Polo-like Kinase 1 Facilitates Chromosome Alignment during Prometaphase through BubR1*

Plk1, an evolutionarily conserved M phase kinase, associates with not only spindle poles but also kinetochores during prometaphase. However, the role of Plk1 at kinetochores has been poorly understood. Here we show that BubR1 mediates the action of Plk1 at kinetochores for proper chromosome alignment. Our results show that BubR1 colocalizes with Plk1 at kinetochores of unaligned chromosomes and physically interacts with Plk1 in prometaphase cells. Down-regulation of Plk1 by small interfering RNA abolished the mobility-shifted, hyperphosphorylated form of BubR1 in the prometaphase-arrested cells. In addition, BubR1 was phosphorylated by Plk1 in vitro at two Plk1 consensus sites in the kinase domain of BubR1. The add-back of either wild-type BubR1 or BubR1 2E, in which the two Plk1 phosphorylation sites were replaced by glutamic acids, but not that of BubR1 2A, an unphosphorylatable mutant, rescued the chromosome alignment defects in BubR1-deficient cells. Moreover, when both Plk1 and BubR1 were down-regulated, the add-back of BubR1 2E, but not that of wild-type BubR1, rescued the chromosome alignment defects. These results taken together suggest that Plk1 facilitates chromosome alignment during prometaphase through BubR1.

The proper segregation of chromosomes in anaphase is essential to prevent aneuploidy (1). All chromosomes are attached to the mitotic spindle and properly aligned to the metaphase plate so that a surveillance mechanism, the spindle checkpoint, is silenced. The chromosome alignment is, thus, critical for accurate chromosome segregation. Numerous protein-protein interactions and chemical reactions including phosphorylation on kinetochores have been implicated in proper microtubule-kinetochore interactions (2,3). Recently it has been shown that Polo-like kinase 1 (Plk1) 2 creates 3F3/2 phospho-epitope on mitotic kinetochores of unaligned or unattached chromosomes that lack tension (4,5).
Plk1, a mammalian ortholog of Drosophila polo, plays a crucial role in multiple stages of mitosis (6 -8). Plk1 localizes to centrosomes and kinetochores during early stages of mitosis and then to midbody during later stages (7,9). Such dynamic changes in subcellular localization of Plk1 are believed to be important for its function. When the kinetochores of a sisterchromatid pair are attached by microtubules from opposite spindle poles, tension develops across the sister kinetochores. Subsequently this tension is thought to stabilize the microtubule-kinetochore interactions, and the kinetochores are allowed to be fully occupied with microtubules (10,11). Recent studies have shown that Plk1 is involved in functional bipolar spindle formation and in the generation of proper tension at kinetochores (12,13). However, how Plk1 regulates these events has remained unknown.
BubR1 is an essential component of the spindle checkpoint (14) and is also shown to function in chromosome alignment (15). BubR1 is able to inhibit APC/C by binding to Cdc20 (14) and associates with the kinetochore motor protein CENP-E (16 -18), which is shown to be essential for both chromosome alignment and spindle checkpoint (19 -21). Most recently, Lampson and Kapoor have demonstrated that BubR1 plays a role in attachment of spindle microtubules to kinetochores (15).
Because both Plk1 and BubR1 are shown to localize to kinetochores of unaligned chromosomes and play a role in the formation of the stable microtubule-kinetochore interactions, we hypothesized that BubR1 might be a target of Plk1 at kinetochores. Here we show several lines of evidence that Plk1 regulates chromosome alignment through phosphorylation and activation of BubR1 at kinetochores.

EXPERIMENTAL PROCEDURES
Cell Culture and Drug Treatment-HeLaS3 cells, HeLa cells expressing a GFP-histone H2B protein, and COS7 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. HeLa cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% bovine calf serum. Cells were synchronized in S phase by the double-thymidine block. Mitotically arrested cells were achieved by treatment with 250 ng/ml nocodazole (Sigma) or 10 M paclitaxel (Sigma) at ϳ8 h after release from the thymidine block. When transfection was performed before the first thymidine block, HeLa cells were transfected in Opti-MEM with the use of Lipofectamine Plus (Invitrogen) for 2 h and washed into fresh medium containing thymidine for the first block. For transfection with DNA vectors and RNA oligonucleotides during synchronization, cells were transfected in Opti-MEM with the use of Lipofectamine Plus (Invitrogen) for 2 h and washed into fresh medium containing thymidine for the first block. Then cells were transfected by the use of Oligo-fectamine (Invitrogen) after the first thymidine block. 50 M proteasome inhibitor, MG132 (Calbiochem), was added 10 h after release from the thymidine block. Expression in COS7 cells was carried out by the use of Lipofectamine Plus (Invitrogen) for 4 h.
