Dissecting Temporal and Spatial Control of Cytokinesis with a Myosin II Inhibitor

See allHide authors and affiliations

Science  14 Mar 2003:
Vol. 299, Issue 5613, pp. 1743-1747
DOI: 10.1126/science.1081412


Completion of cell division during cytokinesis requires temporally and spatially regulated communication from the microtubule cytoskeleton to the actin cytoskeleton and the cell membrane. We identified a specific inhibitor of nonmuscle myosin II, blebbistatin, that inhibited contraction of the cleavage furrow without disrupting mitosis or contractile ring assembly. Using blebbistatin and other drugs, we showed that exit from the cytokinetic phase of the cell cycle depends on ubiquitin-mediated proteolysis. Continuous signals from microtubules are required to maintain the position of the cleavage furrow, and these signals control the localization of myosin II independently of other furrow components.

Cytokinesis is the process by which daughter cells are separated from one another at the end of cell division. During cytokinesis, the cell cycle, cytoskeleton, and membrane systems of the cell undergo a linked, closely coordinated series of changes on a time scale of minutes (1,2). Two outstanding questions dominate the field of cytokinesis: What timing mechanisms control progress through cytokinesis, and how is the cleavage furrow positioned to divide the cell into two equal parts? The cleavage furrow is positioned on a plane that bisects the axis of chromosome segregation by microtubules associated with the mitotic apparatus (3), but the signals used to communicate between the microtubules and the furrow are largely unknown. Pathways that control the timing of progression through cytokinesis to the G1 phase of the cell cycle have been identified with genetic studies in yeast (4, 5). Components of these pathways are conserved, but their relevance to mammalian cytokinesis is unclear.

We took a small molecule–based approach to dissect the logic of cytokinesis in mammalian cells. Previous small-molecule work using microtubule or actin depolymerizers (colchicine, nocodazole, or cytochalasin) has demonstrated the importance of microtubules for both positioning (3) and ingression (6) of the cytokinesis furrow, and of actin for the structure of the furrow (7). Actin and microtubule inhibition have also revealed the existence of an ∼1-hour window after anaphase during which the cell is competent to undergo cytokinesis, called C phase (8, 9). To investigate further, we wanted a small molecule that would arrest furrow ingression, but not perturb furrow assembly. We therefore targeted nonmuscle myosin II, which provides the force for furrow ingression (10, 11), using a high-throughput screening assay (12, 13). Blebbistatin (Fig. 1A, fig. S1) inhibited both the adenosine triphosphatase (ATPase) and gliding motility activities of human platelet nonmuscle myosin II without inhibiting myosin light chain kinase (fig. S2). We separated the enantiomers (13) and found that (–)-blebbistatin is the active species, with an inhibitory concentration for 50% inhibition (IC50) of ∼2 μM, whereas (+)-blebbistatin was inactive (Fig. 1A). Rabbit skeletal and human nonmuscle myosin II were inhibited by blebbistatin, but human myosins Ib, Va, and X were not (Fig. 1B).

Figure 1

(A) Effect on human platelet myosin II S1 Mg · ATPase by active (–) enantiomer (squares) and inactive (+) enantiomer (circles) of blebbistatin. (Inset) The structure of (±)-blebbistatin. (B) Effects of (±)-blebbistatin on the actin-activated ATPase rate of muscle and nonmuscle myosins. Myo X-HMM, myosin X-heavy meromyosin; Myo Va-S1, myosin Va-subfragment 1; Myo 1b, myosin 1b; Sm-HMM, smooth muscle myosin II-HMM, NMIIB-HMM, nonmuscle myosin IIB-HMM; NMII-HMM, nonmuscle myosin II-HMM; Sk-HMM, skeletal muscle myosin II-HMM. (C) Cytokinetic Xenopus tissue culture cells with or without 100 μM (±)-blebbistatin. (D) Comparison of (+) (circles) versus (–) (squares) enantiomers of blebbistatin on binucleate cell formation in HeLa cells. (E) HeLa cells untreated (control) or treated (blebbistatin) with 100 μM (±)-blebbistatin and stained for DNA (blue), α-tubulin (green), and nonmuscle myosin II (red).

