Dynamic acetylation of the kinetochore-associated protein HEC1 ensures accurate microtubule–kinetochore attachment

Faithful chromosome segregation during mitosis is critical for maintaining genome integrity in cell progeny and relies on accurate and robust kinetochore–microtubule attachments. The NDC80 complex, a tetramer comprising kinetochore protein HEC1 (HEC1), NDC80 kinetochore complex component NUF2 (NUF2), NDC80 kinetochore complex component SPC24 (SPC24), and SPC25, plays a critical role in kinetochore–microtubule attachment. Mounting evidence indicates that phosphorylation of HEC1 is important for regulating the binding of the NDC80 complex to microtubules. However, it remains unclear whether other post-translational modifications, such as acetylation, regulate NDC80–microtubule attachment during mitosis. Here, using pulldown assays with HeLa cell lysates and site-directed mutagenesis, we show that HEC1 is a bona fide substrate of the lysine acetyltransferase Tat-interacting protein, 60 kDa (TIP60) and that TIP60-mediated acetylation of HEC1 is essential for accurate chromosome segregation in mitosis. We demonstrate that TIP60 regulates the dynamic interactions between NDC80 and spindle microtubules during mitosis and observed that TIP60 acetylates HEC1 at two evolutionarily conserved residues, Lys-53 and Lys-59. Importantly, this acetylation weakened the phosphorylation of the N-terminal HEC1(1–80) region at Ser-55 and Ser-62, which is governed by Aurora B and regulates NDC80–microtubule dynamics, indicating functional cross-talk between these two post-translation modifications of HEC1. Moreover, the TIP60-mediated acetylation was specifically reversed by sirtuin 1 (SIRT1). Taken together, our results define a conserved signaling hierarchy, involving HEC1, TIP60, Aurora B, and SIRT1, that integrates dynamic HEC1 acetylation and phosphorylation for accurate kinetochore–microtubule attachment in the maintenance of genomic stability during mitosis.

The core function of mitosis is to equally distribute the duplicated genetic materials into two daughter cells. To achieve this, the accurate segregation of the sister chromatids is the crucial step for faithful mitosis. During mitosis, the connection between chromosomes and spindle microtubules is mediated by a proteinaceous supercomplex, kinetochore, which assembles at the centromere region of chromosome (1)(2)(3). NDC80C is a tetramer composed of HEC1/NDC80, NUF2, SPC24, and SPC25 with a 57-nm around dumbbell-shaped structure (4 -11). SPC24 and SPC25 bind to Mis12 complex in the inner kinetochore, thus mediating NUF2 and HEC1 anchorage to the kinetochore (11)(12)(13). Both HEC1 and NUF2 contain the calponin homology (CH) 4 domain at their N termini, through which HEC1 binds to lateral sides of the spindle microtubule. The positively charged and unstructured N-tail (first 80 amino acids; annotated as N80) of HEC1 also binds to microtubule directly and is regulated by Aurora B phosphorylation (14,15). Dynamic phosphorylation of HEC1 by Aurora A and Aurora B destabilizes kinetochore-microtubule attachment and promotes error correction in early mitosis. The later dephosphorylation of HEC1 is required for chromosome biorientation and silencing of the spindle assembly checkpoint (16 -18). It has also been reported that HEC1 N80 phosphorylation regulates microtubule dynamics (19,20). HEC1 also contains an unstructured loop region in the middle, which recruits Ska complex to stabilize kinetochore-attached microtubules. Ska complex is also implicated in promoting the flexibility of NDC80 when connecting with microtubule (21)(22)(23). Acetyltransferase TIP60 has been well-studied in ATM activation, p53 activation, PML stabilization for DNA damage response, and genomic stability control (24 -28). Recently, our study revealed the function of TIP60 in chro- . The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This article contains Figs. S1-S8. 1 Both authors contributed equally to this work. 2 To whom correspondence may be addressed. E-mail: jjhong@ustc.edu.cn. 3 To whom correspondence may be addressed. E-mail: yaoxb@ustc.edu.cn. mosome stability control during mitosis through acetylating Aurora B (29) and RAN GTPase (30,31). However, the precise mechanisms underlying accurate kinetochore microtubule attachment remain obscure. Additionally, it was unclear whether and how TIP60 and NDC80C cooperate at the kinetochore during mitosis.
Here, we identified HEC1 as a novel substrate of TIP60 and the acetylation of HEC1 regulates the interaction between flexible HEC1 N-tail and CH domains of HEC1/ NUF2 and reduces the phosphorylation of HEC1 N80 , revealing a potential cross-talk between the two post-translational modifications in orchestrating kinetochore-microtubule attachment during mitosis.

