Bubr1 Localizes to Centrosomes and Suppresses Centrosome Amplification Via Regulating Plk1 Activity in Interphase Cells

Oncogene (2009) 28, 2806–2820 & 2009 Macmillan Publishers Limited All rights reserved 0950-9232/09 $32.00 www.nature.com/onc ORIGINAL ARTICLE BubR1 localizes to centrosomes and suppresses centrosome ampliﬁcation via regulating Plk1 activity in interphase cells

H Izumi1, Y Matsumoto1, T Ikeuchi2, H Saya3, T Kajii4 and S Matsuura1

1Department of Radiation Biology, Research Institute for Radiation Biology and Medicine (RIRBM), Hiroshima University, Hiroshima, Japan; 2Division of Molecular Cytogenetics, Medical Research Institute, Tokyo Medical and Dental University, Tokyo, Japan; 3Institute for Advanced Medical Research, Keio University School of Medicine, Tokyo, Japan and 4Hachioji, Tokyo, Japan

BubR1 is a critical component of the mitotic checkpoint mitosis, eukaryotic cells employ a dynamic architecture that delays the onset of anaphase until all chromosomes of kinetochores, centrosomes and spindle microtubules have established bipolar attachment to the microtubules. (Nigg, 2002; Rieder and Khodjakov, 2003). We previously reported that mutations of the BUB1B The spindle attachment to kinetochore is robustly gene (encoding BubR1) caused premature chromatid monitored by a mitotic spindle assembly checkpoint separation (PCS) syndrome, a condition characterized (Cleveland et al., 2003; Kops et al., 2005; Musacchio by constitutional aneuploidy and a high risk of childhood and Salmon, 2007), which prevents chromosome segre- cancer. We here report that the cells from PCS syndrome gation error by blocking the onset of anaphase until all patients have loss of regulation of the centrosome chromosomes make proper attachments to mitotic duplication machinery, resulting in centrosome amplifica- spindles (Yu, 2002; Kops et al., 2005). BubR1 is a tion and multipolar mitosis. PCS syndrome cells show central component of the spindle assembly checkpoint, increased activity of Polo-like kinase 1 (Plk1), which localizes to the kinetochore in early metaphase whose knockdown suppresses centrosome amplification. and prevents premature separation of sister chromatids. BubR1 localizes to centrosomes, physically interacts We and another group recently reported constitutional with Plk1 and inhibits Plk1 phosphorylation and its mutations of the BUB1B gene (encoding BubR1) in kinase activity during interphase. These results unravel patients with premature chromatid separation (PCS) a crucial role of BubR1 in preventing centrosome syndrome (Hanks et al., 2004; Matsuura et al., 2006), reduplication through negative regulation of Plk1 in also called mosaic variegated aneuploidy (MVA) interphase cells. syndrome. PCS syndrome is a rare autosomal recessive Oncogene (2009) 28, 2806–2820; doi:10.1038/onc.2009.141; disorder characterized by premature separation of sister published online 8 June 2009 chromatids of all chromosomes in more than 50% of mitotic lymphocytes and a variety of mosaic aneuploi- Keywords: centrosome; BubR1; Plk1; chromosome dies, especially trisomies and monosomies (Kajii et al., instability syndrome 1998). Infants with PCS syndrome show growth retardation, severe microcephaly and a high risk of malignancy, such as Wilms tumor and rhabdomyo- sarcoma (Kawame et al., 1999; Kajii et al., 2001; Plaja et al., 2001; Jacquemont et al., 2002). We previously Introduction reported that the cultured skin fibroblasts from two unrelated infants with PCS syndrome showed defects in Faithful chromosome segregation is maintained by a the spindle assembly checkpoint (Matsuura et al., 2000). mitotic spindle checkpoint, a surveillance mechanism Consequently we identified monoallelic BUB1B muta- that delays the onset of anaphase until all chromosomes tions in seven families with PCS syndrome (Matsuura have established bipolar attachment to the micro- et al., 2006). In each of these seven families there was tubules. Formation of a bipolar spindle is also essential one mutation that resulted in premature protein for chromosome stability, because a numerical abnorm- truncation or functional debilitation of the gene ality of the mitotic spindle poles leads to chromosome product. Although no second mutation was found in segregation errors, resulting in chromosome instability. the allele opposite the one in which a mutation was Thus, to ensure accurate chromosome segregation in detected, the number of transcripts of the second allele was moderately reduced. Thus, the cells in the patients Correspondence: Professor S Matsuura, Department of Radiation show >50% decrease in the expression of BubR1 Biology, Research Institute for Radiation Biology and Medicine protein (Matsuura et al., 2006). (RIRBM), Hiroshima University, Kasumi 1-2-3, Minami-ku, Here, we report that PCS syndrome cells show severe Hiroshima 734-8553, Japan. E-mail: [email protected] chromosome instability not only because of the mitotic Received 5 September 2008; revised 17 April 2009; accepted 28 April checkpoint defect but also because of the centrosome 2009; published online 8 June 2009 amplification. We also show that Polo-like kinase 1 BubR1 prevents Plk1 activity on centrosome H Izumi et al 2807 (Plk1) activity is upregulated in PCS syndrome Results cells, and ectopic expression of BubR1 restores abnormal phenotypes of centrosome amplification in PCS syndrome cells show severe chromosome instability PCS syndrome cells. Thus, BubR1 deficiency causes because of multipolar mitosis centrosome amplification as well as spindle assembly We first carried out a precise chromosome analysis of checkpoint defect. Moreover, BubR1 localizes to the patient’s lymphoblastoid cell line (LCL). The centrosomes, physically interacts with Plk1 and inhibits lymphoblastoid cells from a normal individual (TO) Plk1 phosphorylation and its kinase activity during showed no increase in aneuploidy (Figures 1a and b), interphase. These results suggest that centrosome- whereas mosaic aneuploidy including trisomies and associated BubR1 shields Plk1 from unscheduled monosomies was observed in 80% of a PCS patient’s activation during interphase. Thus, these results cells (MY1L; Figure 1c). This observation could reflect unravel a novel role of BubR1 in preventing centrosome the inaccurate chromosome segregation in mitosis reduplication through negative regulation of Plk1 in because of BubR1 deficiency (Baker et al., 2004; Hanks interphase cells. et al., 2004). In addition, a small fraction (20%) of the

100 100 MY1L TO lymphocyte 90 90 (control) n=44 80 80 n=50 70 70 60 60 50 50 40 40 30

30 Percent of cells Percent of cells 20 20 10 10 TO metaphase MY1L metaphase 0 0 ≤ ≥ ≤40 41 42 43 44 45 46 47 48 ≥49 40 41 42 43 44 45 46 47 48 49 Chromosome number per cell Chromosome number per cell

25 γ-tubulin γ-tubulin Plk1 Plk1 DAPI DAPI 20 TO MY1L

BubR1 15

α-tubulin mitosis 10

Percent of multipolar 5

TO MY1L 0 TO MY1L

25 KM (normal control) MY1 PCS1 KM MY1 PCS1 20 DAPI

BubR1 15 BubR1

mitosis 10 α-tubulin 5 Percent of multipolar -tubulin γ 0 KM MY1 PCS1 Figure 1 Premature chromatid separation (PCS) syndrome cells show severe chromosome instability and multipolar mitosis. (a) Normal lymphocytes do not show chromosome instability in metaphase. (b) Representative image of metaphase (arrows) in normal lymphocytes. Scale bar, 10 mm. (c) Lymphocytes from a PCS syndrome patient (MY1L) show severe chromosome instability in metaphase. In addition, 20% of MY1L cells are severely aneuploid, with chromosome number less than 40 (arrow). (d) Representative image of metaphase (arrowheads) in PCS lymphocytes. Scale bar, 10 mm. (e) Immunoblot of BubR1 in lymphocytes. BubR1 expression was reduced in the lymphocytes from a PCS patient (MY1L) as compared to normal control lymphocytes (TO). Immunoblot of anti-a- tubulin antibody served as a loading control. (f) Representative image of cytokinesis in a normal lymphocyte. Centrosome was stained with anti-g-tubulin antibody (red). Cytoplasm was stained with anti-Polo-like kinase 1 (Plk1) antibody (green). DNA was stained with 40,6-diamidino-2-phenylindole dihydrochloride (DAPI; blue). Scale bar, 10 mm. (g) Representative image of multipolar mitosis (cytokinesis) in a PCS lymphocyte. Scale bar, 10 mm. (h) The proportion of cells with multipolar mitosis in normal lymphocytes (TO) and PCS lymphocytes (MY1L). Approximately 17% of cells showed multipolar mitosis in PCS lymphocytes. For each cell line, X100 cells were examined. The data are shown as average±standard error determined from three separate experiments. (i) BubR1 expression was reduced in the fibroblasts from two unrelated PCS patients (MY1 and PCS1) compared to normal control fibroblasts (KM). Immunoblot of anti-a-tubulin antibody served as a loading control. (j) Representative image of mitosis in normal fibroblast cells (KM). Centrosomes were stained with anti-g-tubulin antibody (red). Kinetochores were stained with anti-BubR1 antibody (green). DNA was stained with DAPI (blue). Scale bar, 10 mm. (k and l) Representative images of multipolar mitosis in two fibroblast cell lines from PCS syndrome patients (MY1 and PCS1). Scale bars, 10 mm. (m) The frequency of cells with multipolar mitosis in PCS syndrome fibroblasts. For each cell line, X100 cells were examined. The data are shown as average±standard error determined from three separate experiments.

