutant appear only at a lower temperature, the exclusive (Figs. 2 and 3). However, the expression of the DRP5B chloro- expression of DRP5A in meristematic cells during the M phase plast division protein in A. thaliana is apparently constant during along with the cell plate localization of DRP5A suggest a the cell cycle, based on the experimental results of cell cycle function in cell division. synchronization in a cultured cell line (ref. 21; http:// www.genevestigator.ethz.ch). In land plants, each cell contains Discussion more than one chloroplast, the division of which is not synchro- Evolutionary Linkage Between Cytokinesis and Chloroplast Division. nous (28). Chloroplasts divide also in postmeristematic cells, Every member of the dynamin family is involved in fission or which expand without cell division (28). These results suggest fusion of distinct eukaryotic membrane systems, including the that cells with multiple chloroplasts have evolved more complex

Miyagishima et al. PNAS ͉ September 30, 2008 ͉ vol. 105 ͉ no. 39 ͉ 15205 Downloaded by guest on October 2, 2021 systems to regulate the expression of the chloroplast division at the site of cytokinesis, and these groups are closely related to dynamin proteins. the mitochondrial division group. These results raise the possibility that mitochondrial division dynamin proteins have evolved Evolution of the Dynamin Family. To date, certain members of the from the cytokinetic machinery of the host cell, as have chlo- dynamin family have been shown to be involved in cytokinesis roplast division proteins. DymA of D. discoideum (30) may (23, 24, 29, 30). In the phylogenetic tree, these proteins (mem- represent an intermediate stage of evolution from the cytoki- bers separately grouped in yellow and gray, DymA of D. netic dynamin to that in endosymbiotic organelles. Investigation discoideum in the mitochondrial division clade in Fig. 1), have a into the evolution of mitochondrial division dynamin proteins relatively close relationship (Fig. 1). The involvement of the may elaborate the story of the chloroplast division dynamin amoeba DlpA (probably DlpB and possibly DlpC also) and plant protein to include another endosymbiotic organelle, mitochon- DRP5A (grouped in purple in Fig. 1) in cytokinesis extends the drion. cytokinetic function of the dynamin family throughout the phylogenetic tree, whereas members involved in other activities Materials and Methods are grouped by their respective functions (Fig. 1). These results Phylogenetic Analyses. The maximum-likelihood tree was constructed with the suggest that the original evolutionary function of dynamin PhyML program (15) using an alignment of amino acid sequences of 78 proteins may have been involved in the cytokinesis of eukaryotic dynamin-like proteins [supporting information (SI) Table S1]. The sequences cells. were collected by BLAST searches in the databases of respective species, Plants and amoebae have several dynamin family members for National Center for Biotechnology Information, the US Department of Energy cytokinesis (Fig. 1). Although the functions of some of these Joint Genome Institute (http://genome.jgi-psf.org/) using DRP5B and DRP3 of members may be partially redundant, single mutants of dlpA, the red alga Cyanidioschyzon merolae as queries. The region of alignment dlpB, dlpC, and dlp5A display defects in cytokinesis, suggesting contained the GTPase domain, middle domain, and GED domain (10) of the sequences. The local bootstrap probabilities were calculated using the PhyML that the function of each of these proteins is not completely (15) and ProtML (14) programs. redundant with those of the others. Moreover, at least in the amoeba D. discoideum (DymA and the others grouped in purple) Cell Cultures and Plant Materials. Wild-type D. discoideum (AX2) and the and plants and algae (grouped in gray and purple), there are mutants derived from AX2 were cultured as previously described (33, 34). The evolutionarily distant cytokinetic dynamin proteins (Fig. 1). dlpA, dlpB, and dlpC genes were disrupted by insertion of the