Functional Interaction between BubR1 and Securin in an Anaphase-Promoting Complex/Cyclosome^Cdc20–Independent Manner

The mitotic spindle checkpoint ensures accurate segregation of mitotic chromosomes by delaying anaphase onset until each kinetochore has correctly attached to the mitotic spindle. Various mitotic checkpoint proteins including Bub1, BubR1, Bub3, and Mad2 are recruited to kinetochores that lack attachments or tension to generate a “wait anaphase” signal through the formation of an inhibitory ternary complex known as the mitotic checkpoint complex (MCC; refs. 1–6). Thus, the mitotic spindle checkpoint ensures that activation of the anaphase-promoting complex/cyclosome (APC/C), an E3 ubiquitin ligase, is delayed until all chromosomes have achieved bipolar kinetochore-microtubule attachment (7–9). MCC formation is facilitated by the binding of BubR1 and Mad2 to Cdc20, which is a crucial cofactor of the APC/C and is the main target of the mitotic spindle checkpoint. Therefore, MCC can inhibit APC/C associated with Cdc20 (APC/C^Cdc20). However, Cdc20 in MCC is unable to activate APC/C, as shown by several experiments in which the addition of Mad2 and/or BubR1 leads to the checkpoint-mediated inhibition of APC/C^Cdc20 activity (2, 10, 11). Silencing of this checkpoint signal is initiated by the release of the inhibitory mitotic checkpoint protein complex from APC/C^Cdc20; APC/C^Cdc20 can drive cells into anaphase by inducing the degradation of securin (pituitary tumor-transforming gene, PTTG), a small protein that inhibits the protease separase and mitotic cyclins (12–15).

In metaphase-to-anaphase transition, APC/C^Cdc20 initiates the process of chromosome segregation through ubiquitination of securin. Once APC/C^Cdc20 is activated, separase cleaves the Scc1/Rad21 subunit of the cohesion complex. This complex holds sister chromatids together, and its cleavage therefore dissolves cohesion between sister chromatids (12–16). Although cells, and even mice, can survive without securin (17), securin destruction mediated by APC/C^Cdc20 is essential for the activation of separase, which in turn, proteolytically degrades cohesin molecules, leading to the onset of anaphase and mitotic exit. A recent study showed that overexpression of securin aberrantly prolongs the progression of mitosis to anaphase (18). In addition, cells lacking securin degradation or expressing a nondegradable mutant form of securin exhibit asymmetrical cytokinesis without chromosome segregation, resulting in macronuclear formation and aneuploidy (19). Securin is overexpressed in most human cancers and transforms cells both in vitro and in vivo (20–22). Therefore, periodic regulation of securin stability or degradation is critical for maintaining the balance between the mitotic checkpoint and chromosome segregation.

Both BubR1 and Mad2 have been shown to interact directly with APC/C^Cdc20 in vivo and to inhibit its ubiquitination activity in vitro. Mad2 can associate with APC/C^Cdc20, but Mad2 does not prevent the binding of Cdc20 to APC/C (2, 23, 24). BubR1 can associate tightly with Cdc20 and/or with another protein that is essential for checkpoint function called Bub3. Binding of BubR1 to Cdc20 can prevent the association of Cdc20 with APC/C (10, 11). It is not known, however, whether this is the physiologic function of BubR1 because if the primary function of BubR1 were to sequester Cdc20 from APC/C, activation of the mitotic checkpoint would also lead to association of BubR1 with APC/C. Furthermore, the mechanisms by which these mitotic checkpoint proteins inhibit APC/C^Cdc20 activity towards some substrates, including securin, have not yet been investigated.

BubR1 is enriched at kinetochores and also associates tightly with Bub3 and another protein that is essential for the mitotic checkpoint activation; BubR1 is then disassociated from the MCC and destabilized following anaphase (10, 11). Importantly, BubR1 monitors the proper attachment of microtubules to kinetochores and links the regulation of chromosome-spindle attachment to mitotic checkpoint signaling (3, 25, 26). Several studies have shown that disruption of BubR1 activity results in a loss of checkpoint control, chromosomal instability (caused by premature anaphase), and/or early onset of malignancy (3, 27–29). These findings indicate that the BubR1 mitotic checkpoint protein plays an essential role in the maintenance of genomic integrity; therefore, defects in BubR1-mediated signaling not only eliminate checkpoint control but are also linked to certain human diseases such as cancer. In our initial efforts to understand the novel molecular mechanisms of BubR1-mediated checkpoint signaling (30), we used proteomic and subsequent matrix-assisted laser desorption ionization-time of flight analyses. In the present study, we used an anti-BubR1 antibody to immunoprecipitate proteins extracted from cells treated with nocodazole and identified securin as a novel BubR1-interacting protein (Supplementary Fig. S1). We found that BubR1 can form different complexes with securin, as it does with Cdc20, and that the interaction with BubR1 contributes to the stability of securin, possibly by preventing Cdc20 binding.

