Inhibition of cyclin-dependent kinase 1 induces cytokinesis without chromosome segregation in an ECT2 and MgcRacGAP-dependent manner.

Cleavage furrow formation marks the onset of cell division during early anaphase. The small GTPase RhoA and its regulators ECT2 and MgcRacGAP have been implicated in furrow ingression in mammalian cells, but the signaling upstream of these molecules remains unclear. We now show that the inhibition of cyclin-dependent kinase (Cdk)1 is sufficient to initiate cytokinesis. When mitotically synchronized cells were treated with the Cdk-specific inhibitor BMI-1026, the initiation of cytokinesis was induced precociously before chromosomal separation. Cytokinesis was also induced by the Cdk1-specific inhibitor purvalanol A but not by Cdk2/Cdk5- or Cdk4-specific inhibitors. Consistent with initiation of precocious cytokinesis by Cdk1 inhibition, introduction of anti-Cdk1 monoclonal antibody resulted in cells with aberrant nuclei. Depolymerization of mitotic spindles by nocodazole inhibited BMI-1026-induced precocious cytokinesis. However, in the presence of a low concentration of nocodazole, BMI-1026 induced excessive membrane blebbing, which appeared to be caused by formation of ectopic cleavage furrows. Depletion of ECT2 or MgcRacGAP by RNA interference abolished both of the phenotypes (precocious furrowing after nocodazole release and excessive blebbing in the presence of nocodazole). RNA interference of RhoA or expression of dominant-negative RhoA efficiently reduced both phenotypes. RhoA was localized at the cleavage furrow or at the necks of blebs. We propose that Cdk1 inactivation is sufficient to activate a signaling pathway leading to cytokinesis, which emanates from mitotic spindles and is regulated by ECT2, MgcRacGAP, and RhoA. Chemical induction of cytokinesis will be a valuable tool to study the initiation mechanism of cytokinesis.

One of the dynamic mitotic processes, anaphase, is also the beginning of cytokinesis. During anaphase, two sister chromatids separate and move to spindle poles, and the cell elongates, as do the pole-to-pole spindles. Simultaneously, a cleavage furrow starts to form in the equatorial plane. Positioning of the cleavage furrow is induced by signals from the mitotic spindle during early anaphase. The molecular mechanism of cleavage furrow positioning and its regulation remains largely unknown. Recent data suggest that a signal emanating from the central spindle is critical to initiate cytokinesis (1). It has been reported that the small GTPase RhoA accumulates at the cleavage furrow (2). ECT2, a guanine nucleotide exchange factor for the Rho family GTPases, is essential for cytokinesis (3,4) as well as epithelial cell polarity (5). ECT2 is regulated by an autoinhibitory mechanism as well as nuclear sequestration in interphase (6). A mutation in the pebble gene, the Drosophila orthologue of human ECT2, inhibits cytokinesis in fly embryos (7). Pebble interacts with RacGAP50C, the Drosophila orthologue of human MgcRacGAP, which connects the contractile ring to cortical microtubules at the site of furrow ingression (8). MgcRacGAP, a GTPase-activating protein for the Rho GTPases, is also involved in cytokinesis (9,10). HeLa cells depleted of MgcRacGAP using siRNA 2 do not assemble well developed central spindles and fail to complete cell division (11). Binucleated blastomeres containing chromatin bridges were observed at early embryonic stages of MgcRacGAP-null mice (12). Therefore, MgcRacGAP is required for both mitosis and cytokinesis.
The molecules in the signaling pathways regulating cytokinesis downstream from RhoA are not well documented, but a Rho effector kinase, ROCK, was found to accumulate at the cleavage furrow and shown to phosphorylate intermediate filaments at the furrow (13). Another Rho effector, citron kinase, is critical for cytokinesis, as dominant-negative citron kinase strongly inhibits cytokinesis (14). Citron kinase induces diphosphorylation of the regulatory light chain of myosin II (15). Myosin phosphorylation by myosin light chain kinase and Rho kinase are also involved in cytokinesis (16). Phosphorylation of the myosin II regulatory light chain regulates actomyosin contractility and is critical for contractile ring constriction (17).
