Cross-regulation between Aurora B and Citron kinase controls midbody architecture in cytokinesis

Cytokinesis culminates in the final separation, or abscission, of the two daughter cells at the end of cell division. Abscission relies on an organelle, the midbody, which forms at the intercellular bridge and is composed of various proteins arranged in a precise stereotypic pattern. The molecular mechanisms controlling midbody organization and function, however, are obscure. Here we show that proper midbody architecture requires cross-regulation between two cell division kinases, Citron kinase (CIT-K) and Aurora B, the kinase component of the chromosomal passenger complex (CPC). CIT-K interacts directly with three CPC components and is required for proper midbody architecture and the orderly arrangement of midbody proteins, including the CPC. In addition, we show that CIT-K promotes Aurora B activity through phosphorylation of the INCENP CPC subunit at the TSS motif. In turn, Aurora B controls CIT-K localization and association with its central spindle partners through phosphorylation of CIT-K's coiled coil domain. Our results identify, for the first time, a cross-regulatory mechanism between two kinases during cytokinesis, which is crucial for establishing the stereotyped organization of midbody proteins.


Introduction
Cytokinesis controls the proper partitioning of cytoplasmic and nuclear material between the two nascent daughter cells. Defects in this process have been associated with various human diseases [1] and can cause polyploidy, which in turn can lead to chromosomal instability, a hallmark of cancer cells [2]. Cytokinesis progresses through a series of sequential events. First, after anaphase onset the mitotic spindle is completely reorganized into an array of interdigitating and antiparallel microtubules, known as the central spindle [3]. These microtubules are tightly bundled at their plus ends in a region known as the spindle midzone. Signals from astral and central spindle microtubules promote the formation of a cleavage furrow at the cell's equator that ingresses to bisect the mother cell [4]. Furrow ingression is driven by the assembly and constriction of an actomyosin contractile ring, which progressively compacts the central spindle to form an organelle known as the midbody. The midbody provides a platform necessary for the recruitment and organization of many proteins that regulate the final abscission of the two daughter cells [5] and it has also been proposed to contribute to cell fate determination [6,7], albeit this has not been confirmed in intact organisms [8,9]. The midbody contains at its centre an electron dense structure known as the 'midbody matrix' or 'midbody ring' (figure 1). The regions flanking the midbody ring are often indicated as 'midbody arms'. Some midbody proteins are former contractile ring and central spindle components, while others are recruited after completion of furrow ingression. All these proteins show a very precise and & 2016 The Authors. Published by the Royal Society under the terms of the Creative Commons Attribution stereotypic distribution along the midbody [10]. For example, some contractile ring and midzone components, such as Citron kinase (CIT-K) and the kinesin KIF23/MKLP1, accumulate at the midbody ring, while other central spindle proteins, such as the chromosomal passenger complex (CPC), localize to the midbody arms. This precise distribution pattern is necessary for midbody assembly and function, but what controls the architecture of the midbody is unclear. CIT-K is emerging as an important midbody organizer. This kinase links a network of contractile ring and central spindle proteins-including actin, Anillin, myosin, KIF23, KIF14 and RhoA-in both Drosophila and human cells and has also recently been shown to be important for the stability of midbody microtubules in some mouse tissues [11][12][13][14][15][16]. CIT-K was initially identified as a RhoA effector and proposed to promote contractile ring constriction through phosphorylation of the myosin regulatory light chain (MRLC) [17,18]. However, CIT-K and its Drosophila orthologue, Sticky (Sti), are not required for furrowing [13,19 -22] and evidence in both Drosophila and human cells has shown that CIT-K is required for proper RhoA localization at the cleavage site during late cytokinesis, thus behaving more like a RhoA regulator than an effector [12,13]. Moreover, Sti is not required for MRLC phosphorylation [12,23]. Together, these findings indicate that CIT-K is not required for the early stages of cleavage furrow formation and ingression, but rather for the proper organization and architecture of the midbody in late cytokinesis. Here we report the characterization of the CIT-K interactome in human cells and present evidence that this kinase plays a key role in organizing the proper distribution of midbody components. We also report that CIT-K binds to three CPC subunits, including its kinase component Aurora B, both in vivo and in vitro. Furthermore, we show that CIT-K is required for correct CPC localization and phosphorylates the INCENP subunit of the CPC at the TSS Aurora B activation site. Finally, Aurora B phosphorylates CIT-K to control its localization and interaction with central spindle partners. Together, our findings reveal a novel cross-regulatory mechanism between CIT-K and the CPC necessary to regulate midbody architecture.