Oncogene BubR1 prevents Plk1 activity on centrosome H Izumi et al 2808 MY1L cells was aneuploid with a chromosome number of regulation of the centrosome duplication machinery. less than 40 (Figures 1c and d). Such severe aneuploidy Hydroxyurea (HU), a DNA synthesis inhibitor, was not observed in the control cells. We inferred that arrests the cell cycle at the G1-S transition and these severely aneuploid cells with chromosome number allows centrosome duplication but not centrosome o40 were not due to the spindle assembly checkpoint reduplication in normal human cells. However, if the defect but due to abnormal mitosis such as multipolar centrosome duplication machinery is deregulated, mitosis. We therefore analysed the mitotic cells by centrosome reduplication (centrosome ampliﬁcation) immunostaining. The PCS lymphoblastoid cells is induced with prolonged HU exposure (Balczon (MY1L), in which BubR1 expression was reduced et al., 1995). We therefore exposed HU to the normal (Figure 1e), showed 17% of cells in multipolar cells and PCS syndrome cells, and the cells were mitosis with abnormal numbers of centrosomes (Figures examined by immunostaining. In the normal cells, 1g and h). On the other hand, the cells from a normal only a low level of centrosome reduplication was control (TO) showed no increase in aberrant mitosis detected following HU treatment (Figure 2i), demon- (Figures 1f and h). The immortalized ﬁbroblast cells strating that centrosome duplication was tightly regu- (termed MY1 and PCS1) from two unrelated patients lated with DNA synthesis. On the other hand, the with PCS syndrome (Figure 1i) also exhibited an frequency of centrosome reduplication was increased in increase in multipolar mitosis with abnormal num- PCS syndrome cells (Figure 2i), suggesting that bers of centrosomes, similar to those in the lympho- the centrosome duplication machinery is deregulated blastoid cells, MY1L (Figures 1j–m). Thus, PCS in PCS syndrome cells (see Supplementary Movies syndrome cells have an abnormal number of centro- S1 and S2). somes in mitosis.

BubR1 localizes to centrosomes and suppresses Centrosome duplication machinery was deregulated centrosome amplification in PCS syndrome cells To examine the role of BubR1 in centrosome function, To analyse whether PCS syndrome cells already have we transferred into PCS1 cells a whole chromosome 15 abnormal numbers of centrosomes in interphase, we (Ch.15) or the minimal chromosomal segment performed immunostaining experiments using anti-g- (BubR1), both of which contain the BUB1B locus, and tubulin and anti-centrin 2 antibodies. PCS lymphocytes analysed their centrosome profiles and multipolar had an increased number of centrosomes in inter- mitosis ratio. The chromosome-transferred PCS1 cells phase (Figures 2c and d), whereas normal lymphocytes showed normal expression level of BubR1 (Figure 3a) did not show such a phenotype (Figures 2a and b). and restoration of abnormal number of centrosomes We next examined the fibroblast cells from the two (Figures 3b and c) and multipolar mitosis (Supplemen- patients. Most of the fibroblast cells from a normal tary Figure S1). These results suggest that BubR1 is individual (termed KM) had one or two centrosomes somehow involved in centrosome functions. Therefore, (two or four centrioles, respectively) in interphase, we examined subcellular localization of BubR1 in and the frequency of centrosome amplification (three normal cells (KM). Interestingly, BubR1 localized to or more centrosomes) was only 3% (Figures 2e and h). centrosomes in interphase cells, but such a signal was In contrast, the fibroblasts from the two patients (MY1 not seen in mitotic spindle poles (Figures 3d and e). and PCS1) showed centrosome amplification (Figures Centrosomal localization of BubR1 was also seen 2f–h). Centrosome fragmentation was not observed in in the other cell lines we examined, such as HeLa either the control cells or the PCS syndrome cells cells (Supplementary Figure S2) and HCT116 cells because centrin staining (a marker for individual (data not shown). To examine whether such BubR1 centrioles) showed two spots for every one g-tubulin signals show a genuine centrosomal component, we spot (Figures 2e–g). carried out centrosomal fractionation analysis using We next examined whether excessive duplication of normal cells. BubR1 was included in interphase centrosomes in PCS syndrome cells was due to loss g-tubulin fractions but not in mitotic g-tubulin fractions

Figure 2 Centrosome duplication machinery is deregulated in premature chromatid separation (PCS) syndrome cells. (a) Representative image of centrosome staining in normal lymphocyte. Scale bar, 10 mm. Centrosomes (arrows) were stained with anti-g-tubulin antibody (red). DNA was stained with 40,6-diamidino-2-phenylindole dihydrochloride (DAPI; blue). (b) Centrosome profile in normal lymphocytes. For each cell line, X100 cells were examined. The data are shown as average±standard error determined from three separate experiments. (c) Representative image of amplified centrosomes (arrowheads) in PCS lymphocyte (MY1L). (d) Centrosome profile in PCS lymphocytes. (e) Representative images of centrosome (arrows) in normal control fibroblasts (KM). Insets show the magnified images of centrosomes. Scale bar, 10 mm. Centrosomes were stained with anti-g-tubulin antibody (red) and centrioles were stained with anti-centrin 2 antibody (green).(f and g) Representative images of centrosome amplification in two PCS syndrome fibroblasts (MY1 and PCS1). Arrowheads show amplified centrosomes. PCS syndrome fibroblasts showed centrosome amplification but not centrosome fragmentation. Insets show magnified images of centrosomes. (h) Summary of centrosome profiles in KM, MY1 and PCS1 cells. The ratios of centrosome amplification in two PCS syndrome cells showed fourfold increase as compared to that of control cells. For each cell line, X100 cells were examined. The data are shown as average±standard error determined from three separate experiments. (i) Summary of centrosome profiles in KM, MY1 and PCS1 cells under hydroxyurea (HU) treatment.

Oncogene BubR1 prevents Plk1 activity on centrosome H Izumi et al 2809 (Figures 3f, g and 4c). We then examined centro- 4c; Supplementary Figure S3A), whereas the chromosome localization of BubR1 in PCS1 cells and the some-transferred cells showed increased centrosome chromosome 15-transferred cells. PCS1 cells showed a signals for BubR1 (Figures 3i, j and 4c; Supplementary reduced signal for BubR1 on centrosomes (Figures 3h, Figure S3A).