During the metaphase to anaphase transition, anaphase-promoting complex/cyclosome inactivates cyclin-dependent kinase (Cdk)1 though cyclin B degradation. In anaphase, sister chromatids separate and move to opposite spindle poles as Cdk1 activity decreases. Expression of a destruction-resistant cyclin B reportedly inhibits cytokinesis, suggesting that cyclin B negatively regulates cytokinesis and M phase exit (18,19). However, Cdk1 activity is required for M phase entry, making it difficult to study the regulation of cytokinesis by Cdk1. Although a number of Cdk inhibitors have been reported, these did not show a narrow specificity to Cdks. Recently, a novel kinase inhibitor, BMI-1026, has been shown to specifically inhibit Cdks, leading to forced mitotic exit (20). BMI-1026 specifically inhibits mitotic Cdks but not other mitotic kinases such as Plk1 and Aurora A. Furthermore, mitotically arrested U2OS cells treated with BMI-1026 exhibit G 1 -like morphology with incomplete chromosome separation following membrane blebbing in the presence of nocodazole (20). In this study, we utilized this novel Cdk1 inhibitor to investigate the mechanism of cytokinesis related to the Rho signaling pathway. We show that cytokinesis can be induced in mitotic cells by BMI-1026. We also found that RhoA and two Rho regulators, ECT2 and MgcRacGAP, are essential to furrow ingression induced by BMI-1026. These observations suggest that inhibition of Cdk1 activity is sufficient to induce cytokinesis, which is regulated through ECT2, MgcRacGAP, and RhoA.

EXPERIMENTAL PROCEDURES
Cell Culture and Synchronization-The HeLa cell line and its derivative, which expresses GFP-histone H2B, were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 100 units/ml penicillin G plus 100 g/ml streptomycin (Invitrogen). Synchronization of the cell cycle was carried out by double thymidine block (21). In brief, cells were arrested for 14 h with 2.5 mM thymidine (Sigma) followed by an 8-h release. After the second thymidine treatment, cells were released into fresh medium containing 40 ng/ml of nocodazole (Sigma) to arrest the cells at prometaphase. In some experiments, mitotic cells were collected by the knock-off (mechanical disruption) procedure after nocodazole treatment at a concentration of 40 ng/ml for 6 -10 h. The remaining adherent cells were used as interphase cells. BMI-1026 has been described (20). Purvalanol A, PNU112455A, and Cdk4 inhibitor were from Calbiochem, San Diego, CA.
Western Blot Analysis-Cell lysates were clarified by centrifugation (14,000 ϫ g, 20 min, 4°C), and the protein concentration was determined using the BCA assay kit (Pierce). Samples were separated and analyzed on 6 or 8 -16% gradient SDS-PAGE gels (Invitrogen), and then transferred to polyvinylidene difluoride membranes. The membranes were blocked in 5% nonfat dry milk/PBS for 30 -60 min at room temperature, and incubated with the primary antibody for 1-6 h at room temperature. Bound antibody was detected using horseradish peroxidase-conjugated anti-mouse and anti-rabbit IgG. The signals were detected using ECL chemiluminescence reagents (Amersham Biosciences).
Immunocytochemistry-Cells grown on coverslips were fixed with 100% methanol for 20 min, permeabilized for 10 min with 0.1% Triton X-100 in PBS, and then washed with PBS. Samples were incubated with PBS containing 2% serum for 1 h and then with a rabbit anti-survivin antibody (Abcam) and a mouse anti-␥-tubulin antibody (Sigma) for 1 h. Samples were then treated with FITC-conjugated anti-rabbit IgG, Alexa 488-conjugated anti-mouse IgG (Molecular Probe, Eugene, OR) and DAPI (Sigma). Images were observed with the Zeiss Axiovert S-100 fluorescence microscopy using Openlab software (Improvision).