CIT-K interacts with the CPC in vivo and in vitro
We previously showed that the midbody matrix appears scarce, fragmented and disorganized in Drosophila cells after CIT-K depletion [11]. We observed an almost identical phenotype in human HeLa cells after CIT-K depletion. Electron microscopy (EM) analysis revealed that the midbody matrix was fragmented, disorganized, detached from the cortex and mis-positioned towards one of the two nascent daughter cells (figure 1). Moreover, cortical blebs were often evident where gaps between the midbody matrix and the membrane occurred ( figure 1, arrowhead). These findings confirmed a key role for CIT-K in midbody formation in human cells and prompted us to identify its interactome to understand the mechanisms that control midbody architecture. We generated a monoclonal HeLa cell line stably expressing a Flag-tagged CIT-K transgene that displayed correct localization pattern and rescued the cytokinesis failure caused by depletion of endogenous CIT-K, indicating that it is fully functional (electronic supplementary material, figure S1). By using affinity purification coupled with mass spectrometry (AP-MS), we identified 350 proteins that were specifically pulled down by Flag::CIT-K, but not Flag alone, in HeLa cells synchronized in telophase (figure 2a; electronic supplementary  material, table S1). Among these proteins we found some expected previously identified partners, such as KIF14 [14] and the two components of the centralspindlin complex, KIF23 and RacGAP1 [11], and also other novel interactors, including three of the four CPC components, Aurora B, Borealin and INCENP (figure 2b; electronic supplementary material, table S1). It is very likely that the other CPC component, Survivin, was not identified simply because of MS limitations. The kinesin KIF20A/MKLP2, responsible for CPC translocation from centromeres to the spindle midzone [24], was also identified, albeit with a low score (figure 2b; electronic supplementary material, table S1). Notably, the microtubule-binding protein PRC1-a well-known direct binding partner of both KIF14 and KIF23 [11,14,25] but not of CIT-K Figure 1. CIT-K is required for midbody architecture. EM images of midbodies in HeLa Kyoto cells treated with an siRNA directed against either a random sequence (control) or CIT-K for 48 h. The bottom panels are magnification of the insets as indicated by dotted lines. The arrowhead marks a membrane bleb and the arrows indicate gaps between the midbody matrix and the cortex. MM, midbody matrix. Scale bars, 1 mm.
[11]-was not found in our survey, suggesting that our analysis preferentially identified proteins strongly associated with CIT-K. These findings were confirmed by western blot (figure 2c) and suggested that an interaction between CIT-K and the CPC might be important for midbody formation.
CIT-K and the CPC both localize to the cleavage furrow [17,26], but unfortunately we failed to observe localization of the CPC to the equatorial cortex using antibodies against Aurora B and INCENP. However, we found that CIT-K colocalized with Venus-tagged Aurora B at the ingressing cortex in a transgenic U2OS cell line [27] (figure 2d), albeit fluorescence intensity profiles indicated that Aurora B-Venus was distributed closer to the inside edge of the furrow (figure 2d). CIT-K has been described to localize predominantly to the cleavage furrow (figure 2d) [17,19], but Flag::CIT-K also weakly accumulated at the spindle midzone (electronic supplementary material, figure S1, arrow). This does not seem to be an over-expression artefact because a similar midzone-associated localization of endogenous CIT-K has already been reported [28], and we could also observe it in isolated midbodies (figure 2e and f). This midzone signal partially overlapped with the CPC in mid-telophase, but in late-telophase CIT-K accumulated at the midbody ring, whereas the CPC localized to the midbody arms (figure 2f).
To assess whether CIT-K could directly bind to the CPC, we purified recombinant CIT-K fragments tagged with glutathione S-transferase (GST) from bacteria and tested their ability to pull down various in vitro translated and radiolabelled polypeptides of the three CPC subunits identified in our AP-MS experiments: Aurora B, Borealin and INCENP (figure 3a-c). All these fragments were chosen on the basis of the structures of the proteins and their known interaction domains. We carried out multiple experiments using increasing salt concentrations during the washes in order to have an indication of the strength of the interactions (figure 3c). The INCENP central region (residues 262-524) was never pulled down by any of the GST::CIT-K fragments (figure 3b,c). The CIT-K CNH domain interacted very strongly with Aurora B, Borealin and the N-terminal region of INCENP (residues 1-261) (figure 3b,c). A weaker interaction was observed between the CIT-K kinase domain and the same CPC polypeptides that bound to the CNH region (figure 3b,c). Finally, the CIT-K CC2 region pulled down very well the INCENP C-terminal region (residues 525-918) and less efficiently  (284) and the proteins that co-purified with Flag::CIT-K (449). Note that 350 proteins specifically associated with Flag::CIT-K. (b) Partial list of the CIT-K specific interactors identified by affinity purification. A full list can be found in the electronic supplementary material, table S1, worksheet 1. The bait, CIT-K, is highlighted in red and previously known CIT-K binding partners are in black. The three subunits of the CPC are in blue and the kinesin responsible for the translocation of the CPC to the spindle midzone, KIF20A, is in green. (c) HeLa Kyoto cells stably expressing either Flag-tagged CIT-K or Flag alone were synchronized in telophase by thymidine/ nocodazole block and release and then protein extracts were used in a pull-down assay using anti-Flag antibodies. The extracts and pull downs were analysed by western blot to detect KIF14, KIF23, Aurora B, KIF20A, PRC1 and Flag. Note that a slower migrating KIF23 band was detected in the pull-down compared with whole cell extracts. This is probably a phosphorylated form that we routinely observe in protein extracts from both telophase cells and purified midbodies. INCENP 525 -918 pulled down the CIT-K kinase and CC2 domains (figure 3d), but the interaction with the CIT-K kinase domain was never observed in the previous reciprocal assay (figure 3b,c). We conclude from all these in vitro binding experiments that the CIT-K CNH C-terminal region binds very strongly to the INCENP N-terminal region, Borealin and possibly Aurora B. The other two CIT-K regions, kinase and CC2, showed weaker interactions with the INCENP N-terminal and C-terminal regions, respectively. Moreover, they also weakly interacted with Borealin. However, it is difficult to imagine that all these interactions occur simultaneously and we speculate that, in vivo, the association between the CNH domain and the CPC components might be predominant. The exact function of the CIT-K CNH domain is unclear, but its predicted structure of a seven-bladed b-propeller (figure 3e), a structure known to function as a hub for multiple protein-protein interactions and found in scaffolding proteins of the WD40 family [29,30], is perfectly consistent with our binding results.