blasticidin These facts suggest that at least two major groups of cytokinetic S-resistant gene via homologous recombination (33). Spores were germinated dynamins have already evolved to play different roles in the by heat shock as described previously (19), and DNase-treated total RNA was cytokinesis of ancestral eukaryotes. used for RT-PCR analyses. Wild-type A. thaliana, its transgenic lines, and Like chloroplasts, mitochondria evolved from an ␣-proteobac- homozygous drp5A T-DNA insertional mutants were grown as previously terial endosymbiont, and members of the dynamin family of described (9), except where indicated. All plants were of Col-0 background, proteins are involved in mitochondrial division (refs. 4, 5, and 10; except for the phenotypic comparison between drp5A-1 and wild-type (Nos-0 background). The homozygous T-DNA insertion lines drp5A-1 (SALK࿝065118) and grouped in orange in Fig. 1). In addition, several lineages of ࿝ eukaryotes still use the FtsZ descended from the ␣-proteobac- and drp5A-2 (RATM12-3406-1 H) were provided by the Arabidopsis Biological Resource Center (ABRC) (35) and the RIKEN BioResource Center (36), respec- terial endosymbiont to regulate mitochondrial division (31). The tively. mitochondrion-dividing ring, which has a structure similar to the plastid-dividing (PD) ring, has been observed in some eukaryotic Immunoblotting and Immunofluorescence. The 6ϫ His-tag fusion of partial species (3, 5). These results suggest that the host cell used a polypeptides of DlpA (511 residues) and DRP5A (445 residues) expressed in strategy to regulate the division of the cyanobacterial endosym- Escherichia coli was used to generate polyclonal antibodies in rabbits. Immu- biont similar to the one it used for the ␣-proteobacterial noblotting was performed using total proteins from amoeba cells, and im- endosymbiont. Disruption of the dynamin gene dymA in D. munoﬂuorescence microscopy was performed on log-phase amoeba cells and discoideum blocks both cytokinesis and mitochondrial division root tips. The full experimental procedures and associated references are (30). Similar results were also reported in a dynamin mutant of available in the SI Materials and Methods. Trypanosoma brucei (32), although it is not known whether these dynamin proteins localize at the cleavage furrow. However, ACKNOWLEDGMENTS. We thank C. Saito, T. Mori, and T. Kuroiwa for their help in microscopic analyses and for useful discussions, and we thank Y. Ono animal dynamin proteins (refs. 24 and 29; and grouped in yellow for technical support. We also thank the RIKEN BioResource Center and the in Fig. 1) and plant-specific cytokinetic dynamin proteins (refs. Arabidopsis Biological Resource Center for providing seeds of drp5A-1 and 23 and 29; and grouped in gray in Fig. 1) were shown to localize drp5A-2. This work was supported by the Sumitomo Foundation (S.-y.M.)

1. McFadden GI (1999) Endosymbiosis and evolution of the plant cell. Curr Opin Plant Biol 11. Pyke KA, Leech RM (1994) A genetic analysis of chloroplast division and expansion in 2:513–519. Arabidopsis thaliana. Plant Physiol 104:201–207. 2. Reyes-Prieto A, Weber AP, Bhattacharya D (2007) The origin and establishment of the 12. Okamoto N, Inouye I (2005) A secondary symbiosis in progress? Science 310:287. plastid in algae and plants. Annu Rev Genet 41:147–168. 13. Rodriguez-Ezpeleta N, Philippe H (2006) Plastid origin: Replaying the tape. Curr Biol 3. Kuroiwa T, et al. (1998) The division apparatus of plastids and mitochondria. Int Rev 16:R53–R56. Cytol 181:1–41. 14. Adachi J, Hasegawa M (1996) MOLPHY version 23: Programs for molecular phyloge- 4. Osteryoung KW, Nunnari J (2003) The division of endosymbiotic organelles. Science netics based on maximum likelihood. Comput Sci Monogr 28:1–150. 302:1698–1704. 15. Guindon S, Gascuel O (2003) A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst Biol 52:696–704. 5. Miyagishima S, Nishida K, Kuroiwa T (2003) An evolutionary puzzle: Chloroplast and 16. Hong Z, et al. (2003) A unified nomenclature for Arabidopsis dynamin-related large mitochondrial division rings. Trends Plant Sci 8:432–438. GTPases based on homology and possible functions. Plant Mol Biol 53:261–265. 