Chromatin Protein HP1α Interacts with the Mitotic Regulator Borealin Protein and Specifies the Centromere Localization of the Chromosomal Passenger Complex*

Background: Borealin is a conserved centromere protein essential for mitosis. Results: Borealin interacts with HP1α, and this interaction recruits the CPC to the centromere. Conclusion: The HP1α-borealin interaction specifies CPC localization in the centromere. Significance: A physical link between CPC and HP1α in the centromere orchestrates chromosome segregation and accurate cell division. Accurate mitosis requires the chromosomal passenger protein complex (CPC) containing Aurora B kinase, borealin, INCENP, and survivin, which orchestrates chromosome dynamics. However, the chromatin factors that specify the CPC to the centromere remain elusive. Here we show that borealin interacts directly with heterochromatin protein 1α (HP1α) and that this interaction is mediated by an evolutionarily conserved PXVXL motif in the C-terminal borealin with the chromo shadow domain of HP1α. This borealin-HP1α interaction recruits the CPC to the centromere and governs an activation of Aurora B kinase judged by phosphorylation of Ser-7 in CENP-A, a substrate of Aurora B. Consistently, modulation of the motif PXVXL leads to defects in CPC centromere targeting and aberrant Aurora B activity. On the other hand, the localization of the CPC in the midzone is independent of the borealin-HP1α interaction, demonstrating the spatial requirement of HP1α in CPC localization to the centromere. These findings reveal a previously unrecognized but direct link between HP1α and CPC localization in the centromere and illustrate the critical role of borealin-HP1α interaction in orchestrating an accurate cell division.

The kinetochore is a supermolecular complex assembled at each centromere in eukaryotic cells. Kinetochores provide a chromosomal attachment point for spindle microtubules and power the spatiotemporal dynamics for initiating, controlling, and monitoring the movements of chromosomes during mitosis. Accurate chromosome segregation is essential for cell plasticity, and aberrant mitosis contributes to tumorigenesis (1,2). The chromosomal passenger complex (CPC), 3 which is comprised of Aurora B, INCENP, survivin, and borealin, regulates multiple events during mitotic progression, including spindle assembly, chromosome alignment, and cytokinesis (3,4). The accurate execution of CPC function depends on its spatiotemporal dynamics, which are highly dynamic and finely regulated during the cell cycle. From later G 2 phase to prophase, the CPC localizes to the pericentromeric heterochromatin and chromosome arms. During prometaphase and metaphase, the CPC localizes at the inner centromere and relocates to the central spindle at anaphase onset (3). In budding yeast, relocation of the CPC from the centromere to the central spindle at the metaphase-to-anaphase transition is triggered by the dephosphorylation of INCENP by phosphatase Cdc14 (5). In mammals, the mitotic kinesin Mklp2 is responsible for translocating the CPC from the centromere to the central spindle at anaphase onset (6). In addition, Mklp2 also recruits Cdc14A to the central spindle, and the dephosphorylation of INCENP by Cdc14A may contribute the relocation of the CPC (6,7). Using a FRET-based kinase sensor, our previous study showed that Aurora B kinase activation requires PLK1-mediated phosphorylation of survivin as a priming factor (8). However, the molecular mechanisms underlying CPC localization to the centromere and the temporal control of its activity have remained elusive.
Several lines of evidence demonstrate that the correct localization of CPC components is interdependent (9 -12). In addition, structural biological analyses of the survivin-borealin 10 -109 -INCENP 1-57 core complex indicate a mutual dependence underlying the localization of CPC components to the centromere (13).
Recently, several lines of evidence have suggested multiple mechanisms that mediate centromere localization of the CPC. Survivin can bind to the histone H3 phosphorylated by haspin kinase at threonine 3 (14,15); Bub1 kinase phosphorylates histone H2A (at Thr-120) and, thereafter, phosphorylated H2A (phospho-Thr-120) recruits Sgo1 protein (16); and Cdk1/Cyclin B phosphorylates borealin to facilitate its binding with Sgo1 (17). Consistent with a previous study (18), borealin 10 -109 can localize to the central spindle and midbody correctly but fail to localize to the centromere (13). These results imply that the C-terminal region of borealin may specify the centromere localization of the CPC.
To explore the molecular mechanism underlying CPC targeting to the centromere, we searched for proteins interacting with HP1␣, given its critical role in centromere specification.
Here we show that borealin contains three P/LXVXI/V motifs, the chromo shadow domain (CSD) binding consensus, at its C terminus. Borealin interacts with the CSD of HP1␣ via its third PXVXL motif. Mutation of the PXVXL motif abolishes the localization of borealin and other CPC components to the centromere and perturbs accurate chromosome segregation in mitosis.

MATERIALS AND METHODS
Cell Culture and Synchronization-HeLa and HEK293T cells (ATCC) were maintained as subconfluent monolayers in Dulbecco's modified Eagle's medium (Invitrogen) with 10% fetal bovine serum (Hyclone, Logan, UT) and 100 units/ml penicillin plus 100 g/ml streptomycin (Invitrogen) at 37°C with 8% CO 2 . Cells were synchronized at G 1 /S phase with 2 mM thymidine for 12ϳ16 h, washed with PBS three times, and cultured in thymidine-free medium. Nocodazole and MG132 were used at a final concentration of 100 ng/ml and 20 M, respectively.