CIT-K is required for proper CPC localization and the orderly arrangement of midbody proteins
To investigate the functional significance of the interaction between CIT-K and the CPC, we first analysed CPC localization after CIT-K depletion. The CPC relocates from centromeres to the spindle midzone after anaphase onset, and then accumulates at the midbody arms [31]. The CPC's subunit INCENP   rsob.royalsocietypublishing.org Open Biol. 6: 160019 visualize not only CPC localization but also the entire midbody architecture, we simultaneously stained CIT-K depleted cells for Aurora B, which marks the midbody arms, and for the kinesin KIF23, a well-documented midbody ring/matrix marker (figure 4e). After CIT-K depletion, KIF23 no longer localized between the two CPC bands and appeared to be 'pushed' towards one side of the midbody in mid-telophase (figure 4e).
In late-telophase CIT-K RNAi cells, KIF23 was often found to one side of the CPC (figure 4e), even when two Aurora B bands were clearly evident (figure 4e, bottom panels). These phenotypes were never observed in control cells. To conclude, our results indicate that CIT-K is required for proper CPC localization and for the orderly arrangement of midbody proteins.

CIT-K phosphorylates INCENP at the TSS Aurora B activation site
To establish whether the components of the CPC could be CIT-K substrates, we developed an in vitro kinase assay using the isolated kinase domains (wild-type and kinase dead; KD) of human and mouse CIT-K purified from bacteria. These CIT-K depletion in HeLa cells caused a 35% reduction in the pTSS signal by immunofluorescence (figure 5e). As phosphorylation of the TSS motif increases Aurora B kinase activity [34][35][36], our data suggested that CIT-K could be important to elicit full Aurora B activation at the midbody. Consistent with this hypothesis, CIT-K depletion caused a significant decrease in Aurora B phosphorylation at T232 (figure 5f), which is essential to achieve full Aurora B activation following INCENP TSS phosphorylation [35,37]. Notably, although CPC components often failed to properly localize to the midbody after CIT-K RNAi, their accumulation was not reduced ( figure 4). Therefore, we conclude that CIT-K is required for proper localization and phosphorylation of INCENP and Aurora B, but not their recruitment, at the midbody.

CIT-K localization during cytokinesis requires Aurora B activity
We next analysed CIT-K localization after perturbation of CPC activity. After treatment with the Aurora B inhibitor ZM447439, CIT-K showed reduced accumulation at the ingressing furrow in early telophase (figure 6a). Treatment with ZM447439 impairs the spindle assembly checkpoint and cells exit mitosis very rapidly, often failing to assemble a proper central spindle and midbody (figure 6a). Thus, to investigate the role of the CPC in CIT-K localization in late cytokinesis, we prevented the translocation of this complex to the spindle midzone by depleting the kinesin KIF20A [24]. As expected, INCENP failed to accumulate at the spindle midzone and persisted at centromeres after KIF20A depletion (figure 6b,c). Consistent with the results obtained after Aurora B inhibition (figure 6a), CIT-K showed reduced accumulation at the cleavage furrow in early telophase KIF20A RNAi cells (figure 6c). Moreover, CIT-K failed to make a compact structure at the midbody ring and appeared diffuse along the cortex in about 47% of late-telophase KIF20A RNAi cells, a phenotype very rarely observed (1.3%) in control cells (figure 6c,d). EM ultrastructural analysis confirmed that, after KIF20A depletion, the midbody matrix was scarce, fragmented and mis-localized, similar to and possibly even slightly more severe than in CIT-K RNAi cells (figure 6e cf. figure 1). Moreover, KIF20A depleted cells also showed KIF23 mislocalization towards one side of the midbody, exactly like after CIT-K RNAi (figure 6f cf. figure 4e). These results indicate that CPC recruitment to the midbody is necessary for proper CIT-K localization and that the CPC and CIT-K cooperate to organize the architecture of the midbody.
To distinguish whether CIT-K localization required the whole CPC or just Aurora B activity, we tested if a chimera composed of the microtubule-binding domain of the spindle midzone protein PRC1 coupled with an Aurora B kinase module (dubbed Baronase and comprising a truncated form of Aurora B and the INCENP activating region) [38] could rescue CIT-K localization after KIF20A depletion. Cells carrying a doxycycline-inducible PRC1::Baronase transgenic chimera were depleted of KIF20A and CIT-K localization was analysed with or without expression of the PRC1::Baronase chimera. In the absence of transgene expression, cells showed a frequency of CIT-K abnormal localization very similar to non-transgenic HeLa cells (figure 6g,h). Induction of PRC1::Baronase expression, however, considerably rescued CIT-K localization (figure 6g,h). These results indicate that CIT-K localization requires primarily Aurora B activity.