80 80 γ-tubulin γ-tubulin 70 DAPI DAPI 70 60 60 50 50 40 40 30 30 20 20 Percent of cells 10 10 0 0 12≥3 12≥3 Centrosome number per cell Centrosome number per cell

KM (control) γ-tubulin Centrin 2 Merge /DAPI

MY1 γ-tubulin Centrin 2 Merge /DAPI

PCS1 γ-tubulin Centrin 2 Merge /DAPI

80 70 KM MY1 PCS1 60 50 40 30 20 Percent of cells 10 0 12≥312≥312≥3 Number of centrosome per cell

80 70 KM MY1 PCS1 (HU) (HU) (HU) 60 50 40 30 20 Percent of cells 10 0 12≥312≥312≥3 Number of centrosome per cell

Oncogene BubR1 prevents Plk1 activity on centrosome H Izumi et al 2810 To determine the BubR1 domain for centrosome activity is upregulated in PCS syndrome cells. Therefore, localization, we constructed several green fluorescent we addressed Plk1 activity in PCS syndrome cells protein (GFP)-fused BubR1 deletion mutants, trans- by an in vitro kinase assay (Figure 4b). The Plk1 fected them into normal cells and analysed their substrate GST-Cdc25C (Barr et al., 2004; van Vugt and centrosome localizations. The full-length BubR1 (wild Medema, 2005) was added to the immunoprecipitates by type), N-terminal deletion mutant (DN), C-terminal anti-Plk1 antibody. We found that the serine residue deletion mutant (DKinase) and kinase-dead mutant with (Ser198) in GST-Cdc25C was highly phosphorylated in K795R (KD) were able to localize to the centrosomes PCS1 cells as compared to that of the control cell lines, (Figure 3k; Supplementary Figure S4). On the other suggesting that PCS syndrome cells have high Plk1 hand, the deletion mutant of kinase only was not able to activity, and that the increased phosphorylated form of localize. These results suggest that the central region of Plk1 reflects Plk1 activity (Toyoshima-Morimoto et al., BubR1 (residues 240–639) is important for centrosomal 2002). localization. We then transfected these deletion mutants To further confirm that phospho-Plk1 (active Plk1) into PCS1 cells and analysed the number of cells with localizes to centrosomes, we carried out additional three or more centrosomes. The PCS1 cells transfected centrosomal fractionation analysis using normal cells, with the mutants that were able to localize to centro- PCS1 cells and PCS1-BubR1 cells. We also examined somes showed restoration of an abnormal number of the immunoblots of total cell lysates in addition to that centrosomes (Figure 3k). On the other hand, the cells of their centrosome fractions. Both proteins were transfected with the mutant of kinase only that was not included in the major peak of g-tubulin fractions of able to localize to centrosomes did not show such a normal cells, which is direct evidence that BubR1 and restoration (Figure 3k). phospho-Plk1 are genuine centrosomal components (Figure 4c). It was also found that an especially high PCS syndrome cells have high Plk1 activity on amount of phospho-Plk1 was included in the g-tubulin centrosomes fractions from PCS1 cells. These results support that To explore the mechanism of centrosome amplification PCS syndrome cells have high Plk1 activity on centro- by BubR1 deficiency, we examined the expression levels somes. It was also found that the band intensity of of several centrosomal proteins, including Plk1, Plk4, BubR1 in the g-tubulin fractions was not as high as that Aurora-A and p53, by immunoblotting (Figure 4a). The in the total cell lysate. These results suggest that only a expression levels of p53 and Aurora-A did not change small fraction of the total BubR1 pool localizes to when compared with those of control cells. Flow centrosomes and the remaining larger BubR1 pool cytometry analyses also showed that these cell lines distributes around cytoplasm during interphase. have similar cell-cycle profiles (data not shown). Plk4 Although Plk1 localizes to centrosomes throughout expression was slightly decreased in PCS syndrome cells. interphase, Plk1 has a maximum activity around G2-M Conversely, we found that the phosphorylated form of phases. Thus, active Plk1 was localized to duplicated Plk1 was higher in PCS syndrome cells (Figure 4a). The centrosomes. To investigate centrosomal Plk1 activity transfer of chromosome 15 normalized the increased during cell cycle in PCS syndrome cells, we used anti- band intensities of the phosphorylated form of Plk1 phospho-Plk1 antibody for immunostaining of centro- (Figure 4a). As it has been reported that the phosphor- somes. Our immunofluorescent studies showed that, in ylation of threonine residue 210 of Plk1 is well normal cells, active Plk1 signal did not detect the correlated with Plk1 activity (Barr et al., 2004; van immature, unduplicated centrosomes, but detected Vugt and Medema, 2005), our results suggest that Plk1 duplicated centrosomes (Figures 4d–f). On the other

Figure 3 BubR1 localizes to centrosomes and suppresses centrosome amplification. (a) Immunoblot analysis of premature chromatid separation (PCS) syndrome cells before and after chromosome 15 transfer. KM is a control fibroblast cell line. MY1 and PCS1 are PCS syndrome fibroblast cell lines. MY1-Ch.15 and PCS1-Ch.15 are the patient cells to which chromosome 15 was transferred. PCS1- BubR1 are the cells to which a minimal fragment of chromosome 15 was transferred, including the BUB1B locus. Immunoblot of anti- a-tubulin antibody was used as a loading control. (b) Centrosome profile of PCS1 cells and chromosome 15 transferred-PCS1 cells (PCS1-Ch.15 and PCS1-BubR1). For each cell line, X100 cells were examined. The data are shown as average±standard error determined from three separate experiments. (c) Summary of centrosome profiles in PCS1, PCS1-Ch.15 and PCS1-BubR1 cells under hydroxyurea (HU) treatment. (d) BubR1 localized to centrosomes in normal interphase cells. BubR1 (green) and g-tubulin (red) colocalized to centrosomes (arrows). DNA was stained with 40,6-diamidino-2-phenylindole dihydrochloride (DAPI; blue). Insets show the magnified images of centrosomes. Scale bar, 10 mm. (e) BubR1 did not localize to spindle poles (arrows) in normal mitotic cells. Scale bar, 10 mm. (f) BubR1 was included in g-tubulin fractions in normal interphase cells. (g) BubR1 was not included in g-tubulin fractions in nocodazole-treated cells. (h) BubR1 hardly localized to centrosomes in PCS syndrome cells. Arrowheads show amplified centrosomes. Insets show the magnified images of centrosomes. Scale bar, 10 mm. (i and j) Chromosome 15 transfer restored centrosome localization of BubR1 in PCS syndrome cells. Arrows show centrosomes. Insets show the magnified images of centrosomes. Scale bars, 10 mm. (k) Schema of GFP-BubR1 deletion mutants. The deletion mutants were transfected into normal cells (KM), and the centrosome localization domain was determined by immunostaining. For each transfectant, X25 cells were examined. Green fluorescent protein (GFP) alone, and GFP-kinase did not localize to centrosomes. Full-length, DN, DKinase and KD (kinase- dead) mutant proteins were able to localize to the centrosomes. These results suggest that residues 240-639 of BubR1 are necessary for the centrosomal localization of BubR1. These deletion mutants were also transfected into PCS1 cells, and the ratio of centrosome amplification was determined by immunostaining. For each transfectant, X50 cells were examined. The data are shown as an average of three separate experiments.

Oncogene BubR1 prevents Plk1 activity on centrosome H Izumi et al 2811 hand, in PCS syndrome cells, active Plk1 signal detected active Plk1 signal (Figures 4d–f). These results suggest both unduplicated and duplicated centrosomes with that Plk1 has already been activated during interphase intense signals (Figure 4d–f). Transfer of chromosome in PCS syndrome cells and that BubR1 may negatively 15 into PCS syndrome cells restored such an abnormal, regulate Plk1 activity in interphase.

80 70 PCS1 PCS1-Ch.15 PCS1-BubR1 60 50 40 30 20 Percent of cells 10 0 KM MY1 MY1-Ch.15KM PCS1 PCS1-Ch.15PCS1-BubR1 12 ≥3 12≥3 12≥3 BubR1 80 70 PCS1 PCS1-Ch.15 PCS1-BubR1 (HU) (HU) (HU) α-tubulin 60 50 40 30 20 Percent of cells 10 0 12 ≥3 12≥3 12≥3 Number of centrosomes per cells

γ-tubulin BubR1 Merge/DAPI γ-tubulin BubR1 Merge/DAPI KM interphase KM mitosis

Centrosome fractions : Asynchronous Centrosome fractions : Nocodazole-arrested 12345678Frac. No. 12345 6 7 8 Frac. No.