6. Yoshida Y, et al. (2006) Isolated chloroplast division machinery can actively constrict 17. Rodríguez-Ezpeleta N, et al. (2007) Toward resolving the eukaryotic tree: The phylo- after stretching. Science 313:1435–1438. genetic positions of jakobids and cercozoans. Curr Biol 17:1420–1425. 7. Miyagishima S, et al. (2003) A plant-specific dynamin-related protein forms a ring at the 18. Yumura S, Uyeda TQ (2003) Myosins and cell dynamics in cellular slime molds. Int Rev chloroplast division site. Plant Cell 15:655–665. Cytol 224:173–225. 8. Gao H, Kadirjan-Kalbach D, Froehlich JE, Osteryoung KW (2003) ARC5, a cytosolic 19. Chen G, Kuspa A (2005) Prespore cell fate bias in G1 phase of the cell cycle in dynamin-like protein from plants, is part of the chloroplast division machinery. Proc Dictyostelium discoideum. Eukaryot Cell 4:1755–1764. Natl Acad Sci USA 100:4328–4333. 20. Coloń-Carmona A, You R, Haimovitch-Gal T, Doerner P (1999) Spatio-temporal analysis 9. Miyagishima S, Froehlich JE, Osteryoung KW (2006) PDV1 and PDV2 mediate recruit- of mitotic activity with a labile cyclin-GUS fusion protein. Plant J 20:503–508. ment of the dynamin-related protein ARC5 to the plastid division site. Plant Cell 21. Menges M, Hennig L, Gruissem W, Murray JA (2003) Genome-wide gene expression in 18:2517–2530. an Arabidopsis cell suspension. Plant Mol Biol 53:423–442. 10. Praefcke GJ, McMahon HT (2004) The dynamin superfamily: Universal membrane 22. So¨llner R, et al. (2002) Cytokinesis-defective mutants of Arabidopsis. Plant Physiol tubulation and fission molecules? Nat Rev Mol Cell Biol 5:133–147. 129:678–690.

15206 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0802412105 Miyagishima et al. Downloaded by guest on October 2, 2021 23. Hong Z, Geisler-Lee CJ, Zhang Z, Verma DP (2003) Phragmoplastin dynamics: Multiple 30. Wienke DC, Knetsch ML, Neuhaus EM, Reedy MC, Manstein DJ (1999) Disruption of a forms, microtubule association and their roles in cell plate formation in plants. Plant dynamin homologue affects endocytosis, organelle morphology, and cytokinesis in Mol Biol 53:297–312. Dictyostelium discoideum. Mol Biol Cell 10:225–243. 24. Thompson HM, Skop AR, Euteneuer U, Meyer BJ, McNiven MA (2002) The large GTPase 31. Kiefel BR, Gilson PR, Beech PL (2006) Cell biology of mitochondrial dynamics. Int Rev dynamin associates with the spindle midzone and is required for cytokinesis. Curr Biol Cytol 254:151–213. 12:2111–2117. 32. Chanez AL, Hehl AB, Engstler M, Schneider A (2006) Ablation of the single dynamin of 25. Okamoto N, Inouye I (2005) The katablepharids are a distant sister group of the T brucei blocks mitochondrial fission and endocytosis and leads to a precise cytokinesis Cryptophyta: A proposal for Katablepharidophyta divisio nova/Kathablepharida arrest. J Cell Sci 119:2968–2974. phylum novum based on SSU rDNA and beta-tubulin phylogeny. Protist 156:163– 33. Kuwayama H, et al. (2002) PCR-mediated generation of a gene disruption construct 179. without the use of DNA ligase and plasmid vectors. Nucleic Acids Res 30:E2. 26. Hackett JD, Anderson DM, Erdner DL, Bhattacharya D (2004) Dinoflagellates: A re- markable evolutionary experiment. Am J Bot 91:1523–1534. 34. Urushihara H (2006) Cultivation, spore production, and mating. Methods Mol Biol 27. Marin B, Nowack EC, Melkonian M (2005) A plastid in the making: Evidence for a second 346:113–124. primary endosymbiosis. Protist 156:425–432. 35. Alonso JM, et al. (2003) Genome-wide insertional mutagenesis of Arabidopsis thaliana. 28. Pyke KA (1999) Plastid division and development. Plant Cell 11:549–556. Science 301:653–657. 29. Konopka CA, Schleede JB, Skop AR, Bednarek SY (2006) Dynamin and cytokinesis. 36. Kuromori T, et al. (2004) A collection of 11800 single-copy Ds transposon insertion lines Traffic 7:239–247. in Arabidopsis. Plant J 37:897–905. PLANT BIOLOGY

Miyagishima et al. PNAS ͉ September 30, 2008 ͉ vol. 105 ͉ no. 39 ͉ 15207 Downloaded by guest on October 2, 2021