Luciferase siRNA (GL-2; accession no. X65324) was used as a control. HeLa cells were seeded in 35-mm dishes and transfected with siRNA as follows. In a tube, Opti-MEM I (250 l) was mixed with 15 l of the stock RNA. In a second tube, 60 l of Opti-MEM I was incubated with 15 l of Oligofectamine (Invitrogen) for 8 min at room temperature. The two mixtures above were combined, gently mixed, and incubated for 22 min at room temperature. The entire mixture was then added to the cells in 2 ml of 10% fetal bovine serum containing Dulbecco's modified Eagle's medium. Knockdown levels of the proteins were estimated with Western blotting 24 -48 h after transfection.

Delivery of FITC Antibody into Cultured Cells-FITC-conjugated
anti-p34 Cdk1 antibody (anti-Cdc2 p34; Santa Cruz Biotechnology) was delivered into HeLa cells using Provectin (Imgenex, San Diego, CA) according to the manufacturer's protocol. HeLa cells were seeded on coverslips in 24-well plates. FITC-conjugated anti-p34 Cdk1 antibody or FITC-labeled goat IgG (control) was diluted in HBS buffer (10 mM Hepes, 150 mM NaCl, pH 7.0) and added into tubes which contained Provectin dry film. The Provectin/antibody mix was incubated at room temperature for 5 min and then brought the volume up to 250 l with serum-free medium. The mixture was transferred onto cells and incubated for 4 h in serum-free medium. The culture was supplemented with serum and continued to incubate for 12 h. Cells were fixed with methanol, and the nuclei were stained with DAPI. Cell morphology was examined using the Zeiss LSM510 confocal microscope. For quantitative analysis, at least 400 cells were counted from three independent slides.
Immunofluorescence Microscopy-HeLa cells synchronized at prometaphase, were treated with BMI-1026 (200 nM) with or without nocodazole (40 ng/ml) or taxol (1 M) for 3 h. Cells were fixed with methanol and followed by acetone for 5 min each at Ϫ20°C. After incubation for 1 h with 3% bovine serum albumin in phosphate-buffered saline, cells were further incubated for 1 h at room temperature with anti-survivin polyclonal antibody at 1:200 (R&D Systems) and anti-␣tubulin monoclonal antibody at 1:500 (Sigma-Aldrich). After washing three times with PBS, cells were labeled with Alexa 488 or Alexa 568conjugated secondary antibodies (Molecular Probe). Nuclei were stained with DAPI and were examined with the Zeiss LSM510 confocal microscope.
Time-lapse Microscopy-HeLa cells expressing GFP-histone H2B were cultured in a 35-mm dish on the stage of a Zeiss Axiovert S-100 inverted microscope equipped with an environmental chamber. Timelapse images were captured at 3-min intervals by the Photometric digital camera controlled by Openlab software.

Cdk1 Inhibition Precociously Induces Cleavage Furrow Ingression in
Mitotic Cells-To test the effects of the Cdk1 inhibition on cytokinesis, HeLa cells expressing GFP-histone H2B were arrested at prometaphase by nocodazole, released from the arrest, and then treated with the Cdk inhibitor BMI-1026. The morphological changes were recorded by time-lapse microscopy ( Fig. 1A). Surprisingly, BMI-1026 potently induced cleavage furrow ingression. Furrow ingression was observed at an earlier time point in BMI-1026-treated cells (10 -15 min after release from nocodazole) than untreated cells (90 min after release from nocodazole). Moreover, although untreated cells started cytokinesis after chromosome separation, furrow ingression was initiated before chromosome separation in BMI-1026-treated cells. This phenotype, induced by BMI-1026 after nocodazole release, was very similar to the Cut phenotype (23). However, a population of BMI-1026-treated cells did not undergo abscission, and the emerging daughter cells subsequently merged to generate cells with aberrant nuclei. As BMI-1026 specifically inhibits mitotic Cdk but not other mitotic kinases such as Plk1 and Aurora A (20), it is likely that Cdk1 inhibition induces cytokinesis initiation. To confirm that BMI-1026 induces cytokinesis by inhibiting the Cdk1/cyclin B complex, we tested other Cdk inhibitors. Purvalanol A displays a narrow specificity to Cdk1/cyclin B (IC 50 ϭ 4 nM) compared with that to Cdk1/cyclin A (IC 50 ϭ 70 nM), Cdk2/cyclin E (IC 50 ϭ 35 nM), and Cdk5/p35 (IC 50 ϭ 75 nM). As shown in Fig. 1B, Purvalanol A induced furrow ingression in mitotic cells as BMI-1026 did. In contrast, the Cdk2/Cdk5-specific inhibitor PNU112455A or Cdk4 inhibitor did not display this activity. These results suggest that Cdk1 inhibition is sufficient to induce cleavage furrow ingression.