Immunofluorescence Analysis-HeLa cells, HeLa GFP-histone H2B cells, and HeLaS3 cells were grown onto coverslips, fixed with 4% formaldehyde in PBS for 10 min, and permeabilized with 0.5% Triton X-100 in PBS for 10 min. For BubR1 staining, cells were fixed with 4% paraformaldehyde in PBS for 7 min and permeabilized with 0.2% Triton X-100 in PBS for 10 min. To destabilize nonkinetochore microtubules, mitotic cells were treated with a calcium-containing buffer for 40 s during permeabilization and then fixed for 10 min with 4% paraformaldehyde in the same buffer as described by Mitchison (25). Primary antibodies included those against BubR1 (mouse; Chemicon), Aurora-B (mouse; Transduction Laboratories), Myc (9E10, mouse; Santa Cruz Biotechnology), ␣-tubulin (DM1A, mouse; Sigma), and ␤-tubulin (TUB2.1, mouse; Sigma). Immune complexes were detected with appropriate secondary antibodies labeled with either Alexa488 or Alexa594 (Molecular Probes). DNA was stained with Hoechst 33342. Fixed samples were observed with a Zeiss Axiophot2, a Photometrics CCD camera, and software IPLab Spectrum (Scanalytics, Inc). Each image in Figs. 5E and 7, E and F, were taken in the same exposure time by software IPLab Spectrum. Images in Figs. 1B, 2B, and 7B were acquired as Z-stacks with 0.2-m spacing using a ϫ100, 1.35 NA objective on a DeltaVision Image Restoration Microscope (Applied Precision Instruments, Olympus and Seki Technotron Corp.) and processed by iterative constrained deconvolution (SoftWoRx, Applied Precision Instruments). Image analysis was performed using either SoftWoRx or Adobe Photoshop (Adobe Systems).

RESULTS
Plk1 Is Involved in Chromosome Alignment-Plk1 associates with kinetochores during prometaphase and dissociates from kinetochores at the onset of anaphase (7,9) and is reported to play a role in functional bipolar spindle formation and generation of proper tension (12,13). To show the involvement of Plk1 in kinetochore-microtubule attachment, we analyzed the kinetics of recovery processes after release from prometaphase arrest by nocodazole treatment. The Plk1 siRNA treatment reduced the Plk1 protein level by about 90% in total in each experiment (see Fig. 4A). A bipolar spindle was formed in most of the cells at 2 h after release in both control and Plk1 siRNA-treated cells (Fig. 1A). As to chromosome alignment, however, the recovery was significantly delayed in Plk1 siRNA-treated cells; more than 40% of the cells showed abnormalities in chromosome alignment even at 2 h after release (Fig. 1A, histograms). In contrast, less than 20% of mock-treated cells showed the abnormalities at 2 h after release (Fig. 1A, histograms). To characterize Plk1-deficient cells in more detail, kinetochores were visualized by transfection of YFP-CENP-A, and cells were treated with calcium-containing buffers during permeabilization. This treatment is known to preferentially destabilize nonkinetochore microtubules. In control cells, each kinetochore appeared to be stably re-captured by kinetochore microtubules (Fig. 1B, YFP-CENP-A, upper, control). In contrast, in Plk1deficient cells two spots of a kinetochore pair were not alongside microtubules and were often unattached to kinetochore microtubules (Fig. 1B, YFP-CENP-A, upper, Plk1 siRNA), suggesting the defect in stable microtubule-kinetochore attachment. In addition, Mad2-positive kinetochores were frequently observed in Plk1-deficient cells, but not in control cells, at metaphase (Fig. 1B, lower). It should be noted that in Plk1-deficient cells a bipolar spindle was smaller than that in control cells and was more sensitive to calcium treatment that causes microtubule destabilization. These observations were consistent with previous reports that Plk1 is involved in functional spindle formation (12,13). Taken together, these results suggest that Plk1 is necessary for not only proper spindle assembly but also the formation of functional kinetochore-microtubule attachments. were synchronized and treated with mock or Plk1 siRNA. 9 h after release from the G 1 -S boundary, cells were treated with nocodazole for 2 h, and then cells were released into fresh medium for nocodazole removal. 2 h after nocodazole removal, cells were treated with a calcium-containing buffer, then fixed and stained for ␣-tubulin (red) and DNA (blue). Insets, control, a and b, and Plk1 siRNA, a, b, and c) detail kinetochore-microtubule interactions. Lower, HeLaS3 cells were synchronized and transfected with siRNAs for Plk1. 9 h after release from the G 1 -S boundary, the cells were treated as above. ␣-Mad2 (green, middle), ␣-tubulin (red) and DNA (blue). The scale bar represents 5 m. Insets show optical sections, 300% magnification.

Plk1
Interacts with BubR1 during Prometaphase-To test our idea that BubR1 might be a target of Plk1 at kinetochores, we first examined localization of Plk1 at kinetochores in detail. HeLa cells were transfected with GFP-fused Plk1, then fixed and stained with anti-Aurora B antibody or anti-BubR1 antibody. BubR1 and GFP-Plk1 colocalized at kinetochores of prometaphase cells, whereas Aurora B localized differently at the inner centromere region between the two spots of GFP-Plk1 ( Fig. 2A). Observations with DeltaVision image restoration microscope also showed that GFP-Plk1 localized at kinetochores of unaligned chromosomes and colocalized with BubR1 there (Fig. 2B, arrowheads). Colocalization of GFP-Plk1 and BubR1 at kinetochores was also observed in prometaphase-arrested cells resulting from nocodazole or taxol treatment (data not shown).