Blebbistatin blocked myosin II–dependent cell processes. The compound's name is derived from its ability to block cell blebbing rapidly (within ∼2 min) and reversibly (fig. S3, movie S1). It also rapidly disrupted directed cell migration (fig. S3, movie S2) and cytokinesis in vertebrate cells (Fig. 1C). Blebbistatin did not affectDrosophila melanogaster cells. Inhibition was specific for the (–) isomer, consistent with myosin II being the relevant drug target (Fig. 1D). Addition of blebbistatin to dividing cells blocked furrow ingression within 5 min and was reversible (movies S3 and S4). Blebbistatin did not block assembly of either the furrow or the microtubules that position it. In arrested cells, two major furrow components, myosin II and anillin (14, 15), were localized to an equatorial ring, and microtubules were organized into two prominent cytokinesis-associated arrays: a bipolar midzone complex between the separated chromosomes and asters radiating from the centrosomes to the furrow (Fig. 1E). This organization resembled that seen in control cells, except for the lack of furrow contraction (Fig. 1E). Thus, blebbistatin can be used to probe the spatial organization of the cytokinesis machinery in the absence of furrow contraction.

We examined the timing of mitotic exit in blebbistatin. Staining for DNA and nuclear lamins showed that chromosome decondensation and nuclear envelope reformation occurred in the presence of blebbistatin (Fig. 2A), and time-lapse microscopy demonstrated that the timing of these events was unaffected by the drug (fig. S4) (13). Blebbistatin-treated HeLa cells initiated DNA replication ∼12 hours after release from mitosis, like untreated cells; HeLa cells lack the checkpoint that prevents entry into S phase if mitosis fails (16, 17). Thus, blocking furrow contraction does not in itself prevent exit from mitosis, although clearly these events must be coordinated in normal cells.

Figure 2

(A) Mitotically synchronized HeLa cells released into medium without (Control) or with (Blebbistatin) 100 μM (±)-blebbistatin. Cells in different stages after anaphase were stained for DNA, α-tubulin, lamins, and nonmuscle myosin II. (B) Quantitation of the presence of localized myosin in binucleate cells treated without (circles) or with (squares) MG132 in the presence of blebbistatin (n = 3 independent experiments; bars represent standard error). Mitotically synchronized HeLa cells were released into blebbistatin for 80 min then treated without or with 100 μM MG132. Cells were fixed at 15 min intervals after release from mitosis and stained for DNA, α-tubulin, nonmuscle myosin II, and anillin. (C) Cells treated as described in (B) and fixed at 170 min after release from metaphase.

We next investigated the timing of C phase. C-phase initiation is triggered by anaphase onset and requires the anaphase-promoting complex (APC) and ubiquitin-mediated proteolysis (18). We measured the duration of C phase in the presence of blebbistatin by allowing cells to initiate anaphase in the presence of blebbistatin, then removing the drug at increasing intervals and monitoring furrowing. Unsynchronized cells remained competent to contract for 45 to 60 min after anaphase onset as determined by time-lapse imaging (13). The ∼1-hour duration of C phase in the presence of blebbistatin is similar to its reported duration in the presence of an actin-depolymerizing drug (8) and the window of cortical contractility after anaphase onset (9), suggesting that an unidentified cell-cycle signal terminates C phase. In synchronized cells, C phase was scored by the presence of the microtubule midzone and of an equatorial ring of myosin II and anillin. Synchronized cells acquired these markers beginning ∼60 min after release from a mitotic block into blebbistatin (13) and lost them over the next 60 to 90 min.

In yeast, APC-dependent proteolysis is required to exit mitosis (19, 20). We therefore tested whether the proteasome inhibitor MG132 (21) affected C-phase duration in synchronized HeLa cells released from mitosis into blebbistatin. MG132 was added after >75% of the cells had initiated anaphase, and samples were fixed at different time points. The percentage of binucleate cells that retain a myosin II ring after 135 min is significantly increased by MG132 (Fig. 2B). Because cells that had not initiated anaphase before MG132 addition remained arrested in metaphase by the drug, our data underestimate the effect of MG132. We estimate that MG132 more than doubles the duration of C phase. Cells fixed at 170 min after treatment with both blebbistatin and MG132 mostly retained a midzone complex, although it was often pushed to one side of the cell by the decondensed nuclei (Fig. 2C), and myosin II and anillin remained localized to regions of the cortex closest to the displaced midzone. In contrast, cells treated with blebbistatin alone showed interphase microtubule organization and delocalized myosin II and anillin (Fig. 2C). These data reveal a role for ubiquitin-dependent proteolysis in C-phase exit in mammalian cells.

