Dissecting the roles of human BUB1 in the spindle assembly checkpoint

During mitosis, all chromosomes have to attach to microtubules of the mitotic spindle and become bi-orientated before cells are allowed to proceed into anaphase. Chromosome attachment is mediated by kinetochores, large multi-protein structures that form the bridge between chromosomes and spindle microtubules (Foley and Kapoor, 2013). Microtubule binding activity at kinetochores is mainly provided by the KMN network, a 10-subunit protein assembly comprising three subcomplexes KNL1-C–MIS12-C–NDC80-C (Cheeseman et al., 2006).

In addition to the formation of kinetochore–microtubule attachments, the KMN network plays a key role in coupling kinetochore attachment status to spindle assembly checkpoint (SAC) activity. The SAC is activated at unattached kinetochores and results in the formation of a soluble inhibitor of the anaphase promoting complex/cyclosome (APC/C) with its co-activator CDC20 (APC/C^CDC20), a multisubunit E3 ligase of which the activity initiates chromosome segregation and mitotic exit by targeting key mitotic regulators for proteasomal destruction (Pines, 2011). The soluble inhibitor of the APC/C is known as the mitotic checkpoint complex (MCC) and comprises MAD2 (also known as MAD2L1), BUBR1 (also known as BUB1B), BUB3 and CDC20 (Chao et al., 2012; Izawa and Pines, 2015; Kulukian et al., 2009; Sudakin et al., 2001).

The SAC response is initiated at the KMN network through phosphorylation of multiple motifs in repeat sequences of KNL1 (also known as CASC5), which is mediated by MPS1 (also known as TTK) (London et al., 2012; Shepperd et al., 2012; Vleugel et al., 2015; Yamagishi et al., 2012). When phosphorylated, these motifs recruit BUB1–BUB3 dimers that are essential for SAC signaling and chromosome bi-orientation (Krenn et al., 2014; Primorac et al., 2013; Vleugel et al., 2013,, 2015; Zhang et al., 2014). Since the discovery of BUB1, the molecular mechanism by which it participates in SAC activation has been a matter of controversy. Studies in mice, budding yeast and fission yeast have shown that BUB1 kinase activity is required for SAC activation (Kawashima et al., 2010; Ricke et al., 2012; Yamaguchi et al., 2003), but this has been contradicted by other studies using the same model organisms (Baker et al., 2009; Fernius and Hardwick, 2007; London and Biggins, 2014; Perera et al., 2007; Rischitor et al., 2007; Warren et al., 2002). Human cell studies have proven equally inconsistent – one study proposes that BUB1 directly phosphorylates and inhibits CDC20 (Kang et al., 2008), whereas another shows that a truncated BUB1 protein lacking the kinase domain significantly restored SAC activity to BUB1-depleted cells (Klebig et al., 2009). Recent studies show that in budding yeast and Caenorhabditis elegans, BUB1 directly binds to MAD1 through a region preceding the kinase domain (S. cerevisiae), or through the kinase domain itself (C. elegans) (London and Biggins, 2014; Moyle et al., 2014). Whether this function of BUB1 is conserved in human cells is unknown. In addition to MAD1 and MAD2 as well as BUB3, BUB1 is essential for recruiting BUBR1 and CDC20 to kinetochores (Johnson et al., 2004; Kang et al., 2008; Klebig et al., 2009; Sharp-Baker and Chen, 2001). It is at present unknown how BUB1 accomplishes this, and it is unclear if this is functionally relevant for SAC signaling.

Here, we set out to identify the molecular role of BUB1 in SAC signaling in human cells. We provide evidence that promoting localization of BUBR1 to kinetochores is not a crucial aspect of the function of BUB1 in the SAC and that BUB1-mediated recruitment of MAD1 to kinetochores is insufficient for SAC function. We instead find a strong correlation between SAC activity, binding of CDC20 to kinetochores and the presence of 50 residues in BUB1 that harbor two motifs that, in other proteins, are known to bind to CDC20.