Plasmid Construction and Recombinant Protein Production-The cDNA of borealin (NM_018101.2) was provided by Dr. William Earnshaw (University of Edinburgh, UK). The shRNA plasmid targeting borealin and shRNA-resistant LAP-borealin plasmids were a gift from Dr. Geert Kops (University Medical Center Utrecht, The Netherlands). To generate GFP-tagged, FLAG-tagged, mCherry-tagged, and bacterial expression plasmid of borealin as well as different deletion mutants, PCR-amplified cDNAs were cloned into pEGFP-C3, p3ϫFLAG-myc-CMV24, pcDNA3.1B-mCherry, and pGEX-5X-3 (Amersham Biosciences) vectors by EcoRI and XhoI digestion, respectively. The PCR-amplified cDNA of the H2B targeting sequence was inserted into pEGFP-C3 by BamHI. The GFP-H2B-HP1␣ W174A mutant was obtained by site-directed mutagenesis by PCR, and the HP1␣ W174A was inserted to the GFP-H2B plasmid constructed above. PCR-amplified cDNA of the HP1␣ targeting sequence was inserted into the pEGFP-C3 vector by XhoI/HindIII digestion. MBP-HP1␣ was obtained by subcloning HP1␣ cDNA into pMAL-C2 by BamHI/HindIII digestion. HP1␣ cDNA was also subcloned into the p3ϫFLAG-myc-CMV24 vector (Sigma). Mutagenesis was performed using the QuikChange site-directed mutagenesis kit (Stratagene) according to the instructions of the manufacturer. All constructs were verified by sequencing.
In Vitro Pulldown Assay-MBP-HP1␣-bound Sepharose beads were used as an affinity matrix to testify and measure the intensity of the interaction between borealin and HP1␣.
Briefly, purified GST-borealin WT , GST-borealin V174E , GSTborealin V179E , GST-borealin V231E , and two deletions, GST-borealin N (amino acids 1-140) and GST-borealin C (amino acids 141-280) were eluted from GST beads with a reduced glutathione solution and incubated with MBP-HP1␣ for 2 h at 4°C. Beads were washed five times with MBP column buffer with 0.1% Triton X-100 and then boiled in SDS-PAGE sample buffer, followed by 6 -14% SDS-PAGE gradient gel and Western blot detection using an anti-GST antibody. Immunoreactive signals were detected with an ECL kit (Pierce) and visualized by autoradiography on Kodak BioMAX film. The intensity of bands was quantified by ImageJ (National Institutes of Health).
Immunoprecipitation and Western Blotting-293T cells transfected with corresponding plasmids were collected and lysed in lysis buffer (50 mM HEPES (pH 7.4), 150 mM NaCl, 2 mM MgCl 2 , 1 mM EGTA, 0.2% Triton X-100, 1 mM DTT, 10% glycerol, and protease inhibitor mixture (Sigma Chemicals)). Cell lysates were clarified using centrifugation at 13,000 rpm for 20 min at 4°C, and the supernatants were mixed with FLAG-M2 affinity matrix (Sigma). After being washed five times with the lysis buffer, beads were boiled in SDS-PAGE sample buffer. Subsequently, the samples were subjected to SDS-polyacrylamide gel, transferred onto nitrocellulose membrane, and probed with the indicated antibodies, followed by detection with ECL (Pierce).
Chromosome Spread-For the chromosome spread, mitotic HeLa cells were collected after 16-h treatment with nocodazole, swollen in PEM buffer (5 mM PIPES (pH 7.2), 0.5 mM EDTA, 5 mM MgCl 2 , and 5 mM NaCl), and then cytospun onto the slides at 1000 rpm for 5 min. The immunofluorescence experiments were carried out as described below.
Chromosome Fractionation-Chromosomes isolation from mitotic HeLa cells for biochemical characterization has been described previously (19). Briefly, logarithmically growing HeLa cells transiently transfected to express GFP-H2B-HP1␣ and GFP-H2B-HP1␣ W174A proteins were treated with 100 ng/ml nocodazole for 18 h. After arrest, mitotic HeLa cells were harvested by mitotic shakeoff and washed with ice-cold PBS. Mitotic HeLa cells were hypotonically swollen for 5 min at room temperature in 10 volumes of PEM buffer containing 5 mM PIPES (pH 7.2), 0.5 mM EDTA, 5 mM MgCl 2 , 5 mM NaCl, and protease inhibitor mixture (1 mM PMSF, 2 g/ml aprotinin, 2 g/ml pepstatin A, and 2 g/ml leupeptin). The hypotonically swollen cells were harvested by centrifugation and homogenized in PEM buffer containing 0.1% digitonin (Sigma). The homogenates were clarified to remove nuclei, and the supernatant was loaded onto a stepwise gradient containing 30, 40, 50, and 60% sucrose in PEM buffer and centrifuged (2500 ϫ g for 15 min) at 4°C. After centrifugation, a visible, flocculent band migrating at the 50 -60% sucrose interphase was harvested and suspended in 3 volumes of PEM buffer.