TIP60 is a novel interacting protein of HEC1
Our previous study showed that acetylation of Aurora B kinase by TIP60 protects an activation of Aurora B from dephosphorylation by the PP2A phosphatase to ensure error-free chromosome segregation during cell division (29). Given the fact that kinetochore localization of TIP60 depends on HEC1, we sought to examine whether HEC1 binds to TIP60 directly. We purified MBP-TIP60 as affinity matrix to pull down HEC1 from cell lysate and found that TIP60 specifically binds to HEC1 from the mitotic HeLa cell lysate. Immunoprecipitation studies also validated that endogenous HEC1 and TIP60 form a complex (Fig. 1, A and B). Because HEC1 forms a tetrameric NDC80C together with NUF2, SPC24, and SPC25 to function in kinetochore-microtubule attachment in mitosis (4,6,8,32), we next tested whether TIP60 interacts with NDC80C. GSTtagged engineered "Bonsai" NDC80C (33) was used as an affinity matrix to pull down purified MBP-TIP60. As shown in Fig.  1C, TIP60 binds to NDC80 Bonsai . To confirm that the binding of NDC80 Bonsai to TIP60 was a direct interaction between TIP60 and HEC1 rather than via other components of NDC80C, MBP-TIP60 was used as an affinity matrix to pull down GFP-HEC1, GFP-NUF2, GFP-SPC24, and SPC25 from cell lysate, respectively. Only GFP-HEC1 was absorbed by the affinity matrix, indicating a specific interaction between HEC1 and TIP60 in Fig. S1A. Thus, TIP60 interacts with HEC1, a major constituent for kinetochore-microtubule attachment.
We then examined the subcellular distribution profile of TIP60 relative to HEC1 to ascertain whether TIP60 activity determines the localization of HEC1 to kinetochore. First, TIP60 protein levels were suppressed by shRNA treatment and were judged by Western blot analyses (Fig. 1D). Next, we examined the relationship of TIP60 and HEC1 in subcellular localization. As shown in Fig. 1E (top), TIP60 and HEC1 co-localized at the kinetochores of HeLa cells, transfected with control shRNA, which is consistent with a previous report (29). However, the localization of HEC1 was altered by neither shRNAmediated TIP60 suppression nor inhibition by TIP60 inhibitor NU9056, as indicated by the intensity of HEC1 immunofluorescence at the kinetochore (Fig. 1E, row 2 and 3). As a control, TIP60 signal at the kinetochore is abolished by shRNA-elicited knockdown but not by NU9056 treatment (Fig. 1, E and F). Statistical analyses from three separate experiments confirmed that TIP60 is not required for HEC1 localization to the kinetochore (Fig. S1B). Thus, we conclude that TIP60 co-localizes and interacts with HEC1 in mitosis.

HEC1 interacts with TIP60 via its N-terminal unstructured tail
Having demonstrated an interaction between TIP60 and HEC1, we wanted to identify which domain of HEC1 is responsible for binding to TIP60. A series of deletion mutants of HEC1 tagged with GFP were constructed according to its structural features (33,34), as illustrated in Fig. 2A. HEK293T cells were transiently transfected with GFPtagged full-length HEC1 and deletion mutants, respectively. Twenty-four hours after the transfection, transfected cells were harvested and lysed for a biochemical assay. After clarification, cell lysates were incubated with MBP-TIP60 affinity beads for the indicated time intervals, followed by three extensive washes before analyses by SDS-PAGE and Western blotting. As shown in Fig. S2A, full-length HEC1 and N-terminal fragments HEC1  and HEC1(1-263) exhibited consistent binding activity to TIP60 acetyltransferase (lanes 9 -11). Therefore we conclude that HEC1  is responsible for interacting with TIP60.
After having demonstrated a direct association between TIP60 and HEC1 N80 (Fig. S2C), we sought to determine the functional relevance of HEC1 N80 during mitosis using livecell imaging of HeLa cells expressing GFP-HEC1 and GFP-HEC1 ⌬N80 (deletion mutant in which amino acids from 1 to 80 were removed). To rule out the interference of endogenous HEC1, we suppressed endogenous HEC1 with a siRNA and introduced expression constructs resistant to siRNA treatment. As shown in Fig. 2C, the exogenous GFP-HEC1 protein was expressed at a level similar to endogenous protein (lane 2). In addition, exogenous expression construct GFP-HEC1 was resistant to siRNA treatment (lane 4). As shown in Fig. 2D, suppression of HEC1 by siRNA treatment exhibited a typical chromosome segregation defect, as chromosomes are scattering around the spindle poles with chronic mitotic arrest (row 2). As predicted, expression of GFP-HEC1 rescued the phenotype observed in siRNA treatment, as cells progressed into anaphase onset in 30 min (row 3). However, expression of GFP-HEC1 ⌬N80 failed to restore accurate chromosome segregation induced by siRNA (bottom panel). Statistical analyses show that cells expressing GFP-HEC1 ⌬N80 exhibited a high proportion of Acetylation of HEC1 ensures accurate mitosis chromosome alignment defect and abnormal anaphase (Fig.  2E). In addition, these GFP-HEC1 ⌬N80 -expressing cells displayed a prolonged interval from nuclear envelope breakdown to anaphase onset (Fig. 2F), consistent with the phenotypes reported in the literature (14). To examine whether the deletion mutant HEC1 ⌬N80 altered NDC80C assembly to the kinetochore, the localization of NUF2, SPC24, and HEC1 was evaluated using immunofluorescence. As shown in Fig. S2D, kinetochore localization of NUF2 and SPC24 was not remarkably changed. Statistical analyses confirmed that the integrity of the NDC80C and its localization to the kinetochore were not affected by the deletion of the N-terminal tail of HEC1 (Fig.   S2E). Therefore, our results are consistent with the previous studies (5, 14 -16). In other words, HEC1 N80 performed an important function in mitosis, and its deletion resulted in severe mitotic defects. Furthermore, we showed that HEC1 N80 interacted with TIP60 (Fig. S2C), which implied that TIP60 might modify HEC1 N80 and thus affect the function of HEC1.

HEC1 N80 is a bona fide substrate of TIP60
TIP60 is an important regulator for spindle plasticity and accurate chromosome segregation (29), whereas mounting evidence demonstrated the regulation of HEC1 N80 by phosphory- Figure 1. TIP60 interacts and co-localizes with HEC1 at kinetochore during mitosis. A, MBP-TIP60 -bound agarose beads were used as affinity matrices to absorb HEC1 proteins from HeLa cell lysate, and bound proteins were then analyzed by Western blotting. Cells were synchronized in G 1 with thymidine for 17 h and released into mitosis (4 h after thymidine release and treatment with STLC for an additional 4 h). Note that HEC1 from mitotic cells was retained on TIP60 affinity matrix. B, HeLa cells were transfected with FLAG-TIP60 for 12 h; in lane 2, cells were blocked with thymidine and STLC as in A. Cell lysate were clarified by centrifugation and subjected to immunoprecipitation with anti-FLAG antibody. Immunoprecipitate, after extensive washes, was fractionated by SDS-PAGE and subsequently analyzed by Western blot analyses. C, GST-NDC80 Bonsai -bound agarose beads were used as affinity matrices to absorb MBP and MBP-TIP60 proteins. Proteins retained on affinity matrix, after extensive washes, were analyzed by SDS-PAGE followed by CBB staining (bottom) and Western blot analysis with an anti-MBP antibody (top). Red asterisk, nonspecific binding protein. D, HeLa cells were transfected with shTIP60 for 48 h followed by Western blot analyses to evaluate the efficiency of shTIP60. A tubulin blot served as loading control. E, HeLa cells were treated with shTIP60 and NU9056. After 24 h of transfection, cells were synchronized with thymidine for 17 h. After they were released from thymidine for 8 h, cells were fixed and co-stained for TIP60 (green), HEC1 (red), and DNA (blue). Scale bars, 10 m. F, scatter plots of the TIP60 intensity at kinetochore in the cells treated as in E (30 kinetochores). Data represent mean Ϯ S.E. (error bars). p values were determined by Student's test. ***, p Ͻ 0.001; ns, not significant.

