The C-terminal helix of BubR1 is essential for CENP-E-dependent chromosome alignment

During cell division, misaligned chromosomes are captured and aligned by motors before their segregation. The CENP-E motor is recruited to polar unattached kinetochores, to facilitate chromosome alignment. The spindle checkpoint protein BubR1 has been reported as a CENP-E interacting partner, but to what extent, if at all, BubR1 contributes to CENP-E localization at kinetochores, has remained controversial. Here we define the molecular determinants that specify the interaction between BubR1 and CENP-E. The basic C-terminal helix of BubR1 is necessary but not sufficient for CENP-E interaction, while a minimal key acidic patch on the kinetochore-targeting domain of CENP-E, is also essential. We then demonstrate that BubR1 is required for the recruitment of CENP-E to kinetochores to facilitate chromosome alignment. This BubR1-CENP-E axis is critical to align chromosomes that have failed to congress through other pathways and recapitulates the major known function of CENP-E. Overall, our studies define the molecular basis and the function for CENP-E recruitment to BubR1 at kinetochores during mammalian mitosis.


Introduction
To maintain their genomic integrity, eukaryotic cells must distribute their DNA equally to the daughter cells. Spindle microtubules mediate the segregation of chromosomes, by associating with the kinetochore of chromosomes, a large protein complex that mediates the end-on attachment of chromosomes to microtubules. At mitotic onset, chromosomes are dispersed throughout the cytoplasm, posing a challenge for their capture by microtubules from opposite poles, a pre-requisite for their accurate segregation. Multiple pathways involving microtubules and motors co-exist to ensure chromosome congression and bi-orientation (Maiato et al., 2017). A subset of chromosomes that lie outside the interpolar region assembling the spindle are dependent on CENP-E for congression. CENP-E is a large 312 kDa plus end-directed kinesin that is recruited to unattached and unaligned kinetochores, and to the outer corona, that expands around kinetochores to maximize microtubule capture (Cooke et al., 1997;Yao et al., 1997). Kinetochore-bound CENP-E moves laterally attached chromosomes to the cell equator along microtubules (Wood et al., 1997). CENP-E may also help sort kinetochore-nucleated microtubules and promote end-on attachments and biorientation (Shrestha and Draviam, 2013;Sikirzhytski et al., 2018). CENP-E then remains at aligned kinetochores, albeit with lower levels, where it plays a role in maintaining a robust connection between kinetochores and microtubules during metaphase, and during anaphase as the kinetochores are pulled to opposite poles by depolymerizing microtubules (Brown et al., 1996;Vitre et al., 2014).
CENP-E is enriched at unattached and misaligned kinetochores in early mitosis.
The human CENP-E kinetochore-targeting domain has previously been mapped (Chan et al., 1998). Over the years, CENP-E has been reported to interact with multiple kinetochore proteins: BubR1, CENP-F, Clasp2, Mad1, and other interactors such as Septin, CKAP5, NPM1 (Akera et al., 2015;Chan et al., 1998;Maffini et al., 2009;Zhu et al., 2008;Maliga et al., 2013). Post-translational modifications may also enhance CENP-E targeting to kinetochores (Ashar et al., 2000;Zhang et al., 2008). Overall, the molecular basis for CENP-E recruitment to kinetochores remains poorly understood. Its recruitment there through a dependency with the spindle-checkpoint proteins, budding uninhibited by benzimidazole 1 (Bub1) and Bub1-related (BubR1) mitotic checkpoint Ser/Thr kinases has been shown; in Xenopus and DLD-1 cells, CENP-E kinetochore levels are strongly reduced upon BubR1 depletion (Johnson et al., 2004;Mao et al., 2003). Other studies however argue CENP-E levels are not affected by BubR1 depletion. (Chan et al., 1999;Ciossani et al., 2018;Kops et al., 2004). These observed differences could be due to distinct experimental setups and different cell types and species. CENP-E recruitment to the outer corona appears independent of CDK1 activity but depends on the presence of the RZZ complex (Pereira et al., 2018;Sacristan et al., 2018). In the absence of the RZZ complex, CENP-E is recruited to kinetochores but not to the expandable corona. CENP-E remains at bioriented kinetochores after removal of checkpoint proteins, disassembly of the outer corona and throughout anaphase, indicating CENP-E has multiple yet unidentified binding partners at the kinetochore (Brown et al., 1996;Cooke et al., 1997;Gudimchuk et al., 2013). Here we characterized the kinetochore targeting domain of CENP-E biophysically and used a non-biased approach to find mitotic partners of CENP-E. We found BubR1 as a major interactor for this domain of CENP-E and defined the molecular requirements for the BubR1-CENP-E interaction. Overall, we demonstrate BubR1 contributes to CENP-E localization to kinetochores at aligned chromosomes and during spindle checkpoint activation and that this axis is essential for facilitating chromosome alignment.