We also utilized another method to inhibit Cdk1. We introduced affinity-purified anti-Cdk1 monoclonal antibody to asynchronous HeLa cells by Provectin. Consistent with the requirement of Cdk1 for the entry to M phase, the appearance of mitotic cells was strongly inhibited by anti-Cdk1 (Fig. 2B), suggesting that anti-Cdk1 effectively inhibited Cdk1 in these cells. A population of anti-Cdk1-transfected cells should enter M phase before Cdk1 was effectively inhibited. The inhibition of Cdk1 by anti-Cdk1 antibody after M phase entry might induce cleavage furrow ingression before chromosome separation as BMI-1026 did. Although control IgG-transfected cells generate cells with single nuclei, anti-Cdk1-introduced cells frequently contained multiple nuclei of different sizes (Fig. 2). These results support that inhibition of Cdk1 after M phase entry induces precocious cleavage furrow ingression before chromosome separation. Efficient generation of multinucleate cells by anti-Cdk1 also support the notion that abscission did not occur frequently after cleavage furrow ingression induced by Cdk1 inhibition. All of these results suggest that Cdk1 inhibition is sufficient to induce cleavage furrow ingression in mitotic cells.
Normal Mitotic Spindles Are Required for Furrowing Induced by Cdk1 Inhibition-There is accumulating evidence that cytokinesis is induced by signals emanating from overlapping mitotic spindles (1). To test whether spindles are required for furrow ingression induced by BMI-1026, we performed similar assays in the presence of the microtubule depolymerizing agent nocodazole. HeLa cells were arrested at prometaphase by nocodazole, and BMI-1026 was added without removal of nocodazole. In the presence of a high concentration of nocodazole (300 ng/ml), mitotically rounded cells spread without cleavage furrow ingression within 30 min after BMI-1026 treatment (Fig. 3A, top panels), suggesting that the induction of furrow ingression by BMI-1026 requires mitotic spindles. Induction of cell spreading by BMI-1026 was  consistent with a previous report that BMI-1026 causes precocious mitotic exit in U2OS cells (20). Next, we examined the effects of partial depolymerization of mitotic spindles on induction of furrow ingression by BMI-1026. In the presence of a low concentration of nocodazole (40 ng/ml), BMI-1026 induced excessive membrane blebbing in most of the cells (Fig. 3A, middle panels). Strong tubulin staining was observed in these blebs (see Fig. 7), but survivin was not concentrated in the blebs (data not shown). To test the possibility that incomplete depolymerization of mitotic spindles is responsible for this phenotype induced by BMI-1026, the microtubule stabilizing agent taxol was utilized in place of nocodazole. As shown in Fig. 3A, BMI-1026 also potently induced membrane blebbing in the presence of taxol. Localization of tubulin, but not survivin, was evident on the blebs induced by BMI-1026 in the presence of taxol (Fig. 3B, see also Fig. 7). However, survivin was concentrated at the structures, which could be identified as the midzone created by abnormal cytokinesis (Fig. 3B). We further tested the localization of two Rho regulators, ECT2 and MgcRacGAP, which are known to localize at the midbody during cytokinesis (4,9). These proteins were localized at structures that were similar to the midbody, although cytokinesis was abnormal in these cells (Fig. 3B). Therefore, BMI-1026 appeared to induce both excessive blebbing and furrowing in the presence of taxol. Presumably, furrowing was initiated at the sites where microtubule bundling was evident and thus chromosomal passenger proteins and Rho regulators accumulated. Taken together, these results suggest that perturbation of normal mitotic spindles, but not their complete disappearance, is responsible for excessive membrane blebbing induced by BMI-1026.