Acetylation of HEC1 ensures accurate mitosis
lation (35). We hypothesized that TIP60 acetylates HEC1 N80 , and acetylation of HEC1 N80 modulates its interaction with microtubule and NDC80C. To test whether HEC1 is a substrate of TIP60, we carried out an in vitro acetylation assay as we recently reported (29). Specifically, FLAG-TIP60 isolated from HEK293T cells was used to acetylate recombinant GST-   in vitro in the presence or absence of Ac-CoA and TIP60 inhibitor NU9056. As shown in Fig. 3A, acetylation of GST-HEC1 was reported by Western blot analyses, which showed that HEC1 is acetylated by TIP60 (lane 2) and that the acetylation is suppressed by TIP60 inhibitor NU9056 (lane 3).  K53R/K59R -His, GST-HEC1-CT, or GST-HEC1-CT K527R , respectively, in the presence of Ac-CoA for an in vitro acetylation assay. The acetylation levels were analyzed by Western blot analyses using an anti-acetyllysine antibody. HEC1-CT contains aa 222-642. D, HEC1 was immunoprecipitated from HeLa cell lysate with an anti-HEC1 antibody and then analyzed with anti-acetyllysine antibody and HEC1 antibody. In the first three lanes, cells were synchronized in interphase by thymidine treatment for 17 h. In the other three lanes, cells were treated with thymidine for 17 h and released for 8 h and then synchronized in mitosis by STLC for 4 h. In lanes 2 and 5, cells were treated with NU9056. In lanes 3 and 6, cells were transfected with shTIP60 for 24 h before synchronization. E, HeLa cells were transfected with FLAG-tagged HEC1 or its mutants (K527R, K53R/K527R, K59R/K527R, K53R/K59R, and K53R/K59R/ K527R), respectively. Cells were also co-transfected with control or TIP60 shRNA. After 24 h of transfection, anti-FLAG antibody were used to immunoprecipitate FLAG-tagged HEC1 from cell lysate and then analyzed with an anti-acetyllysine antibody and an anti-HEC1 antibody.

Acetylation of HEC1 ensures accurate mitosis
We next sought to pinpoint the acetylation sites of HEC1 that are responsible for TIP60 catalysis. To this end, we first searched the PhosphoSitePlus database and found that HEC1 Lys-53, Lys-59, and Lys-527 sites were acetylated based on mass spectrometric analyses of endogenous HEC1 protein (36). We then conducted sequence alignment analyses to determine the evolutionarily conserved lysines in HEC1 and found that the Lys-53, Lys-59, and Lys-527 sites are conserved ( Fig. 3B and Fig.  S8). To further confirm whether the above mentioned three sites are substrates of TIP60, we performed an in vitro acetylation assay and found that the acetylation level of HEC1  was reduced when Lys-53 or Lys-59 was mutated to arginine ( Fig. 3C, lanes 4 and 5). However, the acetylation was totally abolished when both Lys-53 and Lys-59 were mutated to arginine (lane 6). In contrast, the C-terminal HEC1(222-642) (HEC1-CT) was not acetylated by TIP60 in vitro (lanes 8 and 9), indicating that Lys-527 is not acetylated by TIP60.
To further determine whether the acetylation in vivo exhibits similar characteristics seen in the in vitro reaction, HeLa cells were synchronized in interphase using thymidine or prometaphase using Eg5 inhibitor STLC followed by siRNA-mediated suppression of TIP60 or treatment with TIP60 inhibitor NU9056. The treated cells were used to generate cell lysates for an anti-HEC1 immunoprecipitation. As shown in Fig 6), indicating that HEC1 is acetylated by TIP60 in a cell cycle-dependent manner. To clarify the acetylation sites in vivo, a series of nonacetylatable HEC1 mutants were generated and expressed in HeLa cells in the presence or absence of TIP60 or inhibited TIP60 with NU9056, respectively. These exogenously expressed FLAG-HEC1 mutants were then isolated for Western blot analyses of the acetylation levels. As shown in Fig were reduced when cells were co-transfected with shTIP60 or treatment with NU9056, indicating that Lys-53 and Lys-59 were acetylated by TIP60 in vivo. However, the acetylation of HEC1 K53R/K59R (lane 5 versus lane 11) was not influenced by shTIP60 or NU9056, indicating that Lys-527 is not the preferred acetylation site of TIP60. Livecell imaging also revealed no obvious mitotic phenotypes in nonacetylated K527R mutant (Fig. S3, B-D). Thus, we focused on TIP60 substrates of Lys-53 and Lys-59 sites.
Our previous trial experiments showed that HEC1 acetylation levels changed during the cell cycle (Fig. 3D). To study the cell cycle profile of HEC1 acetylation, we collected cells at different time points after release from G 1 phase and detected the acetylation level of HEC1. As shown in Fig. S3E, Western blot analyses indicate that the acetylation of HEC1 is a function of TIP60 in mitosis, as suppression of TIP60 abolished the acety-lation of HEC1 (lanes 9 and 10). Using HEC1 isolated from synchronized cell lysate with nocodazole (which disassembles the microtubules and blocks cells in prometaphase), our Western blot analyses show that acetylation of HEC1 reached a high level at prometaphase (lane 9). The blot analyses of relative intensity of acetylated HEC1 over HEC1 protein level demonstrated that the acetylation level of HEC1 is highest in prometaphase cells (Fig. 3F). Thus, we conclude that the acetylation of HEC1 is dynamic and peaks in mitosis.