Results
To define the regulation of CENP-E targeting to kinetochores, we examined quantitatively endogenous CENP-E levels at kinetochores during distinct stages of cell division, CENP-E levels were maximal during prometaphase and decreased during metaphase ( Fig S1A, B). We also observed endogenous CENP-E on microtubules in early stages of mitosis, although microtubules did not appear necessary for its kinetochore targeting. Indeed CENP-E levels were largely comparable to that in prometaphase upon nocodazole-induced depolymerization of microtubules, creating unattached kinetochores (Fig S1A, B). To analyze the molecular requirements for CENP-E localization, we precisely mapped the regions of CENP-E isoform 1 that target to kinetochores and centrosomes using transient transfection of CENP-E constructs fused to GFP (Fig 1A). CENP-E 2055CENP-E -2608 in the C terminus of CENP-E, largely similar to the previously published kinetochore-targeting construct 1958-2628, was necessary and sufficient for targeting to kinetochores in HeLa cells (Fig 1B, C)(Chan et al., 1998). The shorter CENP-E 2055-2450 still showed kinetochore localization, while an even shorter CENP-E 2055-2356 targeted weakly to a subset of kinetochores (Fig 1B, C). This heterogeneous targeting was previously observed and is likely to reflect different attachment states or kinetochore heterogeneity (Chan et al., 1998). In the absence of the first 35 amino acids in this domain, CENP-E 2090CENP-E -2450 lost the ability to target to kinetochores (Fig 1B, C). However, CENP-E 2260-2608 localized specifically to a region between the two centrioles or associated closely with one centriole both in interphase and mitosis (Fig S1C). The intercentriole interacting proteins remain unknown and this interaction will not be pursued further here. kinetochore proteins necessary for its recruitment (Fig 1D, S1D). However, GFP-CENP-E 2055-2608 was strongly enriched at kinetochores close to spindle poles, suggesting GFP-CENP-E 2055-2608 was recruited to these kinetochore subpopulations independently of endogenous CENP-E, through another binding partner (Fig 1D, S1E). GFP-CENP-E 2055-2608 appeared to compete with CENP-E at kinetochores, causing a reduction in endogenous CENP-E at kinetochores (Fig 1E) in agreement with (Schaar et al., 1997).
Additionally, CENP-E 2055-2608 transfection caused many chromosomes to become misaligned (Fig 1D,  targeting to kinetochores outcompetes endogenous CENP-E but it may also favor the targeting of kinetochores to spindle poles through its spindle pole-targeting region. To define the molecular basis for the CENP-E kinetochore-targeting domain, we expressed and purified recombinant CENP-E 2055-2608 , which robustly targets to both kinetochores and centrosomes, and CENP-E 2055-2358 , which targets weakly to kinetochores. CENP-E 2055-2608 aggregated in 150mM NaCl and was maintained in 500mM NaCl. SEC-MALS analysis revealed that CENP-E 2055-2608 assembles as a dimer in solution, while the minimal kinetochore-targeting domain CENP-E 2055-2358 was monomeric (Fig S2A, B). Circular dichroism further defined the secondary structural elements of CENP-E 2055-2608 .and CENP-E 2055-2358 ( Fig S2C). CENP-E 2055-2608 has a αhelical content of around 50.9%, while the shorter domain CENP-E 2055-2358 is 80% αhelical, with 8.8% containing turns and 12.1% containing unstructured regions ( Fig S2D).