We next examined whether there is any physical association between BubR1 and Plk1. HeLa cells were transfected with HA-tagged Plk1 and Myc-tagged BubR1 during double thymidine block. Then cells were released from the G 1 -S boundary and treated with nocodazole for 2 h at 10 h after release. Then cell lysates were obtained and subjected to immunoprecipitation. As shown in Fig. 2C, BubR1 and Plk1 co-immunoprecipitated, indicating that BubR1 physically interacts with Plk1 at prometaphase. Because it has been reported that Plk1 interacts with its substrates in the C-terminal polo-box domain (residues 410 -604) (26 -28), we examined whether BubR1 also interacts with the C-terminal domain of Plk1 by testing both the C-terminal-truncated form of Plk1 (N, residues 1-364) and the N-terminal-truncated form of Plk1 (C, residues 357-604) for their ability to bind to BubR1. Only the N-terminal-truncated form bound to BubR1 (Fig.  2C), suggesting that BubR1 binds to the C-terminal, complete polo-box domain of Plk1. A co-immunoprecipitation experiment in cell lysates from non-transfected cells demonstrated that endogenous BubR1 binds to endogenous Plk1 in prometaphase (Fig. 2D). The binding between endogenous BubR1 and Plk1 was recently reported (29). Plk1 Is Responsible for Hyperphosphorylation of BubR1-Recent studies have revealed that BubR1 and Aurora B are involved in chromosome alignment as well as mitotic checkpoint during prometaphase (15, 30 -32). Because our results have demonstrated that BubR1 colocalizes with, and interacts with Plk1 during prometaphase, we hypothesized that Plk1 and BubR1 might function cooperatively at kinetochores. It has been shown that BubR1 displays a mobility shifted band in SDS- polyacrylamide gel electrophoresis due to hyperphosphorylation in mitotically arrested cells by nocodazole or taxol treatment (17,33,34). By using synchronized HeLa cells, we confirmed the appearance of this mobility shifted band of endogenous BubR1 in the presence of nocodazole or taxol (Fig.  3A). These mobility-shifted bands of BubR1 in the presence of nocodazole or taxol were due to the hyperphosphorylation, as confirmed by the phosphatase treatment (Fig. 3A). To see when BubR1 is phosphorylated in M phase, HeLa cells were arrested at prometaphase and metaphase by treatment with nocodazole and MG132, the proteasome inhibitor, respectively, after release from the G 1 -S block. Then cells at each phase were further collected by mechanical shake-off from flasks. It is known that cells are arrested at metaphase when synchronized cells are treated with MG132 during prometaphase (see Fig. 7D, control) (30). Immunoblot analysis with anti-BubR1 antibody showed that BubR1 was phosphorylated and dephosphorylated at prometaphase and metaphase, respectively (Fig. 3B, noc. and MG132). To examine whether BubR1 is phosphorylated in M phase during normal cell cycle progression, i.e. under conditions in which no drug treatment is given, we collected mitotic cells at 10 h after release from G 1 -S block by mechanical shake-off. The immunoblot analysis clearly detected the mobility shifted band of BubR1 in these mitotic cells (Fig. 3B, 10 h). Furthermore, to eliminate the possibility that dephosphorylation of BubR1 might be due to some side effect of MG132 treatment, we collected mitotic cells treated with both nocodazole and MG132 after release from the G 1 -S block. The cells were arrested at prometaphase in this case. The immunoblot analysis showed that BubR1 remained phosphorylated and was not dephosphorylated even in the presence of MG132 in those prometaphasearrested cells (Fig. 3B, noc. ϩ MG132), eliminating the possibil-ity. These results taken together suggest that BubR1 is phosphorylated during prometaphase and dephosphorylated at metaphase.