Immunofluorescence Microscopy-Immunofluorescence, image acquisition, and processing were performed as described previously (20). For time-lapse microscopy, cells were cultured in a glass-bottom culture dish (MatTek) with CO 2 -independent medium (Invitrogen) at 37°C and examined with a DeltaVision microscopy system. Images were acquired at 3-min intervals and presented in Photoshop (Adobe).
Protein Preparation for the NMR Study-The HP1␣ CSD (amino acids110 -180) was subcloned into a modified pET-28a plasmid using NdeI and XhoI restriction sites as an N-terminal His 6 -tagged fusion protein. The fusion protein was expressed in Escherichia coli BL21 (DE3) Gold cells induced with 0.5 mM isopropyl 1-thio-␤-D-galactopyranoside at 16°C overnight. Uniformly 15 N-or 15 N/ 13 C-labeled fusion proteins were produced by growing the bacteria in minimal medium using 15 NH 4 Cl (0.5 g/liter) and/or 13 C 6 -glucose (2.5 g/liter) as the sole nitrogen and carbon sources. The fusion proteins were first purified on a nickel-chelating column and further by size exclusion column chromatography. NMR samples (0.8 or 1.4 mM) were buffered in 40 mM sodium phosphate (pH 6.4), 200 mM NaCl, 1.5 mM TCEP, 1 mM EDTA, and 1 mM NaN 3 in 90% H 2 O and 10% D 2 O. The HP1␣ CSD L139K mutant was obtained by QuikChange mutagenesis (Takara) and confirmed by DNA sequencing. The mutant was expressed and purified following the same procedures as those used for the wild type. Borealin 223-240 peptide was synthesized chemically at Gel Biotech Co. Ltd. Peptide was dissolved in NMR buffer at 8 mM and stored at Ϫ80°C.
NMR Spectroscopy-All NMR spectra were recorded at 310 K on a Bruker DMX500 (with a cryoprobe) spectrometer. The backbone resonance assignments of HP1␣ CSD L139K were achieved by using triple-resonance CBCA(CO)NH, CBCANH, C(CO)NH-TOCSY, and 15 N-NOESY spectra recorded on a uniformly 15 N/ 13 C-labeled protein (1.4 mM). NMR titration of HP1␣ CSD L139K with borealin peptide was performed on 15 Nlabeled protein (dimer, 0.8 mM) by recording a series of 1 H, 15 N-HSQC in the presence of different amounts of peptide (0 -3.2 mM). NMR spectra were processed with NMRPipe and NMRDraw (21). The spectra assignment and analysis were performed with Sparky software.
Fluorescence Intensity Quantification-The fluorescence intensity of kinetochore-associated protein labeling was measured using an Applied Precision Deltavision deconvolution microscope as described by Yuan et al. (20). In brief, the average pixel intensities of Aurora B from various treated cells with at least 20 kinetochore pairs from five cells were measured, and background pixel intensities were subtracted. The pixel intensities at each kinetochore pair were then normalized against ACA (anti-centromere antibody) pixel values to account for any variations in staining or image acquisition. The values of specific shRNA-treated and mutant borealin-expressing cells were then plotted as a percentage of the values obtained from cells transfected with a control siRNA duplex.
Data Analyses-To determine significant differences between means, unpaired Student's t test assuming unequal variance was performed and evaluated using Prism software (GraphPad Soft-ware). Statistical analysis was considered significant when the twosided p value was less than 0.05.

Borealin Interacts Directly with HP1␣ via Its PXVXL Motif-
To search for the molecular mechanism underlying CPC localization to the centromere, we computed for a pentapeptide sequence (P/L)XVX(I/V) that would interact with the CSD (22). Our computational analyses of CSD binders in the centromere and the kinetochore proteome identified borealin as a strong interacting protein for HP1␣, an evolutionarily conserved centromere protein essential for mitosis (23).
Borealin is a regulatory component of key mitotic machinery, the CPC, which governs the spatiotemporal dynamics of Aurora B activity in mitosis. Previous studies had shown that both borealin 1-140 and borealin 10 -109 , capable of forming a ternary complex with survivin and INCENP, could localize to the spindle midzone and midbody correctly (13). However, these deletion mutants lost their ability to localize to the inner centromere. We reasoned that the C terminus of borealin, borealin 141-280 , contains an element that specifies the centromere localization of CPC. Our analyses of the primary amino acid sequence of borealin reveal that the borealin C terminus contains three (P/L)XVX(I/V) motifs. The first two (P/L)X-VX(I/V) motifs (PAVGR, 172-176, and LEVSM, 177-181) are less conserved evolutionarily, whereas the third one (LTVPV, 229 -233) is highly conserved among different species (Fig. 1A).