To examine how cytokinesis is spatially controlled, we used blebbistatin to probe microtubule-to-cortex communication. Cells arrested in mitosis with the Eg5 kinesin inhibitor monastrol were released into blebbistatin for ∼1 hour to allow the initiation of anaphase and furrow assembly. In the continued presence of blebbistatin, we then added other drugs (Fig. 3). Spindle assembly was complete before the new drug was added, and thus we could probe the roles of components required for spindle assembly, such as microtubules. Depolymerizing microtubules by adding nocodazole caused delocalization of both myosin II and anillin. Thus, microtubule-to-cortex communication was required continuously to maintain the localization of furrow components. This is consistent with the observation that microtubules are required for furrow ingression as well as furrow establishment (6). Nocodazole treatment also caused the separated nuclei to collapse back together (Fig. 3), suggesting that the midzone is required to keep the nuclei apart until cytokinesis is finished. Actin depolymerization with latrunculin delocalized myosin II and anillin, as expected, and had little effect on the organization of cytokinesis-associated microtubules. In other systems, perturbation of the actin cytoskeleton was reported to cause loss of midzone microtubules (22, 23). We did observe subtle disorganization of microtubules in latrunculin, but overall our data suggest that, in HeLa cells, communication from microtubules to the cortex is primarily unidirectional.

Figure 3

Effect of inhibitors on the microtubule spindle and the contractile ring in blebbistatin-treated cells. Mitotically synchronized HeLa cells were released into the presence of 100 μM (±)-blebbistatin, then blebbistatin alone (control), nocodazole, latrunculin, staurosporine, or Y27632 was added. Cells were stained for DNA, α-tubulin, nonmuscle myosin II, and anillin.

We next investigated the nature of the signals involved. The broad-spectrum kinase inhibitor staurosporine (24) caused extensive disorganization, although not complete removal, of midzone microtubules, indicating a role for kinases in midzone organization. Polo and aurora B are candidates for the kinase involved (25–27), but staurosporine does not substantially inhibit polo kinase at the concentration used in the experiment (28). The effect of staurosporine on aurora B has not yet been determined, so we tested two known ATP competitive inhibitors of aurora B [Compound 1, table 1, of (29) and Compound 1, table 1, of (30)]. These inhibitors gave results similar to those obtained with staurosporine treatment, causing disruption of the microtubule midzone (fig. S5).

Staurosporine also caused delocalization of myosin II but not of anillin. This provides evidence that localization of anillin and myosin II, which are normally colocalized in the furrow (31), are independently controlled. Aurora kinase inhibitors also caused delocalization of myosin II but not of anillin. The Rho-kinase inhibitor Y27632 (32) had no detectable effect on microtubule organization, delocalized myosin II, and again did not perturb anillin. Thus, Rho-kinase is required to localize myosin II to furrows in blebbistatin-arrested cells, which is broadly consistent with previous data (33), and other kinases, probably including aurora B, are required for midzone organization.

Cyclin-dependent kinase (CDK) inhibitors and monastrol do not perturb the cytology of blebbistatin-arrested cells (Table 1). Thus, CDKs do not appear to be directly involved in maintaining the contractile ring after anaphase, although it is possible that they might influence C-phase timing. The target of monastrol, the mitotic kinesin Eg5, is thought to have a central role in the establishment of spindle bipolarity, but our data suggest that it is not responsible for bipolar organization of the midzone during cytokinesis.

Table 1

Effects of drug treatments on nonmuscle myosin II, anillin, and the microtubule midzone in the presence of (±)-blebbistatin as described for Fig. 3. ER, endoplasmic reticulum.

View this table:

Our experiments emphasize the usefulness of fast-acting and reversible drugs in a dynamic cellular pathway such as cytokinesis. It is impossible to determine whether Eg5, for example, is required in cytokinesis with simple ablation experiments, because the ablation of Eg5 disrupts mitosis. New drugs that target guanosine triphosphatases, membrane dynamics, and mitotic motors will be useful in further dissecting the logic of cytokinesis. These proteins and processes are all required for cytokinesis (1, 2), but their precise roles in timing and spatial organization have yet to be defined.

Supporting Online Material

Materials and Methods

Figs. S1 to S6

References and Notes

Movies S1 to S4

  • * To whom correspondence should be addressed. E-mail: aaron_straight{at}


View Abstract

Stay Connected to Science

Navigate This Article