HEC1 acetylation promotes robust kinetochore-microtubule attachment
To probe whether the acetylation of Lys-53 and Lys-59 exhibits any physiological function in mitosis, we carried out real-time analyses to examine chromosome segregation dynamics using mCherry-H2B-expressing HeLa cells. Endogenous HEC1 was knocked down, and exogenous GFP-tagged siRNA-resistant and nonacetylatable HEC1 RR (K53R/K59R) or acetylation-mimicking HEC1 QQ (K53Q/K59Q) was expressed, respectively ( Fig. 4A). Western blot analyses confirmed that the levels of various exogenously expressed GFP-HEC1 proteins (WT, GFP-HEC1 RR , and GFP-HEC1 QQ ) were comparable in the presence of HEC1 siRNA treatment (Fig. S4A). As shown in Fig. 4A, exogenously expressed GFP-HEC1 WT and GFP-HEC1 QQ successfully rescued the phenotypes seen as mitotic arrest and abnormal anaphase resulting from the knockdown of endogenous HEC1. However, nonacetylatable GFP-HEC1 RR failed to rescue the phenotype deficient in endogenous HEC1 (Fig. S4B). Quantitative analyses of the intervals from nuclear envelope breakdown to anaphase, as shown in Fig. 4B, indicate that there was no apparent difference among cells expressing HEC1 WT and HEC1 QQ . However, cells expressing HEC1 RR exhibited mitotic delay in mitosis (Fig. 4, A and B). Interestingly, the exogenous GFP-HEC1 QQ could partially rescue the mitotic defects caused by TIP60 acetyltransferase inhibitor NU9056 (Fig. 4, A (row 8) and B). However, the exogenous GFP-HEC1 WT could not rescue the mitotic defects caused by NU9056 (Fig. 4, A (row 7) and B). These results suggested that the acetylation of HEC1 on Lys-53/Lys-59 by TIP60 involves precise mitosis.
Because HEC1 exhibits critical importance in kinetochoremicrotubule binding and spindle assembly checkpoint signaling (33,34,37), we sought to examine whether HEC1 acetylation modulates the kinetochore-microtubule attachment in mitosis. To this end, HeLa cells were treated with HEC1 siRNA to suppress endogenous HEC1 protein, followed by expressing siRNA-resistant exogenous GFP-HEC1 WT , GFP-HEC1 RR , or GFP-HEC1 QQ . The siRNA-treated cells and exogenous GFP-HEC1 WT -, GFP-HEC1 RR -, or GFP-HEC1 QQ -expressing cells deficient in endogenous HEC1 were subjected to cold treatment followed by immunofluorescence staining to check the stability of kinetochore-microtubule attachment. As shown in Fig. 4C, suppression of endogenous HEC1 resulted in destabilization of spindle (second panel from the top). In addition, inhibition of TIP60 by NU9056 also attenuated spindle microtubule stability. Significantly, expression of exogenous GFP-HEC1 WT or HEC1 QQ rescued the spindle destabilization phenotype, although HEC1 QQ -expressing cells exhibited hyperstabilized Acetylation of HEC1 ensures accurate mitosis spindle (row 6). Of interest, TIP60 inhibitor NU9056 did not attenuate this hyperstabilization with GFP-HEC1 QQ (Fig. 4C, bottom panel), but reduced the hyperstabilization with GFP-HEC1 WT (Fig. 4C, second panel from the bottom). Statistical analyses show that exogenously expressing GFP-HEC1 WT and HEC1 QQ proteins in endogenous HEC1-suppressed cells restored the chromosome alignment (Fig. 4D). However, the exogenous HEC1 RR -expressing cells exhibited high proportion of unaligned chromosomes, similar to cells treated with siHEC1 or TIP60 chemical inhibitor. The results suggest that acetylation of HEC1 promotes kinetochore-microtubule attachment during mitosis.
There are several ways to explain the above results: 1) acetylation of HEC1 could alter the localization of NDC80 components; 2) acetylation may regulate the binding affinity between HEC1 and microtubule; 3) acetylation may modulate the intramolecular interaction of NDC80 components; or 4) acetylation affects the phosphorylation of HEC1 N80 and thus regulates NDC80C-microtubule attachment. To delineate the precise mechanisms underlying acetylation of HEC1 in mitosis, we first assessed whether acetylation of HEC1 modulates the location of NDC80C to the kinetochore. Using immunofluorescence staining, we found that acetylation did not apparently alter the localization of NDC80C components NUF2 and SPC24 to the kinetochore (Fig. S4, C and E). To determine whether acetylation modulates the direct binding of NDC80C to microtubules, we sought to perform an in vitro microtubule co-sedimentation assay using chemically acetylated HEC1 peptide HEC1 N80-K53ac/K59ac -His and nonacetylated HEC1 N80 -His. To perform quantitative analyses of acetylation-elicited binding characteristics, we first determined a linear region of HEC1 peptide detected by chemiluminescence as described previously (e.g. see Refs. 9 and 37). As shown in Fig. S4D, the linear region for quantifying HEC1 peptide is between 0.05 and 0.5 M.
Next, we carried out a co-sedimentation assay using HEC1 peptide (HEC1 N80-K53ac/K59ac -His and HEC1 N80 -His; 200 nM) incubated with preformed microtubules. The Western blot analyses show that the HEC1 peptides were co-sedimented with taxol-stabilized microtubules in a dose-dependent manner (Fig. 4F, lanes 6 -14). Neither protein was pelleted in the absence of microtubules (top and bottom panels; lane 4). Linear transformation of the densitometric measurements into GraphPad Prism version 5 was used to fit the curves and calculate a dissociation constant (K d ) of ϳ0.18 Ϯ 0.05 M for nonacetylated HEC1 binding to microtubule and a K d of ϳ0.42 Ϯ 0.09 M for acetylated HEC1 (Fig. 4G). Thus, we conclude that the acetylation decreased the binding affinity of HEC1 N80 to microtubules. However, acetylation of HEC1 has no effect on the localization of NDC80C components.