To test whether borealin interacts with HP1␣, we carried out an MBP pulldown assay in which MBP-HP1␣ protein was purified on maltose-agarose beads and used as an affinity matrix to absorb recombinant GST-borealin proteins (wild type and mutants) purified from bacteria. As shown in Fig. 1B, lane 11, MBP-HP1␣ binds with GST-borealin but not GST protein, demonstrating a specific binding between HP1␣ and borealin. To map the region of borealin responsible for its interaction with HP1␣, several deletion mutants, such as GST-borealin N (1-140) and GST-borealin C (141-280) were purified and used as inputs ( Because all three CSD readers are located at the C-terminal of borealin (Fig. 1A), we attempted to delineate their requirements for CSD binding. To this end, we generated three Val mutants of borealin, V174E, V179E, and V231E, and tested their ability to bind with HP1␣. As shown in Fig. 1C, all mutant borealin proteins exhibited a similar integrity as that of the wild-type protein (lanes 3-6), and borealin V174E mutant bound to HP1␣ indistinguishably from that of wild-type borealin (lanes 12 and 13). However, the binding of borealin V179E to HP1␣ was reduced, and the binding of borealin V231E to HP1␣ was abolished (Fig. 1C, lanes 14 and 15, respectively). We then quantified the binding efficiency of borealin mutants to HP1␣ from four independent experiments and expressed it as a per-centage of the binding efficiency of wild-type borealin. As shown in Fig. 1D, mutation of borealin Val-231 resulted in a virtual loss of HP1␣-binding, whereas the impact on the borealin V174E -HP1␣ and borealin V179E -HP1␣ interactions was much milder (quantification shown in Fig. 1E). Therefore, we conclude that Val-231 is critical for borealin-HP1␣ interaction.
To validate whether Val-231 is critical for formation of the borealin-HP1␣ complex in vivo, aliquots of HEK293T cells were transiently transfected to express GFP-tagged borealin (wild type and three mutants) and FLAG-tagged HP1␣. 24 h after transfection, transfected cells were harvested and lysed, followed by centrifugation. The aliquots of soluble lysates were incubated with anti-FLAG antibody coupled with agarose beads. The affinity matrix was subjected to extensive washes, followed by boiling in SDS-PAGE sample buffer. The bound proteins were then fractionated by SDS-PAGE, followed by Western blot analyses. As shown in Fig. 1F, lane 5, FLAG-HP1␣ immunoprecipitated wild-type borealin, which is consistent MBP-HP1␣ recombinant protein on maltose-Sepharose beads (New England Biolabs) was used as an affinity matrix to isolate GST-tagged borealin and its deletion mutants. Samples were fractionated by SDS/PAGE, followed by Coomassie Blue staining (top panel). Western blotting using an anti-GST antibody confirmed that both full-length (FL) and C-terminal (C) borealin proteins directly bound to HP1␣ (bottom panel). C, the PXVXL motif mediates a direct borealin-HP1␣ interaction in vitro. MBP-HP1␣ recombinant protein on maltose-Sepharose beads (New England Biolabs) was used as an affinity matrix to isolate GST-tagged borealin 141-280 and its PXVXL motif mutants. Samples were fractionated by SDS/PAGE, followed by Coomassie Blue staining. D, Val-231 of borealin mediates the borealin-HP1␣ interaction in vitro. Purified MBP-HP1␣ recombinant protein on maltose-Sepharose beads (New England Biolabs) was used as an affinity matrix to isolate C-terminal borealin mutants (containing amino acids 141-208 as GST-borealin wt , borealin V174E , borealin V179E , and boreain V231E ). Samples were fractionated by SDS-PAGE and transferred onto a nitrocellulose membrane, followed by probing with MBP (top panel) and GST (bottom panels) antibodies. Aliquots of protein inputs were used to show the qualities of various mutant proteins (lanes 5-8). E, the intensity of bands as shown in D was quantified and normalized to that of borealin WT . The data represent mean Ϯ S.E. (error bars) from three independent experiments. (*, p Ͻ 0.05; **, p Ͻ 0.01; compared with borealin WT ). F, interaction of borealin with HP1␣ in vivo. Cells transiently transfected with FLAG-HP1␣ and GFP-borealin WT , borealin V174E , borealin V179E , or boreain V231E were lysed and incubated with a FLAG antibody-associated affinity matrix (lanes 5-8). Immunoprecipitates (IP) were fractionated by SDS-PAGE and transferred onto a nitrocellulose membrane, followed by probing with FLAG (top panel), GFP (center panel), and tubulin (bottom panel) antibodies.
with the fact that HP1␣ forms a cognate complex with borealin. Importantly, mutation of Val-231, but not Val-174 or Val-179, disrupted the complex (Fig. 1F, lane 8). Therefore, our studies show that HP1␣ interacts physically with borealin in vitro and in vivo.
Nuclear Magnetic Resonance Spectroscopic Analysis Mapping of the Borealin-HP1␣ Binding Interface-To further define the borealin-HP1␣ interaction and its physical contact, we used NMR spectroscopy to monitor the chemical shift perturbation upon borealin peptide (amino acids 223-240) binding to HP1␣. Theoretically, the chemical environment of the interacting surface will be perturbed when two proteins are getting closer, which can lead to substantial shifts of resonance frequencies of neighboring residues as well as spectrum split resonances, which results in broad line widths. Therefore, residues located in interacting surface can be predicted on the basis of chemical shifts of resonance frequencies and line widths. In solution, the HP1␣-CSD molecule existed as a mixture of dimer and higherorder oligomers that obviously decay the NMR signal. After a series of trial experiments with various mutants, we used the single site mutant L139K for the NMR study (Fig. 2D). In solution, HP1␣-CSD L139K mutant proteins were present as consistent dimers and did not affect peptide binding.