Thus the region 2358-2608 responsible for dimerization and centrosome region-targeting, is highly likely an α-helical coiled coil. Rotary shadowing further revealed that CENP-E 2055-2608 is an elongated domain with a globular region at one end and rod-like shape, supporting a coiled coil conformation ( Fig S2E). Overall, these data indicated that the CENP-E 2055-2608 domain can be subdivided into an N-terminal monomeric α-helical rich domain, essential for kinetochore targeting and a C-terminal domain that provides dimerization properties.
We then sought to define the major CENP-E interactors at kinetochores. CENP-E is strongly recruited to unattached kinetochores at mitotic onset (Yen et al., 1991). To identify mitotic interactors of the CENP-E kinetochore-targeting domain, we incubated CENP-E 2055-2358 with clarified mitotic cell lysate, from cells arrested with the microtubuledepolymerizing drug nocodazole. We then pulled down CENP-E 2055-2358 and associated proteins, which were subjected to mass spectrometry for identification (Fig S3A, table   S1). We found CENP-E 2055-2358 specifically interacted with BubR1 and MYPT1, a protein phosphatase 1 regulatory subunit 12A ( Fig S3B). We did not pursue the interaction of  Fig S4A). CENP-E 2055-2358 is monomeric (Fig S2A, B) and not very stable in low salt. To stabilize it while mimicking dimeric CENP-E 2055-2608 , we fused it to a Cterminal GST and removed 14 residues at the N terminus, to stabilize it while mimicking dimeric CENP-E 2055-2608. CENP-E 2069-2358 -GST was more stable in low salt concentration and could then be further used to analyze the BubR1-CENP-E interaction by SEC.
CENP-E 2069-2358 -GST co-eluted with BubR1 705-1050 , as shown by the shift in the elution profile (Fig 2A), while GST alone did not ( Fig S4B). The constructs were monodisperse but we were not able to obtain diffracting crystals of the kinetochore-targeting domain of CENP-E 2055-2358 alone or bound to BubR1.
To investigate the thermodynamics of the BubR1/CENP-E interaction, we performed isothermal titration calorimetry (ITC). The CENP-E 2069-2358 -GST construct had to be optimized slightly to remove some GST contaminants and degradation. We therefore removed a further 22 residues at the N terminus and purified it in complex with BubR1 before separating the complex in high ionic strength using gel filtration. This way, we obtained >95% pure BubR1 705-1050 and CENP-E 2091-2358 -GST. At 37°C the pseudokinase domain of BubR1 bound CENP-E 2091-2358 -GST with mid-nanomolar affinity with a K d =318 ± 90 nM, (Fig 2B). At this temperature, formation of the complex had an exothermic heat signature. The enthalpic and entropic components driving the interaction were of a similar magnitude (ΔH= -5.1 ± 0.2 kcal/mol; -TΔS= -4.1 kcal/mol).
The stoichiometry between BubR1 and CENP-E 2091-2358 was determined to be 1:1. The stoichiometry must be put in the context of full-length dimeric CENP-E. Thus the CENP-E motor is able to bind two molecules of BubR1.
To further map the interaction region specific to BubR1 and CENP-E, we examined the sequence conservation between Bub1 and BubR1 kinase domains. We found a longer loop in the C terminus of BubR1 when compared to Bub1 that showed sequence divergence between human Bub1 and BubR1, but displayed sequence similarity across BubR1 species (Fig 2C). We hypothesized this region may be important for the CENP-E-BubR1 interaction Indeed, CENP-E 2069-2358 -GST did not co-elute with BubR1 705-1030 lacking the last 20 amino acids ( Fig 2D) suggesting that this part of BubR1 is critical for the interaction with CENP-E.. However, on its own this basic helix in BubR1 1030-1050 (pI=10.30), was not sufficient to interact with CENP-E ( Fig 2E). Based on the basic properties of this helix, we also mapped the interaction of BubR1 with CENP-E to the C terminus of CENP-E 2055-2358 . We found a negatively charged region in CENP-E, which we hypothesized could interact with the basic helix of the kinase domain of BubR1.