To uncover the contribution of BUB1 to the SAC, we examined which region of BUB1 is required to localize BUBR1 to kinetochores. We therefore removed various domains and motifs that have been previously implicated in SAC function from GFP-based localization and affinity purification (LAP)-tagged wild-type BUB1 (LAP–BUB1^WT), creating LAP–BUB1^1–318, LAP–BUB1^1–500 and LAP–BUB1^1–778 (Fig. 1A). All of the RNA interference (RNAi)-resistant constructs were stably integrated into HeLa Flp recombination target (HeLa-FRT) cells at a single doxycycline-inducible locus (Klebig et al., 2009). Recruitment of BUBR1 to the kinetochores in nocodazole-treated cells was abolished through RNAi-mediated depletion of BUB1, and this was rescued by expression of LAP–BUB1^WT (Fig. 1B,C). All BUB1 truncation mutants localized to kinetochores to a similar degree (Fig. 1B,C), as expected (Krenn et al., 2012), and were able to restore BUBR1 levels at kinetochores upon BUB1-targeted RNAi (Fig. 1B,C). We thus conclude that the N-terminal 318 amino acids of BUB1, encompassing a TPR domain and GLEBS motif, are sufficient for BUBR1 kinetochore recruitment. This redefines the region in BUB1 that is required for BUBR1 recruitment to a conserved α-helix following the GLEBS motif (amino acid residues 271–318) (Overlack et al., 2015).

We next asked whether the BUB1–BUB3 interaction is required to recruit BUBR1 to kinetochores. To circumvent the complicating issue that disrupting this interaction (by mutating the GLEBS motif) also prevents kinetochore binding of BUB1 (Taylor et al., 1998), BUB1 was artificially tethered to kinetochores by fusing it to the KMN network protein MIS12 (LAP–MIS12–BUB1^WT) (Fig. 1A). BUBR1 localization was recovered when endogenous BUB1 was replaced with MIS12–BUB1, but not when the mutation E252K was introduced into the GLEBS motif (MIS12–BUB1^E252K) (Fig. 1D,E) (Overlack et al., 2015).

To determine whether BUBR1 kinetochore localization is required for SAC signaling, cells expressing MIS12–BUB1^E252K were analyzed for SAC activity by filming mitotic progression in the presence of nocodazole and a low dose of the MPS1 inhibitor reversine (250 nM) (Santaguida et al., 2010). Cells that had been depleted of BUB1 exited mitosis within 1 h, whereas control cells and BUB1-depleted cells that expressed MIS12–BUB1^WT were able to maintain a functional SAC for over 5 h (Fig. 1F). Surprisingly, although MIS12–BUB1^E252K did not recruit detectable levels of BUBR1 to kinetochores (Fig. 1D,E), it fully restored SAC function (Fig. 1F). Conversely, a BUB1 fragment that was sufficient for BUBR1 kinetochore localization (LAP–MIS12–BUB1^1–318, supplementary material Fig. S1A,B) could not support SAC activation (Fig. 1F). Taken together, these observations suggest that a primary role of kinetochore BUBR1 lies not in SAC activation but most likely in other processes, such as chromosome bi-orientation through recruitment of the phosphatase PP2-B56 (Kruse et al., 2013; Suijkerbuijk et al., 2012; Xu et al., 2013).

We next asked whether human BUB1 promotes SAC activation by forming a stable complex with MAD1, as it does in budding yeast and C. elegans (London and Biggins, 2014; Moyle et al., 2014). LAP–BUB1^WT that had been immunoprecipitated from nocodazole-treated cells associated with known interacting proteins – such as BUB3, BUBR1 and KNL1 – as determined by using mass spectrometry (Fig. 2A). However, no MAD1 peptides were identified in these BUB1 purifications (Fig. 2A). Furthermore, ectopic tethering of BUB1 to a LacO array (LacI–BUB1) was insufficient to recruit MAD1, whereas endogenous BUBR1 was readily detected at those LacO foci (Fig. 2B). These observations contradict the notion of the existence of a stable BUB1–MAD1 complex in human cells. We next wanted to determine the extent to which BUB1 contributes to MAD1 kinetochore localization. Cells that had been depleted of BUB1 were treated with nocodazole for 3 h, fixed and stained for MAD1 by using immunofluorescence. Although BUB1 kinetochore levels were reduced to <5% in both HeLa and RPE-1 cells, we observed no difference in MAD1 kinetochore localization compared with that in control cells (Fig. 2C–F). SAC sensitization through partial MPS1 inhibition (250 nM reversine) reduced the levels of MAD1 at kinetochores but, even in this case, RNAi-mediated BUB1 depletion did not further reduce MAD1 levels (Fig. 2C–F). Taken together, these data argue against a role of human BUB1 as the predominant receptor for MAD1 in human cells.

Although our observed lack of correlation between BUB1 and MAD1 at kinetochores agrees with some reports (Hewitt et al., 2010; Vleugel et al., 2013), it contrasts with others (Kim et al., 2012; Klebig et al., 2009). We noted that although we assessed the levels of MAD1 at kinetochores in cells that had been treated with nocodazole for several hours, others did so in unperturbed cells, which are likely to have spent less time in mitosis (Klebig et al., 2009). To test the effects of this, we measured MAD1 kinetochore levels at different times after metaphase kinetochores that were devoid of MAD1 had been forced to detach and to recruit MAD1, through the addition of nocodazole (Fig. 2G,H). In control cells, BUB1 and MAD1 rapidly accumulated at kinetochores, with kinetochore levels peaking at 10 and 20 min (of BUB1 and MAD1, respectively) (Fig. 2G,H). As expected, BUB1-depleted cells had accumulated similar amounts of MAD1 as control cells after 40 min. Strikingly, however, in the first 20 min, MAD1 levels were substantially lower in BUB1-depleted cells (Fig. 2G,H). BUB1 thus accelerates efficient loading of MAD1 onto kinetochores, potentially through the previously identified CD1 motif (Klebig et al., 2009), but is not essential for it. These findings show that the mechanism of MAD1 binding to kinetochores in human cells is, at least in part, different from that in budding yeast and C. elegans, in which BUB1 is the primary kinetochore receptor for MAD1 (London and Biggins, 2014; Moyle et al., 2014). Human BUB1 appears to catalyze recruitment of MAD1 to kinetochores, perhaps by forming transient interactions to bring MAD1 to kinetochores or by licensing kinetochores for MAD1 binding.

To assess whether accelerating MAD1 kinetochore loading through BUB1 affects SAC signaling, we analyzed various BUB1 truncation mutants (BUB1^1–318, BUB1^1–500 and BUB1^1–778) for their ability to rescue SAC function and MAD1 loading after BUB1 depletion. The SAC defect that was induced by BUB1 depletion was rescued by expression of BUB1^WT, BUB1^1–778, BUB1^1–696 and BUB1^1–555, but not by BUB1^1–318 or BUB1^1–500, suggesting that that the region between amino acids 501–555 is essential for SAC function (Fig. 3A). In our analyses, therefore, the BUB1 kinase domain was not required for the SAC. Importantly, although BUB1-depleted cells had reduced levels of MAD1 at kinetochores 10 min after the addition of nocodazole to metaphase cells, cells that expressed BUB1^1–500 recruited normal amounts of MAD1 in that time span (Fig. 3B,C). Poor SAC signaling in BUB1^1–500-expressing cells cannot therefore be explained by impaired MAD1 loading, or, as shown in Fig. 1, by impaired binding of BUBR1 to kinetochores.

Amino acids 501–555 of BUB1 harbor KEN-box motif 1 (Kang et al., 2008). In addition, our alignment of metazoan BUB1 orthologs revealed the presence of a conserved [F/Y]xx[F/Y]x[D/E] motif (amino acids 527–532) (Fig. 3D). The yeast protein Acm1 utilizes a similar motif known as the ‘A motif’ in addition to its KEN box to interact with and thereby inhibit the CDC20 yeast ortholog Cdh1 (Enquist-Newman et al., 2008; He et al., 2013). Because recent reports have shown the existence of a similar motif in BUBR1 (dubbed the Phe box or ICDC20BD) that is important for a stable BUBR1–CDC20 interaction and for SAC activity (Diaz-Martinez et al., 2015; Lischetti et al., 2014), we tested whether this region in BUB1 is involved in the localization of CDC20 to kinetochores. CDC20 localizes strongly to kinetochores in nocodazole-treated cells, and this depends on BUB1 (Fig. 3E,F). Although BUB1^WT and BUB1^1–555 restored CDC20 kinetochore levels, BUB1^1–500 did not (Fig. 3E,F), showing that, indeed, the region between residues 501–555 is responsible for kinetochore-targeting of CDC20. Furthermore, LacI–BUB1 efficiently recruited CDC20 to ectopic sites in U2OS-LacO cells (Fig. 3G). This depended on the GLEBS motif (Fig. 3G) and, therefore, most likely required the presence of BUBR1–BUB3, although we cannot exclude a potential separate function of the GLEBS-motif (Fig. 3G). Thus, the A-box-containing segment, as well as BUBR1, was required for the ability of BUB1 to recruit CDC20. This agrees with a recent study by the Pines laboratory, published while our manuscript was under revision, which shows that recruitment of CDC20 to kinetochores requires the A-box-like motifs (referred to as the ABBA motif in that study) of both BUB1 and BUBR1 (Di Fiore et al., 2015). To retain consistency in nomenclature, we will hereafter refer to the A-box-like motif in BUB1 as the ABBA motif. Although direct (Di Fiore et al., 2015), the BUB1–CDC20 interaction might be transient because CDC20 peptides were not detected in BUB1 immunoprecipitations by using mass spectrometry (Fig. 4). We conclude that, although BUB1 promotes loading of MAD1 to unattached kinetochores, its ability to activate the SAC correlates more strongly with its ability to recruit CDC20 to kinetochores. We thus propose that this constitutes the essential role of BUB1 in SAC signaling.

In order to understand SAC activation and MCC assembly, it will be important to study the ABBA motif and KEN box motifs in BUB1, and the role of CDC20 kinetochore binding in the SAC. CDC20 recruitment to kinetochores through BUB1 might bring CDC20 in close proximity to newly formed closed (C-)MAD2 and thus facilitate the formation of C-MAD2–CDC20 dimers, as has recently been proposed for nuclear-pore-mediated C-MAD2–CDC20 formation before nuclear envelope breakdown (Rodriguez-Bravo et al., 2014). Alternatively, by ensuring CDC20 kinetochore localization, BUB1 might promote SAC signaling by allowing kinetochore-driven modification of CDC20 or by mediating a conformational change in CDC20 that allows it to bind to C-MAD2. Because BUBR1 is not required at unattached kinetochores for SAC activity, we propose a model in which C-MAD2–CDC20 dimers that have formed at kinetochores interact with cytoplasmic BUBR1–BUB3 dimers.

pCDNA5-LAP-BUB1^WT was generated by ligation of LAP–BUB1^WT into the Xho1 and Hpa1 site of pCDNA5-LAP-KNL1 (Vleugel et al., 2013) and encodes full-length siRNA-resistant BUB1. LAP–BUB1-truncation constructs were generated by introducing stop codons into LAP–BUB1^WT. LAP–MIS12–BUB1^WT was generated by PCR-mediated amplification of MIS12 to include two Xho1 sites, and then ligation into the Xho1 site of LAP–BUB1^WT. LAP–MIS12–BUB1^E252K was generated by introduction of a lysine substitution into LAP–MIS12–BUB1^WT using site-directed mutagenesis.

Tissue culture and transfection were performed as described previously (Vleugel et al., 2013). Expression of the constructs was induced by treatment with 2 µg/ml doxycycline for 24 h. The siRNA against BUB1 (5′-GAAUGUAAGCGUUCACGAA-3′) was transfected into cells twice using HiPerfect (Qiagen) for HeLa cells or Lipofectamine RNAiMAX (Life Technologies) for RPE cells for 24 h at 20 nM.

Cells were grown on 12-mm coverslips and treated as indicated in the figure legends with nocodazole (830 nM), MG132 (5 μM) and reversine (250 nM). Cells were fixed and stained as described previously (Vleugel et al., 2013). All images were acquired on a DeltaVision RT microscope (Applied Precision) with a 100×1.40 NA U Plan S Apochromat objective (Olympus) using Softworx software (Applied Precision). Images are maximum intensity projections of deconvolved stacks and were quantified as described previously (Vleugel et al., 2013). The primary antibodies used were GFP-Booster (ChromoTek), rabbit anti-BUB1 and rabbit anti-BUBR1 (Bethyl Laboratories), mouse anti-MAD1 (a gift from Andrea Musacchio, MPI, Dortmund, Germany) rabbit anti-CDC20 (Santa Cruz), guinea pig anti-CENP-C (MBL International) and CREST (anti-centromere antibodies) (Cortex Biochem). Secondary antibodies were Alexa-Fluor-488, -568 and -647 (Molecular Probes).

Cells were grown on 24-well plates and synchronized using 2 mM thymidine for 24 h. Cells were released into nocodazole (830 nM) and reversine (250 nM). Filming started 4–6 h after thymidine release and was performed at 37°C under 5% CO₂ with a 20×0.5 NA UPFLN objective on a microscope (model IX-81; Olympus) controlled by Cell-M software (Olympus). Images were acquired using a CCD camera (ORCA-ER; Hamamatsu Photonics) and processed using ImageJ software.

Cells were synchronized in mitosis by using a 24-h thymidine block, followed by overnight treatment with nocodazole. LAP–BUB1 expression was induced for 24 h, and cells were harvested, followed by immunoprecipitation and mass spectrometry analyses, as described previously (Vleugel et al., 2013).

Conservation of BUB1 residues was determined by a Jackhmmer run with human BUB1 as query on the UniProt database. Only metazoan species were included in the analysis.

We thank members of the Kops laboratory for discussions.