Cells and synchronization. The hSecurin^+/+ (HCT116 securin^+/+) and hSecurin^−/− (HCT116 securin^−/−) cells were kindly provided by Bert Vogelstein (31). Generation of the stable HCT116 BubR1 knock down (HCT116 BubR1^KD) cell lines has been previously described (30). HeLa cells were purchased from the American Type Culture Collection, and LL86, L132, SW-900, HCC-95, SK-MES-1, HCC-1171, HCC-1833, HCC-2108, and Calu-3 cells were purchased from the Korean Cell Line Bank. For synchronization, HCT116 or HCT116 BubR1^KD cells were grown in the presence of 200 ng/mL of nocodazole for 16 h, harvested, and then used in various assays.

Plasmid construction, small interfering RNA synthesis, and transfection. The full-length cDNA sequences of the human BubR1 and Securin genes were amplified by PCR using templates generated by reverse transcription from HeLa cell total mRNA and using oligo-dT as a primer. Glutathione S-transferase (GST) fusion constructs for expression in Escherichia coli cells were generated by in-frame insertion of PCR fragments encoding BubR1 amino acid residues 1 to 300, 1 to 525, 401 to 700, and 526 to 1050 into the pGEX-KG vector (Pharmacia). cDNAs for BubR1 wild-type and mutant versions of BubR1, residues 1 to 525 and K795R, were subcloned into the Myc epitope–encoding pcDNA3.1 vector to generate pcDNA-Myc-BubR1 wild-type, 1-525, and K795R, respectively. The coding sequences of the securin wild-type or deletion mutants (amino acid residues 1–67, 65–122, and 122–202) were cloned into the pGEX-KG vector. For small interfering RNA (siRNA) synthesis, the following gene-specific sequences were used to generate siRNAs (Dharmacon): Cdc20 siRNA no. 1, 5′-ACCUGGCGGUGACCGCUAU-3′; Cdc20 siRNA no. 2, 5′-UGUGUGGCCUAGUGCUCCU-3′; Cdh1 siRNA, 5′-UGAGAAGUC-UCCCAGUCAG-3′; BubR1 siRNA no. 1, 5′-AAGGGUUCAGAGCCAUCAG-3′; BubR1 siRNA no. 2, 5′-GGAGAUCCUCUACAAAGGG-3′; control siRNA, 5′-CCUACGCGGAAUACUUCGA-3′. cDNAs for mutant versions of securin, residues 1 to 67 and 65 to 202, were subcloned into the Myc epitope–encoding pcDNA3.1 vector to generate pcDNA-Myc-securin (1–67) and (65–202), respectively. For transient transfections, electroporation of HCT116 or HeLa cells was performed using a Microporator (Digital Biotechnology, South Korea) according to the manufacturer's instructions.

Immunoblot assays and antibodies. For immunoblot assays, the cells were synchronized as described above or left asynchronized, harvested by scraping, and then washed twice in cold PBS and lysed in a TNN buffer (30) containing a protease inhibitor cocktail (Sigma). Equal amounts of protein (quantified by Bio-Rad assay) from each sample were separated by SDS-PAGE, transferred to a nitrocellulose filter, blocked, and analyzed with anti-BubR1 (BD Biosciences PharMingen), anti-Bub3 (BD Biosciences PharMingen), anti-Mad2 (BD Biosciences PharMingen), anti–Aurora A (BD Biosciences PharMingen), anti-securin (Zymed), anti-separase (Novus), anti-Cdc20 (Santa Cruz Biotechnology), anti-p53 (Santa Cruz Biotechnology), anti–cyclin B (Santa Cruz Biotechnology), anti-Cdh1 (Calbiochem), anti-actin (Sigma), anti-Myc (Abcam), and anti-His (Abcam) antibodies.