Acetylation eliminates HEC1 N80 interference to interaction of NUF2-HEC1 CH domain and weakens the phosphorylation of HEC1 N80
To test whether the acetylation of HEC1 N80 regulates the intramolecular interaction of HEC1-NUF2, GST-NUF2(1-169)-His was used as an affinity matrix to isolate GFP-HEC1 and its mutants (GFP-HEC1 RR and GFP-HEC1 QQ ) from HEK293T cell lysate, respectively. In addition, we performed a reciprocal pulldown assay by using GST-HEC1(1-196)-His and its mutants (HEC1(1-196) RR and HEC1(1-196) QQ ) as affinity matrix to absorb NUF2. Both experiments showed that HEC1 QQ has a higher binding affinity to the CH domain of NUF2 than HEC1 WT or HEC1 RR (Fig. S5, A and B). In addition, when HeLa cells were blocked in interphase or prometaphase and the activity of endogenous TIP60 was inhibited by its specific inhibitor NU9056 or shTIP60, the binding of endogenous HEC1 to recombinant GST-NUF2(1-169)-His was attenuated in a pulldown experiment (Fig. S5, C and D). Moreover, an immunoprecipitation assay indicated that FLAG-HEC1(1-196) QQ had a higher binding affinity to NUF2 in vivo than FLAG-HEC1(1-196) and FLAG-HEC1(1-196) RR (Fig. S5E). Together, the data showed that acetylation of HEC1 by TIP60 at Lys-53 and Lys-59 can promote the interaction of HEC1-NUF2 CH domain.
By structural analyses of the HEC1(80 -196)-NUF2(1-169) interface on the NDC80C structure, we found that the electrostatic surfaces of some areas of interest in both HEC1(80 -196) (indicated by a green circle) and NUF2(1-169) (indicated by an orange circle) were negatively charged, as shown in Fig. 5A. However, HEC1 N80 has a strong positively charged surface based on a predicted model (Fig. S5, F and G), which may have electrostatic interaction between HEC1 N80 and HEC1(80 -196) or NUF2(1-169), and such interaction may affect the interface of HEC1-NUF2 CH domain. More importantly, our pulldown assay ensured that HEC1 N80 can interact with both HEC1(80 -196) and NUF2(1-169) (Fig. S5, H and I), and another pulldown and competition binding assay in vitro supported our hypothesis that HEC1 N80 interferes with the interaction between HEC1(80 -196) and NUF2(1-169) (Fig. S5 (J and K) and Fig. 5 (B and C)). As described earlier, a linear region of anti-SPC25  Western blotting was determined (Fig. S5L). We then carried out a co-sedimentation assay using acetylation-mimicking mutants of NDC80 Bonsai proteins incubated with preformed microtubules. The Western blotting of SPC25 in Fig. S5M shows that the NDC80 Bonsai (1 M) proteins (WT, RR, and QQ) were co-sedimented with taxol-stabilized microtubules in a dose-dependent manner. Neither protein was pelleted in the absence of microtubules (Fig. S5M, lane 4).

Acetylation of HEC1 ensures accurate mitosis
Linear transformation of the densitometric measurements into GraphPad Prism version 5 was used to fit the curves and calculate a K d of 0.01 Ϯ 0.01 M for acetylation-mimicking NDC80 Bonsai (QQ). The values for WT and RR mutant NDC80 Bonsai are 0.05 Ϯ 0.02 and 0.06 Ϯ 0.03 M, respectively ( Fig. S5M and Fig. 5D). Taken together, these studies indicate that HEC1 acetylation at Lys-53 and Lys-59 promotes the association of NDC80 complex with the microtubules.
Mounting evidence demonstrated that the N terminus of HEC1 (N80) is also regulated by Aurora B kinase, and the phosphorylation of HEC1 N80 plays an important role in kinetochore-microtubule attachment. The phosphorylation sites identified at N80 include Ser-4, Ser-5, Ser-8, Ser-15, Ser-44, Thr-49, Ser-55, Ser-62, and Ser-69 (e.g. see Refs. 14, 15, 17, 19, 20, 33, 38, and 39). Because some of these sites are close to the Lys-53 and Lys-59, we hypothesize that the acetylation identified in this study may interact with some of the aforementioned phosphorylation sites. To test this hypothesis, we compared the phosphorylation characteristics of acetylation-mimicking HEC1 with those of unacetylatable HEC1. As shown in Fig. 5E, we found that the phosphorylation of HEC1(1-196) K53Q/K59Q by Aurora B was clearly weaker than that of WT and RR mutants judged by a Phos-tag SDS-PAGE assay. This result implied that acetylation of HEC1 N80 may attenuate the phosphorylation of the N terminus of HEC1. Interestingly, Lys-53 and Lys-59 are just located at the Ϫ2 or Ϫ3 position of Ser-55 and Ser-62, so we sought to examine whether the acetylation on Lys-53 and Lys-59 affects the phosphorylation of Ser-55 and Ser-62. By using the in vitro enzyme kinetics experiment with FLAG-Aurora B and the 15 amino acid peptides containing Ser-55, Lys-53, Lys-59, and Ser-62 as well as the corresponding RR and QQ mutants respectively, we analyzed and compared the K m (Michaelis constant) and K cat of the three peptides, as shown in Fig. 5F and Fig. S5 (N and O), and the results showed that the QQ mutation attenuated the phosphorylation of Ser-55 and Ser-62 by Aurora B. The experimental results of in vitro phosphorylation were in line with our in vivo cell experiments (Fig. 4, A and B). In other words, the inhibition of TIP60 by NU9056 or siRNA will result in a decrease of HEC1 acetylation, which further leads to an increase of HEC1 phosphorylation and thus destabilizes the kinetochore-microtubule connection. On the other hand, exogenously expressed acetylation-mimicking mutants HEC1 K53Q/K59Q weaken the HEC1 phosphorylation by Aurora B and in return stabilize the kinetochore-microtubule connection and promote a normal mitosis process.
In summary, HEC1 N80 may transiently interact with HEC1 CH domain and NUF2 CH domain before or during the assembly of NDC80C (Fig. S5, H and I). The acetylation of HEC1 Lys-53/Lys-59 may reduce such intramolecular interactions but enhance the intermolecular interactions of the HEC1-NUF2 CH domain of NDC80C. In addition, HEC1 Lys-53 and Lys-59 acetylation may weaken the phosphorylation of Ser-55 and Ser-62 of HEC1 by Aurora B, which in turn strengthens the kinetochore-microtubule attachment.