NMR data of the HP1␣ CSD were collected with or without the borealin PXVXL motif and analyzed as described previously (24). Analysis of the 1 H, 15 N-HSQC spectra of HP1␣-CSD L139K in the presence of increasing amounts of borealin peptide revealed that several resonances were perturbed (Fig. 2, A and  E). The line widths of residues Glu-169, Glu-170, Arg-171, Leu-172, Trp-174, and His-175 became so broad that their crosspeaks could not be observed from the beginning of the titration. These residues are likely in intermediate exchange relative to the NMR timescale, confirming a direct interaction with the peptide. Residues Ala-129, Thr-130, Asp-131, Ser-132, and Met-137 exhibited large chemical shift changes in both the proton and nitrogen dimensions. These residues are in fast chemical exchange and may contact the flanking residues of the peptide. The migration pattern of those peaks indicated that the interaction is highly similar to HP1␣/PXVXL motif recognition (the peptide binds in an extended conformation across the symmetry axis of the HP1␣-CSD L139K dimer, sandwiching with the C-terminal tail from each monomer) (24). The information from the chemical shift perturbation was used to map the borealin peptide-binding surface on the HP1␣-CSD L139K dimer (Fig. 2, B and C). All perturbed residues were clustered together spatially and formed a contiguous and extensive binding surface. The denoted residues were all located in monomer A. The C-terminal tail of each monomer formed a hydrophobic pocket at the dimer interface, which was critical for peptide binding.
HP1␣ Is Essential for CPC Localization to the Inner Centromere-During mitosis, most HP1␣ proteins become dissociated from chromosome arms because they do not bind to methylated lysine 9 of histone H3 after the neighbor serine 10 is phosphorylated by Aurora B kinase (25)(26)(27)(28). Interestingly, HP1␣ concentrated at the inner centromere during mitosis, whereas HP1␤ and HP1␥ did not (29,30). Moreover, HP1␣ recruits human Sgo1 to the centromere through its interaction with Sgo1 (31). Therefore, we tested whether HP1␣ is required for the centromere recruitment of the CPC. To this end, we first introduced siRNA oligonucleotides of HP1␣ into HeLa cells to suppress the HP1␣ protein level. We confirmed by immunoblotting that our siRNA treatment typically caused an ϳ87 Ϯ 5% reduction in the amount of HP1␣ protein after 24 h, whereas the levels of both tubulin and Aurora B showed no fluctuations in HP1␣ siRNA-treated cells ( Fig. 3A; quantification shown in Fig. 3B). We next examined CPC localization of chromosome spreads in HP1␣-depleting cells. The chromosome spreads were chosen because centromere localization and quantification can be better achieved in a flattened mitotic chromosome. As shown in Fig. 3E, top row, in chromosome spreads of control siRNA-transfected cells, it is readily apparent that both HP1␣ and Aurora B localized to the centromere, as indicated by arrows in the magnified montage. However, the Aurora B signal at the centromere of the chromosome spreads is almost undetectable in HP1␣ siRNA-transfected cells (Fig. 3E, bottom row,  arrow). However, Aurora B signals were seen occasionally in the chromosome arms (Fig. 3E, bottom row, arrowheads), suggesting that Aurora B failed to reach a stable localization to the centromere in the absence of HP1␣.
Next, we tested whether HP1␣ was a direct recruiting factor of the CPC. To this end, we created an artificial GFP-H2B-HP1␣ construct to see whether expression of H2B-HP1␣ fusion protein will result in a gross distribution of borealin and other CPC components on chromosomes. As shown in Fig. 3C, exogenously expressed GFP-H2B and GFP-H2B-HP1␣ were expressed at the right sizes. In addition, the exogenously expressed GFP-H2B and GFP-H2B-HP1␣ proteins did not alter the expression level of endogenous proteins such as borealin (Fig. 3, C and D). Consistent with our expectations, the CPC components borealin and Aurora B were brought to chromosome arms in these GFP-H2B-HP1␣expressing cells because of the association of borealin with HP1␣ (Fig. 3, F and G). However, the localization of the kinetochore protein Hec1 was intact in GFP-H2B-HP1␣-expressing cells (Fig. 3H), suggesting a physical interaction of borealin with HP1␣. Quantitative analyses of chromosome spreads confirmed that the suppression of HP1␣ by siRNA abolished the localization of Aurora B to the centromere in a similar pattern as that of GFP-H2B-HP1␣ expression (Fig. 3I). As a control, expression of GFP-H2B did not alter the centromere localization of Aurora B, as judged by the chromosome spread assay (Fig. 3G).
To validate whether the CSD of HP1␣ specifies borealin recruitment to the centromere, we also generated a GFP-H2B-HP1␣ W174A mutant that would perturb the HP1␣-borealin interactions via the CSD-LXVXV motif recognition. As shown in Fig. 4A, our immunoblot analyses of the isolated chromosome showed that both the GFP-H2B-HP1␣ and the GFP-H2B-HP1␣ W174A mutant were expressed at comparable levels. We then examined the centromere localization of borealin in GFP-H2B-HP1␣ and GFP-H2B-HP1␣ W174A mutant-expressing cells. As shown in Fig. 4C, center row, GFP-H2B-HP1␣ recruits borealin to the chromosome arms. In contrast, the GFP-H2B-HP1␣ W174A mutant, with a disrupted borealin-HP1␣ interaction, failed to recruit borealin to the chromosome arms. Instead, the localization of borealin remained concentrated in the centromere region in the GFP-H2B-HP1␣ W174A mutant (Fig. 4C, bottom row), indicating a critical role of Trp-174 in mediating a physical interaction between borealin and HP1␣.
As a control, expression of GFP-H2B did not alter the centromere localization of borealin (Fig. 4C, top row).
If the CPC relocates to the chromosome arms from the centromere because of H2B-HP1␣ overexpression, the centromere-associated Aurora B kinase activity would diminish. Because phosphorylation of CENP-A at Ser-7 is an exquisite reporter of Aurora B kinase activity, we performed immunofluorescence of Ser(P)-7-CENP-A. As shown in Fig. 4A, bottom  panel, lanes 1 and 2, our immunoblot analyses of isolated chro-mosome expressing the GFP-H2B-HP1␣ and GFP-H2B-HP1␣ W174A mutants showed that the level of Ser(P)-7-CENP-A was reduced greatly in GFP-H2B-HP1␣-expressing chromosomes compared with those of the GFP-H2B-HP1␣ W174A mutant. Quantitative analyses showed that the relative level of Ser(P)-7-CENP-A (the ratio of band density of Ser(P)-7-CENP-A/CENP-A) was reduced significantly in the GFP-H2B-HP1␣expressing chromosome fraction compared with that of the GFP-H2B-HP1␣ W174A mutant-expressing chromosomes (Fig. 4B).  15 N-HSQC spectra HP1␣ CSD L139K in a free state (black) and in the presence of borealin 223-240 peptide (red). The molar ratio of HP1␣ CSD dimer:borealin 223-240 peptide was ϳ1:4. Some of the peaks affected by the peptide binding are denoted. B, schematic of the borealin peptide binding surface of the HP1␣ CSD L139K dimer. Green indicates residues that significantly shift upon addition of the peptide. Perturbed residues and the N and C termini of monomer A are shown. This symmetrical dimer structure was generated by PyMOL software, which used the predicted monomer structure (from a three-dimensional structure prediction web service (40)) to align to the structure of the HP1C⅐CAF-1 complex (PDB code 1S4Z). C, surface presentation of B exhibiting the hydrophobic pocket of the HP1␣ CSD dimer, which is critical for borealin peptide binding. D, overlay of 1 H, 15 N-HSQC spectra of the 15 N-labeled HP1␣ CSD and the HP1␣ CSD (L139K). The red peaks represent the backbone resonances of the wild-type HP1␣ CSD. The black peaks represent the backbone resonances of the mutant domain, HP1␣ CSD (L139K). Note that the mutant did not change the resonances of residues Thr-130, Ser-132, Met-137, Ile-165, Tyr-168, Glu-169, Arg-171, and Ala-176, which were perturbed by borealin peptide binding. The black peaks of Leu-172, Trp-174, and His-175 could not be observed in the wild-type spectra, but these residues are critical for peptide binding. The crystal structure (PDB code 3I3C) and gel filtration chromatography (data not shown) showed that the HP1␣ CSD molecule exists as a mixture of dimer and higherorder oligomers in solution. This self-association can be a major problem in NMR spectroscopy. Note that the number of red peaks is less than the residues of the protein. The signal intensity of the red peaks is also weak because of self-association. After analysis of the crystal structure, we tested a series of mutants to optimize the sample for NMR study. At last we determined the mutant L139K, which was suggested as an oligomerizing contact site. In solution, the mutant proteins were present as consistent dimers and did not affect peptide binding. It is reasonable to assume that the mutant can be used for demonstrating the interaction between HP1␣ and borealin peptide. E, overlay of 1 H, 15   Consistent with our biochemical characterization, our immunofluorescence analyses confirmed that the Aurora B activity was reduced dramatically in GFP-H2B-HP1␣-expressing cells because little Ser(P)-7-CENP-A signal was detected in the centromere of GFP-H2B-HP1␣-expressing cells (Fig. 4D,  center row). In contrast, Aurora B activity is intact in GFP-H2B-HP1␣ W174A -expressing cells (Fig. 4D, bottom row) and GFP-H2B-expressing cells (Fig. 4D, top row). Quantitative analyses of centromere localization of the Ser(P)-7-CENP-A-positive chromosome spread indicates that expression of GFP-H2B-HP1␣ diminishes Aurora B activity from the centromere, as seen in Fig. 4E. Therefore, we conclude that HP1␣ plays an essential role in recruiting the CPC to the centromere via a direct interaction with borealin and that this function depends on its conserved Trp-174 residue within the CSD.
Borealin Is Required for Correct Localization and Function of the CPC-To test the function of the PXVXL motif of borealin, we first examined the localization of borealin mutants. To this end, HeLa cells were transiently transfected to express mCherry-tagged, wild-type borealin and mutant borealin V231E , followed by immunocytochemical staining. As shown in Fig.  5A, wild-type borealin localized to the centromere correctly in mitotic cells on the basis of its colocalization with centromere marker ACA (Fig. 5A, top row, Merge). In contrast, borealin V231E failed to localize in the centromeres of mitotic cells (Fig. 5A, bottom row). Moreover, borealin V231E exhibits a diffusive staining pattern in the cytoplasm, confirming that the HP1␣-borealin interaction is critical for a stable localization of borealin and the CPC to the centromere.
The CPC is important protein machinery that exhibits dynamic spatiotemporal regulation in mitosis. To test whether the HP1␣borealin interaction is responsible for the localization of borealin to other mitotic structures, such as the midzone, we examined anaphase cells expressing mCherry-borealin WT and the mutant borealin V231E . Although borealin V231E failed to localize to the centromere, it targeted to the midzone correctly and exhibited no difference from borealin WT (Fig. 5B). Examination of other CPC components, such as Aurora B and survivin, in borealin V231E -expressing cells revealed no alteration (data no shown). Therefore, these data demonstrate that the PXVXL motif specifies borealin and CPC localization to the centromere but not to other mitotic structures, such as the midzone. To better illustrate the function of the PXVXL-mediated borealin-HP1␣ interaction in chromosome segregation, we next examined the localization of Aurora B in borealin V231E -expressing cells lacking endogenous borealin, which was achieved by expressing shRNA-resistant borealin (32,33). As shown in Fig. 5C, shRNAmediated knockdown efficiently suppressed the expression of borealin because little borealin signal is visualized at the centromere. Importantly, a very low level of Aurora B is found at the centromere, supporting the role of borealin in targeting Aurora B to the centromere (Fig. 5C, first panel). In contrast, the scrambled mock shRNA did not affect the expression and localization of borealin (Fig. 5C, first panel). As predicted, expression of shRNAresistant, wild-type borealin indeed restored its localization to the centromere (Fig. 5C, third panel). In addition, the centromere localization of Aurora B became apparent in shRNA-resistant cells (Fig. 5C, third panel). Nevertheless, expression of shRNA-resistant borealin V231E failed to rescue Aurora B localization to the centromere of a misaligned chromosome (Fig. 5C, fourth panel). Interestingly, Aurora B appeared to be targeted correctly to the centromeres of some chromosomes in shRNA-resistant borealin V231E -expressing cells (Fig. 5C, fourth panel), validating that there are additional mechanisms responsible for the localization of Aurora B to the centromere. Consistent with the critical role of the CPC in mitosis, Aurora B-positive centromeres appeared to be aligned properly at the equator. We then conducted statistical analyses of four independent experiments to quantify the relative fluorescence intensity of Aurora B and present those analyses in Fig. 5D. Our analyses suggest that centromere targeting of borealin is required for accurate localization of Aurora B because the V231E mutant perturbs the localization of Aurora B.
To test whether the borealin V231E mutant affects borealin association with Aurora B, we carried out an immunoprecipitation experiment in which FLAG-borealin and the V231E mutant were transiently transfected to express borealin proteins. As shown in Fig. 5E, both wild-type and V231E mutant borealin immunoprecipitated Aurora B, whereas V231E borealin failed to retain HP1␣ protein. The statistical analyses from three independent experiments shown in Fig. 5F suggest that borealin mutant V231E remains associated with Aurora B.
If targeting borealin to centromere is necessary for accurate chromosome segregation, delocalization of borealin from the centromere would cause aberrant chromosome segregation, in particular in chromosome congression to the equator, because a functional CPC is required for proper kinetochore-spindle attachments. To validate this hypothesis, HeLa cells were transiently transfected to express the borealin wild type and V231E mutant, as indicated in Fig. 5C. Transfected cells were fixed 24 h after transfection. Treated cells were then stained with the appropriate antibodies before examination (data not shown). As shown in Fig. 5G, suppression of borealin expression by shRNA or expression of V231E altered chromosome alignment. Therefore, we conclude that the HP1␣-borealin interaction specifies the localization of the CPC to the centromere, which governs accurate chromosome segregation in mitosis.

DISCUSSION
In this study, we reported that HP1␣ specifies the inner centromere localization of the CPC through its interaction with the CSD binding motifs within the C terminus of borealin. The most prominent CSD-binding interface is mapped to Val-231 of borealin on the basis of solid phase biochemical characterization and NMR analyses in solution. The functional importance of this HP1␣-borealin interaction was demonstrated by the requirement of borealin for stable CPC localization to the centromere. This HP1␣-dependent localization is essential for Aurora B kinase activity in the centromere and accurate chromosome segregation in mitosis.
The assembly of the centromere in mammalian cells involves several parallel but interactive pathways. We noticed that another CPC component, INCENP, also interacts with HP1␣ in chicken cells (34). However, the function of this interaction for targeting INCENP (and the CPC) was ruled out because the INCENP molecule lacking the HP1␣ binding domain localizes to the centromere correctly (34). Moreover, INCENP 1-58 can effectively target survivin and borealin in the absence of endogenous INCENP (10), suggesting that INCENP may target to the centromere independent of its interaction with HP1␣. Recently, the INCENP-HP1␣ interaction was revisited. In a proteomics search for HP1␣-binding proteins, Nozawa et al. (35) identified the CPC components Aurora B, borealin, and INCENP in HP1␣ immunoprecipitates. Although a PXVXL motif (167-171) of human INCENP was required for its interaction with HP1␣, the INCENP molecule lacking the wild-type PXVXL motif still localized to centromeres in metaphase chromosomes, consistent with our finding that the HP1␣-borealin interaction specifies the CPC to the centromere. Therefore, the identification of borealin in HP1␣ immunoprecipitates by Nozawa et al. (35) supports our finding of a borealin-HP1␣ interaction. Our characterization of a direct physical interaction between borealin and HP1␣ revealed a previously unrecognized link between CPC and HP1␣, and our NMR study of the CSD-PXVXL interaction delineates the structural basis of CPC localization to the centromere. On the other hand, Kang et al. (36) have shown that HP1␣ is targeted to mitotic centromeres by INCENP via binding of the HP1␣ CSD and a PXVXL/I motif in INCENP. Their studies also suggest that the HP1␣-INCENP interaction is required for the recruitment of HP1␣ to mitotic centromeres, which prevents HP1␣-Sgo1 binding (36). This recent study indicates another aspect of HP1␣-CPC interaction that remains complicated and elusive.
During the course of our study, several other studies reported that there are alternative mechanisms responsible for CPC centromere localization. Survivin mediates the centromere localization of the CPC by binding phosphorylated histone pT-H3 (14,15). Tsukahara et al. (17) also showed that phosphorylation of borealin by Cdk1 facilitates its association with Shugoshin proteins, hence its centromere localization . Our study demonstrates that borealin contributes to CPC centromere targeting via a new pathway, binding with CSD of HP1␣. We reason that these multiple pathways of CPC centromere targeting are not mutually exclusive and that eukaryotic cells evolved an elaborate centromere plasticity control machinery to ensure faithful chromosome biorientation, attachment error correction, and accurate segregation in mitosis. It is also likely that these pathways cooperate to orchestrate the spatiotemporal dynamics of CPC distribution and precise control of Aurora B activity, priming, and activation (37).
We envision that a multiple pathway model of the centromere localization of CPC is emerging (Fig. 6). Specifically, HP1␣ contributes to targeting the CPC through its direct interaction with borealin; borealin, phosphorylated by Cdk1, enables its association with Shugoshin, whose centromere localization is also dependent on HP1␣ (17,31); and Haspin kinase-phosphorylated histone H3 recruits survivin (14,15). In line with the key function of mitotic kinases in orchestrating mitosis, Cdk1, Haspin, and Bub1 play important regulatory roles in the centromere targeting of the CPC via the regulation of a survivin-dependent pathway (16,17,37). It would be of great interest to delineate how those pathways act in concert to ensure chromosome dynamics and plasticity during cell division and what happens in response to genotoxic stress. It would be of equal importance to elucidate the temporal orders of those interactions and the cross-talk among those pathways.
Taken together, our study demonstrates that HP1␣ contributes to CPC centromere localization through its physical interaction with borealin. The aforementioned interactive pathways synergize several weak protein-protein interactions for dynamic targeting of the CPC to the inner centromere during the prophase-to-prometaphase transition (38,39). The precise spatiotemporal dynamics of Aurora B activation enable phosphorylation of its vicinity substrates to regulate multiple mitotic events, such as biorientation and error correction, essential for faithful mitosis. The multiple pathway model also underscores the orchestration of mitotic kinase pathways in centromere plasticity and genomic stability. FIGURE 6. Hypothetic model accounting for CPC assembly and plasticity in the centromere. Several pathways have been involved in recruitment of the CPC to centromeres during early mitosis. With our findings in this study, we propose following scenarios. 1) The CPC is targeted to the centromere through the interaction of borealin with HP1␣. HP1␣ binds to adjacent trimethylated Lys-9 on H3 (H3K9me3) via its chromo domain (CD) and, thereafter, recruits borealin through Trp-174 in its CSD and Val-231 in the borealin C terminus. HP1␣ can also recruit Sgo1 via its CSD. 2) Phosphorylated H3T3 is generated by haspin kinase and binds with the baculovirus IAP repeat (BIR) domain of survivin. 3) Phosphorylated borealin by cyclin-dependent kinase 1 (Cdk1) enables its binding with Sgo1, which interacts with phosphorylated histone H2A at Thr-120 generated by Bub1. Future studies will ascertain their respective contributions to chromosome dynamics and plasticity during the cell division cycle.