Since potential parallel pathways targeting CENP-E to kinetochores seem to coexist, we then tested whether CENP-E 2091-2358 -GST-GFP could target to kinetochores in cells and whether this recruitment was only dependent on BubR1. Transiently transfected CENP-E 2091-2358 -GST-GFP is dimeric due to GST and robustly targeted to all kinetochores, however it did not cause chromosome misalignment (Fig 3C) (Fig 4A, S4A). Cells depleted of BubR1 displayed a near complete loss of CENP-E from kinetochores in this situation, suggesting that CENP-E localization to microtubule-attached kinetochores is dependent on the residual pool of BubR1 retained at metaphase chromosomes ( Fig 4A-C, Fig S5 A,B). This was also the case when Bub1, essential for the recruitment of BubR1 ( most chromosomes were still able to form a metaphase plate, consistent with the idea that PP2A-B56 targeting was restored in this construct. However, in comparison to cells expressing GFP-BubR1 WT we observed a significant increase in cells with misaligned chromosomes ( Fig 4K). In these cells, a small number of chromosomes were unable to congress and displayed high levels of GFP-BubR1 1-1030 at kinetochores, indicating spindle checkpoint activation ( Fig 4J, K). This phenotype is very similar to that of CENP-E depletion or knockout suggesting that a pool of CENP-E required for efficient chromosome alignment was missing ( Fig 1D) (Schaar et al., 1997). CENP-E, however, was present on the same kinetochores, presumably through a pathway that does not depend on the C terminus of BubR1. The BubR1 C-terminal helix specifically recruits one pool of CENP-E to kinetochores in mitosis and during spindle checkpoint activation.
This interaction seems to be required for the productive chromosome alignment and biorientation of chromosomes. In this situation, the CENP-E recruitment to kinetochores through another pathway does not seem to enable full chromosome alignment.
Importantly, in our experiments, the GFP-BubR1 construct was expressed with levels similar to endogenous BubR1 ( Fig 4L). Overall our data indicate that BubR1 recruits CENP-E specifically to bioriented chromosomes and is important for rapid recruitment of CENP-E to unattached kinetochores during SAC activation. Yet another pathway also promotes CENP-E localization to kinetochores in the absence of BubR1 and Bub1 during the maintenance of SAC.
CENP-E is an essential motor, targeting to unattached kinetochores and playing a critical role in the congression, maintenance and biorientation of chromosomes kinetochore-targeting domain of CENP-E. However they used a construct, which also has the centrosome-targeting domain ( Fig S1C) and a second microtubule-binding site (Ciossani et al., 2018;Gudimchuk et al., 2013). In this study, we precisely map the domain of CENP-E necessary for kinetochore targeting, separating it from its centrosome and microtubule-binding functions. This domain is monomeric and associates with a 1:1 stoichiometry with the pseudokinase domain of BubR1, suggesting that full-length CENP-E can associate with 2 molecules of BubR1 at one time with nanomolar affinity. Given the high local concentration of BubR1 at unattached kinetochores, this creates an avidity effect for CENP-E binding to kinetochores to promote its recruitment. Thus the affinity of CENP-E for BubR1 may even be higher than that in vitro. There could also be cooperative binding of BubR1 to CENP-E. Our data reveal BubR1 relies on its divergent and basic C-terminal helix for CENP-E binding, creating a unique and specific association to the mitotic motor. Yet this helix is not sufficient. On CENP-E a small acidic patch is critical to specify the interaction with BubR1. Mutation of these amino acids prevents the targeting of this CENP-E 2055CENP-E -2358 domain to kinetochores and consequently compromises chromosome alignment.
Previous work on how CENP-E localizes to kinetochores remains unclear and the extent to which it requires BubR1 is conflicting. It is likely due to experimental differences between protocols. Indeed we show that BubR1 facilitates the rapid and initial recruitment of CENP-E to kinetochores at the onset of SAC signaling. Once the SAC is on for a significant period of time, we found CENP-E levels become identical at kinetochores in the presence or absence of BubR1 in good agreement with previous work (Ciossani et al., 2018). Our data indicate that BubR1 is a major interactor of CENP-E at kinetochores but there are distinct yet redundant pathways to recruit CENP-E.