GST pull-down assay and immunoprecipitation. For the GST pull-down assays, the fusion proteins were adsorbed onto glutathione-protein A/G-Sepharose beads (Amersham Biosciences) and incubated with whole-cell extracts (3 mg) from asynchronized or nocodazole-treated HCT116 cells for 4 h at 4°C. The bound proteins were separated by SDS-PAGE and then analyzed by immunoblotting with the appropriate antibody. For immunoprecipitation from total cell extracts, asynchronized or nocodazole-treated HCT116 cells or HCT116-BubR1^KD cells were resuspended in lysis buffer A (30, 32), incubated at 4°C for 30 min, and centrifuged at 14,000 rpm for 15 min. The supernatants (cytoplasmic fractions) were obtained and the cell pellets were resuspended in lysis buffer B [50 mmol/L Tris-HCl (pH 7.5), 300 mmol/L NaCl, 1% NP40, 1 mmol/L EDTA, 1 mmol/L phenylmethylsulfonyl fluoride, 1 mmol/L DTT, 0.2 mmol/L Na3Vo4, and 50 mmol/L NaF] containing a protease inhibitor cocktail, incubated at 4°C for 30 min, lysed by passing the cell pellets through a 27-gauge needle five times, centrifuged at 14,000 rpm for 15 min, and the supernatants of cell pellets (nuclear fractions) were obtained. The mixed extracts (cytoplasm plus nuclear fractions) were diluted with a no-salt buffer to reduce the salt concentration to 150 mmol/L, and then centrifuged again before being analyzed by immunoprecipitation. For immunoprecipitation, each mixed extract was incubated with antibodies against securin or normal immunoglobulin IgG (control) for 2 h at 4°C as described previously (32).

Recombinant protein purification and in vitro competition assay. Cdc20 cDNAs were cloned into pFastBac vector (Invitrogen) and used to generate a recombinant baculovirus in Sf9 insect cells using the Bac-to-Bac baculovirus system (Invitrogen). Three days after viral infection, Sf9 cells were resuspended with STE buffer [10 mmol/L Tris-HCl (pH 8.0), 150 mmol/L NaCl, 2% Triton X-100, 1 mmol/L EDTA, 1 mmol/L phenylmethylsulfonyl fluoride, 2 mmol/L DTT, 100 μg/mL lysozyme, and 1.5% N-lauroylsarcosine sodium salt] containing a protease inhibitor cocktail (Sigma), and lysed by sonication. Recombinant Cdc20 proteins were purified from Sf9 cell lysates using nickel-NTA agarose-bead columns (Qiagen). After washing with STE buffer containing 60 mmol of imidazole, purified proteins were eluted with STE buffer containing 500 mmol of imidazole. To convert STE buffer into TNN buffer, pooled fractions were passed into the PD10 column (GE Healthcare Bio-Sciences). Eluted His-Cdc20 proteins were frozen in aliquots and stored at −80°C. To increase the expression level of BubR1 gene in E. coli, the full sequence of the BubR1 gene was codon-optimized (OligoEngine). The optimized BubR1 gene was cloned into the pSumo vector (Lifesensors) containing T7 promoter and His-BubR1 proteins were expressed in E. coli BL21 cells. Purification of His-tagged proteins is described above in detail. For in vitro competition assays, asynchronized HCT116 cells were lysed in TNN buffer. Each cell extract was preincubated with His-Cdc20 or His-BubR1 proteins (0.5, 1, and 2 μg) for 1 h at 4°C, followed by incubation with GST-securin protein (4 μg) for a further 4 h.

Immunohistochemistry and statistical analyses. Tissue samples from a total of 117 cases of lung squamous cell carcinomas (SqCC) diagnosed at Seoul National University Hospital were histologically examined. BubR1 and securin expressions were evaluated by immunohistochemistry using anti-BubR1 and anti-securin antibodies, respectively, with the labeled streptavidin-biotin complex staining method (LSAB kit, DAKO). BubR1 and securin expressions were analyzed semiquantitatively by summing the scores that had been obtained by multiplying the staining intensity and proportion of positive tumor cells. Statistical analysis was performed using SPSS 11.5 (Stanford version). Pearson's χ² test used to examine the relationships between BubR1 and securin.