Sirt1 binds and catalyzes the deacetylation of HEC1 at Lys-53/Lys-59
Because the integrity of NDC80C is essential for chromosome segregation, we reason that the HEC1 acetylation is very dynamic to govern the accurate attachment of NDC80Cmicrotubule. To uncover dynamic acetylation of HEC1 in mitosis, we sought to search for the enzyme that deacetylates HEC1 at Lys-53/Lys-59. We used HEC1  as affinity matrix to pull down different deacetylases, including Sirt1, Sirt2, Sirt3, HDAC1, HDAC2, and HDAC3, expressed in HEK293T cells (40,41). Interestingly, only Sirt1 specifically binds to HEC1  based on the pulldown assay (Fig. 6A, lane 7). To further determine whether Sirt1 could deacetylate HEC1, we employed a genetically encoded method, developed by Jason Chin and adopted in our laboratory (42,43), to produce site-specific recombinant acetylated HEC1 on Lys-53 and Lys-59 (schematic illustration shown in Fig. 6B and Fig. S6 (A and B)). As shown in Fig. 6C, our in vitro deacetylation assay demonstrated that Sirt1 deacetylated the recombinantly acetylated HEC1 , and the deacetylation reaction was inhibited by Sirt1 chemical inhibitor Ex527 (lane 3, top).
To probe how the acetyltransferase and deacetylase orchestrate HEC1 acetylation, we blocked cells in three different cell cycle stages (interphase, prometaphase, and metaphase, respectively), followed by immunoprecipitation using anti-FLAG antibody. As shown in Fig. 6D, the binding efficiency of HEC1 to TIP60 is strong in prometaphase and weak in metaphase, whereas the binding of HEC1 to Sirt1 is strong in interphase but weak in metaphase. Quantitative analyses of three independent experiments confirmed this observation (Fig. 6E). In fact, this result was consistent with the HEC1 acetylation levels during cell cycle ( Fig. 3F and Fig.  S3E).

Mathematical modeling of dynamic acetylation of HEC1 in mitosis
To computationally model the dynamic acetylation and deacetylation on HEC1 relative to the regulation of kinetochore-microtubule attachment, we constructed a chemical equilibrium model to make quantitative analyses. As shown in Fig. 7A, TIP60 catalyzes the acetylation reaction and Sirt1 controls the deacetylation reaction; both acetylated HEC1 (HEC1 ac ) and nonacetylated HEC1 could form a complex with NUF2 but with different equilibrium constants. Given the changing of binding affinity of TIP60 and Sirt1 to HEC1 in different phases illustrated above (Fig. 6D), we sought to analyze how the HEC1 acetylation level and amount of HEC1-NUF2 complex changed. Thus, we supposed the amount of HEC1 as a constant, including HEC1, HEC1 ac , HEC1-NUF2, and HEC1 ac -NUF2, and the amount of NUF2 as a constant, including NUF2, HEC1-NUF2, and HEC1 ac -NUF2. We used the Michaelis-Menten equation to describe the enzymic cata-

Acetylation of HEC1 ensures accurate mitosis
lytic reaction, and we took the Michaelis constant K m of the TIP60-and Sirt1-catalyzed reaction as the variable, which could reflect the binding affinity of TIP60 and Sirt1 to HEC1 (44). The functions were thus listed as in Fig. S7A. To analyze the steady-state solution of the equation with the change of K m(Sirt1) and K m(TIP60) , we set up the reaction rates to be zero and generated the function about the concentration of HEC1 ac and HEC1-NUF2 complex ([HEC1-NUF2] ϩ [HEC1 ac -NUF2]) (Fig. S7B). Then we performed quantitative analyses of the trend of [HEC1 ac ] in different cell phases. Because ϳ60% of all K m values are in the range of 10 -1000 M (45), we set the K m(Sirt1) and K m(TIP60) range as 10 -1000 M. Then we plotted the curved surface of [HEC1 ac ] versus K m(Sirt1) and K m(TIP60) (Fig. 7B); with the rise of TIP60-binding affinity and the drop of  K59ac-HEC1(1-196)) in E. coli. C, the recombinant acetylated GST-HEC1 K53/59 -His protein was incubated with NAD ϩ (lane 1), FLAG-Sirt1 ϩ NAD ϩ ϩ Ex527 (lane 2), or FLAG-Sirt1 ϩ NAD ϩ (lane 3). After incubation at 30°C for 2 h, the samples were analyzed with an anti-acetyllysine antibody (acK) and HEC1 antibody by Western blot analyses. D, 293T cells were transfected with FLAG-TIP60 (lanes 1-3) or FLAG-Sirt1 (lanes 4 -6). In lanes 1 and 4, cells were blocked in interphase. In lanes 2 and 5, cells were blocked in prometaphase with STLC for 16 h. In lanes 3 and 6, cells were blocked in metaphase with MG132. Cell lysate was immunoprecipitated with anti-FLAG antibody, analyzed by SDS-PAGE, and probed by anti-HEC1 and anti-FLAG blotting. E, quantification of HEC1-binding intensity with TIP60 and Sirt1 shown in D. Data represent mean Ϯ S.E. (error bars) from three independent experiments.