RhoA Knockdown Reduces BMI-1026-induced Precocious Furrowing and Membrane
Blebbing-It has been reported that Rho GTPases play a critical role in cell division (24,25). To test which Rho GTPase is responsible for furrow ingression induced by BMI-1026, we knocked down representative Rho GTPases by RNAi (Fig. 5A). We found that knockdown of RhoA affected precocious cytokinesis initiation induced by BMI-1026 after nocodazole release, whereas knockdown of Rac1 or Cdc42 did not display significant effects (Fig. 5B). The effect of RhoA knockdown on furrowing was evident 15 min after BMI-1026 addition, but the number of cells with ingression increased at later time points. Thus, RhoA depletion did not prevent BMI-1026-induced furrowing, but effectively delayed it. As the RhoA was efficiently depleted by siRNA transfection (Fig. 5A), weak inhibition of furrowing in RhoA siRNAtransfected cells might be attributed to compensatory mechanisms by other Rho GTPases. In contrast, the knockdown levels of Cdc42 and Rac1 were lower than that of RhoA. It is possible that residual levels of these GTPases are sufficient to induce cytokinesis.
We also investigated which Rho GTPase is responsible for excessive membrane blebbing induced by BMI-1026 in the presence of a low concentration of nocodazole. Although knockdown of Cdc42 or Rac1 did not exhibit significant effects on BMI-1026-induced blebbing, most of the cells transfected with RhoA siRNA spread without exhibiting excessive membrane blebbing (Fig. 6A). Therefore, RhoA appeared to play a role in BMI-1026-induced excessive blebbing in the presence of a low concentration of nocodazole. Although RhoA knockdown did not completely inhibit excessive blebbing, it efficiently delayed it (Fig. 6B). Other Rho GTPases might compensate for this activity when RhoA is knocked down. As the knockdown levels of Cdc42 and Rac1 were lower than RhoA, it is also possible that Cdc42 or Rac1 plays some role in excessive blebbing.
It has been reported that RhoA is localized to the cleavage furrow during cytokinesis (24,25). When mitotically arrested cells were released from nocodazole and then treated with BMI-1026, RhoA was concentrated at regions corresponding to the cleavage furrow and then at the midbody (Fig. 7, left panels). When BMI-1026 was added in the presence of a low concentration of nocodazole, RhoA was localized at the necks of blebs (Fig. 7, middle). BMI-1026-treated cells in the presence of taxol exhibited similar localization of RhoA (Fig. 7, right). Taken together, these results further support that the RhoA signaling pathway is involved in BMI-1026-induced precocious furrowing as well as excessive blebbing.
We also examined whether the Rho effector kinase, ROCK, contributes to BMI-1026-induced precocious cytokinesis in the presence of nocodazole, using the ROCK-specific inhibitor Y-27632 (26). However,  we could not observe any significant effect of Y-27632 on BMI-1026induced furrowing at a final concentration of 100 M, although cytoskeletal reorganization of Swiss 3T3 cells was completely inhibited at this concentration (data not shown). These results suggest that ROCK is not a critical regulator of BMI-1026-induced precocious cytokinesis.

DISCUSSION
Normal cytokinesis is properly coordinated by the spindle function to ensure accurate segregation of chromosomes. We found that treatment of mitotic cells by the Cdk inhibitor BMI-1026 or purvalanol A potently initiates cytokinesis, but without chromosome separation. The effects of the Cdk1 inhibitors on mitotic cells are surprising in two ways. First, the initiation of cytokinesis was induced chemically. Second, BMI-1026 treatment effectively uncoupled cytokinesis from chromosome separation. Similar to normal cytokinesis, two Rho regulators, ECT2 and MgcRacGAP, were required for BMI-1026-induced cytokinesis. We also found that RhoA is involved in BMI-1026-induced cytokinesis. We found that abscission is not always accompanied by BMI-1026-induced cytokinesis. As BMI-1026 causes cytokinesis before the completion of chromosome separation, the central spindles were not well developed at the time of furrowing. It has been shown that the signals for cleavage furrow formation emanate from overlapping mitotic spindles between the separating chromosomes (1). Thus, the initiation of the central spindle assembly that occurs after release from nocodazole may induce furrowing. Presumably, enough signals could not accumulate to induce abscission because of the lack of well developed central spindles in BMI-1026-treated cells.
