Report

Equatorial Retention of the Contractile Actin Ring by Microtubules During Cytokinesis

See allHide authors and affiliations

Science  06 Jun 2003:
Vol. 300, Issue 5625, pp. 1569-1574
DOI: 10.1126/science.1084671

Abstract

In most eukaryotes cytokinesis is brought about by a contractile actin ring located at the division plane. Here, in fission yeast the actin ring was found to be required to generate late-mitotic microtubular structures located at the division plane, and these in turn maintained the medial position of the actin ring. When these microtubular structures were disrupted, the actin ring migrated away from the cell middle in a membrane traffic–dependent manner, resulting in asymmetrical cell divisions that led to genomic instability. We propose that these microtubular structures contribute to a checkpoint control that retains the equatorial position of the ring when progression through cytokinesis is delayed.

Cytokinesis separates the two nuclei formed at mitosis into two new cells. In many eukaryotes this is brought about by a contractile actin ring located at the division plane (13). Microtubular structures termed the “spindle midzone” and “midbody” in metazoan cells are also found at the division plane between the separating sets of chromosomes during late anaphase and telophase (2). In fission yeast the actin ring forms in the cell middle at the onset of mitosis and contracts after anaphase as a division septum is deposited (4). The contractile ring contains other conserved proteins including the myosin II heavy (Myo2) and light (Cdc4 and Rlc1) chains, the formin Cdc12, and the IQGAP homolog Rng2 (3). The site of cell division is specified by the anillin homolog Mid1p (5, 6), which couples the position of the contractile ring to that of the premitotic nucleus (7). The septation initiation network (SIN) regulates actin ring constriction and septation and includes the protein kinases Cdc7, Sid1, and Sid2; the cdc14 and mob1 gene products; and Cdc11 and Sid4, which anchor these components to the spindle pole body (3, 8). In sin mutant cells an actin ring is formed, but this fails to contract and then disassembles at the end of mitosis. At the end of anaphase, cytoplasmic microtubular structures form in the division plane between the separated nuclei (9). Initially these include an equatorial microtubule organizing center (EMTOC), a ringlike structure containing the γ-tubulin complex (911). As the spindle disassembles, this tubulin ring nucleates the aster-like postanaphase array (PAA) of microtubules (9). The SIN is required for the formation of the γ-tubulin EMTOC (11), suggesting that the PAA might be important for cytokinesis. We investigate here the relation between the actin ring and the division-plane microtubular structures in regulating cytokinesis.

Microtubules were monitored in live temperature-sensitive (ts) SIN-defective cdc11-119 fission yeast cells expressing green fluorescent protein (GFP)–α-tubulin (GFPAtb2) (12) and the nuclear marker Cut11-GFP (13, 14). At the permissive temperature, normal PAAs were observed after spindle elongation (Fig. 1A, 25°C, 8 to 28 min). These were followed by an interphase array of microtubules, and movement of nuclei back to the center of the newly formed cells (38 to 48 min). In contrast, at the restrictive temperature, no PAAs were observed (Fig. 1A, 36°C). After mitosis microtubules appeared near the nuclei (1 to 4 min), and within 5 min of spindle disassembly, a normal interphase array had formed (11 to 23 min), establishing that PAAs are not required to generate interphase arrays. Nuclei quickly moved from the cell ends to the cell middle (4 to 23 min). The sin mutants cdc7-24 and cdc14-118 displayed similar behavior (15). Thus, the SIN is required to form PAAs, in agreement with the requirement of SIN for the formation of the EMTOC (11).

Fig. 1.

PAA formation depends on SIN but not on cytokinesis completion. (A) Time-lapse confocal microscopy of live cdc11-119 cells expressing GFP-tubulin and Cut11-GFP as a nuclear marker at 25° and 36°C. (B) Time-lapse confocal microscopy of live cps1-191 GFP-atb2 cells at 36°C. The behavior of cells incubated at 25°C resembled that of the wild type (15). Numbers in (A) and (B) indicate time elapsed (in min) since the first frame. (C) Actin, tubulin, and nuclei staining of cps1-191 cells after 4 hours at 36°C. Arrested cps1-191 cells primarily have PAA-like structures. Except for actin, the images shown are projections of a series of confocal sections spanning the whole cell. Scale bars (A to C), 10 μm.

To examine whether the presence of a PAA and an actin ring are correlated, we monitored ts cps1-191 mutant cells defective in a 1,3-β-glucan synthase essential for division septum assembly, which arrest with a stable actin ring (Fig. 1C) (16). In live cps1-191 GFP-atb2 cells, PAAs were observed (Fig. 1B), and in the majority of fixed, arrested cps1-191 cells, only PAAs were found (Fig. 1C), establishing that a persistent PAA coexists with a stable actin ring.