Live cell imaging and fluorescence intensity measurement of RFP-securin. To estimate the duration of mitosis, HCT116 or HCT116 BubR1^KD cells were cotransfected with an expression plasmid encoding green fluorescent protein (GFP)-H2B and red fluorescent protein (RFP)-securin and then imaged in δT 0.15-mm dishes in McCoy's supplemented medium containing 10% fetal bovine serum. For 4 h, 0.5-s exposures were taken every 3 min using a 20× NA0.75 objective lens and an LSM500 META confocal microscope (Carl Zeiss). Fluorescence intensity of RFP-securin was measured using AxioVision LE software (Carl Zeiss).

BubR1 interacts with securin in vitro and in vivo. To determine whether BubR1 interacts directly with securin, we generated full-length GST-BubR1 and GST-securin fusion proteins, incubated these proteins with cellular proteins extracted from HCT116 cells, and performed immunoblot analysis with anti-securin or anti-BubR1 antibodies (Fig. 1A,, lanes 1–6). Pull-down assays revealed that GST-BubR1 binds to securin and GST-securin binds to BubR1. Similarly, we generated full-length His-BubR1 and His-securin proteins, which were incubated with beads bound to either GST-securin (Fig. 1A,, lanes 7–9) or GST-BubR1 (Fig. 1A,, lanes 10–13). Again, we found that BubR1 and securin were present in the complex formed in vitro. Consistent with these findings, immunoprecipitation with anti-securin antibody from cellular extracts of either HCT116 securin^+/+ or HCT116 securin^−/− (as a negative control) cells and subsequent immunoblotting with an anti-BubR1 antibody showed that securin and BubR1 form a complex in vivo (Fig. 1B).

To define the domains responsible for the BubR1-securin interaction, we incubated a series of GST-BubR1 and GST-securin deletion mutants with extracts from HCT116 cells expressing wild-type BubR1 and securin. As shown in Fig. 1C, a fragment containing the BubR1 NH₂-terminal homology domain (amino acids 1–525) formed a complex with securin, whereas the central (amino acids 401–700) and COOH-terminal (amino acids 526–1050) regions of BubR1 did not. Similarly, a fragment containing the central (amino acids 65–122) and COOH-terminal (amino acids 122–202) regions of securin interacted with BubR1, whereas the NH₂-terminal (amino acids 1–67) region of securin did not (Fig. 1D). In positive-control experiments, we confirmed that BubR1 and securin interact with their well-defined binding partners, Bub3 and p53, respectively, under these experimental conditions. To exclude the possibility that the interaction between BubR1 and securin is mediated by the formation of a supercomplex with Cdc20, HCT116 cell extracts were immunoprecipitated with either normal IgG or anti-securin antibody (Supplementary Fig. S2). Subsequent immunoblotting with anti-BubR1 or anti-Mad2 antibody revealed that securin forms a complex with BubR1 but not with Mad2; under the same conditions, securin strongly interacted with separase and Cdc20 as positive controls (Supplementary Fig. S2). Together, these data indicate that BubR1 directly interacts with securin in vitro and in vivo.