BubR1 BubR1

Plk1 Plk1

γ -tubulin γ -tubulin

Bottom Top Bottom Top

γ-tubulin BubR1 Merge/DAPI PCS1

Bub3 binding (365-426) Cdc20 binding (490-560) MAD3 homology (49-194) Kinase domain (786-919)

hBubR1 1050 a.a. Centrosome 240 639 Centrosome PCS1-Ch.15 ampliﬁcation localization inPCS1cells (%) GFP - 17.6 1050 WT GFP + 8.3 240 ΔN GFP + 8.2 640 Kinase GFP - 15.6 PCS1-BubR1 639 ΔKinase GFP + 9.5

KD GFP + 9.7 K795R

Oncogene BubR1 prevents Plk1 activity on centrosome H Izumi et al 2812 BubR1 physically interacts with Plk1 centrosomal localization status of BubR1 in interphase The functional relationship between BubR1 and Plk1 (Figure 5f). Control siRNA-treated cells showed high suggests a physical interaction of these proteins. To intensities of BubR1 signals on centrosomes (Figures 5g examine whether BubR1 interacts with Plk1, we and h). On the other hand, the Plk1-knockdown cells performed an immunoprecipitation experiment using showed almost abolished signal intensities of BubR1 on anti-BubR1 and anti-Plk1 antibodies. BubR1 was centrosomes (Figures 5g and h). We also examined this detected in the immunoprecipitates by using anti-Plk1 experiment in the presence of HU, and we obtained antibody (Figure 5a), and, conversely, Plk1 was detected similar result (data not shown). These results show that in the immunoprecipitates using anti-BubR1 antibody in Plk1 is necessary for BubR1 to localize to centrosomes asynchronously growing cells (Figure 5b). These results in interphase cells (Supplementary Figure S3B). suggest that endogenous BubR1 interacts with Plk1 in vivo. To determine which domain of BubR1 interacts with BubR1 negatively regulates Plk1 activity in interphase Plk1, GFP-BubR1 deletion mutants were transfected cells into HeLa cells and analysed by immunoprecipitation To evaluate how BubR1 affects Plk1 function, we experiments using anti-Plk1 antibody. The mutant constructed a tetracycline (Tet)-inducible BubR1 BubR1 constructs are shown in Figure 3k. The full- expression vector and transfected it into the PCS length proteins, DN and DKinase were found to be syndrome cells (PCS1). The stably transfected cells were coimmunoprecipitated with endogenous Plk1, whereas synchronized into interphase with HU in the presence or the N-terminal deletion mutant (kinase only) did not absence of Tet, and after 48 h the cells were lysed and coimmunoprecipitate with Plk1 (Figure 5c). These examined by immunoblotting. The induced expression results suggest that the internal region (residues 240– of BubR1 suppressed Plk1 activity in interphase cells 639) of BubR1 is necessary for forming a complex with (Figure 6a). Moreover, the induced expression of BubR1 Plk1 in vivo. suppressed centrosome amplification (Supplementary To determine the domain of Plk1 that interacts with Figure S5). Conversely, we established a stable BubR1, we transfected a Flag-Plk1 wild-type or a BubR1-knockdown HCT116 cell line (a colorectal cell Flag-Plk1 Polo-box inactive mutant (Flag-Plk1 line without chromosome instability (Cahill et al., 1998)) H538A/K540M) into HeLa cells and carried out by siRNA vector and examined Plk1 activity and immunoprecipitation experiments using anti-Flag anti- centrosome number. The BubR1-knockdown cells body. Flag-Plk1 formed a complex with endogenous showed increased Plk1 activity (Figure 6b), and also BubR1, whereas the Flag-Plk1 Polo-box inactive showed increased centrosome amplification and multi- mutants showed a reduced association with BubR1 polar mitosis as compared to control cells (Supplemen- (Figure 5d). These results suggest that the Polo-box tary Figure S6). We showed in Figure 3k that two domains of Plk1 are important for forming a BubR1 mutants (a deletion mutant lacking the kinase complex with BubR1 in vivo. We also examined domain (DKinase), and a kinase-dead mutant with in vitro immunoprecipitation experiment by using K795R substitution) were able to localize to centro- GST-BubR1 and His-Plk1 proteins (Figure 5e). GST- somes and suppress centrosome amplification in PCS1 BubR1 directly bound to His-Plk1, suggesting that cells, suggesting that the kinase activity of BubR1 is not BubR1 physically interacts with Plk1 in vitro. necessary for suppression of Plk1. We then used a Plk1 We showed in Figures 3k and 5c that the BubR1 Polo-box inactive mutant that was not able to interact deletion mutant, which were not able to interact with with BubR1 and transfected it into asynchronous HeLa Plk1, did not localize to centrosomes. These results cells. The Plk1 Polo-box inactive mutant showed an suggest that Plk1 may be necessary for BubR1 to increase in Plk1 phosphorylation as compared with localize to centrosomes. Therefore, we carried out Plk1 the wild-type Plk1 (Figure 6c). These results suggest depletion in HeLa cells by siRNA and analysed that binding of Plk1 with the inhibitor (possibly BubR1)

Figure 4 Premature chromatid separation (PCS) syndrome cells have high Polo-like kinase 1 (Plk1) activity. (a) Immunoblot analyses of Plk1, phospho-Plk1 (phospho-Thr210), Plk4, Aurora-A, phospho-Aurora-A (phospho-Thr288) and p53 expression in PCS syndrome cells. Immunoblot of anti-a-tubulin antibody served as a loading control. (b) in vitro phosphorylation of GST-Cdc25C by Plk1. GST-Cdc25C was incubated with anti-Plk1 immunoprecipitates from the indicated cell lines in the presence of ATP. The reaction products were analysed by immunoblotting using anti-phospho-Cdc25C (Ser 198) antibody. (c) Centrosome fractionation analysis of BubR1 and phospho-Plk1 in normal cells (KM), PCS1 cells and PCS1-BubR1 cells. BubR1 was included in g-tubulin fractions from both normal cells and PCS-BubR1 cells, whereas BubR1 was not in g-tubulin fractions from PCS1 cells. Instead, an especially high amount of phospho-Plk1 was included in the g-tubulin fractions from PCS1 cells. (d) Active Plk1 was detected to both unduplicated and duplicated centrosomes in PCS1 cells, whereas active Plk1 localized to only the duplicated centrosomes in normal cells and PCS1- BubR1 cells. Arrowheads show unduplicated centrosomes, and arrows show duplicated centrosomes. Insets show the magniﬁcation images of centrosomes. Scale bar, 10 mm. (e) Active Plk1 signal intensities on unduplicated centrosome, duplicated centrosomes or spindle poles/ kinetochores in KM, PCS1 and PCS1-BubR1 cells. The quantitative signal intensities were measured as a pixel density. A. U.; Arbitrary Unit. Plk1 activity kept a high level throughout cell cycle in PCS1 cells, whereas Plk1 activity became high around G2/ M phase in KM and PCS1-BubR1 cells. For each cell line, X50 cells were examined. The data are shown as average±standard error determined from three separate experiments. (F) The quantitative analyses of active Plk1 localization for unduplicated or duplicated centrosomes in normal cells (KM), PCS1 cells and PCS1-BubR1 cells. For each cell lines, X100 cells were examined. The data are shown as average±standard error determined from three separate experiments.

Oncogene BubR1 prevents Plk1 activity on centrosome H Izumi et al 2813 is necessary for suppression of Plk1 phosphorylation whereas only GST peptide did not show such an in interphase. To test this idea, recombinant BubR1 was inhibition. These results suggest that BubR1 binds to subjected to an in vitro kinase assay (Figure 6d). Plk1 and inhibits phosphorylation of Plk1 substrate. Increasing amounts of GST-BubR1 full-length Next, we immunoprecipitated BubR1 from interphase peptide or GST-BubR1 (240–639) peptide inhibited lysate and from mitotic lysate, respectively. After that, phosphorylation of GST-Cdc25C by His-Plk1 in vitro, in vitro Plk1 kinase assay was performed (Figure 6e).