Aurora B phosphorylates CIT-K to regulate its interaction with central spindle components
As the results described in the previous paragraph indicated that CIT-K localization depends on Aurora B activity, we next determined whether Aurora B phosphorylated CIT-K. We first used MS to analyse the phosphorylation pattern of our Flag::CIT-K transgene purified from cells synchronized in telophase. Four out of the twelve identified phosphorylated sites matched the consensus for Aurora B kinase: S480, S699, S1385 and S1962 (figure 7a; electronic supplementary  material, table S2). To address if CIT-K could be a direct substrate of Aurora B, we carried out in vitro phosphorylation assays using recombinant Aurora B and bacterially purified rsob.royalsocietypublishing.org Open Biol. 6: 160019 rsob.royalsocietypublishing.org Open Biol. 6: 160019 GST::CIT-K peptides. Aurora B phosphorylated three CIT-K regions containing the kinase, CC1 and C1þPH regions (figure 7b). Consistent with these in vitro results, two residues found to be phosphorylated in vivo, S480 and S699, lie within the CC1 region and one, S1385, lies within the C1 domain (figure 7a). MS analysis confirmed that S699, but not S480, was phosphorylated by Aurora B in vitro and, consistent with this, a CC1 fragment containing an S to A mutation at this residue (S699A) was no longer efficiently phosphorylated by Aurora B in vitro (figure 7c). MS analysis also identified residues in the C1 þ PH region phosphorylated by Aurora B in vitro. The most highly phosphorylated residues were a stretch of two S and one T residue at positions 1385-1387 and S1474, which were also found phosphorylated in vivo (figure 7a). However, only substitution of the residues at positions 1385-1387 into alanines (TripleA mutant) significantly reduced phosphorylation by Aurora B in vitro (figure 7d). Together these results indicate that at least two CIT-K regions are phosphorylated by Aurora B in vitro and in vivo: S699 and the three SST residues at 1385-1387. It is, however, possible that other post-translational modifications and/or factors might be necessary for Aurora B to recognize and phosphorylate other CIT-K residues, such as S480 and S1474, in vivo.
As CIT-K interacts with both KIF14 and KIF23 via its CC1 domain [11,15], our results suggested that Aurora B could regulate the interaction of CIT-K with these two kinesins and possibly other partners. To address this hypothesis, we generated a monoclonal cell line expressing a Flag-tagged CIT-K S699A non-phosphorylatable mutant to test its localization and interaction with central spindle partners after depletion of endogenous CIT-K. The S699A mutation enhanced the association of CIT-K with KIF23 and Aurora B, but only marginally affected the interaction with KIF14 (figure 7e). Longer exposures confirmed that Flag-tagged wild-type CIT-K was also able to bind KIF23 and Aurora B (electronic supplementary material, figure S5), as already shown (figure 2c). Consistent with this result, the CIT-K S699A mutant showed strongly increased accumulation at the spindle midzone and irregular cortical localization in early stages of cytokinesis (figure 7f,g). Together these findings indicate that CIT-K is an Aurora B substrate and that Aurora B phosphorylation at S699 dampens the association of CIT-K with KIF23 and the CPC in order to reduce CIT-K accumulation at the spindle midzone in early cytokinesis.

Discussion
Cross-regulation among mitotic kinases, and in particular between Aurora B and Polo-like kinase 1 (Plk1), is emerging rsob.royalsocietypublishing.org Open Biol. 6: 160019 as an important mechanism to fine-tune important events and to improve fidelity and robustness during cell division [39 -41]. Here we show for the first time that cross-regulation between two kinases, Aurora B and CIT-K, is crucial for regulation of midbody formation. Our findings identify CIT-K as a key factor that controls the orderly arrangement of proteins, including the CPC, at the midbody. These results reinforce and expand the evidence that CIT-K plays a unique role in shaping midbody architecture. Our data support a model in which, thanks to its ability to bind simultaneously to proteins localized to the midbody ring (i.e. KIF23 and KIF14) and to the midbody arms (i.e. the CPC), CIT-K could act as both bridge and organizer (figure 8). However, it is important to point out that KIF23 and the CPC, albeit improperly localized, remained separate in the absence of CIT-K (figure 4e). This indicates that additional mechanisms must exist to prevent the intermingling of midbody ring and midbody proteins. We cannot categorically exclude that aberrant localization of midbody proteins could be an indirect effect of the abnormal midbody structure observed after CIT-K depletion. However, the evidence that CIT-K directly associates with these midbody components (this paper and [11]) and that KIF23 fails to localize to the midbody 'dark region' even when the CPC appears to localize normally (figure 4e), strongly argue against this possibility. CIT-K and the CPC display a complex binding pattern in vitro (figure 3). The very strong interaction observed between the CIT-K CNH region and all the three CPC subunits analysed (figure 3b,c) strongly suggests that this binding might be prevalent in vivo, and it is tempting to speculate that different 'blades' of the CNH propeller could interact with distinct components of the CPC. It is noteworthy that, consistent with these findings, our previous study in Drosophila indicated that a CIT-K mutant lacking the CNH domain could not fully rescue cytokinesis failure after depletion of the endogenous kinase [12]. We cannot exclude, however, that additional minor interactions between other CIT-K domains and the CPC could also occur, and in vivo cross-linking experiments and detailed structural analyses will be required to define this complex interaction.
Our results indicate that Aurora B phosphorylation controls CIT-K localization and its interaction with spindle midzone partners through phosphorylation of the CC1 region (figures 4, 6 and 7). This also implies that the interaction of CIT-K with the CPC is probably not important for CIT-K localization. Our data indicate that Aurora B phosphorylation prevents accumulation of CIT-K at the spindle midzone in early cytokinesis. We speculate that this may act as a timing mechanism to ensure that CIT-K associates strongly with KIF23 and the CPC only in late telophase, when CPC activity diminishes. The localization of CIT-K S669A does not entirely reflect the distribution of CIT-K after Aurora B inactivation or CPC removal from the spindle midzone ( figure 6). This indicates that Aurora B might control CIT-K localization also via phosphorylation of other CIT-K regions and/or other cytokinesis proteins. Other CIT-K residues phosphorylated in vivo that match the Aurora B consensus site were found downstream of the CNH domain and within the C1 region (figure 7a). However, the CNH region was not phosphorylated by Aurora B in vitro (figure 7b), indicating either that another kinase is responsible for this phosphorylation, or that Aurora B might require other post-translational modifications in order to recognize this site. By contrast, the C1 region was strongly phosphorylated in vivo and by Aurora B in vitro (figure 7), suggesting that the function of this domain is tightly controlled by Aurora B and possibly other kinases. The role of CIT-K's C1 domain has not been characterized yet, but it most likely mediates the interaction of CIT-K with other proteins and/or  Figure 8. Cartoon illustrating the cross-regulation between Aurora B and CIT-K in midbody assembly. The red arrows indicate the phosphorylation of the CIT-K CC1 and C1 domains by Aurora B and the phosphorylation of the INCENP TSS motif by CIT-K. In the presence of CIT-K (a), midbody components including the CPC, centralspindlin, KIF20A and KIF14 are properly aligned along the midbody. By contrast, after depletion of CIT-K (b) KIF14 is not recruited to the midbody [12,14], the CPC and KIF20A are dispersed and centralspindlin is mis-positioned on one side of the midbody. rsob.royalsocietypublishing.org Open Biol. 6: 160019 membrane phospholipids like the C1 domain of RacGAP1 [42]. Although the roles of these phosphorylation events require further investigation, our results strongly indicate that Aurora B could regulate multiple CIT-K functions. It is also worth mentioning that most of the other CIT-K phosphorylated sites identified in vivo match the consensus of other important mitotic kinases, such as Plk1 and Cyclin-dependent kinases, raising the possibility that multiple signalling pathways might converge to regulate the function of this key regulator of midbody architecture.
CIT-K kinase activity is necessary for proper CPC localization and to elicit full activation of Aurora B at the midbody through phosphorylation of the INCENP TSS motif (figures 4 and 5). As Aurora B phosphorylation in turn controls CIT-K distribution ( figure 7), this data suggests the existence of a feedback regulatory loop by which CIT-K reinforces both its own localization and proper architecture of the entire midbody (figure 8). Our findings are consistent with previous results that CIT-K kinase activity is necessary for successful cytokinesis in Drosophila and human cells [12,15]. INCENP is the first CIT-K substrate that has been confirmed in vivo. Previous reports indicating that human CIT-K phosphorylated the MRLC [18] were based solely on gain-of-function experiments and have been challenged by the evidence that CIT-K depletion did not affect MRLC phosphorylation in Drosophila cells [12,23]. It will be important in the future to identify the substrates of this important cytokinetic kinase to fully understand its functions during cytokinesis.

Molecular biology
The cDNA clones for Aurora B, INCENP and Borealin were kind gifts of Dr C. Lindon (Department of Genetics, University of Cambridge, UK). Cloning of the full-length open reading frame (ORF) of human CIT-K was performed by sequentially inserting three different fragments of the CIT-K cDNA sequence into the I.M.A.G.E. cDNA clone 4149886 (for simplicity, renamed construct A) (Source Bioscience). This construct encodes the 5 0 -terminal region (nt 4-811) of the CIT-K transcript (RefSeq accession number: NM_001206999.1) cloned within the pCMV-SPORT6 vector. First, poly(A) messenger RNA (mRNA) was extracted from HeLa Kyoto cells using the MicroPoly(A)Purist Kit (Ambion) and transcribed into cDNAs using random primers and the SuperScript First-Strand Synthesis System for RT-PCR (Invitrogen). The resulting cDNA library was then used as a template to PCR-amplify two CIT-K fragments dubbed CIT-K kin (nt 267-1385) and CIT-K mid (nt 1162-4923), using the primers listed in the electronic supplementary material, table S1 and the Platinum Taq DNA Polymerase High Fidelity (Invitrogen). These PCR products were subsequently cloned individually into the pGEM-T Easy vector using the pGEM-T Easy Vector System (Promega). Through digestion with the BamHI and NotI restriction enzymes, the CIT-K kin fragment was excised from the pGEM-T Easy vector and cloned into construct A in frame with the upstream 5 0 -terminal CIT-K portion. The CIT-K mid region was then inserted into the resulting plasmid using the BglII and NotI restriction sites. Finally, the CIT-K C-terminal region (CIT-K Ct), spanning nt 4457-6293, was PCR-amplified from the I.M.A.G.E. cDNA clone 40133624 (Source Bioscience), digested with BstEII and NotI, and ligated to the rest of the sequence to obtain the full-length human CIT-K ORF. CIT-K kinase dead (K126A) [43] and S699A mutants were generated by site-directed mutagenesis using the QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent) following manufacturer's instructions.
For expression of Flag-tagged CIT-K proteins in HeLa Kyoto cells, a Gateway destination vector was generated as follows: the DNA sequence coding for three copies of the Flag epitope and four copies of the Myc epitope (indicated in the text simply as Flag), followed by the Gateway cassette for N-terminal fusion, was PCR-amplified from the pMYFN Gateway destination vector and then cloned into the pIRESpuro3 vector (Clontech) using the NheI and EcoRV restriction enzymes.
For bacterial expression of GST-tagged proteins, the pDEST15 Gateway destination vector (Invitrogen) was used. All DNA constructs were verified by sequencing.

Affinity purification
For large-scale affinity purifications of Flag-tagged CIT-K, 4 Â 10 7 cells were synchronized in telophase using a thymidinenocodazole block and release procedure. They were first arrested in S phase by the addition of 2 mM thymidine (Sigma-Aldrich) for 19 h, washed twice with phosphate-buffered saline (PBS) and released for 5 h in fresh complete medium. After release, cells were cultured for an additional 13 h in fresh complete medium containing 50 ng ml 21 nocodazole (Sigma-Aldrich) and then harvested by mitotic shake-off. Mitotic cells were washed three times with PBS, released in fresh medium for 1.5 h and incubated for a further 15 min with the CDK1 inhibitor RO3306 (Calbiochem) at a final concentration of 10 mM. Cells were then harvested by centrifugation and frozen in liquid nitrogen. The cell pellet was resuspended in 5 ml of extraction buffer (50 mM HEPES pH 7.5, 100 mM KAc, 150 mM NaCl, 2 mM MgCl 2 , 1 mM EGTA, 0.5% (v/v) NP-40, 1 mM DTT, 5% (v/v) glycerol and Roche Complete Protease Inhibitors) and homogenized using a highperformance disperser (Fisher). The homogenate was clarified by centrifugation at 750g for 20 min at 48C and the supernatant was incubated with 200 ml of anti-FLAG M2 Magnetic Beads (Sigma-Aldrich) for 2-4 h on a rotating wheel at 48C. Beads were then washed four times in 10 ml of extraction buffer for 5 min on a rotating wheel, transferred to a new tube and washed one more time in 10 ml of PBS. Proteins were eluted from beads with 0.5 M NH 4 OH and 0.5 mM EDTA, concentrated, acetone precipitated and analysed by LC-MS/MS.
A very similar procedure was used for western blot analyses. Ten million cells were resuspended in 3 ml of extraction buffer and 50 ml of anti-FLAG M2 Magnetic Beads were added to the supernatant. Washes were performed with 10 ml of extraction buffer. After the last wash in PBS, beads were resuspended in 2Â Laemmli sample buffer (Sigma-Aldrich), boiled for 10 min and stored at 2208C. Proteins were separated on an SDS-PAGE gel, transferred onto PVDF membrane and probed to detect the antigens indicated in the figure legends.

Midbody purification
HeLa S3 cells were synchronized in metaphase by thymidinenocodazole block as described in the previous section, harvested by mitotic shake-off and incubated for an additional 2 h in fresh medium supplemented with 10 mM MG132 (Sigma). Cells were then washed three times with PBS, released in normal medium and allowed to progress throughout mitosis. After approximately 1.5 h, when the vast majority of cells had completed furrowing, 5 mg ml 21 taxol was added to the medium for 2-3 min to stabilize microtubules in vivo. Cells were then harvested by centrifugation, washed once with warm H 2 O and gently resuspended in a hypotonic swelling solution containing 1 mM PIPES pH 7, 1 mM MgCl 2 , 5 mg ml 21 taxol (Sigma) and Roche Complete Protease Inhibitors. Cells were immediately centrifuged at 200g for 3 min, resuspended in lysis buffer (1 mM PIPES pH 7, 1 mM EGTA, 1% (v/v) NP-40, 5 mg ml 21 taxol and Roche Complete Protease Inhibitors) and vortexed vigorously. After the addition of 0.3 volumes of cold 50 mM 2-(N-morpholino) ethane sulfonic acid (MES) pH 6.3, cells were incubated on ice for 20 min and then centrifuged at 200g for 10 min at 48C. The supernatant was transferred to a new tube and centrifuged at 650g for 20 min at 48C to pellet midbodies. The pellet was resuspended in 50 mM MES pH 6.3 and centrifuged through a cushion of 40% (w/v) glycerol in 50 mM MES pH 6.3 at 2800g for 45 min at 48C. After a final wash in 50 mM MES pH 6.3, midbodies were plated on poly-lysine-coated coverslips and processed for immunofluorescence as described below.

Fluorescence microscopy
HeLa Kyoto cells were grown on microscope glass coverslips (Menzel-Gläser) and fixed in PHEM buffer (60 mM Pipes, 25 mM HEPES pH 7, 10 mM EGTA, 4 mM MgCl 2 , 3.7% (v/v) formaldehyde) for 12 min. They were then washed three times for 10 min with PBS and incubated in blocking buffer (PBS, 0.5% (v/v) Triton X-100 and 1% (w/v) BSA) for 1 h at room temperature (RT). Coverslips were incubated overnight at 48C with the primary antibodies indicated in the figure legends, diluted in PBT (PBS, 0.1% (v/v) Triton X-100 and 1% (w/v) BSA). The day after, coverslips were washed twice for 5 min in PBT, incubated with secondary antibodies diluted in PBT for 2 h at RT and then washed twice with PBT and once with PBS. Coverslips were mounted on SuperFrost Microscope Slides (VWR) using VECTASHIELD Mounting Medium containing DAPI (Vector Laboratories). Phenotypes were scored blind and by at least two people independently. IMAGEJ software was used to generate RGB fluorescence profiles and intensity values.

Transmission electron microscopy
For electron microscopy analyses, asynchronous HeLa Kyoto cells were plated on microscope glass coverslips (Menzel-Gläser), incubated for 19 h with 2 mM thymidine, washed three times with PBS, released for 5 h in fresh complete medium, and incubated for 13 h with 10 mM RO3306. Mitotic cells were then gently washed with warm PBS, and released in fresh medium for 2 h. Cells were fixed overnight at 48C in 2.5% (v/v) glutaraldehyde in PBS, post-fixed for 1 h in 1% (v/v) OsO 4 in PBS, dehydrated in a graded series of alcohols, embedded in Epon-Araldite resin, and polymerized for 2 days at 608C. Glass slides were separated from the resin after a short immersion in liquid nitrogen. Sections were obtained with an LKB ultratome, stained with uranyl acetate and lead citrate, and observed and photographed with an FEI Tecnai G2 Spirit transmission electron microscope operating at an accelerating voltage of 100 kV and equipped with a Morada CCD camera (Olympus).

In vitro binding assay
DNA fragments coding for CIT-K fragments (kinase, CC1, CC2, C1 þ PH and CNH), Aurora B, Borealin, INCENP 1-261 and INCENP 525-918 were generated by PCR and cloned into pDEST15 (Invitrogen) to express N-terminal GST-tagged polypeptides in Escherichia coli. The GST-tagged products were then purified using Glutathione Sepharose 4B according to manufacturer's instruction (GE Healthcare). [ 35 S] Methionine-labelled Aurora B, Borealin, INCENP (all three fragments) and CIT-K fragments (kinase, CC2 and CNH) were prepared from rsob.royalsocietypublishing.org Open Biol. 6: 160019 corresponding PCR products amplified using primers harbouring a T7 promoter and then in vitro transcribed and translated (IVTT) using the TnT T7 Quick Coupled Transcription/Translation System (Promega) in the presence of [ 35 S] methionine (PerkinElmer). The binding reaction contained 150 mM NaCl and subsequent washes varied from 150 mM to 1 M NaCl. GST pull-down assays were carried out as described [12].

In vitro phosphorylation assay
GST-tagged CIT-K fragments were incubated with 190 ng of recombinant human Aurora B (Invitrogen), 0.1 mM ATP (Sigma-Aldrich), 5 mCi of [g-32 P] ATP (6000 Ci mmol 21 , 10 mCi ml 21 ) (PerkinElmer) and kinase buffer (20 mM HEPES pH 7.5, 2 mM MgCl 2 , 1 mM DTT) in a final reaction volume of 15 ml. After 30 min incubation at 308C with constant agitation, 15 ml of 2Â Laemmli sample buffer was added to stop the reaction. Samples were boiled for 10 min and loaded on a 4-20% Tris-Glycine precast gel (Thermo Scientific). Gels were stained with Quick Coomassie Stain (Generon) to check the protein loading and then proteins were transferred onto a nitrocellulose membrane using the iBlot Dry Blotting System (Invitrogen). Membranes were exposed to Kodak BioMax XAR Films (Sigma-Aldrich) at 2808C. The radioactive CIT-K in vitro kinase assay was performed as described above except that GST-tagged CIT-K and KD-CIT-K were incubated with GST-tagged Aurora B, Borealin and INCENP in a final volume of 25 ml for 1 h at 308C with constant agitation, where 25 ml of 2Â Laemmli sample buffer was added to stop the reaction. The non-radioactive CIT-K in vitro kinase assay was performed as above except using GST-tagged INCENP 783-918 as the substrate, with ATP at a final concentration of 0.5 mM.

Mass spectrometry
For the identification of CIT-K interactors, the raw MS data were analysed using the MASCOT search engine (http:// www.matrixscience.com). Peptides were searched against the SWISS-PROT human sequence database and the following search parameters were employed: enzyme specificity was set to trypsin, a maximum of two missed cleavages were allowed, carbamidomethylation (Cys) was set as a fixed modification, whereas oxidation (Met), phosphorylation (Ser,Thr and Tyr) and ubiquitylation (Lys) were considered as variable modifications. Peptide and MS/MS tolerances were set to 25 parts per million ( ppm) and 0.8 Da, respectively.
For the identification of phosphorylated residues, beads with bound proteins were reduced by incubating in 50 mM Tris(2-carboxyethyl)phosphine hydrochloride for 1 h at 608C, alkylated in 200 mM methyl methanethiosulfonate for 45 min at RT, and then divided into two equal aliquots. One aliquot was digested overnight with trypsin (sequencing grade Modified Trypsin-Promega V5111) in standard conditions, while the second aliquot was digested with protease XIII (Sigma-Aldrich) in 0.1% formic acid at pH 2 for 2 h. We used this second protease because the Aurora B consensus site contains an arginine (R) or a lysine (K) upstream of the phosphorylated S or T residues ([R/K]p[S/T]). Peptide mixtures were then applied to RP-18 precolumns (nanoACQUITY Symmetry w C18-Waters 186003514) using water containing 0.1% trifluoroacetic acid as mobile phase and then transferred to nano-HPLC RP-18 columns (nanoACQUITY BEH C18-Waters 186003545) using an acetonitrile gradient (5-35%) for 180 min in the presence of 0.05% formic acid with a flow rate of 250 nl min 21 . Column outlet was directly coupled to the ion source of the Q Exactive TM Hybrid Quadrupole-Orbitrap Mass Spectrometer (Thermo Electron Corp., San Jose, CA, USA) working in the regime of data dependent on the MS to MS/MS switch. A blank run ensuring lack of cross contamination from previous samples preceded each analysis. Acquired raw data were processed by Mascot Distiller followed by Mascot Search (Matrix Science, London, UK) against the SWISS-PROT database restricted to human sequences. Search parameters for precursor and product ion mass tolerance were 20 ppm and 0.1 Da, respectively. For the enzyme specificity we selected either trypsin with one missed cleavage site allowed, or no-enzyme specificity for protease XIII digested samples. The following variable modifications were also selected: cysteine by methylthio, methionine oxidation, and phosphorylation of serine, threonine and tyrosine. Peptides with Mascot score exceeding the threshold value corresponding to less than 5% false positive rate, calculated by Mascot procedure, and with the Mascot score above 30 were considered to be positive. In addition, fragmentation spectra corresponding to the phosphorylated peptides were manually inspected.

Structure prediction and identification of putative Aurora B consensus phosphorylation sites
The predicted structure of CIT-K CNH 1634 -1948 was completed using the protein homology/analogy recognition engine V 2.0 (Phyre 2 ) [45]. Visualization and analysis was completed using the molecular modelling software UCSF CHIMERA. The GPS v. 2.1 software (http://gps.biocuckoo.org) [46,47] was used to identify consensus sites for Aurora B phosphorylation.
Data accessibility. The MS datasets supporting this article have been uploaded as part of the electronic supplementary material.