We then tested whether PAA formation is dependent on the actin ring, given that maintenance of the EMTOC depends on F-actin (11). Cold-sensitive nda3-KM311 cells defective in the β-tubulin subunit (17) arrest in mitosis with no spindle but with an actin ring (18). These arrested cells were treated with latrunculin A (LatA), an actin polymerization inhibitor (19), and released into mitosis in the continuing presence of drug. The LatA-treated cells were slightly delayed in mitotic progression compared with control cells (Fig. 2A), probably owing to the spindle orientation checkpoint (20). In control cells, the actin rings remained for 30 min and disappeared as septation took place and PAAs appeared (Fig. 2, B to D). However, in LatA-treated cells, no actin rings, septa, or PAAs were observed (Fig. 2, B to D), and 30 min after release, interphase microtubular arrays were observed (Fig. 2D). Thus, the formation of PAAs is dependent on either the actin ring or another F-actin structure.

Fig. 2.

The actin ring is required for PAA formation. Nda3-KM311 cells were arrested at 18°C; LatA was added, and cells were then released into mitosis by shifting to 32°C in LatA or dimethylsulfoxide (DMSO) (control). (A) Nuclear division of released nda3-KM311 cells. (B) Graph showing the number of nda3-KM311 cells with an actin ring (AR) and a septum during release into mitosis. The arrowhead indicates the time of LatA addition, and time 0 marks the shift to a permissive temperature. (C) Mitotic progression and PAA formation in released nda3-KM311 cells. (D) LatA-treated and control nda3-KM311 cells were fixed 15 and 30 min after release into mitosis, and stained with antibodies to α-tubulin. Within 15 min of the shift, 72% of control cells with a long spindle had early PAAs, whereas only 2% of LatA-treated cells had these structures. Thirty minutes after release, PAAs were still observed in control cells but not in LatA-treated cells, which already had interphase arrays. (E) Time-lapse confocal microscopy of live cdc12-112 GFP-atb2 cells at 25°C. (F) Time-lapse confocal microscopy of live cdc12-112 GFP-atb2 cells at 36°C. Numbers in (E) and (F) indicate time elapsed (in min) since the first frame. Scale bars (D to F), 10 μm.

To distinguish between these possibilities, we examined microtubules in the cytokinesis ts mutant cdc12-112, which has normal actin patches during interphase and actin cables at the medial region during mitosis, but fails to form the actin ring (21). At the permissive temperature, cdc12-112 GFP-atb2 cells displayed normal PAAs (Fig. 2E, 9 to 27 min). In contrast, at the restrictive temperature, no PAAs were observed (Fig. 2F). Instead, after spindle disassembly, astral microtubules elongated (Fig. 2F, 4 to 6 min), and later microtubules nucleated near the nuclei (6 to 9 min), reestablishing an interphase array (10 to 30 min). These results establish that PAA formation is dependent on the actin ring.

Next we examined whether actin ring stability is dependent on the PAA by treating arrested cps1-191 cells with the microtubule-depolymerizing drug carbendazim (MBC) (Fig. 3B). After MBC treatment, most cells still had an actin ring (Fig. 3A), but unexpectedly the ring was dramatically displaced in 30 to 40% of cells. Myo2, Cdc4 (16), Cdc12, Rlc1, and Rng2 remained associated with the stable ring in arrested cps1-191 cells (Fig. 3C) and were present in the displaced actin ring (Fig. 3, C and D). Thus, the PAA is required to maintain the contractile actin ring and associated components in the middle of the cell. Mid1-GFP was found to be mainly localized in the nuclei of arrested cps1-191 cells, with 30% of cells also displaying a faint Mid1-GFP ring at the cell equator (Fig. 3E, Pre). Upon MBC treatment, Mid1-GFP rings were offset and colocalized with the displaced actin ring (Fig. 3, E and F). Only 5% of rings were positioned overlying one of the interphase nuclei (Fig. 3G, cell 3), with 25% of rings placed between a nucleus and a cell tip (Fig. 3G, cell 4) and 70% of rings placed between two nuclei (Fig. 3G, cells 1 and 2). Thus, nuclei do not dictate the displaced position of the actin ring in these cells.

Fig. 3.

PAAs are required to keep the contractile ring in the cell middle. cps1-191 cells were arrested at 36°C for 3 hours and then treated with MBC or DMSO (control) for 1 hour. (A) Actin staining in MBC-treated cps1-191 cells. Arrowheads depict cells in which the actin ring has migrated well away from the cell center. (B) Microtubules and DNA staining in control and MBC-treated cps1-191 cells. (C) Other medial ring components in MBC-treated cps1-191 cells. Cdc4, Cdc12, Myo2, and Rng2 were detected by antibody staining, and Rlc1 by visualization of Rlc1-GFP. (D) Colocalization of components in the displaced ring in MBC-treated cps1-191 cells. (E) Mid1-GFP was stained with antibodies to GFP in arrested cps1-191 cells, before (Pre) and after MBC treament (DMSO, MBC). (F) Colocalization of Mid1-GFP and actin in the displaced ring. (G) Position of the actin ring relative to the daughter nuclei in MBC-treated cps1-191 cells. Scale bars (A to G), 10 μm.