KM MY1 MY1-Ch.15KM PCS1 PCS1-Ch.15PCS1-BubR1 BubR1

KM KM PCS1 PCS1-Ch.15PCS1-BubR1 Plk1 GST-Cdc25C Plk1-T210-P -Ser198-P

Plk4 GST-Cdc25C Aurora-A Plk1

Aurora-A-P GST-Cdc25C -+++ + IP: Plk1 + + + + + p53

α-tubulin

Frac. No. 1 23 4 567 8 TCL BubR1

KM Plk1-P

γ-tubulin

Frac. No. 12 3 4 5678 TCL BubR1

PCS1 Plk1-P

γ -tubulin

Frac. No. 12 3 4 5678 TCL BubR1

PCS1 Plk1-P -BubR1 γ -tubulin

γ-tubulin Plk1-P Merge/DAPI 300 KM KM 250 PCS1 PCS1- BubR1 200

150

100 A. U. (Pixel) PCS1 50

0 Unduplicated Duplicated Spindle pole/ centrosome centrosomes Kinetochore

100 90 KM PCS1 PCS1-BubR1 PCS1- 80 BubR1 70 60 50 40 30 20

localization of Plk1-P 10 Percent of Centrosome 0 1 ≥2 1 ≥2 1 ≥2 Number of centrosome per cell

Oncogene BubR1 prevents Plk1 activity on centrosome H Izumi et al 2814 IP IP IgG αPlk1 Input10% IgG αBubR1 Input10%

IB: αBubR1 BubR1 IB: αPlk1 Plk1

IB: αPlk1 Plk1 IB: αBubR1 BubR1

α α IP: Plk1 IB: GFP Input:8% IB: αGFP

inase K inase ΔN Δ N ΔK 1 Δ 1 bR1 ubR bR1 ubR -B FP FP-Bu FP-B FP FP-Bu FP G G GFP-BubR1GFP-BubR1KinaseG G G GFP-BubR1GFP-BubR1KinaseG

GFP-BubR1 GFP-BubR1ΔN GFP-BubR1ΔKinase NS GFP-BubR1Kinase GFP IP: αPlk1 IB: αPlk 1

GST GST-BubR1

His-Plk1

GST-BubR1 HeLa-Flag Plk1 HeLa-FlagHeLa-Flag vector Plk1H538A/K540M IP: αGST IB: αBubR1 IP: αFlag IB: αFlag GST Input 5% IB: αBubR1

Input 25% (His-Plk1)

γ BubR1 Merge/DAPI -tubulin 100 HeLa-control siRNA 90

al 80

ubR1 70 60 No

ntrosom HeLa-controlHeLa-Plk1 siRNA siRNA n of B 50 faint

HeLa-Plk1 siRNA tio Plk1 of ce 40 intense

liza 30

cent α-tubulin 20

loca Per 10 0 Control Plk1 siRNA siRNA Figure 5 BubR1 interacts with Polo-like kinase 1 (Plk1). (a) Cell lysate from asynchronously growing normal cells was immunoprecipitated with anti-Plk1 antibody. As a control, immunoprecipitate with preimmune mouse immunoglobulin-G (IgG) was used. The immunoprecipitates were then subjected to immunoblotting using anti-BubR1 antibody. (b) Conversely, cell lysates were immunoprecipitated with anti-BubR1, and the immunoprecipitates were subjected to immunoblotting using anti-Plk1 antibody. (c) Several GFP-BubR1 deletion mutants were transfected into HeLa cells, and after 48 h, cells were harvested and lysed. The cell lysate was immunoprecipitated using anti-Plk1 antibody, and the immunoprecipitate was then subjected to immunoblotting using anti-GFP antibody. (d) Flag-Plk1 or Flag-Plk1 H538A/K540M was transfected into HeLa cells, and at 48 h after transfection, the cells were harvested and lysed. The cell lysates were immunoprecipitated using anti-Flag antibody, and the immunoprecipitates were then subjected to immunoblotting using anti-BubR1 antibody. (e) BubR1 directly binds to Plk1 in vitro. His-Plk1 was incubated with GST or GST-BubR1 for 2 h. The incubate mixture was immunoprecipitated with anti-GST antibody, and the immunoprecipitates was subjected to immunoblotting by anti-Plk1 antibody. (f) Control siRNA or Plk1 siRNA was transfected into HeLa cells, and the lysates from the transfectants were subjected to immunoblotting using anti-Plk1 antibody. Immunoblot of anti-a-tubulin antibody served as a loading control. (g) Immunostaining of control siRNA (HeLa-control siRNA) or Plk1 siRNA transfected HeLa cells (HeLa-Plk1 siRNA) for BubR1 and g-tubulin. Arrowheads show centrosomes. Insets show the magniﬁcation images of centrosomes. Scale bars, 10 mm. (h) Comparison of BubR1 signal intensity. The intensity of BubR1 staining for centrosomes was evaluated into three levels: no staining, No; faint staining, faint; intense staining, intense. Centrosomal localization of BubR1 was reduced in Plk1 siRNA transfected HeLa cells.

Oncogene BubR1 prevents Plk1 activity on centrosome H Izumi et al 2815 Although the immunoprecipitated-BubR1 was highly therefore examined the centrosome profile in H2B-GFP- phosphorylated from mitotic sample, interphase positive, mononucleated cells. Overexpression of Plk1 BubR1 was not from interphase lysate. Plk1 kinase dramatically induced centrosome amplification (35%) assay showed that the kinase activity of Plk1, which and multipolar mitosis (40%; Figures 7e, f; Supplemen- interacted with interphase BubR1, was basal level tary Figure S7A) in spite of HeLa cells having tight (Figure 6e). On the other hand, the kinase activity centriole duplication machinery (Loncarek et al., 2008). of Plk1, which interacted with mitotic BubR1, was Moreover, overexpression of Plk1 plus knockdown of extraordinarily high level, even when Plk1 protein level BubR1 enhances more dramatically induced centrosome was compensated (Figure 6f). These results suggest that amplification (47%) and multipolar mitosis (52%; Plk1 activity is dependent on the phosphorylation status Figures 7e and f). When these cells were treated with of BubR1. We then constructed a BubR1 deletion HU, the frequency of centrosome amplification was mutant (BubR1DPlk1) lacking the Plk1-binding domain markedly increased (Figures 7g; Supplementary Figure (residues 240–639), which was not able to interact with S7B), suggesting that overexpressed Plk1-mediated Plk1, and transfected it into PCS1 cells. Plk1 activity centrosome amplification is induced during interphase was examined by in vitro kinase assays. The BubR1 and that BubR1 suppresses centrosome amplification deletion mutant was not able to suppress high Plk1 activity of Plk1. activity in PCS1 cells, whereas the full-length BubR1 could suppress Plk1 activity efficiently (Figure 6g). It was also found that the deletion mutant was not able to suppress centrosome amplification in PCS1 cells Discussion (Figure 6h). We found that centrosome amplification and multipolar mitosis result in severe chromosome instability in PCS The knockdown of Plk1 suppresses centrosome syndrome cells. We also found that the frequency of amplification in PCS syndrome cells centrosome amplification increases after HU treatment. As Plk1 was upregulated in PCS syndrome cells, we These results suggest that PCS syndrome cells have lost analysed whether the knockdown of Plk1 influences the ability to regulate the centrosome duplication centrosome amplification in PCS syndrome cells. After machinery. PCS syndrome cells showed increased 72 h of siRNA-mediated Plk1 depletion, we examined activity of Plk1, whose knockdown suppressed centro- the centrosome profile in these PCS1 cells (Figure 7a). some amplification. BubR1 localized to centrosomes The knockdown of Plk1 suppressed centrosome ampli- and negatively regulated Plk1 activity in normal fication in PCS1 cells (Figure 7b, left panel). As it is interphase cells. These results suggest that BubR1 possible that cells with multiple centrosomes could arise prevents centrosome amplification, one mechanism of as a result of cytokinetic failure, we carried out which is negative regulation of Plk1. immunostaining experiment under HU-treated condi- Centrosomes serve as the microtubule-organizing tions. Control siRNA-treated cells showed high rate of center and play a critical role in organizing the bipolar centrosome amplification. On the other hand, the mitotic spindle. It has been reported that Plk1 is knockdown of Plk1 also suppressed centrosome ampli- involved in maturation, separation and duplication of fication in PCS1 cells (Figure 7b, right panel). Thus it is centrosomes (Lane and Nigg, 1996; Liu and Erikson, likely that Plk1 is directly involved in centrosome 2002; Barr et al., 2004; van Vugt and Medema, 2005). amplification in PCS syndrome cells. Plk1 localizes to the centrosome during interphase, but its activity is suppressed until early M phase (Golsteyn Overexpression of Plk1 directly induces centrosome et al., 1995). We showed that overexpression of Plk1- amplification in interphase cells mediated centrosome amplification is induced by To examine whether overexpression of Plk1 directly centrosome reduplication during interphase, and the induces centrosome amplification, the Plk1 expression contribution of cytokinesis failure is not as great as vector with H2B-GFP as a transfection marker reported previously (Meraldi et al., 2002). These findings was cotransfected into HeLa cells (Figure 7c), and suggest that Plk1 activity is strictly regulated during the H2B-GFP-positive cells were examined. In addition, cell cycle, and Plk1 executes the centrosome reduplica- we prepared BubR1-knockdown HeLa cells transfected tion function when prematurely activated before early with Plk1 expression vector and H2B-GFP (Figure 7c; M phase. HeLa-Plk1/H2B-GFP/BubR1i). It was reported pre- BubR1 protein contains several functional domains: a viously that overexpression of Plk1 induces a defect in highly conserved domain with MAD3 homology at the cytokinesis, resulting in the accumulation of binucleated N terminus, a Bub3-binding domain and a Cdc20- cells (Meraldi et al., 2002). Consistent with this, we binding domain in the central region and a kinase observed binucleated cells after overexpression of Plk1 domain at the C terminus. We found that the central (Figure 7d). However, the frequency of binucleated cells region is necessary for centrosomal localization of was only 12%, and the remaining cells (88%) were BubR1 and for suppression of centrosome amplification mononucleated (Figure 7d). Similar result was obtained in PCS syndrome cells, and that BubR1 interacts with in BubR1-knockdown HeLa cells transfected with Plk1 in this region, which overlaps with the region for Plk1 expression vector and H2B-GFP (Figure 7d). We centrosomal localization.