BMI-1026 has been shown to specifically inhibit Cdks both in vitro and in vivo (20). Selective inhibition of Cdk1 among the mitotic kinases in vivo has been demonstrated by induction of a potent G 2 /M arrest, mitotic catastrophe, and precocious mitotic exit. The Cdk1-specific inhibitor purvalanol A also induced a similar phenotypes as BMI-1026, but Cdk2/5-specific and Cdk4-specific inhibitors did not show these phenotypes. Moreover, the introduction of an anti-Cdk1 antibody resulted in the generation of aberrant nuclei, a phenotype expected from furrowing without chromosomal separation. Together with the reports that expression of a destruction-resistant cyclin B inhibits cytokinesis (18,19), the induction of cytokinesis in mitotic cells by BMI-1026 strongly suggests that Cdk1 inactivation is sufficient for the initiation of cytokinesis. Recently, it has been shown that phosphorylation of MKLP1 by Cdk1 inhibits central spindle formation (27). Given that central spindle formation allows the accumulation of signaling molecules that initiate furrowing, dephosphorylation of MKLP1 could induce central spindle assembly and initiate cytokinesis. However, as BMI-1026 induced cytokinesis before the central spindle was fully assembled, dephosphorylation of MKLP1 might not be the sole mechanism of cytokinesis initiation by BMI-1026. Although initiation of central spindle assembly is required for establishment of the cleavage plane, dephosphorylation of components of the Rho signaling pathway that regulates cytokinesis might initiate furrow ingression, which results in cytokinesis without chromosome separation.
When mitotically arrested cells were treated with BMI-1026 in the presence of a high concentration of nocodazole, cells spread without initiation of cytokinesis. Therefore, mitotic spindles might be required for BMI-1026-induced cytokinesis. The induction of cell spreading by BMI-1026 was consistent with the previous report that BMI-1026 induces mitotic exit in U2OS cells (20). We also found that BMI-1026 induced excessive membrane blebbing in the presence of a low concentration of nocodazole. Although most spindles were disrupted and delocalized in cells treated with a low concentration of nocodazole, these abnormal spindles were probably sufficient to initiate furrow formation. Thus, the blebs observed in cells in the presence of a low concentration of nocodazole might represent ectopic cleavage furrows. As excessive blebbing is accompanied by abnormal cytokinesis (28 -30), it is also possible that these blebs might represent cortical activity that was located along the cleavage furrow in normal cytokinesis but was mislocalized and amplified by partial disruption of mitotic spindles in cells treated with BMI-1026 in the presence of a low concentration of nocodazole. As BMI-1026-induced excessive blebbing was inhibited by knockdown of ECT2, MgcRacGAP, or RhoA, bleb formation appeared to be regulated by the Rho signaling pathway, similar to furrow ingression.
We showed that knockdown of RhoA, but not Rac1 or Cdc42, effectively delayed BMI-1026-induced cytokinesis. RhoA was localized at the cleavage furrow and the necks of the blebs induced by BMI-1026. These results suggest that RhoA is involved in cytokinesis. However, despite the high knockdown level of RhoA, RhoA siRNA could not completely inhibit BMI-1026-induced cytokinesis. Therefore, it is possible that other Rho GTPases closely related to RhoA, such as RhoB and RhoC, are also involved in cytokinesis. It has been shown that RhoA is activated during cytokinesis, and this activation is inhibited by dominant-negative ECT2 (21). MgcRacGAP, which is a GAP for Rac1 and Cdc42, is converted to a RhoGAP by phosphorylation by Aurora B in M phase and regulates cytokinesis (10). We showed that knockdown of ECT2 or MgcRacGAP efficiently inhibited BMI-1026-induced cytokinesis. Therefore, the Rho signaling pathway regulated by ECT2 and MgcRac-GAP might be critical to regulate cytokinesis. As the ROCK inhibitor Y-26732 did not affect BMI-1026-induced cleavage furrow ingression, other Rho effectors such as citron kinase might play a critical role in cytokinesis.