The ring component Rlc1-GFP was used to monitor the ring in live cells. In 73% of arrested MBC-treated cps1-191 cells (36/49), the Rlc1-GFP ring moved in a continuous and directed manner from a central position to one cell end (Fig. 4A, Movie S1). The rings mostly migrated intact at right angles to the long axis of the cell, although 30% of rings (13/36) showed an Rlc1-GFP strand separating from the ring (Movie S2). Most rings contracted as they approached the cell end and finally collapsed into an Rlc1-GFP dot (Fig. 4A). Because the ring is associated with the membrane cortex throughout cytokinesis (2), we investigated whether ring movement involved membrane trafficking. Filipin stains sterol-rich rafts (22, 23) and marks central cortical membranous regions of dividing cells (24). In cps1-191 cells undergoing ring displacement, bright filipin-stained membranous regions also became displaced, colocalizing with the contractile ring (Fig. 4B). Furthermore, treatment with brefeldin A, an inhibitor of endoplasmic reticulum–to–Golgi transport (25), completely inhibited actin ring displacement (Fig. 4C). Thus, membrane trafficking seems to be required for contractile ring migration.

Fig. 4.

The contractile ring moves away from the cell middle in the absence of microtubules. (A) Time-lapse confocal microscopy of live, MBC-treated cps1-191 rlc1-GFP cells at 36°C. Numbers indicate time elapsed (in min) since the first frame. (B) Filipin staining of cps1-191 rlc1-GFP cells before (Pre) and after MBC treatment (DMSO, MBC). (C) Ring displacement is dependent on membrane trafficking. Brefeldin A was added 30 min before MBC addition. The graph shows the percentage of cells with a displaced ring in the presence or absence of brefeldin A. (D) Upon removal of MBC, there is little repositioning of the actin ring back to the cell center. The graph shows the number of MBC-treated cps1-191 cells with a clearly displaced actin ring. Cells were treated with MBC or DMSO, and then a sample of the culture was washed (MBC–) and another kept in MBC (MBC+). The position of the actin ring was scored 1 (t1) and 2 (t2) hours after removal of MBC (t0). (E) One hour after removal of MBC, upon microtubule repolymerization, MBC-treated cps1-191 cells have offset PAAs (arrowheads), and nuclei have migrated back to the cell center (arrows). (F) After shifting MBC-treated cps1-191 cells to the permissive temperature, displaced actin rings directed the formation of offset septa (CFW, arrowheads), resulting in the appearance of anucleate and binucleate cells (DNA, arrowheads). Scale bars (A, B, E, F), 10 μm.

Removal of MBC from cells with offset actin rings left the rings displaced (Fig. 4D), and the newly formed PAAs were also offcenter (Fig. 4E, arrowheads), demonstrating that once re-formed, the PAA cannot reposition itself or the actin ring back to the cell center. After removal of MBC, the two nuclei moved together to the cell middle (Fig. 4E, arrows), suggesting that the PAA helps to keep the nuclei apart at cell ends during cytokinesis. On shifting MBC-treated cps1-191 cells back to 25°C, septation frequently took place off-center, generating anucleate and binucleate cells (Fig. 4F, arrowheads). Thus, the PAA helps to immobilize the contractile ring when cytokinesis is delayed.

At the end of mitosis the actin ring brings about the formation of the PAA, perhaps by providing a scaffold for γ-tubulin (11). This microtubular structure immobilizes the contractile ring at the cell equator and contributes to maintaining nuclear separation in the event of a cytokinesis delay, as seen in cps1-191 cells. If the PAA is defective, the contractile ring and its associated sterol-rich cortical membrane compartment migrate away from the cell middle in a membrane traffic–dependent manner. Because the actin ring is highly dynamic (26), it must be constantly re-formed as the structure migrates along the cell. We propose that actin ring immobilization by PAA microtubules is part of a checkpoint control that is triggered when cytokinesis is delayed (27, 28). This control maintains the ring in a medial position while cytokinesis is delayed, ensuring that when septation eventually takes place, it will occur between the two separated nuclei. PAA microtubules might immobilize the actin ring by anchoring it within the cell cortex (29), or might have regulatory roles in either directing membrane traffic to the cleavage plane (30) or preventing cortical membrane reorganization until cytokinesis is completed. Interactions between the actin ring and the spindle midzone and midbody have also been observed in metazoan eukaryotes (2, 31, 32), and microtubules are continuously required to maintain the position of the cleavage furrow and keep nuclei apart during cytokinesis in mammalian cells (33), suggesting that a similar mechanism to the one we have described in fission yeast may operate in other eukaryotes.

Supporting Online Material

www.sciencemag.org/cgi/content/full/300/5625/1569/DC1

Materials and Methods

References and Notes

Movies S1 and S2

References and Notes

View Abstract

Navigate This Article