Oncogene BubR1 prevents Plk1 activity on centrosome H Izumi et al 2816 Two possibilities are conceivable for the mechanism second is that BubR1 simply binds to Plk1 and directly by which BubR1 suppresses Plk1 activity. The ﬁrst is inhibits its phosphorylation and kinase activity of Plk1. that BubR1 phosphorylates Plk1 and inactivates it. The We showed that both the GFP-BubR1 deletion mutants

Ai Ai N N 1 R PCS1-BubR1-Tet ubR HU + + T116 T-control R T-B C C C Tet - + H H H BubR1 BubR1

Plk1 Plk1

Plk1-T210-P Plk1-T210-P

α-tubulin α-tubulin

GST GST-BubR1(240-639) GST-BubR1

GST-Cdc25C eLa-FlagHeLa-Flag Plk1 Plk1 HeLa-FlagH vector H538A/K540M Ser198-P IB: αBubR1 GST-Cdc25C

IP: αFlag IB: Plk1-T210-P His-Plk1 α IB: αFlag IB: GST 0 50 100 200 0 50 100 200 0 50 100 200 (ng)

GST GST-BubR1(240-639) GST-BubR1

IP: IgG αBubR1 αBubR1 7

HU HU Noc 6

5 GST-Cdc25C-S198-P 4

GST-Cdc25C Arbitrary Unit 2 1 IB: αPlk1 Plk1 0 BubR1-P S198-P signal S198-P signal IB: αBubR1 intensity/ interphase intensity/mitotic BubR1 Plk1 protein level Plk1 protein level

lk1 ΔP 1 1 ubR ubR 1-vec 1-B 1-B S S S lk1 C ΔP PC PC P 1 1 IB: αCdc25C-P GST-Cdc25C-P ubR ubR 1-B S1-vec S S1-B (KDa) C PC P PC IB: αGST GST-Cdc25C 150 BubR1 Plk1 100

BubR1 Δ IP: αPlk1 75 BubR1 Plk1

IB (Input 10%): αBubR1

IB: αBubR1

70 70 70 PCS1-vec PCS1-BubR1 PCSz1-BubR1 ΔPlk1 60 60 60 50 50 50 40 40 40 30 30 30 20 20 20

Percent of cells 10 10 10 0 0 0 12≥3 12≥3 12≥3 Number of centrosome per cell

Oncogene BubR1 prevents Plk1 activity on centrosome H Izumi et al 2817 lacking the kinase domain and the BubR1 kinase-dead Plk1. During G2-M transition, once BubR1 is phos- mutant were able to localize to centrosomes and phorylated by Cdk1 kinase, the phosphorylated-BubR1 suppress centrosome amplification in PCS syndrome may dissociate from centrosomes, and Plk1 may be cells. We therefore conclude that the first possibility is easily activated by Aurora-A/Bora system (Macurek unlikely. We then examined the second possibility and et al., 2008; Seki et al., 2008) and/or autophosphoryla- analysed a Plk1 Polo-box inactive mutant that was not tion. Thus, BubR1 may be a bifunctional protein: able to interact with BubR1. The Plk1 mutant showed a BubR1 prevents premature activation of Plk1 in substantial increase in autophosphorylation, as has been interphase, whereas BubR1 acts as a spindle assembly suggested previously (Lee and Erikson, 1997; Mundt checkpoint complex with Plk1 in mitosis (Figure 8). et al., 1997). These results strongly suggest that binding BubR1 was initially identified as a spindle assembly of the Polo-box domain with its inhibitor (possibly checkpoint protein; it generates the ‘wait anaphase’ BubR1) is necessary for Plk1 suppression in interphase. signal by binding to and inhibiting Cdc20, a specific In fact, in vitro kinase assay showed that BubR1 activator of the anaphase-promoting complex/cyclo- inhibited phosphorylation of GST-Cdc25C by Plk1. some (APC/C). Recently, Oikawa et al. (2005) reported From this finding, it is likely that BubR1 simply binds to an indirect role of BubR1 in the suppression of Plk1 and sequesters it from substrates. Moreover, we centrosome amplification in mouse cells lacking p53. examined an in vitro kinase assay using the BubR1 They found that exogenous expression of BubR1 deletion mutant that lacks the Plk1-binding domain eliminated mouse p53À/À cells with amplified centro- (BubR1DPlk1). This deletion mutant was not able to somes by inducing mitotic cell death via execution of the suppress high Plk1 activity in PCS syndrome cells, mitotic spindle checkpoint (Oikawa et al., 2005). These whereas full-length BubR1 could suppress Plk1 activity results show that the mitotic spindle checkpoint may considerably. It was also found that this deletion mutant be an eliminator of any mitotic cells with amplified was not able to suppress centrosome amplification in centrosomes. Alternatively, amplified centrosomes PCS syndrome cells. These results strongly suggest that initially form multipolar spindle, but these spindles BubR1 binds to the Polo-box domain on Plk1 and ultimately become bipolar. This phenomenon required prevents Plk1 from premature activation. Moreover, we a delay in mitosis that is mediated by mitotic spindle examined immunoprecipitation-combined in vitro kinase checkpoint in several culture conditions (Chiba et al., assay whether the interphase BubR1 was biochemically 2000; Basto et al., 2008). On the other hand, in this different from the mitotic BubR1. Our results showed study, we found that BubR1 has an additional role in that Plk1 activity was correlated with the phosphoryla- the suppression of centrosome amplification: BubR1 tion status of BubR1, suggesting that BubR1 function is shields Plk1 activity with physical and functional controlled by phosphorylation. During G2-M transition, binding on centrosomes in interphase cells. Thus, in BubR1 is phosphorylated by the priming kinase, Cdk1/ PCS syndrome cells, high Plk1 activity due to BubR1 Cyclin B (Elowe et al., 2007; Matsumura et al., 2007). deficiency contributes to centrosome amplification. The phosphorylated BubR1, in turn, acts as a spindle PCS syndrome is a rare autosomal recessive disorder assembly checkpoint component on kinetochore. This with a high risk of malignancy, including Wilms tumor regulation must be an important event for function of (Kawame et al., 1999; Kajii et al., 2001; Plaja et al., BubR1 during cell cycle. That is, the interphase 2001; Jacquemont et al., 2002). We examined centro- nonphosphorylated BubR1 acts as an inhibitor of Plk1 some status in Wilms tumor tissue derived from a PCS on centrosome and prevent premature activation of syndrome patient. A high degree of centrosome