Acetylation of HEC1 ensures accurate mitosis
Sirt1-binding affinity, the [HEC1 ac ] significantly rises. To further analyze the [HEC1 ac ] in a different phase, we quantified the binding affinity of TIP60 and Sirt1 from the result in Fig. 6D and used this to estimate the K m(Sirt1) and K m(TIP60) . We assumed in the interphase, when TIP60 had the lowest binding affinity and Sirt1 had the highest binding affinity, the K m(TIP60) ϭ 1000 M and K m(Sirt1) ϭ 100 M; thus, we estimated the value of K m(Sirt1) and K m(TIP60) in prometaphase and metaphase by assuming that K m had an inverse relation with binding affinity quantified from Fig. 6D. We then marked the [HEC1 ac ] in different phases in a two-dimensional picture (Fig. 7C). It showed that HEC1 ac level rose from interphase to prometaphase and dropped in metaphase, consistent with our experiments (Fig. 3F). To analyze how binding affinity of TIP60 and

Acetylation of HEC1 ensures accurate mitosis
Sirt1 influence the formation of HEC1-NUF2 complex, we plotted the curved surface of ([HEC1-NUF2] ϩ [HEC1 ac -NUF2]) versus K m(Sirt1) and K m(TIP60) (Fig. 7D) and marked the state in different cell phase (Fig. 7E), with a similar tendency of [HEC1 ac ]. Thus, with the high binding affinity of TIP60 and significantly reduced binding affinity of Sirt1, the acetylation level of HEC1 was significantly elevated and elicited the chemical equilibrium's shift toward the stabilization of HEC1-NUF2 complex and promoted highly stable attachment to microtubules.

Discussion
The NDC80C of the outer kinetochore is responsible for the connection to microtubules, and the stability of this connection is regulated by various post-translational modifications. Previous studies demonstrate that Aurora B phosphorylates the HEC1 N80 to destabilize the microtubule-kinetochore attachment (16 -20). In this study, HEC1 was demonstrated to be a cognate substrate of acetyltransferase TIP60 at Lys-53 and Lys-59 of HEC1 N80 during mitosis. This TIP60-elicited acetylation plays a regulatory role in the intermolecular interactions between HEC1 and NUF2 CH domain and accurate chromosome segregation by inhibiting intramolecular interaction between N80 and the CH domain. Interestingly, the acetylation of N80 weakens the phosphorylation of Ser-55 and Ser-62 at HEC1 by Aurora B, suggesting an additional layer of cross-talk between TIP60 signaling and Aurora B effectors beyond the direct TIP60 -Aurora B interaction (29). These results presented here establish a previously uncharacterized regulatory mechanism governing NDC80C plasticity control through acetylation-mediated regulation of the CH domain and crosstalk between acetylation and phosphorylation of HEC1. We developed a new mathematical model to account for accurate attachment of NDC80C-microtubule via dynamic post-transcriptional modifications (Fig. 7F).
Previous studies demonstrated that the phosphorylation of HEC1 N80 region weakens the NDC80C-microtubule attachment to enable a prompt correction of aberrant connections, such as syntelic and merotelic attachments. The dephosphorylation process is spatiotemporally coordinated with the phosphorylation to promote the NDC80C-microtubule attachment in a precisely controlled manner (14,15,19,20). Thus, the excitement and challenge ahead is to illuminate the spatiotemporal characteristics underlying the cross-talk between the TIP60-elicited acetylation and Aurora B-mediated phosphorylation of HEC1 N80 . Recently, two independent studies from the Yu (46) and Kops (47) groups, respectively, demonstrated that phosphorylation of the HEC1 N80 enhanced binding of Mps1 to NUF2 through the MR fragment. Mps1 and microtubules bind NDC80C competitively and thus monitored kinetochore-microtubule attachment (46,47).
It is worth noting that the HEC1 N80 acetylation characterized here exhibits a context-dependent function relative to NDC80Cmicrotubule interaction. Using two independent analyses (e.g. see Refs. 33 and 46), our calculation indicated that the TIP60-elicited acetylation weakens the affinity of HEC1 N80 for microtubules in isolation. On the other hand, the acetylation of HEC1 N80 by TIP60 strengthens the interaction of Bonsai complex with microtubules.
We believe the regulatory function of acetylation of HEC1 N80 in Bonsai complex reflects and is consistent with the role of TIP60 -NDC80C interaction in mitosis. It is also interesting to note that the acetylation cross-talks with the Aurora B-elicited phosphorylation. From our enzymatic studies, it appears that the acetylation-mimicking HEC1-QQ mutant exhibits an altered affinity with kinase, and likely the kinetics of ADP generation and/or release or the K m value of kinase to ATP was affected by the QQ mutant. Further characterization of the structure-activity relationship of NDC80C acetylation and phosphorylation will yield information helpful for better understanding of dynamic and complex kinetochoremicrotubule interactions in mitotic control. Collectively, these studies demonstrate that HEC1 N80 plays an important role in regulating NDC80C function, and the dynamic post-transla-tionalmodificationsofHEC1 N80 ,includingacetylationandphosphorylation, provide a homeostatic control of NDC80C activity by tuning HEC1 N80 activity for accurate chromosome segregation in mitosis.
In conclusion, we have demonstrated that the acetylation of HEC1, by TIP60, promotes the dynamic and precise assembly of functional N-terminal NDC80C and has crosstalk with phosphorylation necessary for the robust kinetochore-microtubule attachment to ensure chromosome stability in mitosis.
The plasmids of other GST-or MBP-tagged protein were transformed into E. coli strain BL21 or Rosetta (DE3), and 0.2 mM isopropyl 1-thio-␤-D-galactopyranoside was added to induce protein expression when A 600 reached 0.6, with shaking overnight at 16°C. Protein purification was carried out as described previously (49,50). Briefly, the GST fusion proteins were purified using GSH-agarose chromatography, the MBP-tagged proteins were purified using amylose beads, and histidine-tagged proteins were purified using nickel-nitrilotriacetic acid-agarose beads (Qiagen). HEC1 N80 -His and HEC1 N80K53ac/K59ac -His were synthesized by GL Biochem Ltd.

Pulldown assays
MBP, MBP-tagged proteins, GST, or GST-tagged proteinbound agarose beads were incubated with soluble proteins in TGE buffer or with cell lysate at 4°C for 4 h. After washing with pulldown wash buffer three times (5 min each time), the resins were boiled in SDS sample buffer. Samples were analyzed by Western blotting or CBB staining. TGE buffer contained 50 mM Tris-Cl, 50 mM NaCl, 0.5 mM EDTA, 1 mM DTT, and 5% glycerol, and pH was adjusted to 7.9.

Immunoprecipitation
293T cells were transfected with plasmids of FLAG-tagged proteins for 24 h, and then cells were collected and lysed in immunoprecipitation buffer (50 mM HEPES, 100 mM NaCl, 5 mM MgCl 2 , 2 mM EGTA, 1 mM DTT, 0.1% Triton X-100, pH 7.4) supplemented 1 mM DTT, protease inhibitor mixture (Sigma), and DNase. After centrifugation at 12,000 rpm for 20 min, supernatant was incubated with anti-FLAG M2 beads at 4°C for 4 h. Then beads were washed two times with PBS supplemented with 0.1% Triton X-100 and one time with PBS. Then beads were boiled in SDS sample buffer and analyzed by Western blot analyses or CBB staining.

In vitro acetylation assay
The in vitro acetylation assay was described previously (29,51). Briefly, purified FLAG-TIP60 was incubated with GST-HEC1(1-196)-His in 40 l of HAT buffer (250 mM Tris-HCl, pH 8.0, 50% glycerol, 5 mM DTT, 0.5 mM EDTA) containing 1 mM acetyl-CoA for 2 h at 30°C. The reaction was stopped by adding 5ϫ sample buffer and boiled at 95°C for 5 min. Samples were analyzed by Western blot analyses.

Kinase kinetics characterization
FLAG-Aurora B kinases were subjected to an in vitro kinase assay using the 15-mer substrate peptide (WT, ERKVSLFG-KRTSGHG; RR, ERRVSLFGRRTSGHG; QQ, ERQVSLF-GQRTSGHG) and ATP as the substrates and the ADP assay buffer provided by the Amplite TM universal fluorimetric kinase assay kit *Red Fluorescence* (AAT Bioquest) as the kinase buffer. The produced ADP was quantified using the Amplite TM universal fluorimetric kinase assay kit according to the manufacturer's manuals (53)(54)(55)(56). The velocity of phosphorylation at a certain concentration of ATP and substrate peptide was calculated from the concentration of produced ADP, the concentration of kinase, and the reaction time. Then the kinetic parameters were extracted from various substrate concentrations along with the corresponding velocities of three independent experiments using the Michaelis-Menten equation.

Immunofluorescence microscopy
HeLa cells grown on coverslips were fixed by PTEM (50 mM Pipes, 0.2% Triton X-100, 10 mM EGTA, 1 mM MgCl 2 , pH 6.8) with 3.7% formaldehyde. After blocking with PBST containing 1% BSA for 1 h at room temperature, the fixed cells were incubated with primary antibodies overnight at 4°C, followed by secondary antibodies for 1 h at room temperature. DNA was stained by 4Ј,6-diamidino-2-phenylindole (Sigma). Coverslips were mounted on glass slides with antifade mounting medium and sealed with nail polish. Images were acquired using an

Acetylation of HEC1 ensures accurate mitosis
Olympus ϫ60, numerical aperture 1.42 Plan APO N objective on a DeltaVision microscope (Applied Precision) and processed by deconvolution and z-stack projection (DeltaVision softWoRx software) as described previously (29).

Live-cell imaging
HeLa cells were cultured in glass-bottomed culture dishes (MatTek). Culture medium was changed to CO 2 -independent medium (Gibco) supplemented with 10% fetal bovine serum before imaging. During imaging, the dishes were placed in a sealed chamber at 37°C. Images of living cells were captured with the DeltaVision RT system (Applied Precision).

Microtubule co-sedimentation assays
The microtubule co-sedimentation assay was described previously (48,49,58). Briefly, different concentrations of taxol-stabilized microtubules were incubated with HEC1 N80 -His or HEC1 N80K53ac/K59ac -His, GST-NDC80 WT , GST-NDC80 K53R/K59R , GST-NDC80 K53Q/K59Q for 20 min at 27°C in BRB buffer followed by centrifugation at 80,000 rpm at 25°C for 10 min. The pellets and supernatants were separated and solubilized in SDS sample buffer and visualized by Western blot analyses with the indicated antibodies. Densitometric quantification of co-sedimentation was carried out with ImageJ. The percentage of protein bound to microtubules was expressed as pellet signal divided by total supernatant and pellet signal. Mean binding values from three independent experiments were used to determine the apparent K d by following the quadratic equation, where B max represents the maximal fractional NDC80 Bonsaitubulin complex, K d is the dissociation constant, and X is the concentration of tubulin dimer. Given the fact that the concentrations of the NDC80 complex fragments (peptide or complex) are close to the reported K d values (e.g. see Ref. 33), we used the full quadratic binding equation for evaluating the affinities, and K d was calculated using GraphPad Prism version 5.0 (GraphPad Software, Inc.) as described previously (33). Our calculated K d of WT Bonsai is consistent with the value in the literature (e.g. see Refs. 33 and 46).

Data analyses
To determine significant differences between means, unpaired Student's t test assuming unequal variance was performed and evaluated using GraphPad software (26). Statistical analysis was considered to be significant when the two-sided p value was Ͻ0.05.