BubR1 contributes to the stability of securin in unperturbed mitotic cells. The degradation of securin, which is mediated by APC/C^Cdc20, is required for anaphase onset. Our data indicate that BubR1 directly interacts with securin in vitro and in vivo. Therefore, it is likely that BubR1 not only inhibits the ubiquitination activity of APC/C^Cdc20 (Supplementary Fig. S3A) but also regulates the stability of securin through a direct interaction. To investigate whether the stability of securin is regulated by BubR1 in unperturbed mitotic cells, we transfected parental HCT116 and HCT116 BubR1^KD cells, which exhibit a 60% to 70% reduction of BubR1 levels as described previously (30), with plasmids encoding GFP-tagged H2B (GFP-H2B) to visualize chromosomes and/or RFP-tagged securin (RFP-securin; Fig. 2A–C). First, time-lapse microscopic analyses were used to determine the relative times for the individual cell types to complete chromosome separation from the nuclear envelope breakdown (Fig. 2A). In HCT116 cells, the time to completion of chromosome separation was about 73 minutes. However, HCT116 BubR1^KD cells showed a significantly faster exit from mitosis at about 61 minutes, as has recently been reported (Fig. 2B and C). As expected, both HCT116 and HCT116 BubR1^KD cells transfected with plasmids encoding RFP-securin took longer to reach anaphase compared with the control nontransfected HCT116 and HCT116 BubR1^KD cells, indicating that securin blocks the progression of mitosis to anaphase. Interestingly, our time-lapse analysis at the single-cell level revealed that the levels of RFP-securin were constant for more than 230 to 290 minutes, but were then degraded after anaphase onset in HCT116 cells (Fig. 2B and C). However, the relative duration of RFP-securin in HCT116 BubR1^KD cells was markedly shorter at 85 to 125 minutes after nuclear envelope breakdown. In both cell types, RFP-securin degradation begins after anaphase onset; however, during mitosis, both the duration and inductive level of RFP-securin were markedly reduced in HCT116 BubR1^KD cells compared with those in HCT116 cells. The extent of securin destabilization in HCT116 BubR1^KD cells seemed to correlate with the BubR1 expression levels. Furthermore, we measured the endogenous levels of securin in HCT116 and HCT116 BubR1^KD using immunostaining analysis. As expected, the expression of securin in HCT116 BubR1^KD cells was significantly decreased in both interphase and mitotic (prometaphase) cells, when compared with that in control HCT116 cells (Supplementary Fig. S3B). We confirmed the effects of BubR1 on the stability of securin by transiently transfecting HeLa cells with siRNAs against BubR1 or luciferase. Again, cells transfected with two separate BubR1 siRNAs (which substantially depleted BubR1 protein levels) contained markedly lower levels of endogenous securin in a BubR1 dose-dependent manner (Fig. 2D). These results strongly indicate that BubR1 contributes to the stability of securin.

In vivo correlation between BubR1 and securin expressions in human cancer. To further examine the stability of securin, HCT116 and HCT116 BubR1^KD cells were released from nocodazole control and exposed to the protein translation inhibitor, cycloheximide. Securin levels in these cells were then examined over time by immunoblotting assay (Fig. 3A). Following exposure to cycloheximide for 30 minutes, securin levels decreased by almost 80% in HCT116 BubR1^KD cells, but by only ∼5% in parental cells. At 60 minutes posttreatment, HCT116 BubR1^KD cells showed almost no detectable securin, whereas securin proteins could still be detected at relatively significant levels in parental HCT116 cells. Similarly, these cells were synchronized at the boundary between G₁ and S phases using a double thymidine block (Thy-DB) and then released into the cell cycle. As shown in Supplementary Fig. S3C and D, cells depleted of BubR1 (HCT116 BubR1^KD) did not significantly differ from the control cells (HCT116 con) in terms of progression through the S and G₂ phases, mitosis, or exit from mitosis. However, in the absence of microtubule inhibitor, the levels of securin were also dramatically decreased in HCT116 BubR1^KD, whereas steady state levels were observed in HCT116 con cells. To further characterize this relationship, we tested the effects of introducing exogenous BubR1 wild-type, a kinase-insufficient mutant [BubR1 (K795R)], or a BubR1-deletion mutant [BubR1 (1–525)]. As expected, both BubR1 wild-type and mutants markedly increased the stability of securin (Fig. 3B). These findings suggest that the interaction between BubR1 and securin could play an alternative role in the stability of securin.

Next, we investigated the correlation between BubR1 and securin expression levels in lung cells. We prepared cell extracts from two immortalized nontumor cell lines (LL86 and L132), three SqCC cell lines (SW-900, HCC-95, and SK-MES-1), and five adenocarcinoma cell lines (HCC-1171, HCC-1833, HCC-2108, SK-LU-1, and Calu-3) and performed immunoblot analyses. Interestingly, the levels of securin expression in lung cells markedly correlated with those of BubR1 (Supplementary Fig. S4). We further examined BubR1 and securin levels in 117 human SqCC by immunohistochemical analyses with anti-BubR1 and anti-securin antibodies. Typical immunohistochemical staining of either tumor specimens or adjacent normal tissues is shown in Fig. 3C. BubR1 was significantly expressed in 34% of tumor specimens with positive staining defined as more than one cell among five tumor cells showing intense staining for BubR1 protein; 66% of tumor specimens were BubR1-negative, defined as less than one BubR1-positive cell among 10 tumor cells. Interestingly, the securin-negative staining rate was 54% in BubR1-negative tumor specimens, whereas 12% were securin-positive, revealing a significant positive correlation between BubR1 and securin expressions. Thus, the stability of securin seems to be directly associated with the levels of BubR1 protein.

Interaction between BubR1 and securin is independent of Cdc20. BubR1 binds to the central and COOH-terminal domains of securin, but interacts strongly with full-length securin. However, Cdc20 binds to full-length securin but not with the NH₂-terminal, central, or COOH-terminal domains of securin, indicating that the domains of securin responsible for the interactions with BubR1 and/or Cdc20 overlap (Fig. 1C; data not shown). To examine the effect of Cdc20 on the interaction between BubR1 and securin, HCT116 cells were transfected with two different siRNAs against Cdc20, which reduced Cdc20 expression levels by ∼65% and 40%, respectively, or with siRNA against luciferase (negative control). Although Cdc20 siRNA no. 2 slightly induced the sub-G₁ phase (apoptosis) population, neither Cdc20 siRNAs triggered a marked change in the cell cycle profile compared with the negative control (Supplementary Fig. S5). Therefore, we immunoprecipitated securin using anti-securin antibody or normal IgG (negative control) and performed immunoblot assays (Fig. 4A). The amount of separase bound to the securin immunocomplex was almost constant regardless of the Cdc20 expression levels. However, the levels of BubR1 in this complex were significantly increased in Cdc20-depleted cells (Fig. 4A and B). Interestingly, the amount of BubR1 bound to securin seemed to correlate with the degree to which Cdc20 expression was inhibited. Therefore, these data indicate that the interaction between BubR1 and securin is independent of the APC/C^Cdc20 pathway.

As previously reported, both Cdc20 and Cdh1 have essential roles in stimulating the ubiquitin ligase activity of the APC/C. APC/C^Cdc20 promotes the degradation of securin at the metaphase-to-anaphase transition, whereas APC/C^Cdh1 activity in late mitosis and G₁ is required for the degradation of mitotic cyclins (5, 9, 16), indicating that Cdc20 is not redundant with Cdh1 for the transition from metaphase-to-anaphase. However, securin not only interacts with Cdc20, but also with Cdh1 (16). Therefore, we were interested to determine whether Cdh1 could affect the interaction between BubR1 and securin. We transfected HCT116 cells with a set of siRNAs against Cdh1, Cdc20, or luciferase and performed immunoprecipitation assays using anti-securin antibody or normal IgG (Fig. 4C). Interestingly, in cells deplete of Cdh1, the relative levels of BubR1 that coprecipitated with securin were similar to those in the control luciferase siRNA-transfected cells, whereas cells lacking Cdc20 expression had increased levels of BubR1 (Fig. 4C and D). Again, these results suggest that the interaction between BubR1 and securin is independent of Cdc20.

BubR1 competes with Cdc20 for binding to securin. Cdc20 can bind BubR1 and Mad2 simultaneously and also interact with securin. Moreover, our results indicate that BubR1 may compete with Cdc20 for securin binding, thereby providing an alternative mechanism for controlling the metaphase-to-anaphase transition. To assess whether BubR1 competes with Cdc20 for binding to securin, we performed in vitro competition assays using purified GST-securin in E. coli cells and using His-BubR1 and His-Cdc20 in Sf9 cells (Fig. 5A). Recombinant GST-securin, but not GST alone, formed a complex with endogenous BubR1 extracted from HCT116 cells. However, the addition of recombinant His-Cdc20 significantly blocked the interaction between securin and BubR1. Increasing amounts of recombinant His-Cdc20 gradually reduced the relative levels of BubR1 bound to securin (Fig. 5A,, bottom). This inhibition was specific because the levels of unrelated proteins, such as GST-p53 and GST alone, were not affected by coincubation with His-Cdc20 (data not shown). Additionally, we incubated HCT116 cell extracts with GST-securin to investigate the levels of Cdc20 that bind to securin in the presence of purified recombinant His-BubR1 as a competitor (Fig. 5B). Consistent with this, increasing levels His-BubR1 gradually reduced the level of Cdc20 that bound to securin in a dose-dependent manner (Fig. 5B,, bottom), indicating that BubR1 competes with Cdc20 for binding to securin. Next, we generated a NH₂-terminal deletion mutant of securin that contained the COOH-terminal BubR1-binding region (amino acids 65–202) to use as a competitor for the interactions between endogenous BubR1 and securin proteins (Fig. 5C). HCT116 cells were transfected with Myc-tagged securin (1–67) as a control or with Myc-tagged securin (65–202) expression plasmids. Lysates from the transfected cells were immunoprecipitated with control IgG or anti-securin antibody and immunoblotted with anti-securin, anti-BubR1, or anti-separase antibody. Interestingly, the amount of BubR1 bound to the securin immunocomplex was significantly decreased in cells transfected with Myc-securin (65–202). However, the levels of BubR1 in this complex were almost constant in cells transfected with Myc-securin (1–67) expression plasmid compared with the control Myc vector-transfected cells. These results indicate that COOH-terminal securin (65–202) competes with endogenous securin for binding to endogenous BubR1.

In summary, BubR1 and securin are intimately associated in a stable complex, and interaction with BubR1 is therefore likely to contribute to the stability of securin by competing with APC/C^Cdc20.

BubR1 inhibits the activity of APC/C by blocking formation of the active APC/C^Cdc20 complex. Mad2 also seems to inhibit APC/C by a similar mechanism (5, 9). Therefore, Cdc20 can be inhibited by either BubR1 or Mad2, but both BubR1 and Mad2 can synergistically inhibit Cdc20 (2, 10, 11). Cells that overexpress Cdc20 greatly reduced the mitotic arrest in response to spindle damage, presumably because increased levels of Cdc20 activate APC/C and allow cells to bypass the mitotic checkpoint (10, 32, 33). Interestingly, coexpression of BubR1 restored the mitotic arrest of the Cdc20-overexpressing cells (10), indicating that BubR1 directly inhibits the activity of APC/C^Cdc20. However, it is possible that BubR1 may function independently of APC/C^Cdc20 to inhibit the onset of anaphase. It has recently been reported that loss of Cdc20 causes a securin-dependent metaphase arrest in mouse embryos, but a Cdc20 and securin double mutant could not maintain the metaphase arrest (34), suggesting that securin has both Cdc20-independent and -dependent roles in preventing mitotic exit. In this study, our data provide compelling evidence that the mitotic checkpoint protein BubR1 forms a complex with securin, and that BubR1 contributes to the stability of securin by a Cdc20-independent interaction.

Abnormality in the cellular Cdc20 level or in its function may deregulate APC/C activation and promote premature anaphase. Moreover, recent reports have provided evidence that several human tumors show aberrant overexpression of Cdc20 (33, 35), indicating that Cdc20 overexpression might be characteristic of tumorigenesis. Nevertheless, human cancer cells displaying aberrant Cdc20 levels still activate the checkpoint in response to mitotic spindle damage. Inhibition and/or inactivation of BubR1 in these cancer cells override and/or compromise the mitotic checkpoint rather than completely destroy checkpoint control (3, 30, 36). These results indicate that BubR1 may generate an additional signal to induce mitotic arrest independently of the Cdc20 signaling pathway. However, the role of BubR1 is unlikely to be restricted to checkpoint control. We have recently shown that BubR1 functions as a potent apoptotic molecule for preventing the adaptation of abnormal, chromosomally unstable mitotic cells, indicating that BubR1 plays an important role in subsequent postmitotic adaptation (37). Here, we also propose that BubR1 contributes to the generation of another mitotic checkpoint signal by forming a complex with securin, which has an inhibitory effect on separase.

Both BubR1 and Cdc20 are phosphorylated during mitosis and the phosphorylation of these proteins is required for function of the mitotic checkpoint. It has been reported that Cdc20 associates with the mitotic APC/C more strongly than it does with the interphase APC/C (10, 11, 38). This is likely to be because binding of Cdc20 to APC/C during mitosis may be mediated by phosphorylation. However, interaction with and inhibition of APC/C^Cdc20 is independent of the BubR1 kinase activity because a kinase-dead mutant of BubR1 can still bind and inhibit Cdc20 (10). In addition, phosphorylation of Cdc20 is not required for inhibition by BubR1 (11). Moreover, it seems that small portions of BubR1 form a complex with Cdc20; therefore, a significant population of free BubR1 that is not bound to Cdc20 is able to interact with securin. Although differences in affinity cannot be determined under our experimental conditions, free BubR1 interacts with and stabilizes securin by preventing APC/C^Cdc20-mediated ubiquitination. Interestingly, it has recently been suggested that if the APC/C^Cdc20 complex is formed, it becomes more resistant to the action of BubR1 (10). In addition to BubR1, Mad2 is also thought to bind directly to Cdc20 and inhibit APC/C. Thus, there may be stoichiometric binding between the checkpoint proteins and securin, creating supercomplexes such as BubR1-Cdc20-securin and/or Mad2-Cdc20-securin. However, our data clearly showed that securin forms a complex with BubR1 but not with Mad2 (Supplementary Fig. S2).

BubR1 is a serine/threonine kinase and phosphorylates proteins that are involved in the regulation of mitosis and postmitosis, such as Cdc20 and p53 (30, 37, 39). BubR1 is autophosphorylated and phosphorylated by Aurora B and Plk1 in response to mitotic spindle damage (40, 41). Phosphorylation of BubR1 seems to be important for the regulation of mitotic progress because substitution or deletion of BubR1 kinase residues generates a dominant-negative mutant that competes with endogenous wild-type BubR1. Therefore, we investigated whether the introduction of a BubR1-deletion mutant (BubR1 1–525), which contains the NH₂-terminal securin-binding region, or a kinase-insufficient mutant (BubR1 K795R) contribute to the stability of securin, possibly by sequestration of Cdc20 (Fig. 3B). HeLa cells were transfected with an empty vector (pCMV-Myc), Myc-tagged BubR1 1 to 525 (Myc-BubR1 1–525), Myc-BubR1 K795R or Myc-BubR1 wild-type, and the cells were collected for immunoblot analysis with anti-Myc and anti-securin antibodies. Importantly, cells expressing the BubR1 mutants BubR1 1 to 525 or BubR1 K795R significantly increased the stability of securin to levels that were similar to those expressing BubR1 wild-type, indicating that BubR1 stabilizes securin by competitive inhibition of Cdc20 binding. However, our in vitro kinase assay results using BubR1 immunocomplex and purified GST-securin fusion proteins (as a substrate) showed that BubR1 kinase does not contribute to the phosphorylation of securin (Supplementary Fig. S6C), indicating that the BubR1 kinase activity may not be required for the stability of securin. However, we do not exclude the possibility that the BubR1 kinase activity may regulate securin levels through means other than directly phosphorylating securin.

A recent report has suggested that hyperphosphorylated forms of securin are unstable and that protein phosphatase 2A regulates securin levels by preventing securin phosphorylation (42). Unexpectedly, our in vitro kinase assay failed to show the direct phosphorylation of securin by BubR1 kinase (Supplementary Fig. S6; data not shown). However, the hyperphosphorylation of securin was markedly reduced in BubR1-depleted cells in response to mitotic spindle damage (Supplementary Fig. S6), indicating that BubR1 may be indirectly involved in the checkpoint-mediated phosphorylation of securin. Otherwise, interaction with BubR1 may lead to changes in the biochemical properties or conformational stoichiometry of securin as a phosphorylation substrate by an undetermined kinase.

In summary, BubR1 can form different complexes with securin, as it does with Cdc20, and that the interaction with BubR1 contributes to the stability of securin, possibly by preventing Cdc20 binding. Therefore, we propose that silencing of mitotic checkpoint signaling activates the Cdc20-mediated APC/C E3 ubiquitin ligase, and then activated APC/C may ubiquitinate BubR1 and securin (Supplementary Fig. S7). Given that defects in or inactivation of BubR1-mediated checkpoint signaling could cause fatal errors in mitotic progression, BubR1-dependent securin stability may be an alternative mechanism in regulating the anaphase-inhibitory signal.

No potential conflicts of interest were disclosed.

Grant support: Korea Health 21 R&D Project, Ministry of Health and Welfare (03-PJ10-PG13-GD01-0002), the Korea Research Foundation (KRF-2006-312-C00625), and the 21C Frontier Functional Human Genome Project from the Ministry of Science & Technology in Korea (FG07-21-01).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. Bert Vogelstein (Johns Hopkins University, Baltimore, MD) for securin-null HCT116 cells.