BubR1 increases the kinetics of CENP-E recruitment to kinetochores during spindle checkpoint activation. The other pathways contribute to a slower but robust targeting of CENP-E to kinetochores. However they are not sufficient to restore the CENP-E function in chromosome alignment and biorientation. The BubR1-dependent recruitment of CENP-E to kinetochores is therefore essential for correct alignment and biorientation of kinetochores. In the absence of this CENP-E pool at kinetochores, the kinetochoremicrotubule attachment is compromised, even when high levels of CENP-E are present ( Fig 4J, K). Our work to reveal the molecular basis for BubR1/CENP-E binding at kinetochores will now further facilitate the identification of BubR1-independent pathways that allow CENP-E to associate with kinetochores and the outer corona, to define the contribution of this CENP-E pool to chromosome alignment and biorientation. Future work is needed to address the nature and regulation of the multiple interaction partners of CENP-E to understand its critical role in chromosome congression and maintaining biorientation of kinetochores.

Author contributions
TL and JW designed the project. TL, DH, JW, EAB, CS and AGK performed experiments TL, DH, EAB, AGK, UG and JW performed data analysis and interpretation. JR secured funding for mass spectrometry equipment. JW wrote the manuscript. All revised the manuscript.

Cloning
To assay the localization in cell culture of CENP-E subdomains, various constructs were generated from CENP-E transcript variant 1 (NM_001813.2) and cloned into pBABEpuro containing an N-or C-terminal GFP tag and using restriction enzymes (Cheeseman and Desai, 2005). Bacterially-expressed constructs were cloned in pET-3aTr (Tan, 2001).
Details of constructs and cloning are listed in Table S2.
Protein expression, purification and assays

Statistics and reproducibility
Statistical analyses were performed using GraphPad Prism 6.0. No statistical method was used to predetermine sample size. All experiments were performed and quantified from at least three independent experiments, unless specified and the representative data are shown.
Data availability.
All data and reagents supporting the findings of this study are available from the corresponding author on request.    Numbers of kinetochores analyzed for cells treated with 2.5h MG132 and 2.5h MG132+ 5 minutes nocodazole after RNAi depletion were respectively: n control = 185, 140; n BuBR1 =140, 98; n Bub1 =140, 145; n ZW10 =111, 159; n Bub1/BubR1 =120, 100; n Bub1/BubR1/ZW10 =60, 100. Asterisks indicate significance value performed using a ANOVA one-way test. **** indicate a P-value<0.0001.   K  i  n  e  t  o  c  h  o  r  e  E  x  p  a  n  s  i  o  n  i  n  t  h  e  A  b  s  e  n  c  e  o  f  M  i  c  r  o  t  u  b  u  l  e  A  t  t  a  c  h  m  e  n  t  .   C  u  r  r  B  i  o  l   .   2  8  :  3  4  0  8  -3  4  2  1  e  3  4  0  8  .   R  a  p  p  s  i  l  b  e  r  ,  J  .  ,  Y  .  I  s  h  i  h  a  m  a  ,  a  n  d  M  .  M  a  n  n  .  2  0  0  3  .  S  t  o  p  a  n  d  g  o  e  x  t  r  a  c  t  i  o  n  t  i  p  s  f  o  r   m  a  t  r  i  x  -a  s  s  i  s  t  e  d  l  a  s  e  r  d  e  s  o  r  p  t  i  o  n  /  i  o  n  i  z  a  t  i  o  n  ,  n  a  n  o  e  l  e  c  t  r  o  s  p  r S  e  p  t  i  n  7  i  n  t  e  r  a  c  t  s  w  i  t  h  c  e  n  t  r  o  m  e  r  e  a  s  s  o  c  i  a  t  e  d  p  r  o  t  e  i  n  E  a  n  d  i  s  r  e  q  u  i  r  e  d  f  o  r  i  t  s  k  i  n  e  t  o  c  h  o  r  e  l  o  c  a  l  i  z  a  t  i  o  n  .   J  B  i  o  l  C  h  e  m   .  2  8  3  :  1  8  9  1  6  -1  8  9  2  5 .