We propose the following model for the precocious induction of cytokinesis by BMI-1026 (Fig. 8). The mitotic checkpoint regulates both chromosome separation and cleavage furrow formation by inhibiting anaphase-promoting complex/cyclosome, which degrades securin and cyclin B through ubiquitin-mediated proteolysis. When cells are synchronized at prometaphase by nocodazole, the mitotic checkpoint is activated, and the anaphase-promoting complex/cyclosome becomes inactive for securin and cyclin B degradation. When the checkpoint is silenced by release from nocodazole, the anaphase-promoting complex/ cyclosome is activated, and then cyclin B and securin are degraded. This signaling in turn activates two pathways that regulate cell division. Securin degradation results in cohesion inactivation, which initiates chromosome separation and central spindle formation. Silencing of the mitotic checkpoint also activates another pathway toward cytokinesis. Cyclin B is degraded by active anaphase-promoting complex cyclosome,  and then cleavage furrow ingression is induced. It has been reported that expression of nondestructable forms of cyclins inhibits cytokinesis, indicating that Cdk1/cyclinB negatively regulates cytokinesis (18,19). Recently, it has been shown that Cdk1 also negatively regulates central spindle formation by phosphorylating MKLP1 (27). Therefore, Cdk1 inhibition by BMI-1026 should also stimulate central spindle formation. In Drosophila cells, the Pavarotti/RacGAP50C (Drosophila MKLP1/ MgcRacGAP) complex accumulates at the central spindle, and the association of Pebble (Drosophila ECT2) with the complex activates the Rho signaling pathway to initiate cytokinesis (8). BMI-1026 treatment of prometaphase cells inhibits Cdk1 before cyclin B and securin destruction to induce central spindle formation without chromosome separation. Although the central spindle may not be well established without chromosome separation, this may be sufficient to establish the cleavage plane. In Drosophila embryos, Cdk1 negatively regulates the Rho signaling that emanates from the central spindle (19). BMI-1026 thus precociously induces cleavage furrow ingression by removing the negative regulation. The critical cytokinesis regulator that is regulated by Cdk1/ cyclin B may be ECT2, as suggested by genetic work in Drosophila (19). When cells were treated with BMI-1026 in the presence of nocodazole, no signal for cytokinesis emanates because of the lack of the central spindle, and, therefore, cells exited mitosis without cytokinesis. However, when overlapping spindles were partially retained in the presence of a low concentration of nocodazole, the signal for cytokinesis emanating from these partial spindles induces ectopic cleavage furrows, which are observed as excessive membrane blebbing.
We have utilized the Cdk1 inhibitor BMI-1026 to analyze the initiation of cytokinesis. Although a number of Cdk inhibitors have been identified and characterized, many of them exhibit relatively broad specificities and therefore inhibit multiple biological functions. BMI-1026 is a unique small molecule inhibitor that shows a narrow specificity to Cdk1 (20). As Cdk1 is also required to enter mitosis, analysis of M phase exit alone is difficult using other methods such as expression of dominant-negative-Cdk1 and Cdk1 depletion by RNAi. In addition to other small molecule inhibitors, such as blebbistatin (31), monastrol (32), Aurora kinase inhibitors (33,34), and Y-26732 (26), BMI-1026 will be useful for further studies to clarify the signaling pathways that regulate cell division. As cytokinesis is a spatially and temporally regulated event, chemical induction of cytokinesis by BMI-1026 will be useful for analysis of cytokinesis at the molecular level.