Figure 6 BubR1 negatively regulates Polo-like kinase 1 (Plk1) activity in interphase cells. (a) Immunoblot analysis of Plk1 phosphorylation using anti-Plk1-T210-P antibody in PCS1 cells transfected with tetracycline (Tet)-induced BubR1 expression vector under hydroxyurea (HU) treatment (interphase). Immunoblot of anti-a-tubulin antibody served as a loading control. (b) Immunoblot analysis of Plk1 phosphorylation using anti-Plk1-T210-P antibody in BubR1 siRNA-transfected HCT116 cells (HCT-BubR1 RNAi). Immunoblot of anti-a-tubulin antibody was used as a loading control. (c) BubR1 interacts with Plk1 and suppresses phosphorylation of Plk1. Flag-Plk1 or Flag-Plk1 H538A/K540M was transfected into HeLa cells, and at 48 h after transfection, the cells were harvested and lysed. The cell lysates were immunoprecipitated using anti-Flag antibody, and the immunoprecipitates were then subjected to immunoblotting using anti-Plk1-T210-P and anti-BubR1 antibodies. (d) In vitro kinase assay of Plk1 in the presence of BubR1 or BubR1-deletion mutant (240-639a. a.). His-Plk1 (400 ng) was incubated with 1 mg of GST-Cdc25C in the presence or absence of GST- BubR1, GST-BubR1 (240–639) or GST only (0, 50, 100, 200 ng). After incubation at 30 1C for 30 min, the sample was subjected to immunoblotting by anti-phospho-Cdc25C (Ser198-P) antibody. BubR1 inhibits phosphorylation of Plk1 substrate in vitro.(e) In vitro kinase assay of GST-Cdc25C by Plk1 in the anti-BubR1 immunoprecipitates from HU-treated cell lysates or nocodazole-treated cell lysates. The immunoprecipitates were subjected for the in vitro kinase assay of GST-Cdc25C. The reaction products were analysed by immunoblotting using anti-phospho-Cdc25C (Ser198-P) antibody. (f) The quantitative analysis of Ser198-P signal intensity when Plk1 protein level was compensated. (g) In vitro phosphorylation assay of GST-Cdc25C by Plk1 in PCS1 transfectants. GST-Cdc25C was incubated with anti-Plk1 immunoprecipitates from the indicated transfectants in the presence of ATP. The reaction products were analysed by immunoblotting using anti-phospho-Cdc25C (Ser198-P) antibody. The anti-Plk1 immunoprecipitates from the indicated transfectants were also subjected to immunoblotting using anti-BubR1 antibody. Immunoblot analysis of exogenous BubR1 expression in PCS1 transfectants (right panel; 10% input). (h) Centrosome proﬁle of PCS1 transfectants. The transfected cells were examined for immunostaining by using anti-g-tubulin antibody. For each transfectant, X100 cells were examined. The data are shown as average±standard error determined from three separate experiments.

Oncogene BubR1 prevents Plk1 activity on centrosome H Izumi et al 2818 35 35 Asynchronous HU48h 30 30

25 25

20 20

PCS1-control l siRNAPCS1-Plk1 siRNA 15 15 Plk1 10 10 Percent of cells with amplified centrosomes α-tubulin 5 5

0 0 Control Plk1 Control Plk1 siRNA siRNA siRNA siRNA i

60 60 50 50 eLa-Plk1/H2B-GFP HeLa-vec/H2B-GFPH HeLa-Plk1/H2B-GFP/BubR1 45 45 40 40 Plk1 35 35 30 30 25 25 Plk1-P 20 20 15 15 10 10 5 BubR1 Percent of binucleated cells 5 0 0 Vec/ Plk1/ Plk1/ Percent of multipolar mitotic cells Vec/ Plk1/ Plk1/ H2B-GFP H2B H2B H2B/ H2B H2B H2B/ BubR1i BubR1i α-tubulin

HeLa-vec/H2B HeLa-Plk1/H2B HeLa-Plk1/H2B/BubR1i 70 Asynchronous Asynchronous Asynchronous 60 50 40 30 20 Percent of cells 10 0 12≥3 12≥3 12≥3 Number of centrosome per cell

HeLa-vec/H2B HeLa-Plk1/H2B HeLa-Plk1/H2B/BubR1i 80 HU48h HU48h HU48h 70

60 ls ls 50

40 nt of cel of nt 30

20 Perce 10 0 12≥3 12≥3 12≥3 Number of centrosome per cell Figure 7 Polo-like kinase 1 (Plk1) is involved in centrosome ampliﬁcation. (a) Immunoblotting of Plk1 in control or Plk1 siRNA- transfected PCS1 cells. Plk1 expression level was reduced approximately 65% after Plk1 siRNA transfection. Immunoblot of anti-a- tubulin antibody served as a loading control. (b) After 72 h of control or Plk1 siRNA transfection, the cells were examined by immunostaining using anti-g-tubulin antibody (left panel). After 24 h of control or Plk1 siRNA transfection, the cells were exposed with hydroxyurea (HU) for 48 h, and were examined by immunostaining using anti-g-tubulin antibody (right panel). For each transfectant, X100 cells were examined. The data are shown as average±standard error determined from three separate experiments. (c) The Plk1 expression vector was cotransfected with a transfection marker plasmid (H2B-GFP; 20:1) into HeLa cells or HeLa-BubR1 siRNA cells. At 48 h after transfection, the cells were lysed and subjected to immunoblotting. Immunoblot of anti-a-tubulin antibody served as a loading control. (d) Proportion of H2B-GFP-positive binucleated cells transfected with control or Plk1 expression vector. (e and f) H2B-GFP-positive mononucleated cells were examined by immunostaining using anti-g-tubulin antibody. (g) The Plk1 expression vector was cotransfected with H2B-GFP expression vector into HeLa cells or HeLa-BubR1 siRNA cells, and after 6 h, the transfectants were treated with HU for 48 h. H2B-GFP-positive mononucleated cells were examined by immunostaining using anti-g- tubulin antibody. For each transfectant, X100 cells were examined. The data are shown as average±standard error determined from three separate experiments.

Oncogene BubR1 prevents Plk1 activity on centrosome H Izumi et al 2819

Interphase G2/M transition Mitosis

Aurora-A/Bora and/or Autophosphorylation

Inactive Premature Active activation Active P P Plk1 Bipolar spindle formation Plk1 Cdc25C Cdk1/Cyclin B BubR1 Plk1

Centrosome P

Plk1 P P

BubR1 Plk1 Active P Premature P P Plk1 activation BubR1 BubR1 Chromosome Spindle assembly checkpoint

Centrosome ampliﬁcation

BubR1 : BubR1 for centrosomal function

P P

BubR1 : BubR1 for spindle assembly checkpoint

Figure 8 Schematic model of functional interaction between BubR1 and Polo-like kinase 1 (Plk1) on centrosomes during the cell cycle.

amplification was observed in the tumor tissue (Supple- lines were maintained in Dulbecco’s modified Eagle’s medium mentary Figure S8). These findings suggest that centro- supplemented with 10% FBS in an atmosphere containing 5% some amplification might contribute to chromosome CO2. Mouse hybrid cell line GM11715 containing human instability and influence tumor formation in PCS chromosome 15 was obtained from the NIGMS Human syndrome. Genetic Mutant Cell Repository, and microcell-mediated In conclusion, PCS syndrome is a ‘centrosome chromosome transfer was performed as described previously (Matsuura et al., 1997). dysfunctional syndrome’, in addition to the spindle For remaining materials and methods, see Supplemental checkpoint defect, and our findings reveal a crucial role information on line. for BubR1 in preventing centrosome amplification through negative regulation of Plk1.

Conflict of interest Materials and methods The authors declare no conflict of interest. Cell lines and cell culture Epstein–Barr virus-transformed LCLs were cultured in RPMI1640 with 10% fetal bovine serum (FBS). Control Acknowledgements LCL (TO) was established from a normal individual, and PCS syndrome LBL (MY1L) was from a patient described We are grateful to Dr S Tanaka for providing anti-pericentrin previously (Patient 2; Matsuura et al., 2006). For metaphase antibody and Dr M Ohsugi for providing Flag-Plk1 preparation, LCLs were arrested for 3 h with 0.4 mg/ml of plasmids. We are also grateful to Dr Y Yamamoto and colcemid, treated at 32 1C for 20 min in 0.075 M KCl, and fixed Dr N Watanabe for technical advice. We thank H Ikeda and in acetic acid/methanol (1:3). Chromosomes were spread onto H Hatakeyama for technical assistance, and T Jo and glass slides at room temperature under 50–55% ambient A Kamesako for secretarial assistance. We also thank our humidity and stained with 4% Giemsa stain. Immortalized laboratory members for their continuous encouragement. This cell lines were established by transfecting primary cultured work was partly carried out at the Analysis Center of Life fibroblasts with phTERT retrovirus vector (Carney et al., Science, Hiroshima University. This work was supported by 2002). A control cell line (KM) was established from a normal Grants-in-Aid for Scientific Research from the Japanese individual, and PCS syndrome cell lines (MY1 and PCS1) Ministry of Education, Science, Sports and Culture (to HI were from two patients (Patient 2 and Patient 1, respectively; and to SM), MEXT priority research projects (to HI and to Matsuura et al., 2006). HCT116 and HeLa cell lines were SM) and the Haraguchi Memorial Cancer Grant Foundation obtained from ATCC (Manassas, VA, USA). All of these cell (to HI).