Then we examined the effect of down-regulation of Plk1 by siRNA on the hyperphosphorylation of BubR1 in the mitotically arrested cells. Because down-regulation of the gene expression by siRNA works often very well in HeLa cells expressing a GFP-histone H2B fusion protein, we used these HeLa GFP-histone H2B cells for siRNA experiments. Synchronized HeLa GFP-histone H2B cells were transfected with siRNA for Plk1 and/or Aurora B and then arrested at prometaphase by treatment with nocodazole or taxol after release from the G 1 -S block. Then cells at prometaphase were further collected by the mechanical shake-off. Plk1 siRNA resulted in the disappearance of the mobility-shifted band of BubR1 in mitotically arrested cells, which are induced by either nocodazole or taxol treatment (Fig. 4A, Plk1 siRNA). Because either nocodazole or taxol treatment in Plk1-deficient cells resulted in prometaphase arrest (supplemental Fig. S3), the result suggests that Plk1 is responsible for the hyperphosphorylation of BubR1 in mitotically arrested cells. In contrast, Aurora B siRNA caused almost complete disappearance of the taxol treatment-induced BubR1 mobility-shifted band but did not inhibit completely the nocodazole treatment-induced appearance of the mobility- . BubR1 undergoes hyperphosphorylation during prometaphase. A, HeLa cells were synchronized by double thymidine block and released from the G 1 -S block. 10 h after release, cells were arrested at prometaphase by treatment with nocodazole at 37°C or taxol at 37°C for 2 h. Mitotic cells were collected by mechanical shake-off. The obtained whole cell lysates were treated with or without 300 units (70 l) of protein phosphatase (PPase; New England Biolabs) at 30°C for 45 min and then analyzed by immunoblotting with anti-BubR1 antibody. P-, phosphorylated. B, synchronized HeLa cells were released from the G 1 -S block at 0 h. 10 h after release, mitotic cells were collected by mechanical shake-off, and cells were arrested at prometaphase and metaphase by treatment with nocodazole for 2 h and MG132 for 2 h, respectively. When cells were treated with both nocodazole and MG132, cells were arrested at prometaphase. Then mitotic cells were collected by mechanical shake-off (noc., MG132, and noc. ϩ MG132). The obtained whole cell lysates were subjected to immunoblotting with anti-BubR1 antibody. A, HeLa cells expressing GFP-histone H2B were synchronized and transfected with siRNAs for Plk1 and/or Aurora B. 9 h after release from the G 1 -S boundary, cells were treated with nocodazole or taxol for 2 h. Then mitotic cells were collected by mechanical shake-off, and whole-cell extracts were subjected to immunoblot analysis with antibodies to Plk1, Aurora B, BubR1, and ␣-tubulin. P-. phosphorylated. B, add-back of Plk1. HeLa GFP-histone H2B cells were transfected with HA-mouse Plk1 before the first thymidine block, and Plk1 siRNA treatment was performed after the first thymidine block. Cells were released from G 1 -S block and treated with nocodazole (noc.) for 2 h at 10 h after release. Mitotic cells were collected by mechanical shake-off, and wholecell extracts were subjected to immunoblot analysis with antibodies to Plk1 and BubR1. BubR1 and P-BubR1 bands in the right lane, Plk1 siRNA ϩ HAmouse Plk1, were obtained by a longer exposure time. An HA-tagged Plk1 band in the same lane was obtained by a shorter exposure time. Endo, endogenous. MAY 18, 2007 • VOLUME 282 • NUMBER 20 shifted band (Fig. 4A, Aurora B siRNA, nocodazole and taxol). Nocodazole treatment, but not taxol treatment, in Aurora B-deficient cells resulted in prometaphase arrest (data not shown). This is consistent with the previous observation that Aurora B is specifically involved in the tension checkpoint mechanism, which is caused by taxol, but not nocodazole, treatment (30,31,35,36). Thus, the Aurora B-deficient cells have exited from M phase in taxol treatment. Therefore, it is not surprising that the taxol-induced mobility shifted bands of BubR1 disappear in the Aurora B-deficient cells and that the amount of BubR1 protein decreases in these cells as BubR1 is suggested to be degraded after metaphase. That only the partial inhibition of the nocodazole-induced mobility shifted bands was seen in the Aurora B-deficient cells suggests that Aurora B may not be directly responsible for the hyperphosphorylation of BubR1 during prometaphase. We next examined whether add-back of Plk1 leads to the reappearance of the mobilityshifted band of BubR1 in mitotically arrested cells. Synchronized HeLa GFP-histone H2B cells were transfected with Plk1 siRNA (for human Plk1) and HA-tagged mouse Plk1. Immuno-blot analysis showed that the add-back of HA-tagged mouse Plk1 is able to partially recover the mobility shifted band of BubR1 (Fig. 4B). Because endogenous Plk1 was barely seen in the HA-Plk1 add-back experiment (Fig. 4B, Plk1siRNA ϩ  HA-Plk1), the recovery of the mobility shifted BubR1 band was not due to the insufficient down-regulation of endogenous Plk1. Therefore, this result demonstrates that the disappearance of the mobility-shifted bands of BubR1 in Plk1 siRNA-treated cells results from the specific down-regulation of Plk1. Taken together, our results suggest that Plk1 is responsible for the hyperphosphorylation of BubR1 during prometaphase.

Plk1 Facilitates Chromosome Alignment
Plk1 Phosphorylates BubR1 in Vitro-To address the possibility that BubR1 is a substrate for the kinase activity of Plk1, we performed in vitro kinase assays with immunoprecipitated Myc-tagged BubR1 and immunoprecipitated HA-tagged Plk1 on the same beads. BubR1 underwent marked phosphorylation when incubated with wild-type Plk1, whereas only a low level of phosphorylation of BubR1 was observed with kinase-dead Plk1 (Plk1 KR) (Fig. 5A, left). Essentially the same low level of phos-FIGURE 5. Plk1 phosphorylates (P-) BubR1. A and B, HA-tagged Plk1 constructs, Myc-tagged BubR1 constructs were transiently overexpressed in COS7 cells, and whole-cell extracts were subjected to immunoprecipitation with both antibodies to HA and Myc coupled to the same beads. The resulting precipitates were subjected either to immunoblot (IB) analysis with antibody to HA and Myc or to an in vitro kinase assay. C, Myc-tagged BubR1 (WT, 2E, or 2EKR) was transiently overexpressed in COS7 cells, and whole-cell extracts were subjected to immunoprecipitation with antibody to Myc. The resulting precipitates were subjected either to immunoblot analysis with antibody to Myc or to an in vitro kinase assay. D, HeLa GFP-histone H2B cells were transfected with BubR1 WT (sRr) or BubR1 2A (sRr), then synchronized and treated with BubR1 siRNA. 10 h after release from G 1 -S block, cells were arrested at prometaphase by nocodazole (noc.) treatment, and then mitotic cells at prometaphase were collected by mechanical shake-off. The obtained whole cell lysates were subjected to immunoblotting analysis with anti-BubR1 antibody. In another series of experiments, similar amounts of BubR1 WT and BubR1 2A were loaded, because the expression level of BubR1 2A was nearly identical to that of BubR1 WT at the single cell level (see E). E, HeLa GFP-histone H2B cells were transfected with BubR1 WT (sRr) or BubR1 2A (sRr), then synchronized and treated with BubR1 siRNA. 10 h after release from G 1 -S block, cells were arrested at prometaphase by nocodazole treatment, and cells were fixed and stained with anti-BubR1 antibody. DNA was stained by Hoechst. Each image was taken in the same exposure time (1.0 and 3.0 s). phorylation of BubR1 was seen in the absence of Plk1 (Fig. 5A,  left). To eliminate the possibility that this low level phosphorylation of BubR1 was due to nonspecific binding of unknown kinases to the beads, we used BubR1 KR (a kinase-dead form of BubR1). The result has shown no phosphorylation of BubR1 KR in the absence of Plk1 (Fig. 5A, right). Thus, a low level of BubR1 phosphorylation in the absence of Plk1 is mostly due to autophosphorylation by BubR1. With immunoprecipitated Myctagged Aurora B, BubR1 did not undergo marked phosphorylation (data not shown). Thus, Plk1, but not Aurora B, is able to phosphorylate BubR1 directly. There are four putative Plk1 phosphorylation sites which match a reported consensus motif for Plk1 phosphorylation (37) in the kinase domain of BubR1. When an in vitro kinase assay was performed with Plk1 and wild-type BubR1 or a mutant form of BubR1 (BubR1 2A) in which two (Thr-792 and Thr-1008) of the four putative phosphorylation sites for Plk1 were replaced by alanine, only wildtype BubR1 underwent marked phosphorylation (Fig. 5B). BubR1 2A exhibited only a low level of autophosphorylation (Fig. 5B). This result suggests that these two threonines are Plk1 phosphorylation sites. Then we constructed another mutant of BubR1 (BubR1 2E) in which the two Plk1 sites were replaced by glutamic acid. In an autophosphorylation assay BubR1 2E exhibited a much higher autophosphorylation ability than wildtype BubR1 (Fig. 5C). In addition, we constructed another mutant of BubR1, a kinase-defective mutant of BubR1 2E (BubR1 2EKR). BubR1 2EKR exhibited a lower autophosphorylation ability than BubR1 2E (Fig. 5C). It is likely, therefore, that Plk1 phosphorylation of BubR1 increases its kinase activity.
To examine whether the identified two phosphorylation sites in BubR1 are responsible for its mobility shift in M phase, we performed a depletion and add-back experiment. We constructed those mutants of BubR1 (BubR1 WT (sRr) and BubR1 2A (sRr)) that were resistant to BubR1 siRNA by mutating the target sequence recognized by BubR1 siRNA without altering their amino acid sequence. HeLa cells were transfected with BubR1 WT (sRr) or BubR1 2A (sRr) then synchronized and treated with BubR1 siRNA. 10 h after release from G 1 -S block, cells were arrested at prometaphase by nocodazole treatment, and then cells at prometaphase were further collected by mechanical shake-off. Immunoblot analysis with anti-BubR1 antibody showed that BubR1 WT (sRr), but not BubR1 2A (sRr), displayed the mobility shift (Fig. 5D, left). Endogenous BubR1 almost completely disappeared by this BubR1 siRNA treatment (Fig. 5D, left). Although there was less BubR1 2A than BubR1 WT in total (see Fig. 5D, left), which is due to the higher transfection efficiency of BubR1 WT, the expression level of BubR1 2A was nearly identical to that of BubR1 WT at the single cell level (Fig. 5E), the level similar to or slightly higher than that of endogenous BubR1 in BubR1 siRNA-untreated cells (Fig. 5E). In another series of experiments we loaded similar amounts of BubR1 2A and BubR1 WT onto the same gel. Immunoblot analysis clearly showed that BubR1 WT, but not BubR1 2A, displayed the mobility shift (Fig. 5D, right). These results strongly suggest that the identified two Plk1 phosphorylation sites in BubR1 are responsible for the mobility shift of BubR1 in vivo.
Phosphorylation of BubR1 Does Not Affect Its Ability to Activate the Mitotic Checkpoint-To gain an insight into physiological roles of phosphorylation of BubR1, we first performed depletion and add-back experiments to test these mutant forms of BubR1 for their ability to activate mitotic checkpoint in the presence of nocodazole. We constructed those mutants of BubR1 (Myc-BubR1 WT (sRr), Myc-BubR1 2A (sRr), and Myc-BubR1 2E (sRr)) that were resistant to BubR1 siRNA or pSuper-BubR1 (see Fig. 7, A and C). HeLa cells were transfected with Myc-BubR1 WT (sRr), Myc-BubR1 2A (sRr), or Myc-BubR1 2E (sRr) before the first thymidine block, and after the first thymidine block cells were transfected with siRNA for BubR1. 9 h after release from the second thymidine block, cells were treated with nocodazole for 2 h, and cells were fixed and stained with anti-Myc antibody and anti-␣-tubulin antibody. ϳ100 mitotic cells were counted in each case. In control cells, in which mock treatment instead of BubR1 siRNA transfection was given, about 30% of the cells were arrested in prometaphase after 2 h of nocodazole treatment (Fig. 6, upper, BubR1 siRNA; Ϫ). In BubR1 siRNA-treated cells, only about 6% of the cells were in prometaphase (Fig. 6, upper, BubR1  siRNA; ϩ), confirming the requirement of BubR1 function in mitotic spindle checkpoint. In any of BubR1 WT (sRr), BubR1 2A (sRr), or BubR1 2E (sRr) add-back cells, about 30% of the cells were arrested in prometaphase (Fig. 6, lower). This result has clearly shown that Plk1 phosphorylation sites are not important for the spindle checkpoint function of BubR1. It has been shown that BubR1 acts as a mitotic checkpoint protein to bind to Cdc20 to inhibit APC/C (14,38,39). Our immunoprecipitation experiment showed that BubR1 2A and BubR1 2E as well as BubR1 WT were able to bind to Cdc20 (data not shown). This result also suggests that Plk1 phosphorylation sites in BubR1 are not important for the spindle checkpoint function.
Plk1 Phosphorylation of BubR1 Is Involved in Chromosome Alignment-Then we performed further depletion and add-back experiments to examine possible roles of Plk1 phosphorylation of BubR1 in its function as a regulator of chromosome alignment. HeLa cells were co-transfected with pSuper-BubR1 and with Myc-BubR1 WT (sRr), Myc-BubR1 2A (sRr), or Myc-BubR1 2E (sRr), and then the cells were synchronized. 12 h after release from thymidine block, cells were fixed and stained with anti-Myc antibody and anti-␣-tubulin antibody. ϳ100 mitotic cells were counted in each case. Add-back of Myc-BubR1 WT (sRr) or Myc-BubR1 2E (sRr) was able to rescue the chromosome alignment defect in BubR1-deficient cells. About 50% of mitotic cells were at normal There are statistically significant differences between WT (sRr) and 2A (sRr) or between 2A (sRr) and 2E (sRr) (single asterisk, p Ͻ 0.05, t test). B, synchronized HeLa cells were transfected with BubR1 siRNA and Myc-tagged BubR1 mutants (BubR1 WT (sRr), 2A (sRr), and 2E (sRr)). 9 h after release from the thymidine block, cells were treated with nocodazole (noc.) for 2 h, and then 3 h after nocodazole release, cells were fixed and stained anti-Myc antibody and anti-␣-tubulin antibody. Upper, the number of unaligned chromosomes in a cell was counted in ϳ40 mitotic cells in each case. The histogram shows the distribution of cells with unaligned chromosomes in each case. We excluded mitotic cells whose spindle is abnormal. Lower, a typical picture is shown in each case. ␣-Myc (green) and ␣-tubulin (red). C, HeLa cells were transfected with BubR1 mutants and BubR1 siRNA during a double thymidine block. 10 h after release from G 1 -S block, cells were treated with MG132 for 2 h. The whole cell extracts were subjected to immunoblot analysis with anti-BubR1 antibody. endo, endogenous. D, left, HeLa cells were synchronized and treated with or without BubR1 siRNA. 10 h after release from the G 1 -S boundary, the proteasome inhibitor MG132 was added, and 2 h after the addition of MG132, cells were fixed and stained for ␣-tubulin and DNA. Mitotic cells were examined for their chromosomes. More than 100 cells were counted at each point. Right, HeLa cells were synchronized and transfected with Myc-tagged BubR1 (WT (sRr), 2A (sRr), or 2E (sRr)), p-Super-BubR1, and p-Super-Plk1. 10 h after release from the G 1 -S boundary, the proteasome inhibitor MG132 was added, and 2 h after the addition of MG132, cells were fixed and stained for Myc, ␣-tubulin, and DNA. Mitotic cells expressing respective mutants at kinetochores were examined for their chromosomes. More than 100 cells were counted at each point, and 3 independent experiments were performed. We excluded mitotic cells whose spindle is abnormal. The green bars show the percentages of cells in normal metaphase, and red bars show the percentages of cells in abnormal prometaphase with misaligned chromosomes. Statistically significant differences between the two coupled samples are, respectively, indicated by a single asterisk (p Ͻ 0.05, t test), double asterisks (p Ͻ 0.01, t test), or triple asterisks (p Ͻ 0.001, t test). E and F, HeLa cells were synchronized and transfected with Myc-tagged BubR1 (WT (sRr), BubR1 2A (sRr), or 2E (sRr)) and pSuper-BubR1 and/or pSuper-Plk1. 10 h after release from G 1 -S block, cells were treated with MG132 for 2h, and then cells were fixed and stained with antibody to BubR1 or Myc. DNA was stained by Hoechst. Each image was taken in the same exposure time (1.0 and 3.0 s). G, the staining intensity was measured in ϳ20 cells in each sample and shown as relative intensity. The average intensity in WT (sRr) was set 1.0. H, HeLa cells were synchronized and transfected with Myc-tagged BubR1 (WT (sRr), 2E (sRr), or 2EKR (sRr)), p-Super-BubR1, and p-Super-Plk1. 10 h after release from G 1 -S boundary, the proteasome inhibitor MG132 was added, and 2 h after the addition of MG132, cells were fixed and stained for Myc, ␤-tubulin, and DNA. Mitotic cells expressing respective mutants at kinetochores were examined for their chromosomes. 30 cells were counted at each point, and 3 independent experiments were performed. We excluded mitotic cells whose spindle is abnormal. Green bars show the percentages of cells in normal metaphase, and red bars show the percentages of cells in abnormal prometaphase with misaligned chromosomes. Statistically significant differences between the two coupled samples are, respectively, indicated by a single asterisk (p Ͻ 0.05, t test) or double asterisks (p Ͻ 0.01, t test). A typical picture is shown in each case. Green, ␣-Myc; red, ␤-tubulin; blue, and DNA. The scale bar represents 5 m. metaphase with proper chromosome alignment. In contrast, in Myc-BubR1 2A (sRr) add-back cells only about 30% of mitotic cells were at normal metaphase (Fig. 7A, lower). Next, we compared the ability of each mutant or wild-type (WT) BubR1 to rescue the chromosome alignment defect in BubR1-deficient cells by analyzing the recovery process after release from nocodazole-induced prometaphase arrest. Synchronized HeLa cells were transfected with BubR1 siRNA and Myctagged BubR1 mutants (BubR1 WT (sRr), 2A (sRr), and 2E (sRr)). 9 h after release from thymidine block, cells were treated with nocodazole for 2 h, and then at 3 h after nocodazole release cells were fixed and stained anti-Myc antibody and anti-␣-tubulin antibody. Typical pictures are shown (Fig. 7B,  lower). The number of unaligned chromosomes in a cell was counted in ϳ40 mitotic cells in each case. When Myc-BubR1 WT (sRr) and Myc-BubR1 2E (sRr) were added back, only 10% and 17.5% of cells, respectively, contained more than 10 unaligned chromosomes (Fig.  7B, upper, WT, 2E). In contrast, when Myc-BubR1 2A (sRr) was added back, 43.5% of cells had more than 10 unaligned chromosomes (Fig. 7B, upper, 2A). To assess chromosome misalignment at metaphase in more detail, we performed depletion and add-back experiments in the presence of the proteasome inhibitor, MG132, because it has been reported that BubR1 repression prevents chromosome alignment in the presence of proteasome inhibitor, MG132 (30). When synchronized cells were treated with MG132 at prometaphase, most cells were arrested at metaphase. Immunoblot analysis showed that endogenous BubR1 was almost completely depleted, and the expression of each BubR1 mutant was at similar levels in these addback experiments in which synchronized HeLa cells were transfected with BubR1 siRNA and Myc-tagged BubR1 mutants (BubR1 WT (sRr),  (Fig. 7C). Under our conditions, about 80% of the control cells were in metaphase with proper chromosome alignment (Fig. 7D, left, control). In contrast, only about 40% of the cells were in normal metaphase, and about 60% of cells showed misaligned chromosomes in BubR1-deficient cells (Fig. 7D, left, BubR1 siRNA). When Myc-BubR1 WT (sRr) or Myc-BubR1 2E (sRr) was added back only about 30% of cells displayed chromosome misalignment, and others were in normal metaphase (Fig. 7D, right, BubR1 WT (sRr), BubR1 2E (sRr)). In contrast, the add-back of Myc-BubR1 2A (sRr) resulted in chromosome misalignment in more than 50% of cells (Fig. 7D, right, BubR1 2A (sRr)). All these results taken together suggest that BubR1 2A (sRr), an unphosphorylatable mutant of BubR1, failed to rescue the chromosome alignment defect in BubR1-deficient cells. We then down-regulated both Plk1 and BubR1. When wild-type BubR1 (sRr) or BubR1 2E (sRr) was added-back, BubR1 2E (sRr) was able to rescue the chromosome alignment defect in Plk1/BubR1-deficient cells more efficiently than wild-type BubR1 (sRr) (Fig. 7D, right, pS-Plk1ϩBubR1 WT (sRr) and pS-Plk1ϩBubR1 2E (sRr)). To eliminate the possibility that these differences resulted from the variations in the degree of synchrony, we examined the mitotic index in add-back experiments in which MG132 was added. The result has shown that the degree of synchrony, the rate of cell cycle progression, and the response to MG132 are nearly identical in each case (supplemental Fig. S1). Immunofluorescence analysis showed that under the conditions used (pS-BubR1), endogenous BubR1 was decreased to the undetectable level (Fig. 7E), and the expression levels of each form of the Myc-tagged BubR1 constructs were similar at the single cell level (Fig. 7F). To further demonstrate this, we measured the staining intensity in ϳ20 cells in each sample. The result has shown that the standard deviation in each sample is small and the intensity is nearly identical among BubR1 mutants (Fig.  7G). Although BubR1 2E is expressed at a slightly lower level than others, this does not affect our conclusion that BubR1 2E is more effective than BubR1 WT in the Plk1/BubR1-defficient cells. Moreover, although exogenous BubR1 constructs were overexpressed about 5-10-fold compared with the level of endogenous BubR1 (see Fig. 7E), more than 10-fold overexpression of any of these constructs in control HeLa cells did not interfere with normal cell cycle, the cell cycle progressed normally, and chromosome alignment and segregation occurred normally (supplemental Fig. S2).

DISCUSSION
Consistent with previous reports (12,13), our analysis on the kinetics of recovery process after release from prometaphase arrest by nocodazole treatment confirms that Plk1 regulates the formation of functional bipolar spindle and the stable microtubule-kinetochore attachment. We show that Aurora B siRNA did not completely inhibit the nocodazole treatment-induced appearance of the mobility shifted band of BubR1 (Fig. 4A, Aurora B siRNA, nocodazole and taxol). Ditchfield et al. (30) have recently reported that inhibition of Aurora B prevents phosphorylation of BubR1. The reason for this apparent difference is not known at present. However, in our experiments we used the mechanical shake-off after the drug-induced synchronization and were able to collect specifically prometaphase-arrested cells, in which almost all BubR1 existed as phosphorylated forms. In contrast, in their experiment only a portion (about half) of BubR1 displayed a mobility-shifted band. Therefore, partial inhibition might have been interpreted as complete inhibition in their experiment. However, their result is in accord with ours in that inhibition of Aurora B reduces phosphorylation of BubR1. It is likely that Aurora B indirectly regulates the phosphorylation state of BubR1. Although it has been reported that Aurora B and BubR1 are functionally related in chromosome alignment (15,30,40), our analyses on BubR1 localization and hyperphosphorylation suggest that Aurora B is an indirect regulator of BubR1, and Plk1 is a direct regulator of BubR1 during chromosome alignment. Plk1 and Aurora B at kinetochores of unaligned chromosomes may cooperatively regulate chromosome alignment.
Recently, Gorbsky and co-workers (4) group and Fang and Wong (5) have reported that Plk1 is responsible for the generation of the 3F3/2 phospho-epitope at kinetochores that are not under tension. BubR1 localizes to kinetochores which are unattached with microtubules or not under tension (24,(41)(42)(43)(44). Our finding that Plk1 is responsible for the hyperphosphorylation of BubR1 in the mitotically arrested cells by the treatment with nocodazole or taxol is in good agreement with these observations. We have shown that Plk1 associates with BubR1 in its C-terminal polo-box domain. It has previously been reported that the polo-box domain of Plk1 is responsible for control of its subcellular localization and recognition of its substrates (26 -28, 45). Moreover, the polo-box domain is shown to be important for chromosome congression (46). Thus, it is likely that Plk1 phosphorylates many other substrates at kinetochores that are unattached with microtubules or not under tension. Identification of Plk1 substrates at kinetochores should facilitate further studies for understanding of molecular mechanisms of chromosome alignment.
We employed add-back experiments to examine the role of Plk1 phosphorylation of BubR1 in spindle checkpoint and in chromosome alignment. Our results suggest that Plk1 phosphorylation of BubR1 does not affect the function of BubR1 as a mitotic checkpoint protein. This is in good agreement with previous observations that Plk1 is not involved in the proper activation of the spindle checkpoint (Refs. 12 and 45 and Fig. 1A). It is noted that in our add-back experiments in Plk1/BubR1-deficient cells, the population of the cells that show the spindle defects (ϳ15%) was not significantly reduced by the add-back of BubR1 2E. 3 We, thus, speculate that phosphorylation on BubR1 by Plk1 does not play an important role in spindle formation. Our experiments with Plk1-phosphorylation-site mutants of BubR1 strongly suggested that Plk1 regulates the action of BubR1 in chromosome alignment through phosphorylation. Thus, Plk1 may function in preparing a proper platform for chromosome alignment through phosphorylation of its various substrates at kinetochores.
It has previously been reported that BubR1 associates with the kinetochore motor protein CENP-E, which is shown to be essential for both chromosome alignment and mitotic checkpoint (18,19) and that CENP-E regulates BubR1 activity in spindle checkpoint signaling (20,21). Our results indicate that Plk1 phosphorylation of BubR1 increases the kinase activity of BubR1 and facilitates proper chromosome alignment during prometaphase. We speculate that the BubR1-CENP-E complex has two functions; one is to activate and silence the spindle checkpoint signaling, and the other is to facilitate chromosome alignment. Molecular mechanisms by which BubR1 regulates chromosome alignment as well as mitotic checkpoint should be elucidated in future studies.