Oncogene BubR1 prevents Plk1 activity on centrosome H Izumi et al 2820 References

Baker DJ, Jeganathan KB, Cameron JD, Thompson M, Juneja S, Lane HA, Nigg EA. (1996). Antibody microinjection reveals an Kopecka A et al. (2004). BubR1 insufficiency causes early onset of essential role for human polo-like kinase 1 (Plk1) in the functional aging-associated phenotypes and infertility in mice. Nat Genet 36: maturation of mitotic centrosomes. J Cell Biol 135: 1701–1713. 744–749. Lee KS, Erikson RL. (1997). Plk is a functional homolog of Balczon R, Bao L, Zimmer WE, Brown K, Zinkowski RP, Brinkley Saccharomyces cerevisiae Cdc5, and elevated Plk activity induces BR. (1995). Dissociation of centrosome replication events from multiple septation structures. Mol Cell Biol 17: 3408–3417. cycles of DNA synthesis and mitotic division in hydroxyurea- Liu X, Erikson RL. (2002). Activation of Cdc2/cyclin B and inhibition arrested Chinese hamster ovary cells. J Cell Biol 130: 105–115. of centrosome amplification in cells depleted of Plk1 by siRNA. Barr FA, Sillje HH, Nigg EA. (2004). Polo-like kinases and the Proc Natl Acad Sci USA 99: 8672–8676. orchestration of cell division. Nat Rev Mol Cell Biol 5: 429–440. Loncarek J, Hergert P, Magidson V, Khodjakov A. (2008). Control of Basto R, Brunk K, Vinadogrova T, Peel N, Franz A, Khodjakov A daughter centriole formation by the pericentriolar material. Nat Cell et al. (2008). Centrosome amplification can initiate tumorigenesis in Biol 10: 322–328. flies. Cell 133: 1032–1042. Macurek L, Lindqvist A, Lim D, Lampson MA, Lompmaker R, Bornens M, Moudjou M. (1999). Studying the composition Freire R et al. (2008). Polo-like kinase-1 is activated by aurora A to and function of centrosomes in vertebrates. Methods Cell Biol 61: promote checkpoint recovery. Nature 455: 119–123. 13–34. Matsumura S, Toyoshima F, Nishida E. (2007). Polo-like kinase 1 Cahill DP, Lengauer C, Yu J, Riggins GJ, Willson JK, Markowitz SD facilitates chromosome alignment during prometaphase through et al. (1998). Mutations of mitotic checkpoint genes in human BubR1. J Biol Chem 282: 15217–15227. cancers. Nature 392: 300–303. Matsuura S, Ito E, Tauchi H, Komatsu K, Ikeuchi T, Kajii T. (2000). Carney SA, Tahara H, Swartz CD, Risinger JI, He H, Moore AB et al. Chromosomal instability syndrome of total premature chromatid (2002). Immortalization of human uterine leiomyoma and myome- separation with mosaic variegated aneuploidy is defective in mitotic- trial cell lines after induction of telomerase activity: molecular and spindle checkpoint. Am J Hum Genet 67: 483–486. phenotypic characteristics. Lab Invest 82: 719–728. Matsuura S, Matsumoto Y, Morishima K, Izumi H, Matsumoto H, Chiba S, Okuda M, Mussman JG, Fukasawa K. (2000). Genomic Ito E et al. (2006). Monoallelic BUB1B mutations and defective convergence and suppression of centrosome hyperamplification mitotic-spindle checkpoint in seven families with premature chromatid in primary p53À/À cells in prolonged culture. Exp Cell Res 258: separation (PCS) syndrome. Am J Med Genet A 140: 358–367. 310–321. Matsuura S, Weemaes C, Smeets D, Takami H, Kondo N, Sakamoto Cleveland DW, Mao Y, Sullivan KF. (2003). Centromeres and S et al. (1997). Genetic mapping using microcell-mediated chromo- kinetochores: from epigenetics to mitotic checkpoint signaling. some transfer suggests a locus for Nijmegen breakage syndrome at Cell 112: 407–421. chromosome 8q21-24. Am J Hum Genet 60: 1487–1494. Elowe S, Hummer S, Uldschmid A, Li X, Nigg A. (2007). Tension- Meraldi P, Honda R, Nigg EA. (2002). Aurora-A overexpression sensitive Plk1 phosphorylation on BubR1 regulates the reveals tetraploidization as a major route to centrosome amplifica- stability of kinetochore microtubule interactions. Genes Dev 21: tion in p53À/À cells. EMBO J 21: 483–492. 2205–2219. Mundt KE, Golsteyn RM, Lane HA, Nigg EA. (1997). On the Golsteyn RM, Mundt KE, Fry AM, Nigg EA. (1995). Cell cycle regulation and function of human polo-like kinase 1 (PLK1): effects regulation of the activity and subcellular localization of Plk1, a of overexpression on cell cycle progression. Biochem Biophys Res human protein kinase implicated in mitotic spindle function. J Cell Commun 239: 377–385. Biol 129: 1617–1628. Musacchio A, Salmon ED. (2007). The spindle-assembly checkpoint in Hanks S, Coleman K, Reid S, Plaja A, Firth H, Fitzpatrick D et al. space and time. Nat Rev Mol Cell Biol 8: 379–393. (2004). Constitutional aneuploidy and cancer predisposition Nigg EA. (2002). Centrosome aberrations: cause or consequence of caused by biallelic mutations in BUB1B. Nat Genet 36: cancer progression? Nat Rev Cancer 2: 815–825. 1159–1161. Oikawa T, Okuda M, Ma Z, Goorha R, Tsujimoto H, Inokuma H Jacquemont S, Boceno M, Rival JM, Mechinaud F, David A. (2002). et al. (2005). Transcriptional control of BubR1 by p53 and High risk of malignancy in mosaic variegated aneuploidy syndrome. suppression of centrosome amplification by BubR1. Mol Cell Biol Am J Med Genet 109: 17–21; discussion 16. 25: 4046–4061. Kajii T, Ikeuchi T, Yang ZQ, Nakamura Y, Tsuji Y, Yokomori K Plaja A, Vendrell T, Smeets D, Sarret E, Gili T, Catala V et al. (2001). et al. (2001). Cancer-prone syndrome of mosaic variegated Variegated aneuploidy related to premature centromere division aneuploidy and total premature chromatid separation: report of (PCD) is expressed in vivo and is a cancer-prone disease. Am J Med five infants. Am J Med Genet 104: 57–64. Genet 98: 216–223. Kajii T, Kawai T, Takumi T, Misu H, Mabuchi O, Takahashi Y et al. Rieder CL, Khodjakov A. (2003). Mitosis through the microscope: (1998). Mosaic variegated aneuploidy with multiple congenital advances in seeing inside live dividing cells. Science 300: 91–96. abnormalities: homozygosity for total premature chromatid separa- Seki A, Coppinger JA, Jang CY, Yates JR, Fang G. (2008). Bora and tion trait. Am J Med Genet 78: 245–249. the kinase Aurora a cooperatively activate the kinase Plk1 and Kawame H, Sugio Y, Fuyama Y, Hayashi Y, Suzuki H, Kurosawa K control mitotic entry. Science 320: 1655–1658. et al. (1999). Syndrome of microcephaly, Dandy-Walker malforma- Toyoshima-Morimoto F, Taniguchi E, Nishida E. (2002). Plk1 tion, and Wilms tumor caused by mosaic variegated aneuploidy promotes nuclear translocation of human Cdc25C during prophase. with premature centromere division (PCD): report of a new case and EMBO Rep 3: 341–348. review of the literature. J Hum Genet 44: 219–224. van Vugt MA, Medema RH. (2005). Getting in and out of mitosis with Kops GJ, Weaver BA, Cleveland DW. (2005). On the road to Polo-like kinase-1. Oncogene 24: 2844–2859. cancer: aneuploidy and the mitotic checkpoint. Nat Rev Cancer 5: Yu H. (2002). Regulation of APC-Cdc20 by the spindle checkpoint. 773–785. Curr Opin Cell Biol 14: 706–714.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene